fix: preserve typed expression boundaries

fix(python): preserve typed expression round trips
Merge remote-tracking branch 'origin/main' into gatekeeper/fix-1869-1
2026-08-26 16:08:43 +00:00 · 2026-08-06 12:16:54 +00:00 · 2026-08-06 10:53:21 +00:00 · 2026-08-06 09:24:30 +00:00 · 2026-08-06 09:12:57 +00:00 · 2026-08-06 16:54:59 +08:00
90 changed files with 10018 additions and 1871 deletions
@@ -601,7 +601,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "hex",
- "http 1.4.2",
+ "http 1.5.0",
 "sha1 0.10.6",
 "time",
 "tokio",
@@ -664,7 +664,7 @@ dependencies = [
 "bytes-utils",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 0.4.6",
 "http-body 1.1.0",
 "percent-encoding",
@@ -694,7 +694,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body-util",
 "regex-lite",
 "tracing",
@@ -719,7 +719,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "regex-lite",
 "tracing",
 ]
@@ -743,7 +743,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "regex-lite",
 "tracing",
 ]
@@ -773,7 +773,7 @@ dependencies = [
 "hex",
 "hmac 0.13.0",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "lru",
 "percent-encoding",
@@ -802,7 +802,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "regex-lite",
 "tracing",
 ]
@@ -826,7 +826,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "regex-lite",
 "tracing",
 ]
@@ -851,7 +851,7 @@ dependencies = [
 "aws-types",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "regex-lite",
 "tracing",
 ]
@@ -873,7 +873,7 @@ dependencies = [
 "hex",
 "hmac 0.13.0",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "p256",
 "percent-encoding",
 "ring",
@@ -906,7 +906,7 @@ dependencies = [
 "bytes",
 "crc-fast",
 "hex",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "md-5 0.11.0",
@@ -940,7 +940,7 @@ dependencies = [
 "bytes-utils",
 "futures-core",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "percent-encoding",
@@ -961,7 +961,7 @@ dependencies = [
 "h2 0.3.27",
 "h2 0.4.14",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 0.4.6",
 "hyper 0.14.32",
 "hyper 1.9.0",
@@ -1023,7 +1023,7 @@ dependencies = [
 "bytes",
 "fastrand",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 0.4.6",
 "http-body 1.1.0",
 "http-body-util",
@@ -1044,7 +1044,7 @@ dependencies = [
 "aws-smithy-types",
 "bytes",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "pin-project-lite",
 "tokio",
 "tracing",
@@ -1070,7 +1070,7 @@ checksum = "7442cb268338f0eb8278140a107c046756aa01093d8ef5e99628d34ae09c94f5"
 dependencies = [
 "aws-smithy-runtime-api",
 "aws-smithy-types",
- "http 1.4.2",
+ "http 1.5.0",
 ]

 [[package]]
@@ -1084,7 +1084,7 @@ dependencies = [
 "bytes-utils",
 "futures-core",
 "http 0.2.12",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 0.4.6",
 "http-body 1.1.0",
 "http-body-util",
@@ -1132,7 +1132,7 @@ dependencies = [
 "axum-core",
 "bytes",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "hyper 1.9.0",
@@ -1165,7 +1165,7 @@ dependencies = [
 "async-trait",
 "bytes",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "mime",
@@ -3421,8 +3421,8 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"

 [[package]]
 name = "fsst"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "rand 0.9.5",
@@ -3842,7 +3842,7 @@ dependencies = [
 "fnv",
 "futures-core",
 "futures-sink",
- "http 1.4.2",
+ "http 1.5.0",
 "indexmap 2.14.0",
 "slab",
 "tokio",
@@ -3953,7 +3953,7 @@ checksum = "629d8f3bbeda9d148036d6b0de0a3ab947abd08ce90626327fc3547a49d59d97"
 dependencies = [
 "dirs",
 "futures",
- "http 1.4.2",
+ "http 1.5.0",
 "indicatif",
 "libc",
 "log",
@@ -3976,7 +3976,7 @@ checksum = "430b33fa84f92796d4d263070b6c0d3ca219df7b9a0e1853ee431029b1612bcd"
 dependencies = [
 "async-trait",
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "more-asserts",
 "serde",
 "thiserror 2.0.18",
@@ -4041,9 +4041,9 @@ dependencies = [

 [[package]]
 name = "http"
-version = "1.4.2"
+version = "1.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6970f50e31d6fc17d3fa27329444bfa74e196cf62e95052a3f6fee181dba6425"
+checksum = "918d3568bebf352712bc2ef3d46a8bcf1a75b373be6539de198e9105cbbf9ce0"
 dependencies = [
 "bytes",
 "itoa",
@@ -4067,7 +4067,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ca2a8f2913ee65f60facd6a5905613afaa448497a0230cc41ce022d93290bc2c"
 dependencies = [
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 ]

 [[package]]
@@ -4078,7 +4078,7 @@ checksum = "b021d93e26becf5dc7e1b75b1bed1fd93124b374ceb73f43d4d4eafec896a64a"
 dependencies = [
 "bytes",
 "futures-core",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "pin-project-lite",
 ]
@@ -4145,7 +4145,7 @@ dependencies = [
 "futures-channel",
 "futures-core",
 "h2 0.4.14",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "httparse",
 "httpdate",
@@ -4177,7 +4177,7 @@ version = "0.27.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "33ca68d021ef39cf6463ab54c1d0f5daf03377b70561305bb89a8f83aab66e0f"
 dependencies = [
- "http 1.4.2",
+ "http 1.5.0",
 "hyper 1.9.0",
 "hyper-util",
 "rustls 0.23.40",
@@ -4211,7 +4211,7 @@ dependencies = [
 "bytes",
 "futures-channel",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "hyper 1.9.0",
 "ipnet",
@@ -4777,8 +4777,8 @@ checksum = "e037a2e1d8d5fdbd49b16a4ea09d5d6401c1f29eca5ff29d03d3824dba16256a"

 [[package]]
 name = "lance"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arc-swap",
 "arrow",
@@ -4852,8 +4852,8 @@ dependencies = [

 [[package]]
 name = "lance-arrow"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4875,7 +4875,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-scalar"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4889,7 +4889,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-stats"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-schema",
@@ -4898,8 +4898,8 @@ dependencies = [

 [[package]]
 name = "lance-bitpacking"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrayref",
 "crunchy",
@@ -4909,14 +4909,15 @@ dependencies = [

 [[package]]
 name = "lance-core"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
 "arrow-data",
 "arrow-schema",
 "async-trait",
+ "blake3",
 "byteorder",
 "bytes",
 "datafusion-common",
@@ -4949,8 +4950,8 @@ dependencies = [

 [[package]]
 name = "lance-datafusion"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4980,8 +4981,8 @@ dependencies = [

 [[package]]
 name = "lance-datagen"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4998,8 +4999,8 @@ dependencies = [

 [[package]]
 name = "lance-derive"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "proc-macro2",
 "quote",
@@ -5008,8 +5009,8 @@ dependencies = [

 [[package]]
 name = "lance-encoding"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-arith",
 "arrow-array",
@@ -5044,12 +5045,13 @@ dependencies = [

 [[package]]
 name = "lance-file"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-arith",
 "arrow-array",
 "arrow-buffer",
+ "arrow-cast",
 "arrow-data",
 "arrow-schema",
 "arrow-select",
@@ -5075,8 +5077,8 @@ dependencies = [

 [[package]]
 name = "lance-index"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arc-swap",
 "arrow",
@@ -5143,8 +5145,8 @@ dependencies = [

 [[package]]
 name = "lance-index-core"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-schema",
@@ -5166,18 +5168,12 @@ dependencies = [

 [[package]]
 name = "lance-io"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
- "arrow-arith",
 "arrow-array",
- "arrow-buffer",
- "arrow-cast",
- "arrow-data",
 "arrow-schema",
- "arrow-select",
- "async-recursion",
 "async-trait",
 "aws-config",
 "aws-credential-types",
@@ -5186,7 +5182,7 @@ dependencies = [
 "chrono",
 "futures",
 "goosefs-sdk",
- "http 1.4.2",
+ "http 1.5.0",
 "io-uring",
 "lance-arrow",
 "lance-core",
@@ -5210,8 +5206,8 @@ dependencies = [

 [[package]]
 name = "lance-linalg"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -5227,8 +5223,8 @@ dependencies = [

 [[package]]
 name = "lance-namespace"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
 "async-trait",
@@ -5240,8 +5236,8 @@ dependencies = [

 [[package]]
 name = "lance-namespace-impls"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
 "arrow-ipc",
@@ -5295,8 +5291,8 @@ dependencies = [

 [[package]]
 name = "lance-select"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -5311,8 +5307,8 @@ dependencies = [

 [[package]]
 name = "lance-table"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -5351,8 +5347,8 @@ dependencies = [

 [[package]]
 name = "lance-testing"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "arrow-array",
 "arrow-schema",
@@ -5365,8 +5361,8 @@ dependencies = [

 [[package]]
 name = "lance-tokenizer"
-version = "10.0.0-beta.5"
-source = "git+https://github.com/lance-format/lance.git?tag=v10.0.0-beta.5#ddb8e28ca238f29628b8e1795ddccbb7bf75e5c8"
+version = "10.1.0-beta.1"
+source = "git+https://github.com/lance-format/lance.git?tag=v10.1.0-beta.1#68f4d4c1d0c4871b067557c61fc405078f1ab3b7"
 dependencies = [
 "icu_segmenter",
 "jieba-rs",
@@ -5418,7 +5414,7 @@ dependencies = [
 "goosefs-sdk",
 "half",
 "hf-hub",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "lance",
 "lance-arrow",
@@ -6084,15 +6080,15 @@ dependencies = [

 [[package]]
 name = "napi-build"
-version = "2.3.2"
+version = "2.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c9c366d2c8c60b86fa632df75f745509b52f9128f91a6bad4c796e44abb505e1"
+checksum = "5282704fbe8d49b0cf8b08e3f33233416a528658f205c7e5ace63b582de0b11c"

 [[package]]
 name = "napi-derive"
-version = "3.6.0"
+version = "3.6.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a49c513341a61a16a10af6efcce46b30d0822ba2d4fb197d24d33dfc199c78d5"
+checksum = "4d5c9c02556ea6dc99dffd36c1ce60141411657438501a125b675776d011ce92"
 dependencies = [
 "convert_case",
 "ctor 1.0.5",
@@ -6104,9 +6100,9 @@ dependencies = [

 [[package]]
 name = "napi-derive-backend"
-version = "6.0.0"
+version = "6.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4747005fa3e2c9989ac45a723a514c5db2411238b72981a3cda4c701a9dfea17"
+checksum = "d60b5d773ad46c698c8cc2cd9fde0b283d39cbb7f71c04bee633c7bdba4423bd"
 dependencies = [
 "convert_case",
 "proc-macro2",
@@ -6360,7 +6356,7 @@ dependencies = [
 "futures-channel",
 "futures-core",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body-util",
 "httparse",
 "humantime",
@@ -6481,7 +6477,7 @@ dependencies = [
 "base64 0.22.1",
 "bytes",
 "futures",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "jiff",
 "log",
@@ -6506,7 +6502,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0d6f81ba6960e3fae1882f253b114b21d7e444e1534f209c7737a79f6243eb6f"
 dependencies = [
 "futures",
- "http 1.4.2",
+ "http 1.5.0",
 "mea",
 "opendal-core",
 ]
@@ -6550,7 +6546,7 @@ checksum = "0030644366ef5d8cbe3a4a5822bf99a4aafddc1666e9d24b44d158d9062fc76a"
 dependencies = [
 "base64 0.22.1",
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "opendal-service-azure-common",
@@ -6571,7 +6567,7 @@ checksum = "6dea4908d490143a9b0b7f7a790e139ff829b06a023f670455ed3d44f664b361"
 dependencies = [
 "base64 0.22.1",
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "opendal-service-azure-common",
@@ -6589,7 +6585,7 @@ version = "0.57.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9b489f13c42e69d69bdd72952b634356ec43a7881a20259b38b540fcecdf4051"
 dependencies = [
- "http 1.4.2",
+ "http 1.5.0",
 "opendal-core",
 ]

@@ -6600,7 +6596,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "aa8cafe9729213375c7331019b0cb756ad3e1aff7f45cd32c45eae91ebde8901"
 dependencies = [
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "quick-xml 0.39.4",
@@ -6618,7 +6614,7 @@ checksum = "48de101aac565ed06af4b47903c24eafd249075553ec1fb18256751c45148d47"
 dependencies = [
 "async-trait",
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "percent-encoding",
@@ -6653,7 +6649,7 @@ checksum = "c4922661976a1d40794a2adfbdb888cc3c23097690f825a92f773af38908a848"
 dependencies = [
 "bytes",
 "hf-xet",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "percent-encoding",
@@ -6669,7 +6665,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "328fa55e8888cbdfe00826bfea2a79042422b720e8369e9e021e46121dea5ace"
 dependencies = [
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "opendal-core",
 "quick-xml 0.39.4",
@@ -6688,7 +6684,7 @@ dependencies = [
 "base64 0.22.1",
 "bytes",
 "crc32c",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "md-5 0.11.0",
 "opendal-core",
@@ -7587,7 +7583,7 @@ version = "0.14.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "343d3bd7056eda839b03204e68deff7d1b13aba7af2b2fd16890697274262ee7"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
 "itertools 0.14.0",
 "log",
 "multimap",
@@ -8230,7 +8226,7 @@ checksum = "57ac2757f3140aa2e213b554148ae0b52733e624fc6723f0cc6bb3d440176c95"
 dependencies = [
 "anyhow",
 "form_urlencoded",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "percent-encoding",
 "reqsign-core",
@@ -8248,7 +8244,7 @@ dependencies = [
 "anyhow",
 "bytes",
 "form_urlencoded",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "percent-encoding",
 "quick-xml 0.39.4",
@@ -8270,7 +8266,7 @@ dependencies = [
 "base64 0.22.1",
 "bytes",
 "form_urlencoded",
- "http 1.4.2",
+ "http 1.5.0",
 "jsonwebtoken",
 "log",
 "pem",
@@ -8295,7 +8291,7 @@ dependencies = [
 "futures",
 "hex",
 "hmac 0.12.1",
- "http 1.4.2",
+ "http 1.5.0",
 "jiff",
 "log",
 "percent-encoding",
@@ -8322,7 +8318,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "35cc609b49c69e76ecaceb775a03f792d1ed3e7755ab3548d4534fd801e3242e"
 dependencies = [
 "form_urlencoded",
- "http 1.4.2",
+ "http 1.5.0",
 "jsonwebtoken",
 "log",
 "percent-encoding",
@@ -8342,7 +8338,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e128f19525861dbded59e1e7c17653a8ed63d573ca04aed708d552dbef5bb32a"
 dependencies = [
 "anyhow",
- "http 1.4.2",
+ "http 1.5.0",
 "log",
 "percent-encoding",
 "reqsign-core",
@@ -8362,7 +8358,7 @@ dependencies = [
 "futures-core",
 "futures-util",
 "h2 0.4.14",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "hyper 1.9.0",
@@ -8406,7 +8402,7 @@ dependencies = [
 "bytes",
 "futures-core",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "hyper 1.9.0",
@@ -8460,7 +8456,7 @@ checksum = "199dda04a536b532d0cc04d7979e39b1c763ea749bf91507017069c00b96056f"
 dependencies = [
 "anyhow",
 "async-trait",
- "http 1.4.2",
+ "http 1.5.0",
 "reqwest 0.13.3",
 "thiserror 2.0.18",
 "tower-service",
@@ -8502,9 +8498,9 @@ dependencies = [

 [[package]]
 name = "rkyv"
-version = "0.8.16"
+version = "0.8.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "73389e0c99e664f919275ab5b5b0471391fe9a8de61e1dff9b1eaf56a90f16e3"
+checksum = "815cc8a37159a463064825246cadb07961e25cd9885908606f6d08a98d8f8874"
 dependencies = [
 "bytecheck",
 "bytes",
@@ -8521,9 +8517,9 @@ dependencies = [

 [[package]]
 name = "rkyv_derive"
-version = "0.8.16"
+version = "0.8.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5d2ed0b54125315fb36bd021e82d314d1c126548f871634b483f46b31d13cac6"
+checksum = "c0ed1a78a1b19d184b0daa629dd9a024573173ec7d485b287cb369fb3607cc1c"
 dependencies = [
 "proc-macro2",
 "quote",
@@ -9281,7 +9277,7 @@ version = "0.8.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c1c97747dbf44bb1ca44a561ece23508e99cb592e862f22222dcf42f51d1e451"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
 "proc-macro2",
 "quote",
 "syn 2.0.117",
@@ -9293,7 +9289,7 @@ version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "54254b8531cafa275c5e096f62d48c81435d1015405a91198ddb11e967301d40"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
 "proc-macro2",
 "quote",
 "syn 2.0.117",
@@ -9734,7 +9730,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "32497e9a4c7b38532efcdebeef879707aa9f794296a4f0244f6f69e9bc8574bd"
 dependencies = [
 "fastrand",
- "getrandom 0.4.2",
+ "getrandom 0.3.4",
 "once_cell",
 "rustix",
 "windows-sys 0.61.2",
@@ -10062,7 +10058,7 @@ dependencies = [
 "base64 0.22.1",
 "bytes",
 "h2 0.4.14",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "hyper 1.9.0",
@@ -10118,7 +10114,7 @@ checksum = "1e9cd434a998747dd2c4276bc96ee2e0c7a2eadf3cae88e52be55a05fa9053f5"
 dependencies = [
 "bitflags 2.11.1",
 "bytes",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "pin-project-lite",
@@ -10138,7 +10134,7 @@ dependencies = [
 "bytes",
 "futures-core",
 "futures-util",
- "http 1.4.2",
+ "http 1.5.0",
 "http-body 1.1.0",
 "http-body-util",
 "pin-project-lite",
@@ -11159,7 +11155,7 @@ dependencies = [
 "clap",
 "crc32fast",
 "futures",
- "http 1.4.2",
+ "http 1.5.0",
 "hyper 1.9.0",
 "lazy_static",
 "more-asserts",
@@ -11233,7 +11229,7 @@ dependencies = [
 "chrono",
 "clap",
 "gearhash",
- "http 1.4.2",
+ "http 1.5.0",
 "itertools 0.14.0",
 "lazy_static",
 "more-asserts",
@@ -13,20 +13,20 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"

 [workspace.dependencies]
-lance = { "version" = "=10.0.0-beta.5", default-features = false, "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-core = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-datagen = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-file = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-io = { "version" = "=10.0.0-beta.5", default-features = false, "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-index = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-linalg = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace-impls = { "version" = "=10.0.0-beta.5", default-features = false, "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-table = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-testing = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-datafusion = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-encoding = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
-lance-arrow = { "version" = "=10.0.0-beta.5", "tag" = "v10.0.0-beta.5", "git" = "https://github.com/lance-format/lance.git" }
+lance = { "version" = "=10.1.0-beta.1", default-features = false, "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=10.1.0-beta.1", default-features = false, "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=10.1.0-beta.1", default-features = false, "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=10.1.0-beta.1", "tag" = "v10.1.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
 ahash = "0.8"
 # Note that this one does not include pyarrow
 arrow = { version = "58.0.0", optional = false }
@@ -25,6 +25,27 @@ the underlying connection has been closed.

 ## Methods

+### cancelJob()
+
+```ts
+abstract cancelJob(jobId): Promise<boolean>
+```
+
+Request cancellation of a server-side job by id.
+
+Resolves to true if the server accepted the cancellation, false if no
+such job exists. Cancelling an already-terminal job is a no-op success.
+
+#### Parameters
+
+* **jobId**: `string`
+
+#### Returns
+
+`Promise`&lt;`boolean`&gt;
+
+***
+
 ### cloneTable()

 ```ts
@@ -365,6 +386,26 @@ Drop an existing table.

 ***

+### getJob()
+
+```ts
+abstract getJob(jobId): Promise<null | JobDescription>
+```
+
+Describe a single server-side job by id.
+
+Resolves to `null` when the server has no such job.
+
+#### Parameters
+
+* **jobId**: `string`
+
+#### Returns
+
+`Promise`&lt;`null` \| [`JobDescription`](../interfaces/JobDescription.md)&gt;
+
+***
+
 ### isOpen()

 ```ts
@@ -379,6 +420,62 @@ Return true if the connection has not been closed

 ***

+### job()
+
+```ts
+abstract job(jobId): Job
+```
+
+A [Job](Job.md) handle for a server-side job by id.
+
+The handle is constructed without a server round trip; an unknown id
+surfaces when the handle is used. Dropping the handle has no effect on
+the job itself.
+
+#### Parameters
+
+* **jobId**: `string`
+
+#### Returns
+
+[`Job`](Job.md)
+
+***
+
+### jobHistory()
+
+```ts
+abstract jobHistory(jobId?): Promise<Table<any>>
+```
+
+The lifecycle event history of a server-side job, as an Arrow table.
+
+Lists history across all jobs when `jobId` is omitted.
+
+#### Parameters
+
+* **jobId?**: `string`
+
+#### Returns
+
+`Promise`&lt;`Table`&lt;`any`&gt;&gt;
+
+***
+
+### listJobs()
+
+```ts
+abstract listJobs(): Promise<JobInfo[]>
+```
+
+List server-side jobs across the database's tables.
+
+#### Returns
+
+`Promise`&lt;[`JobInfo`](../interfaces/JobInfo.md)[]&gt;
+
+***
+
 ### listNamespaces()

 ```ts
@@ -0,0 +1,83 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / Job
+
+# Class: Job
+
+A handle to an operation that may still be running.
+
+## Constructors
+
+### new Job()
+
+```ts
+new Job(): Job
+```
+
+#### Returns
+
+[`Job`](Job.md)
+
+## Accessors
+
+### id
+
+```ts
+get id(): null | string
+```
+
+Identifies the operation on the server that is running it. Operations
+that run in this process have no server id. The value is opaque.
+
+#### Returns
+
+`null` \| `string`
+
+## Methods
+
+### cancel()
+
+```ts
+cancel(): Promise<void>
+```
+
+Request cancellation. Cancelling a finished operation is a no-op.
+
+#### Returns
+
+`Promise`&lt;`void`&gt;
+
+***
+
+### status()
+
+```ts
+status(): Promise<string>
+```
+
+The operation's current lifecycle state: "running", "finished",
+"failed", or "cancelled".
+
+A point snapshot; unlike [Job.wait](Job.md#wait) it does not block or reject
+on a terminal failure state. States a newer server reports that this
+client version does not know pass through as-is.
+
+#### Returns
+
+`Promise`&lt;`string`&gt;
+
+***
+
+### wait()
+
+```ts
+wait(): Promise<void>
+```
+
+Wait until the operation reaches a terminal state.
+
+#### Returns
+
+`Promise`&lt;`void`&gt;
@@ -295,6 +295,29 @@ await table.createIndex("my_float_col");

 ***

+### createIndexAsync()
+
+```ts
+abstract createIndexAsync(column, options?): Promise<Job>
+```
+
+Create an index, returning a handle to the indexing job.
+
+The job may already be complete when returned; callers must not assume
+the index exists until [Job.wait](Job.md#wait) resolves.
+
+#### Parameters
+
+* **column**: `string`
+
+* **options?**: `Partial`&lt;[`IndexOptions`](../interfaces/IndexOptions.md)&gt;
+
+#### Returns
+
+`Promise`&lt;[`Job`](Job.md)&gt;
+
+***
+
 ### currentBranch()

 ```ts
@@ -25,6 +25,7 @@
 - [Connection](classes/Connection.md)
 - [HeaderProvider](classes/HeaderProvider.md)
 - [Index](classes/Index.md)
+- [Job](classes/Job.md)
 - [MakeArrowTableOptions](classes/MakeArrowTableOptions.md)
 - [MatchQuery](classes/MatchQuery.md)
 - [MergeInsertBuilder](classes/MergeInsertBuilder.md)
@@ -88,6 +89,9 @@
 - [IvfFlatOptions](interfaces/IvfFlatOptions.md)
 - [IvfPqOptions](interfaces/IvfPqOptions.md)
 - [IvfRqOptions](interfaces/IvfRqOptions.md)
+- [JobDescription](interfaces/JobDescription.md)
+- [JobFailureInfo](interfaces/JobFailureInfo.md)
+- [JobInfo](interfaces/JobInfo.md)
 - [ListNamespacesOptions](interfaces/ListNamespacesOptions.md)
 - [ListNamespacesResponse](interfaces/ListNamespacesResponse.md)
 - [LsmWriteSpec](interfaces/LsmWriteSpec.md)
@@ -0,0 +1,66 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / JobDescription
+
+# Interface: JobDescription
+
+A described job from `Connection.getJob`.
+
+## Properties
+
+### creationMs
+
+```ts
+creationMs: number;
+```
+
+When the job was created, in milliseconds since the epoch.
+
+***
+
+### failure?
+
+```ts
+optional failure: JobFailureInfo;
+```
+
+Why the job failed, when the job is failed and the server reports a
+reason.
+
+***
+
+### jobId
+
+```ts
+jobId: string;
+```
+
+***
+
+### jobType
+
+```ts
+jobType: string;
+```
+
+***
+
+### specJson?
+
+```ts
+optional specJson: string;
+```
+
+The job-type-specific specification as a JSON string, when present.
+
+***
+
+### state
+
+```ts
+state: string;
+```
+
+Lifecycle state: "running", "finished", "failed", or "cancelled".
@@ -0,0 +1,33 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / JobFailureInfo
+
+# Interface: JobFailureInfo
+
+The server's account of why a job failed.
+
+## Properties
+
+### message?
+
+```ts
+optional message: string;
+```
+
+***
+
+### phase?
+
+```ts
+optional phase: string;
+```
+
+***
+
+### retryable?
+
+```ts
+optional retryable: boolean;
+```
@@ -0,0 +1,58 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / JobInfo
+
+# Interface: JobInfo
+
+A row from `Connection.listJobs`: one server-side job.
+
+## Properties
+
+### createdAtMillis
+
+```ts
+createdAtMillis: number;
+```
+
+When the job was created, in milliseconds since the epoch.
+
+***
+
+### jobId
+
+```ts
+jobId: string;
+```
+
+The job id -- what `Connection.getJob` and `Connection.cancelJob`
+accept.
+
+***
+
+### jobType
+
+```ts
+jobType: string;
+```
+
+***
+
+### state
+
+```ts
+state: string;
+```
+
+Lifecycle state: "running", "finished", "failed", or "cancelled".
+
+***
+
+### table
+
+```ts
+table: string;
+```
+
+The table the job runs against, without URI or namespace.
@@ -28,7 +28,7 @@
    <properties>
        <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <arrow.version>15.0.0</arrow.version>
-        <lance-core.version>10.0.0-beta.5</lance-core.version>
+        <lance-core.version>10.1.0-beta.1</lance-core.version>
        <spotless.skip>false</spotless.skip>
        <spotless.version>2.30.0</spotless.version>
        <spotless.java.googlejavaformat.version>1.7</spotless.java.googlejavaformat.version>
@@ -197,6 +197,35 @@ describe.each([arrow15, arrow16, arrow17, arrow18])(
        expect(table.getChild("d")?.toJSON()).toEqual([9n, 10n, null]);
      });

+      it("will use a provided FixedSizeList schema with typed array values", function () {
+        const schema = new Schema([
+          new Field("text", new Utf8(), false),
+          new Field(
+            "vector",
+            new FixedSizeList(3, new Field("item", new Float32(), false)),
+            false,
+          ),
+        ]);
+
+        const table = makeArrowTable(
+          [
+            {
+              text: "foo",
+              vector: new Float32Array([1, 2, 3]),
+            },
+          ],
+          { schema },
+        );
+
+        expect(table.getChild("text")?.toJSON()).toEqual(["foo"]);
+        expect(
+          table
+            .getChild("vector")
+            ?.toJSON()
+            .map((value) => value.toJSON()),
+        ).toEqual([[1, 2, 3]]);
+      });
+
      it("will assume the column `vector` is FixedSizeList<Float32> by default", async function () {
        const schema = new Schema([
          new Field("a", new Float(Precision.DOUBLE), true),
@@ -170,6 +170,38 @@ describe("remote connection", () => {
    );
  });

+  it("surfaces JSON server errors from remote table operations", async () => {
+    await withMockDatabase(
+      (req, res) => {
+        const path = req.url ?? "";
+        if (path.endsWith("/describe/")) {
+          res.writeHead(200, { "Content-Type": "application/json" }).end(
+            JSON.stringify({
+              name: "broken_table",
+              version: 1,
+              schema: { fields: [] },
+            }),
+          );
+          return;
+        }
+
+        if (path.endsWith("/count_rows/")) {
+          res
+            .writeHead(400, { "Content-Type": "application/json" })
+            .end(JSON.stringify({ error: "count rows failed" }));
+          return;
+        }
+
+        res.writeHead(404).end();
+      },
+      async (db) => {
+        const table = await db.openTable("broken_table");
+
+        await expect(table.countRows()).rejects.toThrow("count rows failed");
+      },
+    );
+  });
+
  it("should pass on requested extra headers", async () => {
    await withMockDatabase(
      (req, res) => {
@@ -877,3 +909,96 @@ describe("remote connection", () => {
    });
  });
 });
+
+describe("remote connection jobs surface", () => {
+  it("lists, describes, cancels, and reads history", async () => {
+    const { tableFromArrays, tableToIPC } = await import("apache-arrow");
+    const eventsTable = tableFromArrays({ state: ["created", "succeeded"] });
+    const eventsBody = Buffer.from(tableToIPC(eventsTable, "stream"));
+
+    await withMockDatabase(
+      (req, res) => {
+        let body = "";
+        req.on("data", (chunk) => {
+          body += chunk;
+        });
+        req.on("end", () => {
+          const payload = body.length > 0 ? JSON.parse(body) : {};
+          if (req.url === "/v1/jobs/list") {
+            if (payload["page_token"] === undefined) {
+              res
+                .writeHead(200, { "Content-Type": "application/json" })
+                .end(
+                  '{"jobs": [{"job_id": "job-1", "table": "t1", ' +
+                    '"job_type": "create_index", "state": "in_progress", ' +
+                    '"created_at_millis": 1000}], "page_token": "next"}',
+                );
+            } else {
+              res
+                .writeHead(200, { "Content-Type": "application/json" })
+                .end(
+                  '{"jobs": [{"job_id": "job-2", "table": "t2", ' +
+                    '"job_type": "create_index", "state": "succeeded", ' +
+                    '"created_at_millis": 2000}]}',
+                );
+            }
+          } else if (req.url === "/v1/jobs/describe") {
+            if (payload["job_id"] !== "job-1") {
+              res.writeHead(404).end("no such job");
+              return;
+            }
+            res
+              .writeHead(200, { "Content-Type": "application/json" })
+              .end(
+                '{"job_id": "job-1", "job_type": "create_index", ' +
+                  '"job_state": "FAILED", "creation_ms": 1000, ' +
+                  '"spec": {"column": "vec"}, "failure": {"phase": "execute", ' +
+                  '"message": "worker died", "retryable": true}}',
+              );
+          } else if (req.url === "/v1/jobs/cancel") {
+            if (payload["job_id"] !== "job-1") {
+              res.writeHead(404).end("no such job");
+              return;
+            }
+            res
+              .writeHead(200, { "Content-Type": "application/json" })
+              .end('{"job_id": "job-1"}');
+          } else if (req.url === "/v1/jobs/query_events") {
+            res
+              .writeHead(200, {
+                "Content-Type": "application/vnd.apache.arrow.stream",
+              })
+              .end(eventsBody);
+          } else {
+            res.writeHead(404).end();
+          }
+        });
+      },
+      async (db) => {
+        const jobs = await db.listJobs();
+        expect(jobs.map((job) => job.jobId)).toEqual(["job-1", "job-2"]);
+        expect(jobs[0].state).toEqual("running");
+        expect(jobs[1].state).toEqual("finished");
+
+        const description = await db.getJob("job-1");
+        expect(description?.state).toEqual("failed");
+        expect(JSON.parse(description?.specJson ?? "")).toEqual({
+          column: "vec",
+        });
+        expect(description?.failure?.message).toEqual("worker died");
+        expect(await db.getJob("missing")).toBeNull();
+
+        expect(await db.cancelJob("job-1")).toBe(true);
+        expect(await db.cancelJob("missing")).toBe(false);
+
+        const history = await db.jobHistory("job-1");
+        expect(history.numRows).toEqual(2);
+
+        const job = db.job("job-1");
+        expect(job.id).toEqual("job-1");
+        expect(await job.status()).toEqual("failed");
+        await expect(job.wait()).rejects.toThrow("worker died");
+      },
+    );
+  });
+});
@@ -86,6 +86,44 @@ describe.each([arrow15, arrow16, arrow17, arrow18])(
      await expect(table.countRows()).resolves.toBe(3);
    });

+    it("should support a foreign Float64 vector schema end to end", async () => {
+      const conn = await connect(tmpDir.name);
+      const schema = new arrow.Schema([
+        new arrow.Field("resource_id", new arrow.Int32(), false),
+        new arrow.Field(
+          "vector",
+          new arrow.FixedSizeList(
+            3,
+            new arrow.Field("value", new arrow.Float64(), true),
+          ),
+          false,
+        ),
+      ]);
+      const data = [
+        {
+          // biome-ignore lint/style/useNamingConvention: matches the reported schema
+          resource_id: 0,
+          vector: [0.1, 0.1, 0.1],
+        },
+      ];
+
+      const resources = await conn.createTable("resources", data, { schema });
+
+      const existing = await resources
+        .query()
+        .where("resource_id = 0")
+        .limit(1)
+        .toArray();
+      expect(existing).toHaveLength(1);
+
+      const matched = await resources
+        .search(Float64Array.from(data[0].vector))
+        .limit(1)
+        .toArray();
+      expect(matched).toHaveLength(1);
+      expect(matched[0]["resource_id"]).toBe(0);
+    });
+
    it("should support branches", async () => {
      await table.add([{ id: 1 }]);
      expect(await table.countRows()).toBe(1);
@@ -851,7 +889,11 @@ describe("When creating an index", () => {
  afterEach(() => tmpDir.removeCallback());

  it("should create a vector index on vector columns", async () => {
-    await tbl.createIndex("vec");
+    const job = await tbl.createIndexAsync("vec");
+    expect(job.id).toBeNull();
+    await job.wait();
+    // Cancelling a job that already finished succeeds and does nothing.
+    await job.cancel();

    // check index directory
    const indexDir = path.join(tmpDir.name, "test.lance", "_indices");
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

+import { tableFromIPC } from "apache-arrow";
 import {
  Data,
  SchemaLike,
@@ -20,6 +21,9 @@ import type {
  CreateNamespaceResponse,
  DescribeNamespaceResponse,
  DropNamespaceResponse,
+  Job,
+  JobDescription,
+  JobInfo,
  ListNamespacesResponse,
 } from "./native";
 export type {
@@ -436,6 +440,40 @@ export abstract class Connection {
    newName: string,
    options?: RenameTableOptions,
  ): Promise<void>;
+
+  /**
+   * A {@link Job} handle for a server-side job by id.
+   *
+   * The handle is constructed without a server round trip; an unknown id
+   * surfaces when the handle is used. Dropping the handle has no effect on
+   * the job itself.
+   */
+  abstract job(jobId: string): Job;
+
+  /** List server-side jobs across the database's tables. */
+  abstract listJobs(): Promise<JobInfo[]>;
+
+  /**
+   * Describe a single server-side job by id.
+   *
+   * Resolves to `null` when the server has no such job.
+   */
+  abstract getJob(jobId: string): Promise<JobDescription | null>;
+
+  /**
+   * Request cancellation of a server-side job by id.
+   *
+   * Resolves to true if the server accepted the cancellation, false if no
+   * such job exists. Cancelling an already-terminal job is a no-op success.
+   */
+  abstract cancelJob(jobId: string): Promise<boolean>;
+
+  /**
+   * The lifecycle event history of a server-side job, as an Arrow table.
+   *
+   * Lists history across all jobs when `jobId` is omitted.
+   */
+  abstract jobHistory(jobId?: string): Promise<ArrowTable>;
 }

 /** @hideconstructor */
@@ -722,6 +760,30 @@ export class LocalConnection extends Connection {
      options?.newNamespacePath,
    );
  }
+
+  job(jobId: string): Job {
+    return this.inner.job(jobId);
+  }
+
+  async listJobs(): Promise<JobInfo[]> {
+    return this.inner.listJobs();
+  }
+
+  async getJob(jobId: string): Promise<JobDescription | null> {
+    return this.inner.getJob(jobId);
+  }
+
+  async cancelJob(jobId: string): Promise<boolean> {
+    return this.inner.cancelJob(jobId);
+  }
+
+  async jobHistory(jobId?: string): Promise<ArrowTable> {
+    const buf = await this.inner.jobHistory(jobId);
+    if (buf.length === 0) {
+      return new ArrowTable();
+    }
+    return tableFromIPC(buf);
+  }
 }

 /**
@@ -85,7 +85,13 @@ export {
  RenameTableOptions,
 } from "./connection";

-export { Session } from "./native.js";
+export {
+  Job,
+  JobDescription,
+  JobFailureInfo,
+  JobInfo,
+  Session,
+} from "./native.js";

 export {
  ExecutableQuery,
@@ -30,6 +30,7 @@ import {
  DropColumnsResult,
  IndexConfig,
  IndexStatistics,
+  Job,
  Branches as NativeBranches,
  OptimizeStats,
  TableStatistics,
@@ -358,6 +359,17 @@ export abstract class Table {
    options?: Partial<IndexOptions>,
  ): Promise<void>;

+  /**
+   * Create an index, returning a handle to the indexing job.
+   *
+   * The job may already be complete when returned; callers must not assume
+   * the index exists until {@link Job.wait} resolves.
+   */
+  abstract createIndexAsync(
+    column: string,
+    options?: Partial<IndexOptions>,
+  ): Promise<Job>;
+
  /**
   * Drop an index from the table.
   *
@@ -940,6 +952,22 @@ export class LocalTable extends Table {
    );
  }

+  async createIndexAsync(
+    column: string,
+    options?: Partial<IndexOptions>,
+  ): Promise<Job> {
+    // biome-ignore lint/suspicious/noExplicitAny: skip
+    const nativeIndex = (options?.config as any)?.inner;
+    return await this.inner.createIndexAsync(
+      nativeIndex,
+      column,
+      options?.replace,
+      options?.waitTimeoutSeconds,
+      options?.name,
+      options?.train,
+    );
+  }
+
  async dropIndex(name: string): Promise<void> {
    await this.inner.dropIndex(name);
  }
@@ -340,6 +340,69 @@ impl Connection {
        self.get_inner()?.drop_all_tables(&ns).await.default_error()
    }

+    /// A `Job` handle for a server-side job by id.
+    ///
+    /// The handle is constructed without a server round trip; an unknown id
+    /// surfaces when the handle is used.
+    #[napi]
+    pub fn job(&self, job_id: String) -> napi::Result<crate::job::Job> {
+        let job = self.get_inner()?.job(job_id).default_error()?;
+        Ok(crate::job::Job::new(job))
+    }
+
+    /// List server-side jobs across the database's tables.
+    #[napi(catch_unwind)]
+    pub async fn list_jobs(&self) -> napi::Result<Vec<crate::job::JobInfo>> {
+        let jobs = self.get_inner()?.list_jobs().await.default_error()?;
+        Ok(jobs.into_iter().map(Into::into).collect())
+    }
+
+    /// Describe a single server-side job by id. `null` when the server has
+    /// no such job.
+    #[napi(catch_unwind)]
+    pub async fn get_job(
+        &self,
+        job_id: String,
+    ) -> napi::Result<Option<crate::job::JobDescription>> {
+        let description = self.get_inner()?.get_job(&job_id).await.default_error()?;
+        Ok(description.map(Into::into))
+    }
+
+    /// Request cancellation of a server-side job by id. Returns true if the
+    /// server accepted the cancellation, false if no such job exists.
+    #[napi(catch_unwind)]
+    pub async fn cancel_job(&self, job_id: String) -> napi::Result<bool> {
+        self.get_inner()?.cancel_job(&job_id).await.default_error()
+    }
+
+    /// The lifecycle event history of a server-side job (all jobs when
+    /// `job_id` is null), as an Arrow IPC stream buffer. Empty when there is
+    /// no history.
+    #[napi(catch_unwind)]
+    pub async fn job_history(&self, job_id: Option<String>) -> napi::Result<Buffer> {
+        let batches = self
+            .get_inner()?
+            .job_history(job_id.as_deref())
+            .await
+            .default_error()?;
+        let Some(first) = batches.first() else {
+            return Ok(Buffer::from(Vec::<u8>::new()));
+        };
+        let mut out = Vec::new();
+        let mut writer = arrow_ipc::writer::StreamWriter::try_new(&mut out, &first.schema())
+            .map_err(|e| napi::Error::from_reason(e.to_string()))?;
+        for batch in &batches {
+            writer
+                .write(batch)
+                .map_err(|e| napi::Error::from_reason(e.to_string()))?;
+        }
+        writer
+            .finish()
+            .map_err(|e| napi::Error::from_reason(e.to_string()))?;
+        drop(writer);
+        Ok(Buffer::from(out))
+    }
+
    #[napi(catch_unwind)]
    /// Describe a namespace and return its properties.
    pub async fn describe_namespace(
@@ -0,0 +1,123 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+use std::sync::Arc;
+
+use napi_derive::napi;
+
+use crate::error::NapiErrorExt;
+
+/// A handle to an operation that may still be running.
+#[napi]
+pub struct Job {
+    inner: Arc<lancedb::Job>,
+}
+
+impl Job {
+    pub(crate) fn new(inner: lancedb::Job) -> Self {
+        Self {
+            inner: Arc::new(inner),
+        }
+    }
+}
+
+#[napi]
+impl Job {
+    /// Identifies the operation on the server that is running it. Operations
+    /// that run in this process have no server id. The value is opaque.
+    #[napi(getter)]
+    pub fn id(&self) -> Option<String> {
+        self.inner.id().map(str::to_string)
+    }
+
+    /// The operation's current lifecycle state: "running", "finished",
+    /// "failed", or "cancelled".
+    ///
+    /// A point snapshot; unlike {@link Job.wait} it does not block or reject
+    /// on a terminal failure state. States a newer server reports that this
+    /// client version does not know pass through as-is.
+    #[napi(catch_unwind)]
+    pub async fn status(&self) -> napi::Result<String> {
+        self.inner.status().await.default_error()
+    }
+
+    /// Wait until the operation reaches a terminal state.
+    #[napi(catch_unwind)]
+    pub async fn wait(&self) -> napi::Result<()> {
+        self.inner.wait().await.default_error()
+    }
+
+    /// Request cancellation. Cancelling a finished operation is a no-op.
+    #[napi(catch_unwind)]
+    pub async fn cancel(&self) -> napi::Result<()> {
+        self.inner.cancel().await.default_error()
+    }
+}
+
+/// A row from `Connection.listJobs`: one server-side job.
+#[napi(object)]
+pub struct JobInfo {
+    /// The job id -- what `Connection.getJob` and `Connection.cancelJob`
+    /// accept.
+    pub job_id: String,
+    /// The table the job runs against, without URI or namespace.
+    pub table: String,
+    pub job_type: String,
+    /// Lifecycle state: "running", "finished", "failed", or "cancelled".
+    pub state: String,
+    /// When the job was created, in milliseconds since the epoch.
+    pub created_at_millis: i64,
+}
+
+impl From<lancedb::database::JobInfo> for JobInfo {
+    fn from(info: lancedb::database::JobInfo) -> Self {
+        Self {
+            job_id: info.job_id,
+            table: info.table,
+            job_type: info.job_type,
+            state: info.state,
+            created_at_millis: info.created_at_millis,
+        }
+    }
+}
+
+/// The server's account of why a job failed.
+#[napi(object)]
+pub struct JobFailureInfo {
+    pub phase: Option<String>,
+    pub message: Option<String>,
+    pub retryable: Option<bool>,
+}
+
+/// A described job from `Connection.getJob`.
+#[napi(object)]
+pub struct JobDescription {
+    pub job_id: String,
+    pub job_type: String,
+    /// Lifecycle state: "running", "finished", "failed", or "cancelled".
+    pub state: String,
+    /// When the job was created, in milliseconds since the epoch.
+    pub creation_ms: i64,
+    /// The job-type-specific specification as a JSON string, when present.
+    pub spec_json: Option<String>,
+    /// Why the job failed, when the job is failed and the server reports a
+    /// reason.
+    pub failure: Option<JobFailureInfo>,
+}
+
+impl From<lancedb::database::JobDescription> for JobDescription {
+    fn from(description: lancedb::database::JobDescription) -> Self {
+        Self {
+            job_id: description.job_id,
+            job_type: description.job_type,
+            state: description.state,
+            creation_ms: description.creation_ms,
+            spec_json: (!description.spec.is_null()).then(|| description.spec.to_string()),
+            failure: description.failure.map(|failure| JobFailureInfo {
+                phase: failure.phase,
+                message: failure.message,
+                retryable: failure.retryable,
+            }),
+        }
+    }
+}
@@ -11,6 +11,7 @@ mod error;
 mod header;
 mod index;
 mod iterator;
+mod job;
 pub mod merge;
 pub mod otel;
 pub mod permutation;
@@ -168,6 +168,39 @@ impl Table {
        builder.execute().await.default_error()
    }

+    #[napi(catch_unwind)]
+    pub async fn create_index_async(
+        &self,
+        index: Option<&Index>,
+        column: String,
+        replace: Option<bool>,
+        wait_timeout_s: Option<i64>,
+        name: Option<String>,
+        train: Option<bool>,
+    ) -> napi::Result<crate::job::Job> {
+        let lancedb_index = if let Some(index) = index {
+            index.consume()?
+        } else {
+            lancedb::index::Index::Auto
+        };
+        let mut builder = self.inner_ref()?.create_index(&[column], lancedb_index);
+        if let Some(replace) = replace {
+            builder = builder.replace(replace);
+        }
+        if let Some(timeout) = wait_timeout_s {
+            builder =
+                builder.wait_timeout(std::time::Duration::from_secs(timeout.try_into().unwrap()));
+        }
+        if let Some(name) = name {
+            builder = builder.name(name);
+        }
+        if let Some(train) = train {
+            builder = builder.train(train);
+        }
+        let job = builder.execute_async().await.default_error()?;
+        Ok(crate::job::Job::new(job))
+    }
+
    #[napi(catch_unwind)]
    pub async fn drop_index(&self, index_name: String) -> napi::Result<()> {
        self.inner_ref()?
@@ -306,7 +339,9 @@ impl Table {
        let transforms = NewColumnTransform::SqlExpressions(transforms);
        let res = self
            .inner_ref()?
-            .add_columns(transforms, None)
+            .add_columns()
+            .transform(transforms)
+            .execute()
            .await
            .default_error()?;
        Ok(res.into())
@@ -323,7 +358,9 @@ impl Table {
        let transforms = NewColumnTransform::AllNulls(schema);
        let res = self
            .inner_ref()?
-            .add_columns(transforms, None)
+            .add_columns()
+            .transform(transforms)
+            .execute()
            .await
            .default_error()?;
        Ok(res.into())
@@ -26,7 +26,7 @@ lance-namespace-impls.workspace = true
 lance-io.workspace = true
 env_logger.workspace = true
 log.workspace = true
-pyo3 = { version = "0.28", features = ["extension-module", "abi3-py39", "chrono"] }
+pyo3 = { version = "0.28", features = ["extension-module", "abi3-py310", "chrono"] }
 chrono = { version = "0.4", default-features = false, features = ["clock"] }
 pyo3-async-runtimes = { version = "0.28", features = [
    "attributes",
@@ -43,7 +43,7 @@ libc = "0.2"
 [build-dependencies]
 pyo3-build-config = { version = "0.28", features = [
    "extension-module",
-    "abi3-py39",
+    "abi3-py310",
 ] }

 [features]
@@ -20,6 +20,7 @@ from .remote import ClientConfig
 from .remote.db import RemoteDBConnection
 from .expr import Expr, col, lit, func
 from .schema import blob, vector, BlobType
+from .job import AsyncJob, Job
 from .table import AsyncTable, Table
 from .types import BaseTokenizerType
 from ._lancedb import Session
@@ -500,6 +501,7 @@ __all__ = [
    "connect_namespace",
    "connect_namespace_async",
    "AsyncConnection",
+    "AsyncJob",
    "AsyncLanceNamespaceDBConnection",
    "AsyncTable",
    "FtsToken",
@@ -513,6 +515,7 @@ __all__ = [
    "BlobType",
    "vector",
    "DBConnection",
+    "Job",
    "LanceDBConnection",
    "LanceNamespaceDBConnection",
    "RemoteDBConnection",
@@ -14,14 +14,10 @@ import pyarrow as pa
 from .expr import Expr
 from .schema import blob_v2_column_paths
 from .types import BlobMode, QueryProjection, QueryProjectionSpec
-from .util import get_uri_scheme

 if TYPE_CHECKING:
    from _typeshed import WriteableBuffer

-    from .remote.table import RemoteTable
-    from .table import AsyncTable, Table
-
 BLOB_MODE_TO_HANDLING = {
    "lazy": "blobs_descriptions",
    "bytes": "all_binary",
@@ -104,22 +100,6 @@ def validate_blob_mode(blob_mode: BlobMode) -> None:
        raise ValueError(f"blob_mode must be one of {modes}, got {blob_mode!r}")


-def supports_blob_auto_row_id(table: Table | AsyncTable | RemoteTable) -> bool:
-    """Blob auto row-id applies to native tables, not LanceDB Cloud."""
-    from .remote.table import RemoteTable
-
-    if isinstance(table, RemoteTable):
-        return False
-
-    inner = getattr(table, "_inner", None)
-    if inner is not None:
-        uri = inner.database().uri
-        if isinstance(uri, str) and get_uri_scheme(uri) == "db":
-            return False
-
-    return True
-
-
 def projection_includes_blob_column(
    projection: QueryProjection,
    blob_columns: Iterable[str],
@@ -164,16 +144,14 @@ def v2_projection_needs_row_id(


 def blob_auto_row_id_for_scan(
-    table: Table | AsyncTable | RemoteTable,
    schema: pa.Schema,
    projection: QueryProjection,
    *,
    with_row_id: bool | None,
 ) -> bool:
+    """Auto row-id only applies when the caller said nothing about row ids."""
    if with_row_id is not None:
        return False
-    if not supports_blob_auto_row_id(table):
-        return False
    return v2_projection_needs_row_id(schema, projection, with_row_id=False)


@@ -186,6 +164,11 @@ def finalize_blob_query_table(
 ) -> pa.Table:
    if user_requested_row_id or not blob_auto_row_id:
        return tbl
+    if "_rowid" not in tbl.column_names:
+        # A backend that ignores the row-id request leaves nothing to stash. Hand
+        # back the projection as-is so fetch_blobs raises the error that names the
+        # ways to supply row ids, rather than failing here about a hidden column.
+        return tbl
    return stash_auto_row_ids(tbl, blob_paths)


@@ -287,7 +270,8 @@ def _iter_projection_pairs(
            if isinstance(expr, str):
                yield name, expr
            elif isinstance(expr, Expr):
-                yield name, expr.to_sql()
+                source = expr._column_name()
+                yield name, source if source is not None else expr.to_sql()
        return
    for column in projection:
        if isinstance(column, str):
@@ -297,7 +281,8 @@ def _iter_projection_pairs(
            if isinstance(expr, str):
                yield name, expr
            elif isinstance(expr, Expr):
-                yield name, expr.to_sql()
+                source = expr._column_name()
+                yield name, source if source is not None else expr.to_sql()


 def _set_blob_column(tbl: pa.Table, output_name: str, blobs: pa.Array) -> pa.Table:
@@ -87,6 +87,7 @@ class PyExpr:
    def contains(self, substr: "PyExpr") -> "PyExpr": ...
    def isin(self, values: List["PyExpr"]) -> "PyExpr": ...
    def cast(self, data_type: pa.DataType) -> "PyExpr": ...
+    def column_name(self) -> Optional[str]: ...
    def to_sql(self) -> str: ...

 def expr_col(name: str) -> PyExpr: ...
@@ -146,6 +147,13 @@ class Connection(object):
        start_after: Optional[str],
        limit: Optional[int],
    ) -> list[str]: ...  # Deprecated: Use list_tables instead
+    def job(self, job_id: str) -> Job: ...
+    async def list_jobs(self) -> List[JobInfo]: ...
+    async def get_job(self, job_id: str) -> Optional[JobDescription]: ...
+    async def cancel_job(self, job_id: str) -> bool: ...
+    async def job_history(
+        self, job_id: Optional[str] = None
+    ) -> List[pa.RecordBatch]: ...
    async def create_table(
        self,
        name: str,
@@ -209,6 +217,47 @@ class BlobFile:
    def read_range(self, offset: int, length: int) -> bytes: ...
    def read_up_to(self, length: int) -> bytes: ...

+class Job:
+    @property
+    def id(self) -> Optional[str]: ...
+    async def status(self) -> str: ...
+    async def wait(self) -> None: ...
+    async def cancel(self) -> None: ...
+
+class JobInfo:
+    @property
+    def job_id(self) -> str: ...
+    @property
+    def table(self) -> str: ...
+    @property
+    def job_type(self) -> str: ...
+    @property
+    def state(self) -> str: ...
+    @property
+    def created_at_millis(self) -> int: ...
+
+class JobFailureInfo:
+    @property
+    def phase(self) -> Optional[str]: ...
+    @property
+    def message(self) -> Optional[str]: ...
+    @property
+    def retryable(self) -> Optional[bool]: ...
+
+class JobDescription:
+    @property
+    def job_id(self) -> str: ...
+    @property
+    def job_type(self) -> str: ...
+    @property
+    def state(self) -> str: ...
+    @property
+    def creation_ms(self) -> int: ...
+    @property
+    def spec_json(self) -> Optional[str]: ...
+    @property
+    def failure(self) -> Optional[JobFailureInfo]: ...
+
 class Table:
    def name(self) -> str: ...
    def __repr__(self) -> str: ...
@@ -248,6 +297,28 @@ class Table:
        name: Optional[str],
        train: Optional[bool],
    ): ...
+    async def create_index_async(
+        self,
+        column: str,
+        index: Union[
+            IvfFlat,
+            IvfSq,
+            IvfPq,
+            HnswPq,
+            HnswSq,
+            HnswFlat,
+            BTree,
+            Bitmap,
+            LabelList,
+            Fm,
+            FTS,
+        ],
+        replace: Optional[bool],
+        wait_timeout: Optional[object],
+        *,
+        name: Optional[str],
+        train: Optional[bool],
+    ) -> Job: ...
    async def list_versions(self) -> List[Dict[str, Any]]: ...
    async def version(self) -> int: ...
    async def checkout(self, version: Union[int, str]): ...
@@ -508,6 +579,7 @@ class PyQueryRequest:
    filter: Optional[Union[str, bytes]]
    full_text_search: Optional[FullTextQuery]
    select: Optional[Union[str, List[str]]]
+    select_source_columns: Optional[Dict[str, str]]
    fast_search: Optional[bool]
    with_row_id: Optional[bool]
    use_lsm: Optional[bool]
@@ -45,6 +45,7 @@ from lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError

 from . import __version__
 from ._lancedb import connect as lancedb_connect  # type: ignore
+from .job import AsyncJob, Job
 from .table import (
    AsyncTable,
    LanceTable,
@@ -63,6 +64,7 @@ if TYPE_CHECKING:
    from .pydantic import LanceModel

    from ._lancedb import Connection as LanceDbConnection
+    from ._lancedb import JobDescription, JobInfo
    from .common import DATA, URI
    from .embeddings import EmbeddingFunctionConfig
    from ._lancedb import Session
@@ -178,6 +180,51 @@ class DBConnection(EnforceOverrides):
            "Namespace operations are not supported for this connection type"
        )

+    def namespace_exists(self, namespace_id: List[str]) -> bool:
+        """Check if a namespace exists.
+
+        Parameters
+        ----------
+        namespace_id: List[str]
+            The namespace identifier to check.
+
+        Returns
+        -------
+        bool
+            True if the namespace exists, False otherwise.
+
+        Raises
+        ------
+        NotImplementedError
+            If the connection type does not support namespace operations.
+        """
+        raise NotImplementedError(
+            "Namespace operations are not supported for this connection type"
+        )
+
+    def table_exists(self, table_id: List[str]) -> bool:
+        """Check if a table exists.
+
+        Parameters
+        ----------
+        table_id: List[str]
+            The table identifier to check (full path including namespace
+            segments and table name).
+
+        Returns
+        -------
+        bool
+            True if the table exists, False otherwise.
+
+        Raises
+        ------
+        NotImplementedError
+            If the connection type does not support namespace operations.
+        """
+        raise NotImplementedError(
+            "Namespace operations are not supported for this connection type"
+        )
+
    def list_tables(
        self,
        namespace_path: Optional[List[str]] = None,
@@ -563,6 +610,46 @@ class DBConnection(EnforceOverrides):
        """
        raise NotImplementedError("serialize is not supported for this connection type")

+    def job(self, job_id: str) -> Job:
+        """A [Job][lancedb.job.Job] handle for a server-side job by id.
+
+        The handle is constructed without a server round trip; an unknown id
+        surfaces when the handle is used. Dropping the handle has no effect
+        on the job itself.
+        """
+        raise NotImplementedError("job is not supported for this connection type")
+
+    def list_jobs(self) -> List[JobInfo]:
+        """List server-side jobs across the database's tables."""
+        raise NotImplementedError("list_jobs is not supported for this connection type")
+
+    def get_job(self, job_id: str) -> Optional[JobDescription]:
+        """Describe a single server-side job by id.
+
+        Returns None when the server has no such job.
+        """
+        raise NotImplementedError("get_job is not supported for this connection type")
+
+    def cancel_job(self, job_id: str) -> bool:
+        """Request cancellation of a server-side job by id.
+
+        Returns True if the server accepted the cancellation, False if no
+        such job exists. Cancelling an already-terminal job is a no-op
+        success.
+        """
+        raise NotImplementedError(
+            "cancel_job is not supported for this connection type"
+        )
+
+    def job_history(self, job_id: Optional[str] = None) -> List[pa.RecordBatch]:
+        """The lifecycle event history of a server-side job, as Arrow batches.
+
+        Lists history across all jobs when `job_id` is None.
+        """
+        raise NotImplementedError(
+            "job_history is not supported for this connection type"
+        )
+

 class LanceDBConnection(DBConnection):
    """
@@ -620,6 +707,9 @@ class LanceDBConnection(DBConnection):
        self._namespace_client_properties = namespace_client_properties
        if _inner is not None:
            self._conn = _inner
+            # Native-derived wrappers resolve this in their async reconstruction
+            # path so construction never synchronously re-enters LOOP.
+            self._read_consistency_interval = read_consistency_interval
            self._cached_namespace_client = None
            return

@@ -669,11 +759,14 @@ class LanceDBConnection(DBConnection):
        # storage_options.  Also, this class really shouldn't be holding any state
        # beyond _conn.
        self._conn = AsyncConnection(LOOP.run(do_connect()))
+        # Keep property access synchronous so debugger introspection cannot wait on
+        # the background loop while that thread is suspended at a breakpoint.
+        self._read_consistency_interval = read_consistency_interval
        self._cached_namespace_client: Optional[LanceNamespace] = None

    @property
    def read_consistency_interval(self) -> Optional[timedelta]:
-        return LOOP.run(self._conn.get_read_consistency_interval())
+        return self._read_consistency_interval

    @property
    def session(self) -> Optional[Session]:
@@ -684,15 +777,19 @@ class LanceDBConnection(DBConnection):
        return self._conn.uri

    @classmethod
-    def from_inner(cls, inner: LanceDbConnection):
-        return cls(None, _inner=inner)
+    def from_inner(
+        cls,
+        inner: LanceDbConnection,
+        read_consistency_interval: Optional[timedelta],
+    ):
+        return cls(
+            None,
+            read_consistency_interval=read_consistency_interval,
+            _inner=inner,
+        )

    def __repr__(self) -> str:
-        val = f"{self.__class__.__name__}(uri={self._conn.uri!r}"
-        if self.read_consistency_interval is not None:
-            val += f", read_consistency_interval={repr(self.read_consistency_interval)}"
-        val += ")"
-        return val
+        return f"{self.__class__.__name__}(uri={self._conn.uri!r})"

    @override
    def serialize(self) -> str:
@@ -1129,6 +1226,47 @@ class LanceDBConnection(DBConnection):
            )
        )

+    @override
+    def job(self, job_id: str) -> Job:
+        """A [Job][lancedb.job.Job] handle for a server-side job by id.
+
+        The handle is constructed without a server round trip; an unknown id
+        surfaces when the handle is used. Dropping the handle has no effect
+        on the job itself.
+        """
+        return Job(self._conn.job(job_id))
+
+    @override
+    def list_jobs(self) -> List[JobInfo]:
+        """List server-side jobs across the database's tables."""
+        return LOOP.run(self._conn.list_jobs())
+
+    @override
+    def get_job(self, job_id: str) -> Optional[JobDescription]:
+        """Describe a single server-side job by id.
+
+        Returns None when the server has no such job.
+        """
+        return LOOP.run(self._conn.get_job(job_id))
+
+    @override
+    def cancel_job(self, job_id: str) -> bool:
+        """Request cancellation of a server-side job by id.
+
+        Returns True if the server accepted the cancellation, False if no
+        such job exists. Cancelling an already-terminal job is a no-op
+        success.
+        """
+        return LOOP.run(self._conn.cancel_job(job_id))
+
+    @override
+    def job_history(self, job_id: Optional[str] = None) -> List[pa.RecordBatch]:
+        """The lifecycle event history of a server-side job, as Arrow batches.
+
+        Lists history across all jobs when `job_id` is None.
+        """
+        return LOOP.run(self._conn.job_history(job_id))
+
    @override
    def namespace_client(self) -> LanceNamespace:
        """Get the equivalent namespace client for this connection.
@@ -1838,6 +1976,43 @@ class AsyncConnection(object):
            namespace_path = []
        await self._inner.drop_all_tables(namespace_path=namespace_path)

+    def job(self, job_id: str) -> AsyncJob:
+        """An [AsyncJob][lancedb.job.AsyncJob] handle for a server-side job
+        by id.
+
+        The handle is constructed without a server round trip; an unknown id
+        surfaces when the handle is used. Dropping the handle has no effect
+        on the job itself.
+        """
+        return AsyncJob(self._inner.job(job_id))
+
+    async def list_jobs(self) -> List[JobInfo]:
+        """List server-side jobs across the database's tables."""
+        return await self._inner.list_jobs()
+
+    async def get_job(self, job_id: str) -> Optional[JobDescription]:
+        """Describe a single server-side job by id.
+
+        Returns None when the server has no such job.
+        """
+        return await self._inner.get_job(job_id)
+
+    async def cancel_job(self, job_id: str) -> bool:
+        """Request cancellation of a server-side job by id.
+
+        Returns True if the server accepted the cancellation, False if no
+        such job exists. Cancelling an already-terminal job is a no-op
+        success.
+        """
+        return await self._inner.cancel_job(job_id)
+
+    async def job_history(self, job_id: Optional[str] = None) -> List[pa.RecordBatch]:
+        """The lifecycle event history of a server-side job, as Arrow batches.
+
+        Lists history across all jobs when `job_id` is None.
+        """
+        return await self._inner.job_history(job_id)
+
    async def namespace_client(self) -> LanceNamespace:
        """Get the equivalent namespace client for this connection.

@@ -101,8 +101,7 @@ class InstructorEmbeddingFunction(TextEmbeddingFunction):

    @weak_lru(maxsize=1)
    def ndims(self):
-        model = self.get_model()
-        return model.encode("foo").shape[0]
+        return len(self.generate_embeddings([[self.source_instruction, "foo"]])[0])

    def compute_query_embeddings(self, query: str, *args, **kwargs) -> List[np.array]:
        return self.generate_embeddings([[self.query_instruction, query]])
@@ -23,3 +23,15 @@ class MissingColumnError(KeyError):
        return (
            f"Error: Column '{self.column_name}' does not exist in the DataFrame object"
        )
+
+
+class JobFailedError(RuntimeError):
+    """Exception raised when an asynchronous job reaches the failed state."""
+
+    pass
+
+
+class JobCancelledError(RuntimeError):
+    """Exception raised when an asynchronous job was cancelled."""
+
+    pass
@@ -248,6 +248,10 @@ class Expr:

    # ── utilities ────────────────────────────────────────────────────────────

+    def _column_name(self) -> str | None:
+        """Return the source name when this is a bare column expression."""
+        return self._inner.column_name()
+
    def to_sql(self) -> str:
        """Render the expression as a SQL string (useful for debugging)."""
        return self._inner.to_sql()
@@ -311,7 +315,7 @@ def func(name: str, *args: ExprLike) -> Expr:
    --------
    >>> from lancedb.expr import col, func
    >>> func("lower", col("name"))
-    Expr(lower(name))
+    Expr(lower(`name`))
    """
    inner_args = [_coerce(a)._inner for a in args]
    return Expr(expr_func(name, inner_args))
@@ -0,0 +1,105 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+"""Handles to operations a server may run asynchronously."""
+
+import asyncio
+from datetime import timedelta
+from typing import Optional
+
+from lancedb.background_loop import LOOP
+
+from . import _lancedb
+
+
+class AsyncJob:
+    """A handle to an operation that may still be running.
+
+    The operation may already be complete when the handle is created.
+    """
+
+    def __init__(self, inner: Optional["_lancedb.Job"]):
+        self._inner = inner
+
+    @property
+    def id(self) -> Optional[str]:
+        """Identifies the operation on the server that is running it.
+
+        Returned for correlating with server logs or the jobs API. Operations
+        that run in this process have no server id and return `None`. The value
+        is opaque: parsing it or storing it to resume the job later is not
+        supported.
+        """
+        return self._inner.id if self._inner is not None else None
+
+    async def status(self) -> str:
+        """The operation's current lifecycle state: "running", "finished",
+        "failed", or "cancelled".
+
+        A point snapshot; unlike `wait` it does not block or raise on a
+        terminal failure state. States a newer server reports that this
+        client version does not know pass through as-is.
+        """
+        if self._inner is None:
+            return "finished"
+        return await self._inner.status()
+
+    async def wait(self, timeout: Optional[timedelta] = None):
+        """Wait until the operation reaches a terminal state.
+
+        Raises `JobFailedError` if the operation failed, `JobCancelledError`
+        if it was cancelled, and `TimeoutError` if `timeout` elapses first.
+        """
+        if self._inner is None:
+            return
+        if timeout is None:
+            await self._inner.wait()
+        else:
+            await asyncio.wait_for(self._inner.wait(), timeout.total_seconds())
+
+    async def cancel(self):
+        """Request cancellation. Cancelling a finished operation is a no-op."""
+        if self._inner is None:
+            return
+        await self._inner.cancel()
+
+
+class Job:
+    """Synchronous counterpart of `AsyncJob`."""
+
+    def __init__(self, inner: Optional[AsyncJob]):
+        self._inner = inner
+
+    @property
+    def id(self) -> Optional[str]:
+        """Identifies the operation on the server that is running it.
+
+        See :attr:`AsyncJob.id`.
+        """
+        return self._inner.id if self._inner is not None else None
+
+    def status(self) -> str:
+        """The operation's current lifecycle state: "running", "finished",
+        "failed", or "cancelled".
+
+        See :meth:`AsyncJob.status`.
+        """
+        if self._inner is None:
+            return "finished"
+        return LOOP.run(self._inner.status())
+
+    def wait(self, timeout: Optional[timedelta] = None):
+        """Block until the operation reaches a terminal state.
+
+        Raises `JobFailedError` if the operation failed, `JobCancelledError`
+        if it was cancelled, and `TimeoutError` if `timeout` elapses first.
+        """
+        if self._inner is None:
+            return
+        LOOP.run(self._inner.wait(timeout))
+
+    def cancel(self):
+        """Request cancellation. Cancelling a finished operation is a no-op."""
+        if self._inner is None:
+            return
+        LOOP.run(self._inner.cancel())
@@ -92,8 +92,10 @@ class LanceMergeInsertBuilder(object):
        self._when_not_matched_by_source_delete = True
        if isinstance(condition, Expr):
            self._when_not_matched_by_source_condition_expr = condition._inner
-        elif condition is not None:
+            self._when_not_matched_by_source_condition = None
+        else:
            self._when_not_matched_by_source_condition = condition
+            self._when_not_matched_by_source_condition_expr = None
        return self

    def use_index(self, use_index: bool) -> LanceMergeInsertBuilder:
@@ -38,7 +38,11 @@ from lance_namespace_urllib3_client.models.query_table_request_vector import (
    QueryTableRequestVector,
 )
 from lance_namespace_urllib3_client.models.string_fts_query import StringFtsQuery
-from lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError
+from lance_namespace.errors import (
+    NamespaceNotEmptyError,
+    NamespaceNotFoundError,
+    TableNotFoundError,
+)
 from lancedb._lancedb import (
    connect_namespace as _connect_namespace,
    connect_namespace_client as _connect_namespace_client,
@@ -53,6 +57,8 @@ from lance_namespace import (
    DropNamespaceResponse,
    ListNamespacesResponse,
    ListTablesResponse,
+    NamespaceExistsRequest,
+    TableExistsRequest,
 )
 from lancedb.table import AsyncTable, LanceTable, Table
 from lancedb.util import validate_table_name
@@ -780,6 +786,51 @@ class LanceNamespaceDBConnection(DBConnection):
        """
        return LOOP.run(self._inner.describe_namespace(namespace_path))

+    @override
+    def namespace_exists(self, namespace_id: List[str]) -> bool:
+        """
+        Check if a namespace exists.
+
+        Parameters
+        ----------
+        namespace_id : List[str]
+            The namespace identifier to check.
+
+        Returns
+        -------
+        bool
+            True if the namespace exists, False otherwise.
+        """
+        request = NamespaceExistsRequest(id=namespace_id)
+        try:
+            self._namespace_client.namespace_exists(request)
+            return True
+        except NamespaceNotFoundError:
+            return False
+
+    @override
+    def table_exists(self, table_id: List[str]) -> bool:
+        """
+        Check if a table exists.
+
+        Parameters
+        ----------
+        table_id : List[str]
+            The table identifier to check (full path including namespace
+            segments and table name).
+
+        Returns
+        -------
+        bool
+            True if the table exists, False otherwise.
+        """
+        request = TableExistsRequest(id=table_id)
+        try:
+            self._namespace_client.table_exists(request)
+            return True
+        except TableNotFoundError:
+            return False
+
    @override
    def list_tables(
        self,
@@ -1233,6 +1284,49 @@ class AsyncLanceNamespaceDBConnection:
        """
        return await self._inner.describe_namespace(namespace_path)

+    async def namespace_exists(self, namespace_id: List[str]) -> bool:
+        """
+        Check if a namespace exists.
+
+        Parameters
+        ----------
+        namespace_id : List[str]
+            The namespace identifier to check.
+
+        Returns
+        -------
+        bool
+            True if the namespace exists, False otherwise.
+        """
+        request = NamespaceExistsRequest(id=namespace_id)
+        try:
+            self._namespace_client.namespace_exists(request)
+            return True
+        except NamespaceNotFoundError:
+            return False
+
+    async def table_exists(self, table_id: List[str]) -> bool:
+        """
+        Check if a table exists.
+
+        Parameters
+        ----------
+        table_id : List[str]
+            The table identifier to check (full path including namespace
+            segments and table name).
+
+        Returns
+        -------
+        bool
+            True if the table exists, False otherwise.
+        """
+        request = TableExistsRequest(id=table_id)
+        try:
+            self._namespace_client.table_exists(request)
+            return True
+        except TableNotFoundError:
+            return False
+
    async def list_tables(
        self,
        namespace_path: Optional[List[str]] = None,
@@ -226,7 +226,7 @@ class PermutationBuilder:

        async def do_execute():
            inner_tbl = await self._async.execute()
-            return LanceTable.from_inner(inner_tbl)
+            return await LanceTable.from_inner(inner_tbl)

        return LOOP.run(do_execute())

@@ -52,7 +52,6 @@ from ._blob import (
    finalize_blob_query_table,
    replace_v2_blob_columns_with_bytes,
    replace_v2_blob_columns_with_bytes_sync,
-    supports_blob_auto_row_id,
    validate_blob_mode,
 )
 from .types import BlobMode, QueryProjection
@@ -170,6 +169,12 @@ def _projection_to_scanner_kwargs(columns: QueryProjection) -> Dict[str, Any]:
    return {"columns": projection}


+def _query_request_projection(req: "PyQueryRequest") -> QueryProjection:
+    if req.select_source_columns is not None:
+        return req.select_source_columns
+    return req.select
+
+
 def _scanner_kwargs_for_query(
    query: Query,
    blob_mode: BlobMode,
@@ -1280,10 +1285,7 @@ class LanceQueryBuilder(ABC):
        return self._with_row_id is True

    def _blob_auto_row_id_enabled(self) -> bool:
-        if not supports_blob_auto_row_id(self._table):
-            return False
        return blob_auto_row_id_for_scan(
-            self._table,
            self._table.schema,
            self._columns,
            with_row_id=self._with_row_id,
@@ -2701,7 +2703,7 @@ class LanceHybridQueryBuilder(LanceQueryBuilder):
            self._fts_query.phrase_query(True)
        if self._distance_type:
            self._vector_query.metric(self._distance_type)
-        if self._minimum_nprobes:
+        if self._minimum_nprobes is not None:
            self._vector_query.minimum_nprobes(self._minimum_nprobes)
        if self._maximum_nprobes is not None:
            self._vector_query.maximum_nprobes(self._maximum_nprobes)
@@ -2774,23 +2776,23 @@ class AsyncQueryBase(object):
        )

    async def _maybe_add_blob_row_id(self) -> None:
-        if self._table is None or not supports_blob_auto_row_id(self._table):
+        if self._table is None:
            self._blob_auto_row_id = False
            self._blob_paths = ()
            return

        req = self._inner.to_query_request()
        schema = await self._table.schema()
+        projection = _query_request_projection(req)
        self._blob_auto_row_id = blob_auto_row_id_for_scan(
-            self._table,
            schema,
-            req.select,
+            projection,
            with_row_id=self._with_row_id,
        )
        if not self._blob_auto_row_id:
            self._blob_paths = ()
            return
-        self._blob_paths = tuple(blob_v2_projection_sources(schema, req.select).keys())
+        self._blob_paths = tuple(blob_v2_projection_sources(schema, projection).keys())
        self._inner.with_row_id()

    def select(self, columns: Union[List[str], dict[str, str]]) -> Self:
@@ -3034,7 +3036,6 @@ class AsyncQueryBase(object):

        schema = await self._table.schema()
        blob_auto_row_id = blob_auto_row_id_for_scan(
-            self._table,
            schema,
            query.columns,
            with_row_id=self._with_row_id,
@@ -3880,17 +3881,17 @@ class AsyncHybridQuery(AsyncStandardQuery, AsyncVectorQueryBase):
        req = fts_query._inner.to_query_request()
        blob_auto_row_id = False
        blob_paths: tuple[str, ...] = ()
-        if self._table is not None and supports_blob_auto_row_id(self._table):
+        if self._table is not None:
            schema = await self._table.schema()
+            projection = _query_request_projection(req)
            blob_auto_row_id = blob_auto_row_id_for_scan(
-                self._table,
                schema,
-                req.select,
+                projection,
                with_row_id=self._with_row_id,
            )
            if blob_auto_row_id:
                blob_paths = tuple(
-                    blob_v2_projection_sources(schema, req.select).keys()
+                    blob_v2_projection_sources(schema, projection).keys()
                )
        self._blob_auto_row_id = blob_auto_row_id
        self._blob_paths = blob_paths
@@ -7,7 +7,7 @@ import json
 import logging
 from concurrent.futures import ThreadPoolExecutor
 import sys
-from typing import Any, Dict, Iterable, List, Optional, Union
+from typing import TYPE_CHECKING, Any, Dict, Iterable, List, Optional, Union
 from urllib.parse import urlparse
 import warnings

@@ -23,6 +23,10 @@ import pyarrow as pa

 from ..common import DATA
 from ..db import DBConnection, LOOP
+from ..job import Job
+
+if TYPE_CHECKING:
+    from .._lancedb import JobDescription, JobInfo
 from ..embeddings import EmbeddingFunctionConfig
 from lance_namespace import (
    LanceNamespace,
@@ -415,6 +419,11 @@ class RemoteDBConnection(DBConnection):

        if namespace_path is None:
            namespace_path = []
+        if storage_options is not None:
+            logging.info(
+                "storage_options is ignored in LanceDb Cloud"
+                " (storage is managed; set storage_options on connect() instead)"
+            )
        if index_cache_size is not None:
            logging.info(
                "index_cache_size is ignored in LanceDb Cloud"
@@ -684,6 +693,47 @@ class RemoteDBConnection(DBConnection):
            )
        )

+    @override
+    def job(self, job_id: str) -> Job:
+        """A [Job][lancedb.job.Job] handle for a server-side job by id.
+
+        The handle is constructed without a server round trip; an unknown id
+        surfaces when the handle is used. Dropping the handle has no effect
+        on the job itself.
+        """
+        return Job(self._conn.job(job_id))
+
+    @override
+    def list_jobs(self) -> List["JobInfo"]:
+        """List server-side jobs across the database's tables."""
+        return LOOP.run(self._conn.list_jobs())
+
+    @override
+    def get_job(self, job_id: str) -> Optional["JobDescription"]:
+        """Describe a single server-side job by id.
+
+        Returns None when the server has no such job.
+        """
+        return LOOP.run(self._conn.get_job(job_id))
+
+    @override
+    def cancel_job(self, job_id: str) -> bool:
+        """Request cancellation of a server-side job by id.
+
+        Returns True if the server accepted the cancellation, False if no
+        such job exists. Cancelling an already-terminal job is a no-op
+        success.
+        """
+        return LOOP.run(self._conn.cancel_job(job_id))
+
+    @override
+    def job_history(self, job_id: Optional[str] = None) -> List[pa.RecordBatch]:
+        """The lifecycle event history of a server-side job, as Arrow batches.
+
+        Lists history across all jobs when `job_id` is None.
+        """
+        return LOOP.run(self._conn.job_history(job_id))
+
    @override
    def namespace_client(self) -> LanceNamespace:
        """Get the equivalent namespace client for this connection.
@@ -20,6 +20,7 @@ from typing import (
 import warnings

 from lancedb import __version__
+from lancedb._blob import BlobFile

 from lancedb._lancedb import (
    AddColumnsResult,
@@ -35,6 +36,7 @@ from lancedb._lancedb import (
    UpdateResult,
 )
 from lancedb.embeddings.base import EmbeddingFunctionConfig
+from lancedb.expr import Expr
 from lancedb.index import (
    FTS,
    BTree,
@@ -47,6 +49,7 @@ from lancedb.index import (
    IvfSq,
    LabelList,
 )
+from lancedb.job import Job
 from lancedb.remote.db import LOOP
 from lancedb.table import IndexConfigType, KNOWN_METRICS
 import pyarrow as pa
@@ -540,6 +543,34 @@ class RemoteTable(Table):
            )
        )

+    def create_index_async(
+        self,
+        column: str,
+        *,
+        config: IndexConfigType,
+        replace: Optional[bool] = None,
+        wait_timeout: Optional[timedelta] = None,
+        name: Optional[str] = None,
+        train: bool = True,
+    ) -> Job:
+        """Create an index, returning a handle to the indexing job.
+
+        The job may already be complete when returned; callers must not assume
+        the index exists until :meth:`Job.wait` returns.
+        """
+        return Job(
+            LOOP.run(
+                self._table.create_index_async(
+                    column,
+                    replace=replace,
+                    config=config,
+                    wait_timeout=wait_timeout,
+                    name=name,
+                    train=train,
+                )
+            )
+        )
+
    def _is_legacy_create_index_call(
        self,
        first_arg: str,
@@ -825,7 +856,7 @@ class RemoteTable(Table):

    def update(
        self,
-        where: Optional[str] = None,
+        where: Optional[Union[str, Expr]] = None,
        values: Optional[dict] = None,
        *,
        values_sql: Optional[Dict[str, str]] = None,
@@ -836,9 +867,11 @@ class RemoteTable(Table):

        Parameters
        ----------
-        where: str, optional
-            The SQL where clause to use when updating rows. For example, 'x = 2'
-            or 'x IN (1, 2, 3)'. The filter must not be empty, or it will error.
+        where: str or [Expr][lancedb.expr.Expr], optional
+            The filter condition. Can be a SQL string or a type-safe
+            [Expr][lancedb.expr.Expr] built with [col][lancedb.expr.col] and
+            [lit][lancedb.expr.lit]. The filter must not be empty, or it will
+            error.
        values: dict, optional
            The values to update. The keys are the column names and the values
            are the values to set.
@@ -1039,22 +1072,22 @@ class RemoteTable(Table):
        )

    def blob_columns(self) -> list[str]:
-        raise NotImplementedError(
-            "blob_columns() is not yet supported on the LanceDB Cloud"
-        )
+        return LOOP.run(self._table.blob_columns())

-    def fetch_blobs(self, column: str, row_ids) -> pa.LargeBinaryArray:
-        raise NotImplementedError("fetch_blobs() is not supported on LanceDB Cloud")
+    def fetch_blobs(
+        self, column: str, row_ids: Union[list[int], pa.Table]
+    ) -> pa.LargeBinaryArray:
+        return LOOP.run(self._table.fetch_blobs(column, row_ids))

    def fetch_blob_ranges(self, column: str, requests) -> pa.LargeBinaryArray:
        raise NotImplementedError(
            "fetch_blob_ranges() is not supported on LanceDB Cloud"
        )

-    def fetch_blob_files(self, column: str, row_ids):
-        raise NotImplementedError(
-            "fetch_blob_files() is not supported on LanceDB Cloud"
-        )
+    def fetch_blob_files(
+        self, column: str, row_ids: Union[list[int], pa.Table]
+    ) -> "list[Optional[BlobFile]]":
+        return LOOP.run(self._table.fetch_blob_files(column, row_ids))

    def head(self, n=5) -> pa.Table:
        """
@@ -40,6 +40,7 @@ from ._blob import (
 from .types import BlobMode
 from lancedb.arrow import peek_reader
 from lancedb.background_loop import LOOP, embedding_executor
+from lancedb.job import AsyncJob, Job
 from .dependencies import (
    _check_for_hugging_face,
    _check_for_lance,
@@ -977,6 +978,24 @@ class Table(ABC):
        """
        raise NotImplementedError

+    def create_index_async(
+        self,
+        column: str,
+        *,
+        config: IndexConfigType,
+        replace: Optional[bool] = None,
+        wait_timeout: Optional[timedelta] = None,
+        name: Optional[str] = None,
+        train: bool = True,
+    ) -> Job:
+        """Create an index, returning a handle to the indexing job.
+
+        Takes the same arguments as :meth:`create_index`. The job may already
+        be complete when returned; callers must not assume the index exists
+        until :meth:`Job.wait` returns.
+        """
+        raise NotImplementedError
+
    def drop_index(self, name: str) -> None:
        """
        Drop an index from the table.
@@ -1574,8 +1593,10 @@ class Table(ABC):
        """Open lazy, seekable :class:`~lancedb._blob.BlobFile` handles.

        Prefer this over :meth:`fetch_blobs` for large payloads. ``row_ids`` is
-        a ``list[int]`` or query ``pyarrow.Table`` with ``_rowid`` (or stashed
-        row-id metadata). Null rows are ``None``. Local tables only.
+        a ``list[int]`` or a query ``pyarrow.Table`` carrying row identity via
+        ``_rowid`` or a ``_lance_row_id`` field on the blob descriptor. Null
+        rows are ``None``. Remote tables require LanceDB Cloud server 0.5.0 or
+        newer.
        """

    @abstractmethod
@@ -1671,7 +1692,7 @@ class Table(ABC):
    @abstractmethod
    def update(
        self,
-        where: Optional[str] = None,
+        where: Optional[Union[str, Expr]] = None,
        values: Optional[dict] = None,
        *,
        values_sql: Optional[Dict[str, str]] = None,
@@ -1686,9 +1707,11 @@ class Table(ABC):

        Parameters
        ----------
-        where: str, optional
-            The SQL where clause to use when updating rows. For example, 'x = 2'
-            or 'x IN (1, 2, 3)'. The filter must not be empty, or it will error.
+        where: str or [Expr][lancedb.expr.Expr], optional
+            The filter condition. Can be a SQL string or a type-safe
+            [Expr][lancedb.expr.Expr] built with [col][lancedb.expr.col] and
+            [lit][lancedb.expr.lit]. The filter must not be empty, or it will
+            error.
        values: dict, optional
            The values to update. The keys are the column names and the values
            are the values to set.
@@ -1706,6 +1729,7 @@ class Table(ABC):
        Examples
        --------
        >>> import lancedb
+        >>> from lancedb.expr import col
        >>> import pandas as pd
        >>> data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1.0, 2], [3, 4], [5, 6]]})
        >>> db = lancedb.connect("./.lancedb")
@@ -1715,7 +1739,7 @@ class Table(ABC):
        0  1  [1.0, 2.0]
        1  2  [3.0, 4.0]
        2  3  [5.0, 6.0]
-        >>> table.update(where="x = 2", values={"vector": [10.0, 10]})
+        >>> table.update(where=col("x") == 2, values={"vector": [10.0, 10]})
        UpdateResult(rows_updated=1, version=2)
        >>> table.to_pandas()
           x        vector
@@ -2161,11 +2185,15 @@ class LanceTable(Table):
        return self.name

    @classmethod
-    def from_inner(cls, tbl: LanceDBTable):
-        from .db import LanceDBConnection
+    async def from_inner(cls, tbl: LanceDBTable):
+        from .db import AsyncConnection, LanceDBConnection

        async_tbl = AsyncTable(tbl)
-        conn = LanceDBConnection.from_inner(tbl.database())
+        inner_conn = tbl.database()
+        read_consistency_interval = await AsyncConnection(
+            inner_conn
+        ).get_read_consistency_interval()
+        conn = LanceDBConnection.from_inner(inner_conn, read_consistency_interval)
        return cls(
            conn,
            async_tbl.name,
@@ -2469,13 +2497,7 @@ class LanceTable(Table):
        return LOOP.run(self._table.count_rows(filter))

    def __repr__(self) -> str:
-        val = f"{self.__class__.__name__}(name={self.name!r}"
-        if self._conn.read_consistency_interval is not None:
-            val += ", read_consistency_interval={!r}".format(
-                self._conn.read_consistency_interval
-            )
-        val += f", _conn={self._conn!r})"
-        return val
+        return f"{self.__class__.__name__}(name={self.name!r}, _conn={self._conn!r})"

    def __str__(self) -> str:
        return self.__repr__()
@@ -2784,6 +2806,34 @@ class LanceTable(Table):
            )
        )

+    def create_index_async(
+        self,
+        column: str,
+        *,
+        config: IndexConfigType,
+        replace: Optional[bool] = None,
+        wait_timeout: Optional[timedelta] = None,
+        name: Optional[str] = None,
+        train: bool = True,
+    ) -> Job:
+        """Create an index, returning a handle to the indexing job.
+
+        The job may already be complete when returned; callers must not assume
+        the index exists until :meth:`Job.wait` returns.
+        """
+        return Job(
+            LOOP.run(
+                self._table.create_index_async(
+                    column,
+                    replace=replace,
+                    config=config,
+                    wait_timeout=wait_timeout,
+                    name=name,
+                    train=train,
+                )
+            )
+        )
+
    def _is_legacy_create_index_call(
        self,
        first_arg: str,
@@ -3604,7 +3654,7 @@ class LanceTable(Table):

    def update(
        self,
-        where: Optional[str] = None,
+        where: Optional[Union[str, Expr]] = None,
        values: Optional[dict] = None,
        *,
        values_sql: Optional[Dict[str, str]] = None,
@@ -3615,9 +3665,11 @@ class LanceTable(Table):

        Parameters
        ----------
-        where: str, optional
-            The SQL where clause to use when updating rows. For example, 'x = 2'
-            or 'x IN (1, 2, 3)'. The filter must not be empty, or it will error.
+        where: str or [Expr][lancedb.expr.Expr], optional
+            The filter condition. Can be a SQL string or a type-safe
+            [Expr][lancedb.expr.Expr] built with [col][lancedb.expr.col] and
+            [lit][lancedb.expr.lit]. The filter must not be empty, or it will
+            error.
        values: dict, optional
            The values to update. The keys are the column names and the values
            are the values to set.
@@ -3635,6 +3687,7 @@ class LanceTable(Table):
        Examples
        --------
        >>> import lancedb
+        >>> from lancedb.expr import col
        >>> import pandas as pd
        >>> data = pd.DataFrame({"x": [1, 2, 3], "vector": [[1.0, 2], [3, 4], [5, 6]]})
        >>> db = lancedb.connect("./.lancedb")
@@ -3644,7 +3697,7 @@ class LanceTable(Table):
        0  1  [1.0, 2.0]
        1  2  [3.0, 4.0]
        2  3  [5.0, 6.0]
-        >>> table.update(where="x = 2", values={"vector": [10.0, 10]})
+        >>> table.update(where=col("x") == 2, values={"vector": [10.0, 10]})
        UpdateResult(rows_updated=1, version=2)
        >>> table.to_pandas()
           x        vector
@@ -4871,6 +4924,46 @@ class AsyncTable:
                )
            raise e

+    async def create_index_async(
+        self,
+        column: str,
+        *,
+        replace: Optional[bool] = None,
+        config: Optional[
+            Union[
+                IvfFlat,
+                IvfPq,
+                IvfRq,
+                HnswPq,
+                HnswSq,
+                HnswFlat,
+                BTree,
+                Bitmap,
+                LabelList,
+                Fm,
+                FTS,
+            ]
+        ] = None,
+        wait_timeout: Optional[timedelta] = None,
+        name: Optional[str] = None,
+        train: bool = True,
+    ) -> AsyncJob:
+        """Create an index, returning a handle to the indexing job.
+
+        Takes the same arguments as :meth:`create_index`. The job may already
+        be complete when returned; callers must not assume the index exists
+        until :meth:`AsyncJob.wait` resolves.
+        """
+        job = await self._inner.create_index_async(
+            column,
+            index=config,
+            replace=replace,
+            wait_timeout=wait_timeout,
+            name=name,
+            train=train,
+        )
+        return AsyncJob(job)
+
    async def drop_index(self, name: str) -> None:
        """
        Drop an index from the table.
@@ -5603,7 +5696,7 @@ class AsyncTable:
        self,
        updates: Optional[Dict[str, Any]] = None,
        *,
-        where: Optional[str] = None,
+        where: Optional[Union[str, Expr]] = None,
        updates_sql: Optional[Dict[str, str]] = None,
    ) -> UpdateResult:
        """
@@ -5618,9 +5711,11 @@ class AsyncTable:
            The updates to apply.  The keys should be the name of the column to
            update.  The values should be the new values to assign.  This is
            required unless updates_sql is supplied.
-        where: str, optional
-            An SQL filter that controls which rows are updated. For example, 'x = 2'
-            or 'x IN (1, 2, 3)'.  Only rows that satisfy this filter will be udpated.
+        where: str or [Expr][lancedb.expr.Expr], optional
+            The filter condition. Can be a SQL string or a type-safe
+            [Expr][lancedb.expr.Expr] built with [col][lancedb.expr.col] and
+            [lit][lancedb.expr.lit]. Only rows that satisfy this filter will
+            be updated.
        updates_sql: dict, optional
            The updates to apply, expressed as SQL expression strings.  The keys should
            be column names. The values should be SQL expressions.  These can be SQL
@@ -5638,13 +5733,14 @@ class AsyncTable:
        --------
        >>> import asyncio
        >>> import lancedb
+        >>> from lancedb.expr import col
        >>> import pandas as pd
        >>> async def demo_update():
        ...     data = pd.DataFrame({"x": [1, 2], "vector": [[1, 2], [3, 4]]})
        ...     db = await lancedb.connect_async("./.lancedb")
        ...     table = await db.create_table("my_table", data)
        ...     # x is [1, 2], vector is [[1, 2], [3, 4]]
-        ...     await table.update({"vector": [10, 10]}, where="x = 2")
+        ...     await table.update({"vector": [10, 10]}, where=col("x") == 2)
        ...     # x is [1, 2], vector is [[1, 2], [10, 10]]
        ...     await table.update(updates_sql={"x": "x + 1"})
        ...     # x is [2, 3], vector is [[1, 2], [10, 10]]
@@ -5658,7 +5754,8 @@ class AsyncTable:
        if updates is not None:
            updates_sql = {k: value_to_sql(v) for k, v in updates.items()}

-        return await self._inner.update(updates_sql, where)
+        predicate = where.to_sql() if isinstance(where, Expr) else where
+        return await self._inner.update(updates_sql, predicate)

    async def add_columns(
        self, transforms: dict[str, str] | pa.field | List[pa.field] | pa.Schema
@@ -395,6 +395,11 @@ def _(value: dict):
    )


+@value_to_sql.register(pa.Scalar)
+def _(value: pa.Scalar):
+    return value_to_sql(value.as_py())
+
+
@value_to_sql.register(np.ndarray)
 def _(value: np.ndarray):
    return value_to_sql(value.tolist())
@@ -8,7 +8,12 @@ import pyarrow.compute as pc
 import pytest

 import lancedb
-from lancedb._blob import read_row_ids_from_hits, stash_auto_row_ids
+from lancedb._blob import (
+    blob_v2_projection_sources,
+    read_row_ids_from_hits,
+    stash_auto_row_ids,
+)
+from lancedb.expr import col
 from lancedb.index import FTS
 from lancedb.schema import blob_column_paths, blob_v2_column_paths

@@ -70,6 +75,14 @@ def test_blob_v2_column_paths_include_list_children():
    ]


+def test_blob_v2_projection_sources_use_typed_column_name():
+    schema = pa.schema([lancedb.blob("blob")])
+
+    assert blob_v2_projection_sources(schema, {"blob_alias": col("blob")}) == {
+        "blob_alias": "blob"
+    }
+
+
 def _legacy_v1_table(name):
    db = lancedb.connect("memory:///")
    schema = pa.schema(
@@ -166,6 +179,20 @@ async def test_async_table_to_pandas_descriptions_mode_omits_row_id():
    assert set(descriptor.keys()) == {"kind", "position", "size", "blob_id", "blob_uri"}


+@pytest.mark.asyncio
+async def test_async_typed_blob_projection_preserves_source_column():
+    db = await lancedb.connect_async("memory:///typed_blob_projection")
+    schema = pa.schema([pa.field("id", pa.int64()), lancedb.blob("blob")])
+    table = await db.create_table("typed_blob_projection", schema=schema)
+    await table.add([{"id": 1, "blob": b"alpha"}])
+
+    hits = await table.query().select({"blob_alias": col("blob")}).to_arrow()
+
+    assert "_lance_row_id" in hits.schema.field("blob_alias").type.names
+    blobs = await table.fetch_blobs("blob", hits)
+    assert blobs.to_pylist() == [b"alpha"]
+
+
 def test_fetch_blobs_round_trip():
    table = _blob_table(
        "round_trip",
@@ -226,13 +253,13 @@ def test_fetch_blob_ranges_validates_requests():
    table = _blob_table("range_validation", [{"id": 1, "image": b"abc"}])
    row_id = _row_ids_by_id(table)[1]

-    with pytest.raises(RuntimeError, match="exceeds blob size"):
+    with pytest.raises(ValueError, match="exceeds blob size"):
        table.fetch_blob_ranges("image", [(row_id, 2, 2)])

-    with pytest.raises(RuntimeError, match="offset \\+ length overflowed"):
+    with pytest.raises(ValueError, match="offset \\+ length overflowed"):
        table.fetch_blob_ranges("image", [(row_id, 2**64 - 1, 1)])

-    with pytest.raises(ValueError, match="row ids"):
+    with pytest.raises(ValueError, match="row IDs"):
        table.fetch_blob_ranges("image", [(2**64 - 1, 0, 1)])


@@ -403,6 +430,50 @@ async def test_blob_v2_hybrid_fetch_blobs_async():
    assert {blobs[i].as_py() for i in range(len(blobs))} == {b"alpha", b"beta"}


+@pytest.mark.asyncio
+async def test_async_hybrid_typed_blob_projection_preserves_source_column():
+    db = await lancedb.connect_async("memory:///hybrid_typed_blob")
+    schema = pa.schema(
+        [
+            pa.field("id", pa.int64()),
+            pa.field("text", pa.utf8()),
+            pa.field("vector", pa.list_(pa.float32(), list_size=2)),
+            lancedb.blob("blob"),
+        ]
+    )
+    table = await db.create_table("hybrid_typed_blob", schema=schema)
+    await table.add(
+        [
+            {
+                "id": 1,
+                "text": "hello alpha",
+                "vector": [1.0, 0.0],
+                "blob": b"alpha",
+            },
+            {
+                "id": 2,
+                "text": "hello beta",
+                "vector": [0.9, 0.1],
+                "blob": b"beta",
+            },
+        ]
+    )
+    await table.create_index("text", config=FTS(with_position=False))
+
+    hits = await (
+        table.query()
+        .nearest_to([1.0, 0.0])
+        .nearest_to_text("hello")
+        .select({"blob_alias": col("blob")})
+        .limit(2)
+        .to_arrow()
+    )
+
+    assert "_lance_row_id" in hits.schema.field("blob_alias").type.names
+    blobs = await table.fetch_blobs("blob", hits)
+    assert {blobs[i].as_py() for i in range(len(blobs))} == {b"alpha", b"beta"}
+
+
 def test_blob_file_seek_read_and_read_range():
    payload = _identifiable_payload(1024)
    table = _blob_table("seek_read", [{"id": 1, "image": payload}])
@@ -2,6 +2,7 @@
 # SPDX-FileCopyrightText: Copyright The LanceDB Authors


+import inspect
 import re
 import sys
 from datetime import timedelta
@@ -62,6 +63,44 @@ def test_basic(tmp_path):
    assert db.open_table("test").name == db["test"].name


+def test_sync_debugger_inspection_does_not_use_background_loop(tmp_path, monkeypatch):
+    from lancedb.background_loop import LOOP
+
+    db = lancedb.connect(tmp_path)
+    table = db.create_table("test", data=[{"id": 1}])
+
+    def fail_run(*args, **kwargs):
+        raise AssertionError("debugger inspection should not use the background loop")
+
+    monkeypatch.setattr(LOOP, "run", fail_run)
+
+    # Debuggers enumerate and evaluate every exposed attribute when expanding a
+    # variable. This must remain safe while their breakpoint suspends LOOP's thread.
+    members = dict(inspect.getmembers(db))
+
+    assert members["uri"] == str(tmp_path)
+    assert members["read_consistency_interval"] is None
+    assert repr(db) == f"LanceDBConnection(uri={str(tmp_path)!r})"
+    assert repr(table) == f"LanceTable(name='test', _conn={db!r})"
+
+
+def test_read_consistency_interval_does_not_use_background_loop(tmp_path, monkeypatch):
+    from lancedb.background_loop import LOOP
+    from lancedb.db import LanceDBConnection
+
+    consistency_interval = timedelta(seconds=5)
+    db = lancedb.connect(tmp_path, read_consistency_interval=consistency_interval)
+    db_from_inner = LanceDBConnection.from_inner(db._inner, consistency_interval)
+
+    def fail_run(*args, **kwargs):
+        raise AssertionError("properties should not use the Python background loop")
+
+    monkeypatch.setattr(LOOP, "run", fail_run)
+
+    assert db.read_consistency_interval == consistency_interval
+    assert db_from_inner.read_consistency_interval == consistency_interval
+
+
 def test_ingest_pd(tmp_path):
    db = lancedb.connect(tmp_path)

@@ -64,6 +64,23 @@ def test_embedding_function(tmp_path):
    assert np.allclose(actual, expected)


+def test_instructor_ndims_uses_instruction():
+    instructor = get_registry().get("instructor").create()
+    model = MagicMock()
+    model.encode.return_value = np.zeros((1, 384))
+
+    with patch.object(type(instructor), "get_model", return_value=model):
+        assert instructor.ndims() == 384
+
+    model.encode.assert_called_once_with(
+        [[instructor.source_instruction, "foo"]],
+        batch_size=instructor.batch_size,
+        show_progress_bar=instructor.show_progress_bar,
+        normalize_embeddings=instructor.normalize_embeddings,
+        device=instructor.device,
+    )
+
+
 def test_embedding_function_variables():
    @register("variable-testing")
    class VariableTestingFunction(TextEmbeddingFunction):
@@ -115,34 +132,16 @@ def test_embedding_function_variables():
    assert func.safe_model_dump()["secret_key"] == "$var:secret"


-def test_parse_functions_with_variables():
-    @register("variable-parsing-test")
-    class VariableParsingFunction(TextEmbeddingFunction):
-        api_key: str
-        base_url: Optional[str] = None
-
-        @staticmethod
-        def sensitive_keys():
-            return ["api_key"]
-
-        def ndims(self):
-            return 10
-
-        def generate_embeddings(self, texts):
-            # Mock implementation that just returns random embeddings
-            # In real usage, this would use the api_key to call an API
-            return [np.random.rand(self.ndims()).tolist() for _ in texts]
-
+def test_openai_variables_survive_metadata_round_trip():
    registry = EmbeddingFunctionRegistry.get_instance()

    registry.set_var("test_api_key", "sk-test-key-12345")
-    registry.set_var("test_base_url", "https://api.example.com")

    conf = EmbeddingFunctionConfig(
        source_column="text",
        vector_column="vector",
-        function=registry.get("variable-parsing-test").create(
-            api_key="$var:test_api_key", base_url="$var:test_base_url"
+        function=registry.get("openai").create(
+            api_key="$var:test_api_key", base_url="https://api.example.com"
        ),
    )

@@ -150,7 +149,10 @@ def test_parse_functions_with_variables():

    # Create a mock arrow table with the metadata
    schema = pa.schema(
-        [pa.field("text", pa.string()), pa.field("vector", pa.list_(pa.float32(), 10))]
+        [
+            pa.field("text", pa.string()),
+            pa.field("vector", pa.list_(pa.float32(), 1536)),
+        ]
    )
    table = pa.table({"text": [], "vector": []}, schema=schema)
    table = table.replace_schema_metadata(metadata)
@@ -164,13 +166,15 @@ def test_parse_functions_with_variables():

    assert parsed_func.api_key == "sk-test-key-12345"
    assert parsed_func.base_url == "https://api.example.com"
-
-    embeddings = parsed_func.generate_embeddings(["test text"])
-    assert len(embeddings) == 1
-    assert len(embeddings[0]) == 10
-
    assert parsed_func.safe_model_dump()["api_key"] == "$var:test_api_key"

+    with patch("lancedb.embeddings.openai.attempt_import_or_raise") as import_openai:
+        parsed_func._openai_client
+
+    import_openai.return_value.OpenAI.assert_called_once_with(
+        api_key="sk-test-key-12345", base_url="https://api.example.com"
+    )
+

 def test_embedding_with_bad_results(tmp_path):
    @register("null-embedding")
@@ -52,7 +52,7 @@ class TestExprConstruction:
    def test_func(self):
        e = func("lower", col("name"))
        assert isinstance(e, Expr)
-        assert e.to_sql() == "lower(name)"
+        assert e.to_sql() == "lower(`name`)"

    def test_func_unknown_raises(self):
        with pytest.raises(Exception):
@@ -115,7 +115,7 @@ class TestExprOperators:
    def test_and_operator(self):
        e = (col("age") > lit(18)) & (col("status") == lit("active"))
        assert isinstance(e, Expr)
-        assert e.to_sql() == "((age > 18) AND (status = 'active'))"
+        assert e.to_sql() == "((age > 18) AND (`status` = 'active'))"

    def test_or_operator(self):
        e = (col("a") == lit(1)) | (col("b") == lit(2))
@@ -166,7 +166,7 @@ class TestExprOperators:
    def test_coerce_plain_str(self):
        e = col("name") == "alice"
        assert isinstance(e, Expr)
-        assert e.to_sql() == "(name = 'alice')"
+        assert e.to_sql() == "(`name` = 'alice')"

    def test_reflexive_comparisons(self):
        # 10 < col("age") swaps to col("age") > 10
@@ -198,85 +198,85 @@ class TestExprBytesLiteral:

    def test_bytes_equality_expr_sql(self):
        e = col("data") == lit(b"\xca\xfe")
-        assert e.to_sql() == "(data = X'CAFE')"
+        assert e.to_sql() == "(`data` = X'CAFE')"

    def test_bytes_ne_expr_sql(self):
        e = col("data") != lit(b"\xff")
-        assert e.to_sql() == "(data <> X'FF')"
+        assert e.to_sql() == "(`data` <> X'FF')"

    def test_bytes_compound_expr_sql(self):
        e = (col("data") == lit(b"\x01")) & (col("id") > lit(5))
-        assert e.to_sql() == "((data = X'01') AND (id > 5))"
+        assert e.to_sql() == "((`data` = X'01') AND (id > 5))"

    def test_bytes_in_function_call(self):
        # Regression test: binary literals inside scalar function calls
        # used to fail because DataFusion's unparser does not support Binary
        # scalars.  Now handled via a placeholder-substitution rewrite.
        e = func("contains", col("data"), lit(b"\xff"))
-        assert e.to_sql() == "contains(data, X'FF')"
+        assert e.to_sql() == "contains(`data`, X'FF')"

    def test_bytes_in_not(self):
        e = ~(col("data") == lit(b"\xff"))
-        assert e.to_sql() == "NOT (data = X'FF')"
+        assert e.to_sql() == "NOT (`data` = X'FF')"


 class TestExprStringMethods:
    def test_lower(self):
        e = col("name").lower()
        assert isinstance(e, Expr)
-        assert e.to_sql() == "lower(name)"
+        assert e.to_sql() == "lower(`name`)"

    def test_upper(self):
        e = col("name").upper()
        assert isinstance(e, Expr)
-        assert e.to_sql() == "upper(name)"
+        assert e.to_sql() == "upper(`name`)"

    def test_contains(self):
        e = col("text").contains(lit("hello"))
        assert isinstance(e, Expr)
-        assert e.to_sql() == "contains(text, 'hello')"
+        assert e.to_sql() == "contains(`text`, 'hello')"

    def test_contains_with_str_coerce(self):
        e = col("text").contains("hello")
        assert isinstance(e, Expr)
-        assert e.to_sql() == "contains(text, 'hello')"
+        assert e.to_sql() == "contains(`text`, 'hello')"

    def test_chained_lower_eq(self):
        e = col("name").lower() == lit("alice")
        assert isinstance(e, Expr)
-        assert e.to_sql() == "(lower(name) = 'alice')"
+        assert e.to_sql() == "(lower(`name`) = 'alice')"


 class TestExprCast:
    def test_cast_string(self):
        e = col("id").cast("string")
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(id AS VARCHAR)"
+        assert e.to_sql() == "arrow_cast(id, 'Utf8')"

    def test_cast_int32(self):
        e = col("score").cast("int32")
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(score AS INTEGER)"
+        assert e.to_sql() == "arrow_cast(score, 'Int32')"

    def test_cast_float64(self):
        e = col("val").cast("float64")
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(val AS DOUBLE)"
+        assert e.to_sql() == "arrow_cast(val, 'Float64')"

    def test_cast_pyarrow_type(self):
        e = col("score").cast(pa.int32())
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(score AS INTEGER)"
+        assert e.to_sql() == "arrow_cast(score, 'Int32')"

    def test_cast_pyarrow_float64(self):
        e = col("val").cast(pa.float64())
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(val AS DOUBLE)"
+        assert e.to_sql() == "arrow_cast(val, 'Float64')"

    def test_cast_pyarrow_string(self):
        e = col("id").cast(pa.string())
        assert isinstance(e, Expr)
-        assert e.to_sql() == "CAST(id AS VARCHAR)"
+        assert e.to_sql() == "arrow_cast(id, 'Utf8')"

    def test_cast_pyarrow_and_string_equivalent(self):
        # pa.int32() and "int32" should produce equivalent SQL
@@ -597,14 +597,14 @@ class TestExprIsin:
    def test_isin_strs(self):
        assert (
            col("status").isin(["active", "pending"]).to_sql()
-            == "status IN ('active', 'pending')"
+            == "`status` IN ('active', 'pending')"
        )

    def test_isin_coerces_and_mixes(self):
        assert col("id").isin([lit(1), 2]).to_sql() == "id IN (1, 2)"

    def test_isin_empty(self):
-        assert col("id").isin([]).to_sql() == "id IN ()"
+        assert col("id").isin([]).to_sql() == "false"

    def test_isin_filter(self, simple_table):
        result = simple_table.search().where(col("id").isin([1, 3, 5])).to_arrow()
@@ -12,7 +12,7 @@ import pyarrow.compute as pc
 import pytest
 import pytest_asyncio

-from lancedb.index import FTS
+from lancedb.index import BTree, FTS, IvfPq
 from lancedb.table import AsyncTable, Table


@@ -99,6 +99,86 @@ async def test_async_hybrid_query_filters(table: AsyncTable):
    assert result["text"].to_pylist() == ["cat", "b"]


+@pytest.mark.asyncio
+async def test_hybrid_query_with_stale_fixed_size_binary_prefilter(
+    tmpdir_factory,
+):
+    tmp_path = str(tmpdir_factory.mktemp("stale_scalar_prefilter"))
+    db = await lancedb.connect_async(tmp_path)
+
+    def fixed_size_binary(value: int) -> bytes:
+        return value.to_bytes(16, byteorder="big")
+
+    num_rows = 1000
+    data = pa.table(
+        {
+            "space_id": pa.array(
+                [fixed_size_binary(i) for i in range(num_rows)],
+                type=pa.binary(16),
+            ),
+            "text": ["book"] * num_rows,
+            "vector": pa.array(
+                [[float(i), float(i)] for i in range(num_rows)],
+                type=pa.list_(pa.float32(), 2),
+            ),
+        }
+    )
+    table = await db.create_table("test", data)
+    await table.create_index(
+        "vector", config=IvfPq(num_partitions=4, num_sub_vectors=2)
+    )
+    await table.create_index("space_id", config=BTree())
+    await table.create_index("text", config=FTS(with_position=False))
+
+    # Advance the search indices without advancing the scalar index. This is the
+    # state that previously let hybrid search use an incomplete scalar prefilter.
+    await table.add(data)
+    lance_dataset = await table.to_lance()
+    lance_dataset.optimize.optimize_indices(index_names=["vector_idx", "text_idx"])
+    await table.checkout_latest()
+
+    scalar_stats = await table.index_stats("space_id_idx")
+    assert scalar_stats is not None
+    assert scalar_stats.num_indexed_rows == num_rows
+    assert scalar_stats.num_unindexed_rows == num_rows
+
+    for index_name in ["vector_idx", "text_idx"]:
+        search_stats = await table.index_stats(index_name)
+        assert search_stats is not None
+        assert search_stats.num_indexed_rows == num_rows * 2
+        assert search_stats.num_unindexed_rows == 0
+
+    matching_ids = [5, 10, 15, 20, 25, 30]
+    literals = [
+        f"arrow_cast(0x{fixed_size_binary(i).hex()}, 'FixedSizeBinary(16)')"
+        for i in matching_ids
+    ]
+    predicate = f"space_id IN ({', '.join(literals)})"
+    expected_ids = sorted(fixed_size_binary(i) for i in matching_ids for _ in range(2))
+
+    vector_query = (
+        table.query().where(predicate).nearest_to([5.0, 5.0]).limit(num_rows * 2)
+    )
+    vector_results = await vector_query.to_arrow()
+    assert sorted(vector_results["space_id"].to_pylist()) == expected_ids
+
+    fts_query = (
+        table.query().where(predicate).nearest_to_text("book").limit(num_rows * 2)
+    )
+    fts_results = await fts_query.to_arrow()
+    assert sorted(fts_results["space_id"].to_pylist()) == expected_ids
+
+    hybrid_results = await (
+        table.query()
+        .where(predicate)
+        .nearest_to([5.0, 5.0])
+        .nearest_to_text("book")
+        .limit(num_rows * 2)
+        .to_arrow()
+    )
+    assert sorted(hybrid_results["space_id"].to_pylist()) == expected_ids
+
+
@pytest.mark.asyncio
 async def test_async_hybrid_query_default_limit(table: AsyncTable):
    # add 10 new rows
@@ -123,6 +203,19 @@ async def test_async_hybrid_query_default_limit(table: AsyncTable):
    assert texts.count("a") == 1


+def test_hybrid_query_minimum_nprobes_zero_raises(sync_table: Table):
+    # minimum_nprobes(0) must raise the same validation error a plain vector
+    # query raises, not silently no-op because 0 is falsy.
+    with pytest.raises(ValueError, match="minimum_nprobes must be greater than 0"):
+        (
+            sync_table.search(query_type="hybrid")
+            .vector([0.0, 0.4])
+            .text("dog")
+            .minimum_nprobes(0)
+            .to_arrow()
+        )
+
+
 def test_hybrid_query_distance_range(sync_table: Table):
    reranker = RRFReranker(return_score="all")
    result = (
@@ -0,0 +1,33 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+import re
+import shutil
+import subprocess
+import sys
+
+import lancedb._lancedb as _lancedb
+import pytest
+
+
+@pytest.mark.skipif(sys.platform != "linux", reason="ldd is Linux-specific")
+def test_native_extension_does_not_link_openssl():
+    """OpenSSL-linked wheels abort when imported on RHEL hosts in FIPS mode."""
+    ldd = shutil.which("ldd")
+    if ldd is None:
+        pytest.skip("ldd is not installed")
+
+    result = subprocess.run(
+        [ldd, _lancedb.__file__],
+        check=True,
+        capture_output=True,
+        text=True,
+    )
+    openssl_libraries = re.findall(
+        r"^\s*(lib(?:crypto|ssl)\S*)\s+=>", result.stdout, flags=re.MULTILINE
+    )
+
+    assert not openssl_libraries, (
+        "the LanceDB native extension must use rustls instead of linking OpenSSL: "
+        f"{openssl_libraries}"
+    )
@@ -84,6 +84,15 @@ async def binary_table(db_async):
    )


+@pytest.mark.asyncio
+async def test_create_index_async_returns_done_job(some_table: AsyncTable):
+    job = await some_table.create_index_async("id", config=BTree())
+    assert job.id is None
+    await job.wait()
+    assert len(await some_table.list_indices()) == 1
+    await job.cancel()
+
+
@pytest.mark.asyncio
 async def test_create_scalar_index(some_table: AsyncTable):
    # Can create
@@ -363,6 +372,31 @@ async def test_create_vector_index(some_table: AsyncTable):
    assert stats.num_indices == 1


+@pytest.mark.asyncio
+async def test_create_ivf_index_reports_unsplittable_partitions(db_async):
+    dim = 8
+    num_partitions = 300  # More than 256 selects hierarchical k-means.
+    base_vectors = [[float(row == column) for column in range(dim)] for row in range(5)]
+    vectors = pa.array(base_vectors * 200, pa.list_(pa.float32(), dim))
+    table = await db_async.create_table(
+        "unsplittable_partitions",
+        pa.table({"vector": vectors}),
+    )
+
+    error_pattern = (
+        rf"Cannot create {num_partitions} IVF partitions: k-means could only form"
+    )
+    with pytest.raises(RuntimeError, match=error_pattern):
+        await table.create_index(
+            "vector",
+            config=IvfFlat(
+                distance_type="dot",
+                num_partitions=num_partitions,
+                max_iterations=10,
+            ),
+        )
+
+
@pytest.mark.asyncio
 async def test_create_4bit_ivfpq_index(some_table: AsyncTable):
    # Can create
@@ -18,6 +18,7 @@ Tests verify:
 """

 import copy
+import os
 import shutil
 import sys
 import tempfile
@@ -239,7 +240,7 @@ def create_tracking_namespace(

    dir_props = {f"storage.{k}": v for k, v in storage_options_with_refresh.items()}

-    if bucket_name.startswith("/") or bucket_name.startswith("file://"):
+    if os.path.isabs(bucket_name) or bucket_name.startswith("file://"):
        dir_props["root"] = f"{bucket_name}/namespace_root"
    else:
        dir_props["root"] = f"s3://{bucket_name}/namespace_root"
@@ -767,3 +768,70 @@ def test_namespace_with_schema_only(s3_bucket: str, use_custom: bool):

    # Verify data was added
    assert table.count_rows() == 2
+
+
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_exists(use_custom: bool):
+    """
+    Test namespace_exists returns True for existing and False for non-existent.
+    """
+    temp_dir = tempfile.mkdtemp()
+    try:
+        ns_client, _ = create_tracking_namespace(
+            bucket_name=temp_dir,
+            storage_options={},
+            credential_expires_in_seconds=3600,
+            use_custom=use_custom,
+        )
+        db = LanceNamespaceDBConnection(ns_client)
+
+        namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
+        db.create_namespace([namespace_name])
+
+        # Existing namespace should return True
+        assert db.namespace_exists(namespace_id=[namespace_name]) is True
+
+        # Non-existent namespace should return False
+        assert db.namespace_exists(namespace_id=["nonexistent_ns"]) is False
+    finally:
+        shutil.rmtree(temp_dir, ignore_errors=True)
+
+
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_table_exists(use_custom: bool):
+    """
+    Test table_exists returns True for existing table and False for non-existent.
+    """
+    temp_dir = tempfile.mkdtemp()
+    try:
+        ns_client, _ = create_tracking_namespace(
+            bucket_name=temp_dir,
+            storage_options={},
+            credential_expires_in_seconds=3600,
+            use_custom=use_custom,
+        )
+        db = LanceNamespaceDBConnection(ns_client)
+
+        namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
+        db.create_namespace([namespace_name])
+
+        table_name = f"test_table_{uuid.uuid4().hex}"
+        namespace_path = [namespace_name]
+        schema = pa.schema(
+            [
+                pa.field("id", pa.int64()),
+                pa.field("vector", pa.list_(pa.float32(), 2)),
+                pa.field("text", pa.string()),
+            ]
+        )
+
+        db.create_table(table_name, schema=schema, namespace_path=namespace_path)
+
+        # Existing table should return True
+        table_id = namespace_path + [table_name]
+        assert db.table_exists(table_id=table_id) is True
+
+        # Non-existent table should return False
+        assert db.table_exists(table_id=namespace_path + ["nonexistent_table"]) is False
+    finally:
+        shutil.rmtree(temp_dir, ignore_errors=True)
@@ -0,0 +1,42 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+import importlib
+import re
+import sys
+from pathlib import Path
+
+import pytest
+
+
+def test_pyo3_abi_matches_minimum_supported_python():
+    project_dir = Path(__file__).parents[2]
+    pyproject = (project_dir / "pyproject.toml").read_text()
+    cargo_manifest = (project_dir / "Cargo.toml").read_text()
+
+    minimum_python = re.search(
+        r'^requires-python\s*=\s*">=(\d+)\.(\d+)"$', pyproject, re.MULTILINE
+    )
+    assert minimum_python is not None
+
+    major, minor = minimum_python.groups()
+    expected_abi = f"abi3-py{major}{minor}"
+    configured_abis = re.findall(r'"(abi3-py\d+)"', cargo_manifest)
+
+    assert configured_abis == [expected_abi, expected_abi], (
+        "the pyo3 runtime and build ABI features must both match requires-python"
+    )
+
+
+@pytest.mark.skipif(sys.platform != "win32", reason="Windows wheel regression test")
+def test_windows_wheel_tag_and_native_import():
+    project_dir = Path(__file__).parents[2]
+    wheels = list((project_dir.parent / "target" / "wheels").glob("lancedb-*.whl"))
+    if not wheels:
+        pytest.skip("no wheel artifact is available in this development environment")
+
+    assert len(wheels) == 1
+    assert wheels[0].name.endswith("-cp310-abi3-win_amd64.whl")
+
+    native_module = importlib.import_module("lancedb._lancedb")
+    assert Path(native_module.__file__).suffix == ".pyd"
@@ -6,6 +6,7 @@ import math
 import pytest

 from lancedb import DBConnection, Table, connect
+from lancedb.background_loop import LOOP
 from lancedb.permutation import Permutation, Permutations, permutation_builder


@@ -31,6 +32,25 @@ def test_split_random_ratios(mem_db):
    assert 65 <= split_1_count <= 75  # ~70% ± tolerance


+def test_execute_does_not_reenter_background_loop(tmp_path, monkeypatch):
+    import threading
+
+    db = connect(tmp_path)
+    tbl = db.create_table("test_table", pa.table({"x": range(10)}))
+    original_run = LOOP.run
+
+    def fail_on_reentry(future):
+        assert threading.current_thread() is not LOOP.thread
+        return original_run(future)
+
+    monkeypatch.setattr(LOOP, "run", fail_on_reentry)
+
+    permutation_tbl = permutation_builder(tbl).execute()
+
+    assert permutation_tbl.count_rows() == 10
+    assert permutation_tbl._conn.read_consistency_interval is None
+
+
 def test_split_random_counts(mem_db):
    """Test random splitting with absolute counts."""
    tbl = mem_db.create_table(
@@ -35,6 +35,12 @@ def make_mock_http_handler(handler):
    return MockLanceDBHandler


+@pytest.mark.parametrize("db_name", ["a" * 64, "invalid..database"])
+def test_connect_rejects_invalid_cloud_dns_hostname(db_name):
+    with pytest.raises(ValueError, match="DNS labels must contain 1 to 63 bytes"):
+        lancedb.connect(f"db://{db_name}", api_key="fake")
+
+
@contextlib.contextmanager
 def mock_lancedb_connection(handler):
    with http.server.HTTPServer(
@@ -812,6 +818,121 @@ def test_table_create_indices():
        table.drop_index("custom_fts_idx")


+def test_remote_create_index_async_returns_job():
+    from lancedb.index import BTree
+
+    describe_calls = []
+
+    def handler(request):
+        content_len = int(request.headers.get("Content-Length", 0))
+        body = request.rfile.read(content_len) if content_len > 0 else b""
+        if request.path == "/v1/table/test/create_index/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b'{"job_id": "job-1"}')
+        elif request.path == "/v1/jobs/describe":
+            assert json.loads(body)["job_id"] == "job-1"
+            describe_calls.append(1)
+            state = "IN_PROGRESS" if len(describe_calls) == 1 else "DONE"
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(
+                json.dumps(dict(job_id="job-1", job_state=state)).encode()
+            )
+        elif request.path == "/v1/jobs/cancel":
+            assert json.loads(body)["job_id"] == "job-1"
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b"{}")
+        elif request.path == "/v1/table/test/create/?mode=create":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b"{}")
+        elif request.path == "/v1/table/test/describe/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(
+                json.dumps(
+                    dict(
+                        version=1,
+                        schema=dict(
+                            fields=[
+                                dict(name="id", type={"type": "int64"}, nullable=False),
+                            ]
+                        ),
+                    )
+                ).encode()
+            )
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        table = db.create_table("test", [{"id": 1}])
+        job = table.create_index_async("id", config=BTree())
+        assert job.id == "job-1"
+        job.wait(timeout=timedelta(seconds=30))
+        assert len(describe_calls) == 2
+        job.cancel()
+
+
+def test_remote_job_wait_raises_on_failure():
+    from lancedb.exceptions import JobFailedError
+    from lancedb.index import BTree
+
+    def handler(request):
+        content_len = int(request.headers.get("Content-Length", 0))
+        body = request.rfile.read(content_len) if content_len > 0 else b""
+        if request.path == "/v1/table/test/create_index/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b'{"job_id": "job-2"}')
+        elif request.path == "/v1/jobs/describe":
+            assert json.loads(body)["job_id"] == "job-2"
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(
+                json.dumps(dict(job_id="job-2", job_state="FAILED")).encode()
+            )
+        elif request.path == "/v1/table/test/create/?mode=create":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b"{}")
+        elif request.path == "/v1/table/test/describe/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(
+                json.dumps(
+                    dict(
+                        version=1,
+                        schema=dict(
+                            fields=[
+                                dict(name="id", type={"type": "int64"}, nullable=False),
+                            ]
+                        ),
+                    )
+                ).encode()
+            )
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        table = db.create_table("test", [{"id": 1}])
+        job = table.create_index_async("id", config=BTree())
+        with pytest.raises(JobFailedError, match="job-2"):
+            job.wait()
+
+
 def test_remote_create_index_new_api():
    received_requests = []

@@ -1020,7 +1141,7 @@ def query_test_table(query_handler, *, server_version=Version("0.1.0")):
            request.send_header("Content-Type", "application/json")
            request.send_header("phalanx-version", str(server_version))
            request.end_headers()
-            request.wfile.write(b"{}")
+            request.wfile.write(b'{"version": 1, "schema": {"fields": []}}')
        elif request.path == "/v1/table/test/query/":
            content_len = int(request.headers.get("Content-Length"))
            body = request.rfile.read(content_len)
@@ -1858,3 +1979,330 @@ def test_inherited_remote_table_reopens_after_fork():
    finally:
        server.shutdown()
        server_thread.join()
+
+
+BLOB_DESCRIBE_RESPONSE = {
+    "table": "test",
+    "version": 1,
+    "schema": {
+        "fields": [
+            {"name": "id", "type": {"type": "int64"}, "nullable": False},
+            {
+                "name": "image",
+                "type": {
+                    "type": "struct",
+                    "fields": [
+                        {
+                            "name": "data",
+                            "type": {"type": "large_binary"},
+                            "nullable": True,
+                        },
+                        {"name": "uri", "type": {"type": "string"}, "nullable": True},
+                    ],
+                },
+                "nullable": True,
+                "metadata": {
+                    "ARROW:extension:name": "lance.blob.v2",
+                    "ARROW:extension:metadata": "",
+                },
+            },
+        ]
+    },
+}
+
+
+def blob_query_response_table():
+    image_field = pa.field(
+        "image",
+        pa.struct(
+            [
+                pa.field("kind", pa.uint8(), nullable=False),
+                pa.field("position", pa.uint64(), nullable=False),
+                pa.field("size", pa.uint64(), nullable=False),
+                pa.field("blob_id", pa.uint32(), nullable=False),
+                pa.field("blob_uri", pa.string(), nullable=False),
+            ]
+        ),
+        metadata={"lance-encoding:blob": "true"},
+    )
+    images = pa.StructArray.from_arrays(
+        [
+            pa.array([1, 0, 0], type=pa.uint8()),
+            pa.array([0, 0, 0], type=pa.uint64()),
+            pa.array([5, 0, 5], type=pa.uint64()),
+            pa.array([1, 0, 2], type=pa.uint32()),
+            pa.array(["", "", ""], type=pa.string()),
+        ],
+        fields=image_field.type,
+        mask=pa.array([False, True, False]),
+    )
+    return pa.Table.from_arrays(
+        [
+            pa.array([1, 2, 3], type=pa.int64()),
+            images,
+            pa.array([10, 20, 30], type=pa.uint64()),
+        ],
+        schema=pa.schema(
+            [
+                pa.field("id", pa.int64(), nullable=False),
+                image_field,
+                pa.field("_rowid", pa.uint64()),
+            ]
+        ),
+    )
+
+
+@contextlib.contextmanager
+def blob_remote_table(*, server_version=Version("0.5.0")):
+    def handler(request):
+        if request.path == "/v1/table/test/describe/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.send_header("phalanx-version", str(server_version))
+            request.end_headers()
+            request.wfile.write(json.dumps(BLOB_DESCRIBE_RESPONSE).encode())
+        elif request.path.startswith("/v1/table/test/blob/image/"):
+            path = request.path.partition("?")[0]
+            row_id = int(path.split("/")[-2])
+            payload = {10: b"alpha", 20: None, 30: b"gamma"}[row_id]
+            if payload is None:
+                request.send_response(204)
+                request.end_headers()
+                return
+            byte_range = request.headers["Range"].removeprefix("bytes=")
+            start_text, end_text = byte_range.split("-", maxsplit=1)
+            start = int(start_text)
+            end = int(end_text) if end_text else len(payload) - 1
+            chunk = payload[start : end + 1]
+            request.send_response(206)
+            request.send_header("Content-Range", f"bytes {start}-{end}/{len(payload)}")
+            request.send_header("Content-Length", str(len(chunk)))
+            request.end_headers()
+            request.wfile.write(chunk)
+        elif request.path == "/v1/table/test/query/":
+            content_len = int(request.headers.get("Content-Length", 0))
+            body = json.loads(request.rfile.read(content_len))
+            assert body["columns"] == ["id", "image"]
+            assert body["with_row_id"] is True
+            response_table = blob_query_response_table()
+            request.send_response(200)
+            request.send_header("Content-Type", "application/vnd.apache.arrow.file")
+            request.end_headers()
+            with pa.ipc.new_file(request.wfile, response_table.schema) as writer:
+                writer.write_table(response_table)
+        elif request.path == "/v1/table/test/fetch_blobs/":
+            content_len = int(request.headers.get("Content-Length", 0))
+            body = json.loads(request.rfile.read(content_len))
+            assert body["column"] == "image"
+            assert body["row_ids"] == [10, 20, 30]
+            response_table = pa.table(
+                {"image": pa.array([b"alpha", None, b"gamma"], type=pa.large_binary())}
+            )
+            request.send_response(200)
+            request.send_header("Content-Type", "application/vnd.apache.arrow.stream")
+            request.end_headers()
+            with pa.ipc.new_stream(request.wfile, response_table.schema) as writer:
+                writer.write_table(response_table)
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        yield db.open_table("test")
+
+
+def test_remote_blob_columns_and_fetch():
+    with blob_remote_table() as table:
+        assert table.blob_columns() == ["image"]
+        blobs = table.fetch_blobs("image", [10, 20, 30])
+        assert blobs.to_pylist() == [b"alpha", None, b"gamma"]
+
+
+def test_remote_blob_files_are_lazy_seekable_handles():
+    with blob_remote_table() as table:
+        files = table.fetch_blob_files("image", [10, 20, 30])
+
+        assert len(files) == 3
+        alpha, null_row, gamma = files
+        assert null_row is None
+        assert alpha is not None
+        assert gamma is not None
+        assert alpha.size() == 5
+        assert alpha.read_range(1, 3) == b"lph"
+        gamma.seek(2)
+        assert gamma.read() == b"mma"
+
+
+def test_remote_blob_fetch_accepts_query_table():
+    hits = pa.table({"_rowid": pa.array([10, 20, 30], type=pa.uint64())})
+
+    with blob_remote_table() as table:
+        blobs = table.fetch_blobs("image", hits)
+
+    assert blobs.to_pylist() == [b"alpha", None, b"gamma"]
+
+
+def test_remote_blob_query_stashes_row_ids_for_fetch():
+    with blob_remote_table() as table:
+        hits = table.search().select(["id", "image"]).limit(3).to_arrow()
+        assert "_rowid" not in hits.column_names
+        assert "_lance_row_id" in hits.schema.field("image").type.names
+        blobs = table.fetch_blobs("image", hits)
+
+    assert blobs.to_pylist() == [b"alpha", None, b"gamma"]
+
+
+def test_remote_blob_query_survives_a_server_that_ignores_the_row_id_request():
+    def handler(request):
+        if request.path == "/v1/table/test/describe/":
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.send_header("phalanx-version", "0.5.0")
+            request.end_headers()
+            request.wfile.write(json.dumps(BLOB_DESCRIBE_RESPONSE).encode())
+        elif request.path == "/v1/table/test/query/":
+            content_len = int(request.headers.get("Content-Length", 0))
+            assert json.loads(request.rfile.read(content_len))["with_row_id"] is True
+            response_table = blob_query_response_table().drop_columns(["_rowid"])
+            request.send_response(200)
+            request.send_header("Content-Type", "application/vnd.apache.arrow.file")
+            request.end_headers()
+            with pa.ipc.new_file(request.wfile, response_table.schema) as writer:
+                writer.write_table(response_table)
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        table = db.open_table("test")
+        hits = table.search().select(["id", "image"]).limit(3).to_arrow()
+
+        assert hits.column_names == ["id", "image"]
+        assert "_lance_row_id" not in hits.schema.field("image").type.names
+        with pytest.raises(ValueError, match="pass a list of row ids"):
+            table.fetch_blobs("image", hits)
+
+
+def test_remote_blob_byte_apis_not_supported_on_old_server():
+    with blob_remote_table(server_version=Version("0.1.0")) as table:
+        assert table.blob_columns() == ["image"]
+        with pytest.raises(NotImplementedError, match="not supported"):
+            table.fetch_blobs("image", [1])
+        with pytest.raises(NotImplementedError, match="not supported"):
+            table.fetch_blob_files("image", [1])
+
+
+def test_remote_connection_jobs_surface():
+    from lancedb.exceptions import JobFailedError
+
+    schema = pa.schema([("state", pa.string())])
+    batch = pa.record_batch([pa.array(["created", "done"])], schema=schema)
+    sink = pa.BufferOutputStream()
+    with pa.ipc.new_stream(sink, schema) as writer:
+        writer.write_batch(batch)
+    events_body = sink.getvalue().to_pybytes()
+
+    def handler(request):
+        content_len = int(request.headers.get("Content-Length", 0))
+        body = request.rfile.read(content_len) if content_len > 0 else b""
+        payload = json.loads(body) if body else {}
+        if request.path == "/v1/jobs/list":
+            if payload.get("page_token") is None:
+                rsp = dict(
+                    jobs=[
+                        dict(
+                            job_id="job-1",
+                            table="t1",
+                            job_type="create_index",
+                            state="in_progress",
+                            created_at_millis=1000,
+                        )
+                    ],
+                    page_token="next",
+                )
+            else:
+                assert payload["page_token"] == "next"
+                rsp = dict(
+                    jobs=[
+                        dict(
+                            job_id="job-2",
+                            table="t2",
+                            job_type="create_index",
+                            state="succeeded",
+                            created_at_millis=2000,
+                        )
+                    ]
+                )
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(json.dumps(rsp).encode())
+        elif request.path == "/v1/jobs/describe":
+            if payload["job_id"] != "job-1":
+                request.send_response(404)
+                request.end_headers()
+                return
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(
+                json.dumps(
+                    dict(
+                        job_id="job-1",
+                        job_type="create_index",
+                        job_state="FAILED",
+                        creation_ms=1000,
+                        spec=dict(column="vec"),
+                        failure=dict(
+                            phase="execute", message="worker died", retryable=True
+                        ),
+                    )
+                ).encode()
+            )
+        elif request.path == "/v1/jobs/cancel":
+            if payload["job_id"] != "job-1":
+                request.send_response(404)
+                request.end_headers()
+                return
+            request.send_response(200)
+            request.send_header("Content-Type", "application/json")
+            request.end_headers()
+            request.wfile.write(b'{"job_id": "job-1"}')
+        elif request.path == "/v1/jobs/query_events":
+            assert payload["job_id"] == "job-1"
+            request.send_response(200)
+            request.send_header("Content-Type", "application/vnd.apache.arrow.stream")
+            request.end_headers()
+            request.wfile.write(events_body)
+        else:
+            request.send_response(404)
+            request.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        jobs = db.list_jobs()
+        assert [job.job_id for job in jobs] == ["job-1", "job-2"]
+        assert jobs[0].state == "running"
+        assert jobs[0].table == "t1"
+        assert jobs[1].state == "finished"
+
+        description = db.get_job("job-1")
+        assert description.job_type == "create_index"
+        assert description.state == "failed"
+        assert json.loads(description.spec_json) == {"column": "vec"}
+        assert description.failure.message == "worker died"
+        assert description.failure.retryable is True
+        assert db.get_job("missing") is None
+
+        assert db.cancel_job("job-1") is True
+        assert db.cancel_job("missing") is False
+
+        batches = db.job_history("job-1")
+        assert len(batches) == 1
+        assert batches[0].num_rows == 2
+        assert batches[0].column("state").to_pylist() == ["created", "done"]
+
+        job = db.job("job-1")
+        assert job.id == "job-1"
+        assert job.status() == "failed"
+        with pytest.raises(JobFailedError, match="worker died"):
+            job.wait(timeout=timedelta(seconds=5))
@@ -2,11 +2,16 @@
 # SPDX-FileCopyrightText: Copyright The LanceDB Authors


+import ctypes
+import gc
 import os
 import sys
 import threading
 import warnings
+import weakref
+from concurrent.futures import ThreadPoolExecutor
 from datetime import date, datetime, timedelta
+from decimal import Decimal
 from time import sleep
 from typing import List
 from unittest.mock import patch
@@ -98,6 +103,30 @@ def test_basic(mem_db: DBConnection):
    assert table.to_arrow() == expected_data


+def test_search_preserves_nulls_from_sliced_arrow_table(mem_db: DBConnection):
+    data = pa.table(
+        {
+            "id": [0, 1, 2, 3, 4],
+            "score_cn": [None, 22, None, 5, 8],
+            "score_mt": [None, 42, None, 5, 8],
+            "vector": [
+                [20, 19, -1, -1],
+                [41, 38, 22, 42],
+                [10, 10, -1, -1],
+                [5, 5, 5, 5],
+                [8, 8, 8, 8],
+            ],
+        }
+    ).slice(1)
+
+    table = mem_db.create_table("sliced_nullable", data=data)
+    result = table.search([41, 38, 22, 42]).limit(1).to_arrow()
+
+    assert result["id"].to_pylist() == [1]
+    assert result["score_cn"].to_pylist() == [22]
+    assert result["score_mt"].to_pylist() == [42]
+
+
 def test_table_to_pandas_default_matches_arrow(tmp_db: DBConnection):
    pd = pytest.importorskip("pandas")
    data = pa.table({"id": [1, 2], "text": ["one", "two"]})
@@ -308,6 +337,21 @@ async def test_update_async(mem_db_async: AsyncConnection):
    assert await table.count_rows("id == 10") == 1


+@pytest.mark.asyncio
+async def test_update_expr_filter_literals_async(mem_db_async: AsyncConnection):
+    values = ["5", "4.66e-84", "it's"]
+    table = await mem_db_async.create_table(
+        "update_expr_literals",
+        data=[{"field": value, "result": "original"} for value in values],
+    )
+
+    for value in values:
+        update_res = await table.update({"result": value}, where=col("field") == value)
+        assert update_res.rows_updated == 1
+
+    assert (await table.to_arrow())["result"].to_pylist() == values
+
+
 def test_create_table(mem_db: DBConnection):
    schema = pa.schema(
        {
@@ -434,6 +478,38 @@ def test_add(mem_db: DBConnection):
    _add(table, schema)


+def test_add_releases_arrow_buffers_without_gc(mem_db: DBConnection):
+    """Regression test for https://github.com/lancedb/lancedb/issues/2512."""
+    schema = pa.schema([pa.field("x", pa.int64())])
+    table = mem_db.create_table("test_add_releases_arrow_buffers", schema=schema)
+
+    class BufferOwner:
+        def __init__(self, size: int):
+            self.memory = ctypes.create_string_buffer(size)
+
+    owner_refs = []
+    gc_was_enabled = gc.isenabled()
+    gc.disable()
+    try:
+        for _ in range(3):
+            size = 8 * 1024
+            owner = BufferOwner(size)
+            arrow_buffer = pa.foreign_buffer(
+                ctypes.addressof(owner.memory), size, owner
+            )
+            array = pa.Array.from_buffers(pa.int64(), 1024, [None, arrow_buffer])
+            batch = pa.RecordBatch.from_arrays([array], schema=schema)
+            owner_refs.append(weakref.ref(owner))
+
+            table.add(batch)
+            del batch, array, arrow_buffer, owner
+
+        assert all(owner_ref() is None for owner_ref in owner_refs)
+    finally:
+        if gc_was_enabled:
+            gc.enable()
+
+
 def test_add_write_parallelism(mem_db: DBConnection):
    schema = pa.schema([pa.field("id", pa.int64())])
    table = mem_db.create_table("test", schema=schema)
@@ -1402,6 +1478,15 @@ async def test_async_open_table_with_branch_version(tmp_path):
    assert await pinned.count_rows() == 4  # writable again


+def test_create_index_async_returns_done_job(mem_db: DBConnection):
+    table = mem_db.create_table("job_test", [{"id": i} for i in range(10)])
+    job = table.create_index_async("id", config=BTree())
+    assert job.id is None
+    job.wait()
+    assert len(table.list_indices()) == 1
+    job.cancel()
+
+
@patch("lancedb.table.AsyncTable.create_index")
 def test_create_index_method(mock_create_index, mem_db: DBConnection):
    table = mem_db.create_table(
@@ -1815,6 +1900,33 @@ def test_add_nullable_struct_with_none(mem_db: DBConnection):
    assert result.column("data").to_pylist() == [{"x": 1.0}, None]


+def test_read_mostly_null_list_v2_2_page_boundary(tmp_path):
+    # Regression test for #3194. This row/value count crosses a v2.2 structural
+    # encoding page boundary where Lance 3.0.0 sliced repetition/definition
+    # levels by row offset and decoded child arrays at different lengths.
+    num_rows = 64_885
+    num_values = 217
+    list_type = pa.list_(pa.float32())
+    source = pa.table(
+        {
+            "id": np.arange(num_rows, dtype=np.int64),
+            "coords": pa.array(
+                [[1.0, 2.0, 3.0, 4.0]] * num_values + [None] * (num_rows - num_values),
+                type=list_type,
+            ),
+        }
+    )
+    db = lancedb.connect(
+        tmp_path,
+        storage_options={"new_table_data_storage_version": "2.2"},
+    )
+    table = db.create_table("test_sparse_nullable_list", data=source)
+
+    result = table.search().select(["id", "coords"]).limit(num_rows).to_arrow()
+
+    assert result.equals(source)
+
+
 def test_add_with_integer_embeddings_preserves_casting(mem_db: DBConnection):
    class Schema(LanceModel):
        text: str
@@ -2115,6 +2227,27 @@ def test_delete(mem_db: DBConnection):
    assert table.to_arrow()["id"].to_pylist() == [1]


+def test_concurrent_deletes_are_thread_safe(mem_db: DBConnection):
+    num_workers = 8
+    table = mem_db.create_table(
+        "my_table", data=[{"id": row_id} for row_id in range(num_workers)]
+    )
+    barrier = threading.Barrier(num_workers)
+
+    def delete(row_id: int):
+        barrier.wait()
+        return table.delete(f"id = {row_id}")
+
+    with ThreadPoolExecutor(max_workers=num_workers) as pool:
+        results = list(pool.map(delete, range(num_workers)))
+
+    assert all(result.num_deleted_rows == 1 for result in results)
+    assert sorted(result.version for result in results) == list(
+        range(2, num_workers + 2)
+    )
+    assert table.count_rows() == 0
+
+
 def test_delete_expr(mem_db: DBConnection):
    table = mem_db.create_table(
        "my_table",
@@ -2165,6 +2298,162 @@ def test_update(mem_db: DBConnection):
    assert np.allclose(v, np.array([[1.2, 1.9], [1.1, 1.1]]))


+def test_update_expr_filter_literals(mem_db: DBConnection):
+    values = ["5", "4.66e-84", "it's"]
+    table = mem_db.create_table(
+        "update_expr_literals",
+        data=[{"field": value, "result": "original"} for value in values],
+    )
+
+    for value in values:
+        update_res = table.update(where=col("field") == value, values={"result": value})
+        assert update_res.rows_updated == 1
+
+    assert table.to_arrow()["result"].to_pylist() == values
+
+
+def test_update_expr_filter_preserves_typed_semantics(mem_db: DBConnection):
+    low = Decimal("1.234567890123456789")
+    high = Decimal("1.234567890123456790")
+    decimal_schema = pa.schema(
+        [("val", pa.decimal128(19, 18)), ("result", pa.string())]
+    )
+    decimal_table = mem_db.create_table(
+        "update_expr_decimal",
+        pa.table(
+            {"val": [low, high], "result": ["old", "old"]},
+            schema=decimal_schema,
+        ),
+    )
+    predicate = col("val") < lit(high)
+    assert decimal_table.search().where(predicate).to_arrow().num_rows == 1
+    result = decimal_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    keyword_table = mem_db.create_table(
+        "update_expr_keyword", [{"null": 1, "result": "old"}]
+    )
+    predicate = col("null") == 1
+    assert keyword_table.search().where(predicate).to_arrow().num_rows == 1
+    result = keyword_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    empty_in_table = mem_db.create_table(
+        "update_expr_empty_in", [{"id": 1, "result": "old"}]
+    )
+    predicate = col("id").isin([])
+    assert empty_in_table.search().where(predicate).to_arrow().num_rows == 0
+    result = empty_in_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 0
+
+    marker = "__lancedb_binary_placeholder_0__"
+    binary_schema = pa.schema(
+        [("payload", pa.binary()), ("text", pa.string()), ("result", pa.string())]
+    )
+    binary_table = mem_db.create_table(
+        "update_expr_binary",
+        pa.table(
+            {
+                "payload": [b"\x01", b"\x02"],
+                "text": ["other", marker],
+                "result": ["old", "old"],
+            },
+            schema=binary_schema,
+        ),
+    )
+    predicate = (col("payload") == lit(b"\x01")) | (col("text") == marker)
+    assert binary_table.search().where(predicate).to_arrow().num_rows == 2
+    result = binary_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 2
+
+    nonfinite_table = mem_db.create_table(
+        "update_expr_nonfinite",
+        [{"x": 1.0, "result": "old"}, {"x": 2.0, "result": "old"}],
+    )
+    predicate = col("x") < float("inf")
+    assert nonfinite_table.search().where(predicate).to_arrow().num_rows == 2
+    result = nonfinite_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 2
+
+    float16_table = mem_db.create_table(
+        "update_expr_float16",
+        [{"x": 1.0, "result": "old"}, {"x": 3.0, "result": "old"}],
+    )
+    predicate = col("x").cast(pa.float16()) < 2.0
+    assert float16_table.search().where(predicate).to_arrow().num_rows == 1
+    result = float16_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    string_cast_table = mem_db.create_table(
+        "update_expr_string_cast",
+        [{"x": 1, "result": "old"}, {"x": 2, "result": "old"}],
+    )
+    predicate = col("x").cast("string") == "1"
+    assert string_cast_table.search().where(predicate).to_arrow().num_rows == 1
+    result = string_cast_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    quoted_identifier_schema = pa.schema(
+        [("payload", pa.binary()), ("odd'name", pa.int64()), ("result", pa.string())]
+    )
+    quoted_identifier_table = mem_db.create_table(
+        "update_expr_quoted_identifier",
+        pa.table(
+            {"payload": [b"\x01"], "odd'name": [1], "result": ["old"]},
+            schema=quoted_identifier_schema,
+        ),
+    )
+    predicate = (col("payload") == lit(b"\x01")) & (col("odd'name") == 1)
+    assert quoted_identifier_table.search().where(predicate).to_arrow().num_rows == 1
+    result = quoted_identifier_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    decimal256_schema = pa.schema(
+        [("val", pa.decimal256(40, 2)), ("result", pa.string())]
+    )
+    decimal256_table = mem_db.create_table(
+        "update_expr_decimal256",
+        pa.table(
+            {
+                "val": [Decimal("1.00"), Decimal("3.00")],
+                "result": ["old", "old"],
+            },
+            schema=decimal256_schema,
+        ),
+    )
+    predicate = col("val") < lit(Decimal("2.00")).cast(pa.decimal256(40, 2))
+    assert decimal256_table.search().where(predicate).to_arrow().num_rows == 1
+    result = decimal256_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 1
+
+    binary_empty_table = mem_db.create_table(
+        "update_expr_binary_empty",
+        pa.table(
+            {"payload": [b"\x01", b"\x02"], "result": ["old", "old"]},
+            schema=pa.schema([("payload", pa.binary()), ("result", pa.string())]),
+        ),
+    )
+    predicate = (col("payload") == lit(b"\x01")).isin([])
+    assert binary_empty_table.search().where(predicate).to_arrow().num_rows == 0
+    assert predicate.to_sql() == "false"
+    result = binary_empty_table.update(where=predicate, values={"result": "new"})
+    assert result.rows_updated == 0
+
+
+def test_update_with_arrow_scalar(mem_db: DBConnection):
+    schema = pa.schema({"id": pa.int64(), "vector": pa.list_(pa.float32(), 4)})
+    table = mem_db.create_table("my_table", schema=schema)
+    table.add([{"id": 1, "vector": [1.0, 2.0, 3.0, 4.0]}])
+
+    value = table.search().select(["vector"]).limit(1).to_arrow()["vector"][0]
+    assert isinstance(value, pa.FixedSizeListScalar)
+
+    result = table.update(where="id == 1", values={"vector": value})
+
+    assert result.rows_updated == 1
+    assert table.to_arrow()["vector"].to_pylist() == [[1.0, 2.0, 3.0, 4.0]]
+
+
 def test_update_types(mem_db: DBConnection):
    table = mem_db.create_table(
        "my_table",
@@ -2332,6 +2621,55 @@ def test_merge_insert(mem_db: DBConnection):
        )


+def test_merge_insert_nullable_pandas_into_pydantic_schema(mem_db: DBConnection):
+    # Regression test for https://github.com/lancedb/lancedb/issues/2366
+    pd = pytest.importorskip("pandas")
+
+    class Document(LanceModel):
+        id: int
+        title: str
+        content: str
+
+    table = mem_db.create_table("documents", schema=Document)
+    table.add(
+        pd.DataFrame(
+            {
+                "title": ["Old title", "Unchanged"],
+                "id": [2, 3],
+                "content": ["Old content", "Keep this"],
+            }
+        )
+    )
+
+    # Pandas produces nullable Arrow fields, in an order that differs from the
+    # non-nullable Pydantic schema. This is valid as long as the data has no nulls.
+    new_data = pd.DataFrame(
+        {
+            "title": ["Inserted", "Updated"],
+            "id": [1, 2],
+            "content": ["New row", "New content"],
+        }
+    )
+    result = (
+        table.merge_insert("id")
+        .when_matched_update_all()
+        .when_not_matched_insert_all()
+        .execute(new_data)
+    )
+
+    assert result.num_inserted_rows == 1
+    assert result.num_updated_rows == 1
+    expected = pa.Table.from_pylist(
+        [
+            {"id": 1, "title": "Inserted", "content": "New row"},
+            {"id": 2, "title": "Updated", "content": "New content"},
+            {"id": 3, "title": "Unchanged", "content": "Keep this"},
+        ],
+        schema=Document.to_arrow_schema(),
+    )
+    assert table.to_arrow().sort_by("id") == expected
+
+
 def test_merge_insert_by_source_delete_expr(mem_db: DBConnection):
    table = mem_db.create_table(
        "my_table",
@@ -2355,6 +2693,29 @@ def test_merge_insert_by_source_delete_expr(mem_db: DBConnection):
    assert table.to_arrow().sort_by("a") == expected


+def test_merge_insert_by_source_delete_reconfigure(mem_db: DBConnection):
+    # Calling when_not_matched_by_source_delete() again with no condition must
+    # widen the delete to unconditional, not keep the earlier condition around.
+    table = mem_db.create_table(
+        "my_table",
+        data=pa.table({"a": [1, 2, 3], "b": ["a", "b", "c"]}),
+    )
+    new_data = pa.table({"a": [2, 4], "b": ["x", "z"]})
+
+    merge_insert_res = (
+        table.merge_insert("a")
+        .when_matched_update_all()
+        .when_not_matched_insert_all()
+        .when_not_matched_by_source_delete("a > 2")
+        .when_not_matched_by_source_delete()
+        .execute(new_data)
+    )
+    assert merge_insert_res.num_deleted_rows == 2
+
+    expected = pa.table({"a": [2, 4], "b": ["x", "z"]})
+    assert table.to_arrow().sort_by("a") == expected
+
+
@pytest.mark.asyncio
 async def test_merge_insert_by_source_delete_expr_async(
    mem_db_async: AsyncConnection,
@@ -2409,6 +2770,36 @@ def test_merge_insert_subschema(mem_db: DBConnection, data_format):
    assert table.to_arrow().sort_by("id") == expected


+def test_repeated_partial_merge_insert_with_scalar_index(mem_db: DBConnection):
+    def make_batch(start: int) -> pa.Table:
+        return pa.table(
+            {
+                "id": [f"id-{i:04}" for i in range(start, start + 100)],
+                "category": ["A"] * 100,
+                "value_a": [float(i) for i in range(start, start + 100)],
+                "value_b": [float(i) / 10 for i in range(100)],
+            }
+        )
+
+    table = mem_db.create_table("my_table", data=make_batch(0))
+    table.add(make_batch(100))
+    table.add(make_batch(200))
+    table.create_index("id", config=BTree())
+
+    ids = [f"id-{i:04}" for i in range(100, 200)]
+    for value in (999.0, 888.0):
+        result = (
+            table.merge_insert("id")
+            .when_matched_update_all()
+            .execute(pa.table({"id": ids, "value_a": [value] * 100}))
+        )
+        assert result.num_updated_rows == 100
+
+    actual = table.to_arrow().sort_by("id")
+    assert actual.num_rows == 300
+    assert actual["value_a"].to_pylist()[100:200] == [888.0] * 100
+
+
@pytest.mark.asyncio
 async def test_merge_insert_async(mem_db_async: AsyncConnection):
    data = pa.table({"a": [1, 2, 3], "b": ["a", "b", "c"]})
@@ -3087,9 +3478,6 @@ def test_consistency(tmp_path, consistency_interval):

    db2 = lancedb.connect(tmp_path, read_consistency_interval=consistency_interval)
    table2 = db2.open_table("my_table")
-    if consistency_interval is not None:
-        assert "read_consistency_interval=datetime.timedelta(" in repr(db2)
-        assert "read_consistency_interval=datetime.timedelta(" in repr(table2)
    assert table2.version == table.version

    table.add([{"id": 1}])
@@ -3438,8 +3826,8 @@ def test_create_table_empty_list_no_schema_error(mem_db: DBConnection):
        mem_db.create_table("test_empty_no_schema", data=[])


-def test_add_table_with_empty_embeddings(tmp_path):
-    """Test exact scenario from issue #1968
+def test_create_table_without_data_with_vector_schema(tmp_path):
+    """Test exact scenario from issue #1968.

    Regression test for issue #1968:
    https://github.com/lancedb/lancedb/issues/1968
@@ -3451,6 +3839,9 @@ def test_add_table_with_empty_embeddings(tmp_path):
        embedding: Vector(16)

    table = db.create_table("test", schema=MySchema)
+    assert table.count_rows() == 0
+    assert table.schema == MySchema.to_arrow_schema()
+
    table.add(
        [{"text": "bar", "embedding": [0.1] * 16}],
        on_bad_vectors="drop",
@@ -75,6 +75,22 @@ class TestVoyageAIModelRegistration:
        with pytest.raises(ValueError, match="not supported"):
            func.ndims()

+    def test_voyage3_source_embeddings_use_text_api(self, mock_voyageai_client):
+        """Regression test for text table data being sent to the multimodal API."""
+        mock_voyageai_client.tokenize.return_value = [["hello", "world"]]
+        mock_voyageai_client.embed.return_value.embeddings = [[0.1] * 1024]
+
+        registry = get_registry()
+        func = registry.get("voyageai").create(name="voyage-3")
+
+        embeddings = func.compute_source_embeddings("hello world")
+
+        assert embeddings == [[0.1] * 1024]
+        mock_voyageai_client.embed.assert_called_once_with(
+            texts=["hello world"], model="voyage-3", input_type="document"
+        )
+        mock_voyageai_client.multimodal_embed.assert_not_called()
+
    @pytest.mark.parametrize(
        "model_name",
        [
@@ -13,7 +13,11 @@ use crate::{
    runtime::future_into_py,
    table::Table,
 };
-use arrow::{datatypes::Schema, ffi_stream::ArrowArrayStreamReader, pyarrow::FromPyArrow};
+use arrow::{
+    datatypes::Schema,
+    ffi_stream::ArrowArrayStreamReader,
+    pyarrow::{FromPyArrow, ToPyArrow},
+};
 use lancedb::{
    connection::Connection as LanceConnection,
    connection::NamespaceClientPushdownOperation,
@@ -24,7 +28,7 @@ use pyo3::{
    Bound, FromPyObject, Py, PyAny, PyRef, PyResult, Python,
    exceptions::{PyRuntimeError, PyValueError},
    pyclass, pyfunction, pymethods,
-    types::{PyDict, PyDictMethods},
+    types::{PyDict, PyDictMethods, PyList, PyListMethods},
 };

 #[pyclass]
@@ -536,6 +540,55 @@ impl Connection {
            })
        })
    }
+
+    pub fn job(&self, job_id: String) -> PyResult<crate::job::Job> {
+        let inner = self.get_inner()?.clone();
+        Ok(crate::job::Job::new(inner.job(job_id).infer_error()?))
+    }
+
+    pub fn list_jobs(self_: PyRef<'_, Self>) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.get_inner()?.clone();
+        future_into_py(self_.py(), async move {
+            let jobs = inner.list_jobs().await.infer_error()?;
+            Ok(jobs
+                .into_iter()
+                .map(crate::job::JobInfo::from)
+                .collect::<Vec<_>>())
+        })
+    }
+
+    pub fn get_job(self_: PyRef<'_, Self>, job_id: String) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.get_inner()?.clone();
+        future_into_py(self_.py(), async move {
+            let description = inner.get_job(&job_id).await.infer_error()?;
+            Ok(description.map(crate::job::JobDescription::from))
+        })
+    }
+
+    pub fn cancel_job(self_: PyRef<'_, Self>, job_id: String) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.get_inner()?.clone();
+        future_into_py(self_.py(), async move {
+            inner.cancel_job(&job_id).await.infer_error()
+        })
+    }
+
+    #[pyo3(signature = (job_id=None))]
+    pub fn job_history(
+        self_: PyRef<'_, Self>,
+        job_id: Option<String>,
+    ) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.get_inner()?.clone();
+        future_into_py(self_.py(), async move {
+            let batches = inner.job_history(job_id.as_deref()).await.infer_error()?;
+            Python::attach(|py| {
+                let list = PyList::empty(py);
+                for batch in batches {
+                    list.append(batch.to_pyarrow(py)?)?;
+                }
+                Ok(list.unbind())
+            })
+        })
+    }
 }

 #[pyfunction]
@@ -102,6 +102,18 @@ impl<T> PythonErrorExt<T> for std::result::Result<T, LanceError> {
                    err.setattr(intern!(py, "__cause__"), cause_err)?;
                    Err(PyErr::from_value(err))
                }),
+                LanceError::JobFailed { .. } => Python::attach(|py| {
+                    let cls = py
+                        .import(intern!(py, "lancedb.exceptions"))?
+                        .getattr(intern!(py, "JobFailedError"))?;
+                    Err(PyErr::from_value(cls.call1((err.to_string(),))?))
+                }),
+                LanceError::JobCancelled { .. } => Python::attach(|py| {
+                    let cls = py
+                        .import(intern!(py, "lancedb.exceptions"))?
+                        .getattr(intern!(py, "JobCancelledError"))?;
+                    Err(PyErr::from_value(cls.call1((err.to_string(),))?))
+                }),
                _ => self.runtime_error(),
            },
        }
@@ -130,6 +130,14 @@ impl PyExpr {

    // ── utilities ────────────────────────────────────────────────────────────

+    /// Return the referenced column name for a bare column expression.
+    fn column_name(&self) -> Option<String> {
+        match &self.0 {
+            DfExpr::Column(column) if column.relation.is_none() => Some(column.name.clone()),
+            _ => None,
+        }
+    }
+
    /// Render the expression as a SQL string (useful for debugging).
    fn to_sql(&self) -> PyResult<String> {
        lancedb::expr::expr_to_sql_string(&self.0).map_err(|e| PyValueError::new_err(e.to_string()))
@@ -0,0 +1,145 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+use std::sync::Arc;
+
+use crate::runtime::future_into_py;
+use pyo3::{Bound, PyAny, PyRef, PyResult, pyclass, pymethods};
+
+use crate::error::PythonErrorExt;
+
+#[pyclass]
+pub struct Job {
+    inner: Arc<lancedb::Job>,
+}
+
+impl Job {
+    pub(crate) fn new(inner: lancedb::Job) -> Self {
+        Self {
+            inner: Arc::new(inner),
+        }
+    }
+}
+
+#[pymethods]
+impl Job {
+    #[getter]
+    pub fn id(&self) -> Option<String> {
+        self.inner.id().map(str::to_string)
+    }
+
+    pub fn status(self_: PyRef<'_, Self>) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.inner.clone();
+        future_into_py(
+            self_.py(),
+            async move { inner.status().await.infer_error() },
+        )
+    }
+
+    pub fn wait(self_: PyRef<'_, Self>) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.inner.clone();
+        future_into_py(self_.py(), async move {
+            inner.wait().await.infer_error()?;
+            Ok(())
+        })
+    }
+
+    pub fn cancel(self_: PyRef<'_, Self>) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.inner.clone();
+        future_into_py(self_.py(), async move {
+            inner.cancel().await.infer_error()?;
+            Ok(())
+        })
+    }
+}
+
+/// A row from `Connection.list_jobs`: one server-side job.
+#[pyclass(get_all, skip_from_py_object)]
+#[derive(Clone)]
+pub struct JobInfo {
+    job_id: String,
+    table: String,
+    job_type: String,
+    state: String,
+    created_at_millis: i64,
+}
+
+#[pymethods]
+impl JobInfo {
+    fn __repr__(&self) -> String {
+        format!(
+            "JobInfo(job_id={:?}, table={:?}, job_type={:?}, state={:?}, created_at_millis={})",
+            self.job_id, self.table, self.job_type, self.state, self.created_at_millis
+        )
+    }
+}
+
+impl From<lancedb::database::JobInfo> for JobInfo {
+    fn from(info: lancedb::database::JobInfo) -> Self {
+        Self {
+            job_id: info.job_id,
+            table: info.table,
+            job_type: info.job_type,
+            state: info.state,
+            created_at_millis: info.created_at_millis,
+        }
+    }
+}
+
+/// The server's account of why a job failed.
+#[pyclass(get_all, skip_from_py_object)]
+#[derive(Clone)]
+pub struct JobFailureInfo {
+    phase: Option<String>,
+    message: Option<String>,
+    retryable: Option<bool>,
+}
+
+#[pymethods]
+impl JobFailureInfo {
+    fn __repr__(&self) -> String {
+        format!(
+            "JobFailureInfo(phase={:?}, message={:?}, retryable={:?})",
+            self.phase, self.message, self.retryable
+        )
+    }
+}
+
+/// A described job from `Connection.get_job`.
+#[pyclass(get_all, skip_from_py_object)]
+#[derive(Clone)]
+pub struct JobDescription {
+    job_id: String,
+    job_type: String,
+    state: String,
+    creation_ms: i64,
+    spec_json: Option<String>,
+    failure: Option<JobFailureInfo>,
+}
+
+#[pymethods]
+impl JobDescription {
+    fn __repr__(&self) -> String {
+        format!(
+            "JobDescription(job_id={:?}, job_type={:?}, state={:?}, creation_ms={})",
+            self.job_id, self.job_type, self.state, self.creation_ms
+        )
+    }
+}
+
+impl From<lancedb::database::JobDescription> for JobDescription {
+    fn from(description: lancedb::database::JobDescription) -> Self {
+        Self {
+            job_id: description.job_id,
+            job_type: description.job_type,
+            state: description.state,
+            creation_ms: description.creation_ms,
+            spec_json: (!description.spec.is_null()).then(|| description.spec.to_string()),
+            failure: description.failure.map(|failure| JobFailureInfo {
+                phase: failure.phase,
+                message: failure.message,
+                retryable: failure.retryable,
+            }),
+        }
+    }
+}
@@ -25,6 +25,7 @@ pub mod error;
 pub mod expr;
 pub mod header;
 pub mod index;
+pub mod job;
 pub mod namespace;
 pub mod oauth;
 pub mod otel;
@@ -44,6 +45,10 @@ pub fn _lancedb(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
    m.add_class::<Connection>()?;
    m.add_class::<Session>()?;
    m.add_class::<Table>()?;
+    m.add_class::<crate::job::Job>()?;
+    m.add_class::<crate::job::JobInfo>()?;
+    m.add_class::<crate::job::JobDescription>()?;
+    m.add_class::<crate::job::JobFailureInfo>()?;
    m.add_class::<PyBlobFile>()?;
    m.add_class::<IndexConfig>()?;
    m.add_class::<Query>()?;
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

+use std::collections::HashMap;
 use std::sync::Arc;
 use std::time::Duration;

@@ -292,6 +293,7 @@ pub struct PyQueryRequest {
    pub filter: Option<PyQueryFilter>,
    pub full_text_search: Option<PyLanceDB<FtsQuery>>,
    pub select: PySelect,
+    pub select_source_columns: Option<HashMap<String, String>>,
    pub fast_search: Option<bool>,
    pub with_row_id: Option<bool>,
    pub use_lsm: Option<bool>,
@@ -322,6 +324,7 @@ impl From<AnyQuery> for PyQueryRequest {
                full_text_search: query_request
                    .full_text_search
                    .map(|fts| PyLanceDB(fts.query)),
+                select_source_columns: PySelect::source_columns(&query_request.select),
                select: PySelect(query_request.select),
                fast_search: Some(query_request.fast_search),
                with_row_id: Some(query_request.with_row_id),
@@ -347,6 +350,7 @@ impl From<AnyQuery> for PyQueryRequest {
                offset: vector_query.base.offset,
                filter: vector_query.base.filter.map(PyQueryFilter),
                full_text_search: None,
+                select_source_columns: PySelect::source_columns(&vector_query.base.select),
                select: PySelect(vector_query.base.select),
                fast_search: Some(vector_query.base.fast_search),
                with_row_id: Some(vector_query.base.with_row_id),
@@ -379,6 +383,25 @@ impl From<AnyQuery> for PyQueryRequest {
 #[derive(Clone)]
 pub struct PySelect(Select);

+impl PySelect {
+    fn source_columns(select: &Select) -> Option<HashMap<String, String>> {
+        match select {
+            Select::Expr(pairs) => Some(
+                pairs
+                    .iter()
+                    .filter_map(|(output, expr)| match expr {
+                        lancedb::expr::DfExpr::Column(column) if column.relation.is_none() => {
+                            Some((output.clone(), column.name.clone()))
+                        }
+                        _ => None,
+                    })
+                    .collect(),
+            ),
+            _ => None,
+        }
+    }
+}
+
 impl<'py> IntoPyObject<'py> for PySelect {
    type Target = PyAny;
    type Output = Bound<'py, Self::Target>;
@@ -426,9 +426,11 @@ pub struct PyBlobFile {
 impl PyBlobFile {
    fn read_bytes(self_: PyRef<'_, Self>) -> PyResult<Py<PyBytes>> {
        let inner = self_.inner.clone();
-        let bytes = block_on(async move { inner.read().await })
+        let py = self_.py();
+        let bytes = py
+            .detach(move || block_on(async move { inner.read().await }))
            .map_err(|e| PyRuntimeError::new_err(format!("blob read failed: {e}")))?;
-        Ok(PyBytes::new(self_.py(), bytes.as_ref()).unbind())
+        Ok(PyBytes::new(py, bytes.as_ref()).unbind())
    }

    pub fn read(self_: PyRef<'_, Self>) -> PyResult<Bound<'_, PyAny>> {
@@ -444,24 +446,32 @@ impl PyBlobFile {

    fn close(self_: PyRef<'_, Self>) -> PyResult<()> {
        let inner = self_.inner.clone();
-        block_on(async move { inner.close().await })
+        self_
+            .py()
+            .detach(move || block_on(async move { inner.close().await }))
            .map_err(|e| PyRuntimeError::new_err(format!("blob close failed: {e}")))
    }

    fn is_closed(self_: PyRef<'_, Self>) -> bool {
        let inner = self_.inner.clone();
-        block_on(async move { inner.is_closed().await })
+        self_
+            .py()
+            .detach(move || block_on(async move { inner.is_closed().await }))
    }

    fn seek(self_: PyRef<'_, Self>, position: u64) -> PyResult<()> {
        let inner = self_.inner.clone();
-        block_on(async move { inner.seek(position).await })
+        self_
+            .py()
+            .detach(move || block_on(async move { inner.seek(position).await }))
            .map_err(|e| PyRuntimeError::new_err(format!("blob seek failed: {e}")))
    }

    fn tell(self_: PyRef<'_, Self>) -> PyResult<u64> {
        let inner = self_.inner.clone();
-        block_on(async move { inner.tell().await })
+        self_
+            .py()
+            .detach(move || block_on(async move { inner.tell().await }))
            .map_err(|e| PyRuntimeError::new_err(format!("blob tell failed: {e}")))
    }

@@ -475,16 +485,20 @@ impl PyBlobFile {
            .checked_add(length as u64)
            .ok_or_else(|| PyValueError::new_err("offset + length overflowed"))?;
        let inner = self_.inner.clone();
-        let bytes = block_on(async move { inner.read_range(offset..end).await })
+        let py = self_.py();
+        let bytes = py
+            .detach(move || block_on(async move { inner.read_range(offset..end).await }))
            .map_err(|e| PyRuntimeError::new_err(format!("blob read_range failed: {e}")))?;
-        Ok(PyBytes::new(self_.py(), bytes.as_ref()).unbind())
+        Ok(PyBytes::new(py, bytes.as_ref()).unbind())
    }

    fn read_up_to(self_: PyRef<'_, Self>, length: usize) -> PyResult<Py<PyBytes>> {
        let inner = self_.inner.clone();
-        let bytes = block_on(async move { inner.read_up_to(length).await })
-            .map_err(|e| PyRuntimeError::new_err(format!("blob read failed: {e}")))?;
-        Ok(PyBytes::new(self_.py(), bytes.as_ref()).unbind())
+        let py = self_.py();
+        let bytes = py
+            .detach(move || block_on(async move { inner.read_up_to(length).await }))
+            .map_err(|e| PyRuntimeError::new_err(format!("blob read_up_to failed: {e}")))?;
+        Ok(PyBytes::new(py, bytes.as_ref()).unbind())
    }
 }

@@ -731,6 +745,9 @@ impl Table {

    #[allow(private_interfaces)]
    pub fn delete(self_: PyRef<'_, Self>, condition: PredicateArg) -> PyResult<Bound<'_, PyAny>> {
+        // Do not hold the Python borrow across the await.  The cloned Rust table
+        // handle is thread-safe and allows deletes on the same Python table to
+        // run concurrently without PyO3 reporting "Already borrowed".
        let inner = self_.inner_ref()?.clone();
        future_into_py(self_.py(), async move {
            let result = match &condition {
@@ -805,6 +822,37 @@ impl Table {
        })
    }

+    #[pyo3(signature = (column, index=None, replace=None, wait_timeout=None, *, name=None, train=None))]
+    pub fn create_index_async<'a>(
+        self_: PyRef<'a, Self>,
+        column: String,
+        index: Option<Bound<'_, PyAny>>,
+        replace: Option<bool>,
+        wait_timeout: Option<Bound<'_, PyAny>>,
+        name: Option<String>,
+        train: Option<bool>,
+    ) -> PyResult<Bound<'a, PyAny>> {
+        let index = extract_index_params(&index)?;
+        let timeout = wait_timeout.map(|t| t.extract::<std::time::Duration>().unwrap());
+        let mut op = self_
+            .inner_ref()?
+            .create_index_with_timeout(&[column], index, timeout);
+        if let Some(replace) = replace {
+            op = op.replace(replace);
+        }
+        if let Some(name) = name {
+            op = op.name(name);
+        }
+        if let Some(train) = train {
+            op = op.train(train);
+        }
+
+        future_into_py(self_.py(), async move {
+            let job = op.execute_async().await.infer_error()?;
+            Ok(crate::job::Job::new(job))
+        })
+    }
+
    pub fn drop_index(self_: PyRef<'_, Self>, index_name: String) -> PyResult<Bound<'_, PyAny>> {
        let inner = self_.inner_ref()?.clone();
        future_into_py(self_.py(), async move {
@@ -1330,7 +1378,12 @@ impl Table {

        let inner = self_.inner_ref()?.clone();
        future_into_py(self_.py(), async move {
-            let result = inner.add_columns(definitions, None).await.infer_error()?;
+            let result = inner
+                .add_columns()
+                .transform(definitions)
+                .execute()
+                .await
+                .infer_error()?;
            Ok(AddColumnsResult::from(result))
        })
    }
@@ -1344,7 +1397,12 @@ impl Table {

        let inner = self_.inner_ref()?.clone();
        future_into_py(self_.py(), async move {
-            let result = inner.add_columns(transform, None).await.infer_error()?;
+            let result = inner
+                .add_columns()
+                .transform(transform)
+                .execute()
+                .await
+                .infer_error()?;
            Ok(AddColumnsResult::from(result))
        })
    }
@@ -1,2 +1,2 @@
 [toolchain]
-channel = "1.95.0"
+channel = "1.97.0"
@@ -75,6 +75,8 @@ reqwest = { version = "0.12.0", default-features = false, features = [
    "http2",
    "json",
    "macos-system-configuration",
+    # Avoid linking OpenSSL into Python wheels, which breaks on FIPS hosts.
+    "rustls-tls-native-roots",
    "stream",
 ], optional = true }
 http = { version = "1", optional = true } # Matching what is in reqwest
@@ -9,6 +9,7 @@
 //!
 //! Blob tables require Lance file format >= 2.2 and stable row ids at create.

+use std::ops::Range;
 use std::sync::Arc;

 use arrow_array::LargeBinaryArray;
@@ -17,10 +18,202 @@ use arrow_schema::{DataType, Field, Schema};
 use lance::dataset::{BlobRangeRequest as LanceBlobRangeRequest, Dataset, WriteParams};
 use lance_arrow::FieldExt;
 use lance_encoding::version::LanceFileVersion;
+use lance_io::object_store::ObjectStore;
+use object_store::path::Path;

 use crate::error::{Error, Result};

-pub use lance::dataset::BlobFile;
+/// Seekable handle for one blob value, backed by local storage or a remote
+/// HTTP byte-range endpoint.
+#[derive(Debug)]
+pub struct BlobFile {
+    inner: BlobFileInner,
+}
+
+#[derive(Debug)]
+enum BlobFileInner {
+    Native(lance::dataset::BlobFile),
+    #[cfg(feature = "remote")]
+    Remote(Box<crate::remote::table::blobs::RemoteBlobFile>),
+}
+
+impl From<lance::dataset::BlobFile> for BlobFile {
+    fn from(value: lance::dataset::BlobFile) -> Self {
+        Self {
+            inner: BlobFileInner::Native(value),
+        }
+    }
+}
+
+#[cfg(feature = "remote")]
+impl From<crate::remote::table::blobs::RemoteBlobFile> for BlobFile {
+    fn from(value: crate::remote::table::blobs::RemoteBlobFile) -> Self {
+        Self {
+            inner: BlobFileInner::Remote(Box::new(value)),
+        }
+    }
+}
+
+impl BlobFile {
+    /// Inline reader over a data-file slice.
+    pub fn new_inline(
+        object_store: Arc<ObjectStore>,
+        path: Path,
+        position: u64,
+        size: u64,
+    ) -> Self {
+        lance::dataset::BlobFile::new_inline(object_store, path, position, size).into()
+    }
+
+    /// Dedicated sidecar-file reader.
+    pub fn new_dedicated(object_store: Arc<ObjectStore>, path: Path, size: u64) -> Self {
+        lance::dataset::BlobFile::new_dedicated(object_store, path, size).into()
+    }
+
+    /// Packed reader for a slice in a shared sidecar.
+    pub fn new_packed(
+        object_store: Arc<ObjectStore>,
+        path: Path,
+        position: u64,
+        size: u64,
+    ) -> Self {
+        lance::dataset::BlobFile::new_packed(object_store, path, position, size).into()
+    }
+
+    /// External reader at a resolved object location.
+    pub fn new_external(
+        object_store: Arc<ObjectStore>,
+        path: Path,
+        uri: String,
+        position: u64,
+        size: u64,
+    ) -> Self {
+        lance::dataset::BlobFile::new_external(object_store, path, uri, position, size).into()
+    }
+
+    /// Close the handle.
+    pub async fn close(&self) -> lance_core::Result<()> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.close().await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.close().await,
+        }
+    }
+
+    /// Whether the handle is closed.
+    pub async fn is_closed(&self) -> bool {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.is_closed().await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.is_closed(),
+        }
+    }
+
+    /// Read a range without moving the cursor.
+    pub async fn read_range(&self, range: Range<u64>) -> lance_core::Result<bytes::Bytes> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.read_range(range).await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.read_range(range).await,
+        }
+    }
+
+    /// Read ranges without moving the cursor.
+    pub async fn read_ranges(
+        &self,
+        ranges: &[Range<u64>],
+    ) -> lance_core::Result<Vec<bytes::Bytes>> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.read_ranges(ranges).await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.read_ranges(ranges).await,
+        }
+    }
+
+    /// Read from the cursor to the end.
+    pub async fn read(&self) -> lance_core::Result<bytes::Bytes> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.read().await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.read().await,
+        }
+    }
+
+    /// Read up to `len` bytes and advance the cursor.
+    pub async fn read_up_to(&self, len: usize) -> lance_core::Result<bytes::Bytes> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.read_up_to(len).await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.read_up_to(len).await,
+        }
+    }
+
+    /// Move the cursor to `new_cursor`.
+    pub async fn seek(&self, new_cursor: u64) -> lance_core::Result<()> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.seek(new_cursor).await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.seek(new_cursor).await,
+        }
+    }
+
+    /// Current cursor position.
+    pub async fn tell(&self) -> lance_core::Result<u64> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.tell().await,
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.tell().await,
+        }
+    }
+
+    /// Blob length in bytes.
+    pub fn size(&self) -> u64 {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.size(),
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(file) => file.size(),
+        }
+    }
+
+    /// Physical byte offset in the data file. `None` on remote handles. The
+    /// Cloud byte-range route does not expose storage layout.
+    pub fn position(&self) -> Option<u64> {
+        match &self.inner {
+            BlobFileInner::Native(file) => Some(file.position()),
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(_) => None,
+        }
+    }
+
+    /// Path of the data file holding the blob. `None` on remote handles. The
+    /// Cloud byte-range route does not expose storage layout.
+    pub fn data_path(&self) -> Option<&Path> {
+        match &self.inner {
+            BlobFileInner::Native(file) => Some(file.data_path()),
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(_) => None,
+        }
+    }
+
+    /// Native storage layout. `None` on remote handles. The Cloud byte-range
+    /// route does not expose layout.
+    pub fn kind(&self) -> Option<lance_core::datatypes::BlobKind> {
+        match &self.inner {
+            BlobFileInner::Native(file) => Some(file.kind()),
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(_) => None,
+        }
+    }
+
+    /// External URI for native handles. Remote handles do not expose storage URIs.
+    pub fn uri(&self) -> Option<&str> {
+        match &self.inner {
+            BlobFileInner::Native(file) => file.uri(),
+            #[cfg(feature = "remote")]
+            BlobFileInner::Remote(_) => None,
+        }
+    }
+}

 /// One row-specific blob range read request.
 ///
@@ -264,7 +457,10 @@ pub(crate) async fn take_blob_files_aligned(

    let handles = dataset.take_blobs(row_ids, column).await?;
    ensure_all_row_ids_resolved(column, row_ids.len(), handles.len())?;
-    Ok(handles)
+    Ok(handles
+        .into_iter()
+        .map(|handle| handle.map(Into::into))
+        .collect())
 }

 #[cfg(test)]
@@ -23,8 +23,8 @@ use crate::connection::create_table::CreateTableBuilder;
 use crate::data::scannable::Scannable;
 use crate::database::listing::ListingDatabase;
 use crate::database::{
-    CloneTableRequest, Database, DatabaseOptions, OpenTableRequest, ReadConsistency,
-    TableNamesRequest,
+    CloneTableRequest, Database, DatabaseOptions, JobDescription, JobInfo, OpenTableRequest,
+    ReadConsistency, TableNamesRequest,
 };
 use crate::embeddings::{EmbeddingRegistry, MemoryRegistry};
 use crate::error::{Error, Result};
@@ -456,6 +456,10 @@ impl Connection {
    ///
    /// # Returns
    /// Created [`TableRef`], or [`Error::TableNotFound`] if the table does not exist.
+    /// If the table's storage is present but holds no readable dataset (for example a
+    /// `<name>.lance` directory left behind by an interrupted drop and re-create, which
+    /// [`Self::table_names`] still lists) this returns [`Error::TableCorrupted`]
+    /// instead.
    pub fn open_table(&self, name: impl Into<String>) -> OpenTableBuilder {
        OpenTableBuilder::new(
            self.internal.clone(),
@@ -513,6 +517,39 @@ impl Connection {
        self.internal.read_consistency().await
    }

+    /// A [`crate::job::Job`] handle for a server-side job by id, suitable for
+    /// waiting on or cancelling the job.
+    ///
+    /// The handle is constructed without a server round trip; an unknown id
+    /// surfaces when the handle is used. Only server-backed databases support
+    /// job handles by id.
+    pub fn job(&self, job_id: impl AsRef<str>) -> Result<crate::job::Job> {
+        self.internal.job(job_id.as_ref())
+    }
+
+    /// List server-side jobs across the database's tables.
+    pub async fn list_jobs(&self) -> Result<Vec<JobInfo>> {
+        self.internal.list_jobs().await
+    }
+
+    /// Describe a single server-side job by id. `None` when the server has no
+    /// such job.
+    pub async fn get_job(&self, job_id: impl AsRef<str>) -> Result<Option<JobDescription>> {
+        self.internal.get_job(job_id.as_ref()).await
+    }
+
+    /// Request cancellation of a server-side job by id. Returns true if the
+    /// server accepted the cancellation, false if no such job exists.
+    pub async fn cancel_job(&self, job_id: impl AsRef<str>) -> Result<bool> {
+        self.internal.cancel_job(job_id.as_ref()).await
+    }
+
+    /// The lifecycle event history of a server-side job (all jobs when
+    /// `job_id` is `None`), as recorded Arrow batches.
+    pub async fn job_history(&self, job_id: Option<&str>) -> Result<Vec<RecordBatch>> {
+        self.internal.job_history(job_id).await
+    }
+
    /// Drop a table in the database.
    ///
    /// # Arguments
@@ -202,6 +202,17 @@ mod tests {
        assert_eq!(table.count_rows(None).await.unwrap(), 0);
    }

+    #[tokio::test]
+    async fn create_table_in_named_memory_database() {
+        let db = connect("memory://foo").execute().await.unwrap();
+        let batch = record_batch!(("id", Int64, [1, 2, 3])).unwrap();
+
+        let table = db.create_table("my_table", batch).execute().await.unwrap();
+
+        assert_eq!(table.uri().await.unwrap(), "memory://foo/my_table.lance");
+        assert_eq!(table.count_rows(None).await.unwrap(), 3);
+    }
+
    async fn test_create_table_with_data<T>(data: T)
    where
        T: Scannable + 'static,
@@ -18,6 +18,8 @@ use std::collections::HashMap;
 use std::sync::Arc;
 use std::time::Duration;

+use arrow_array::RecordBatch;
+
 use lance::dataset::ReadParams;
 use lance_namespace::LanceNamespace;
 use lance_namespace::models::{
@@ -200,6 +202,45 @@ pub enum ReadConsistency {
    Strong,
 }

+/// A row from [`Database::list_jobs`]: one server-side job (index build,
+/// compaction, column refresh, ...).
+#[derive(Debug, Clone)]
+pub struct JobInfo {
+    /// The job id -- what [`Database::get_job`] and [`Database::cancel_job`]
+    /// accept.
+    pub job_id: String,
+    /// The table the job runs against, without URI or namespace.
+    pub table: String,
+    pub job_type: String,
+    /// Lifecycle state: "running", "finished", "failed", or "cancelled".
+    pub state: String,
+    /// When the job was created, in milliseconds since the epoch.
+    pub created_at_millis: i64,
+}
+
+/// A described job from [`Database::get_job`]: lifecycle state plus the
+/// job-type-specific specification.
+#[derive(Debug, Clone)]
+pub struct JobDescription {
+    pub job_id: String,
+    pub job_type: String,
+    /// Lifecycle state: "running", "finished", "failed", or "cancelled".
+    pub state: String,
+    /// When the job was created, in milliseconds since the epoch.
+    pub creation_ms: i64,
+    /// The job-type-specific specification. Null when the server omits it.
+    pub spec: serde_json::Value,
+    /// Why the job failed, when the job is failed and the server reports a
+    /// reason.
+    pub failure: Option<crate::error::JobFailure>,
+}
+
+fn job_op_not_supported<T>(what: &str) -> Result<T> {
+    Err(crate::error::Error::NotSupported {
+        message: format!("{} is not supported by this database", what),
+    })
+}
+
 /// The `Database` trait defines the interface for database implementations.
 ///
 /// A database is responsible for managing tables and their metadata.
@@ -245,6 +286,31 @@ pub trait Database:
    ///
    /// See [`CloneTableRequest`] for detailed documentation and examples.
    async fn clone_table(&self, request: CloneTableRequest) -> Result<Arc<dyn BaseTable>>;
+    /// A [`crate::job::Job`] handle for a server-side job by id, suitable for
+    /// waiting on or cancelling the job. The handle is constructed without a
+    /// server round trip; an unknown id surfaces when the handle is used.
+    fn job(&self, _job_id: &str) -> Result<crate::job::Job> {
+        job_op_not_supported("job")
+    }
+    /// List server-side jobs across the database's tables.
+    async fn list_jobs(&self) -> Result<Vec<JobInfo>> {
+        job_op_not_supported("list_jobs")
+    }
+    /// Describe a single job by id. `None` when the server has no such job.
+    async fn get_job(&self, _job_id: &str) -> Result<Option<JobDescription>> {
+        job_op_not_supported("get_job")
+    }
+    /// Request cancellation of a job by id. Returns true if the server
+    /// accepted the cancellation, false if no such job exists. Cancelling an
+    /// already-terminal job is a no-op success.
+    async fn cancel_job(&self, _job_id: &str) -> Result<bool> {
+        job_op_not_supported("cancel_job")
+    }
+    /// The lifecycle event history of a job (all jobs when `job_id` is
+    /// `None`), as recorded Arrow batches.
+    async fn job_history(&self, _job_id: Option<&str>) -> Result<Vec<RecordBatch>> {
+        job_op_not_supported("job_history")
+    }
    /// Open a table in the database
    async fn open_table(&self, request: OpenTableRequest) -> Result<Arc<dyn BaseTable>>;
    /// Rename a table in the database
@@ -1376,6 +1376,68 @@ mod tests {
        assert!(!tempdir.path().join("__manifest").exists());
    }

+    /// Regression test for https://github.com/lancedb/lancedb/issues/1600.
+    ///
+    /// Opening a table used to create a separate object-store client instead of
+    /// reusing the one that successfully connected to the database.  Repeating
+    /// credential discovery made S3 table opens intermittent, especially in AWS
+    /// Lambda, and the failed open was reported as `TableNotFound`.
+    #[tokio::test]
+    async fn test_open_table_reuses_connection_object_store() {
+        let tempdir = tempdir().unwrap();
+        let uri = tempdir.path().to_str().unwrap();
+        let registry = Arc::new(lance_io::object_store::ObjectStoreRegistry::default());
+        let session = Arc::new(lance::session::Session::new(16, 16, registry.clone()));
+
+        let request = ConnectRequest {
+            uri: uri.to_string(),
+            #[cfg(feature = "remote")]
+            client_config: Default::default(),
+            options: Default::default(),
+            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
+            read_consistency_interval: None,
+            session: Some(session),
+        };
+        let db = ListingDatabase::connect_with_options(&request)
+            .await
+            .unwrap();
+
+        let schema = Arc::new(Schema::new(vec![Field::new("id", DataType::Int32, false)]));
+        db.create_table(CreateTableRequest {
+            name: "test".to_string(),
+            namespace_path: vec![],
+            data: Box::new(RecordBatch::new_empty(schema)) as Box<dyn Scannable>,
+            mode: CreateTableMode::Create,
+            write_options: Default::default(),
+            location: None,
+            namespace_client: None,
+        })
+        .await
+        .unwrap();
+
+        let before_open = registry.stats();
+        for _ in 0..3 {
+            let table = db
+                .open_table(OpenTableRequest {
+                    name: "test".to_string(),
+                    namespace_path: vec![],
+                    index_cache_size: None,
+                    lance_read_params: None,
+                    location: None,
+                    namespace_client: None,
+                    managed_versioning: None,
+                })
+                .await
+                .unwrap();
+            assert_eq!(table.count_rows(None).await.unwrap(), 0);
+        }
+
+        let after_open = registry.stats();
+        assert_eq!(after_open.misses, before_open.misses);
+        assert!(after_open.hits >= before_open.hits + 3);
+    }
+
    #[tokio::test]
    async fn test_clone_table_basic() {
        let (_tempdir, db) = setup_database().await;
@@ -11,7 +11,9 @@ use lance_core::{cache::LanceCache, utils::futures::FinallyStreamExt};
 use lance_encoding::decoder::{DecoderPlugins, FilterExpression};
 use lance_file::{
    reader::{FileReader, FileReaderOptions},
-    writer::{FileWriter, FileWriterOptions},
+    version::ConcreteFileVersion,
+    versions,
+    writer::FileWriterOptions,
 };
 use lance_io::{
    ReadBatchParams,
@@ -152,8 +154,12 @@ impl Shuffler {
                    source: None,
                })?;
            let object_writer = object_store.create(&path).await?;
-            let writer =
-                FileWriter::try_new(object_writer, schema.clone(), FileWriterOptions::default())?;
+            let writer = versions::create_writer(
+                ConcreteFileVersion::V2_1,
+                object_writer,
+                schema.clone(),
+                FileWriterOptions::default(),
+            )?;
            file_writers.push(writer);
        }

@@ -264,7 +264,7 @@ pub fn compute_output_schema(
        let field_name = ed
            .dest_column
            .clone()
-            .unwrap_or_else(|| format!("{}_embedding", &ed.source_column));
+            .unwrap_or_else(|| format!("{}_embedding", ed.source_column));

        sb.push(Field::new(
            field_name,
@@ -291,7 +291,7 @@ pub fn compute_embeddings_for_batch(
        let dst_field_name = fld
            .dest_column
            .clone()
-            .unwrap_or_else(|| format!("{}_embedding", &fld.source_column));
+            .unwrap_or_else(|| format!("{}_embedding", fld.source_column));

        let dst_field = Field::new(
            dst_field_name,
@@ -315,7 +315,7 @@ impl<R: RecordBatchReader> WithEmbeddings<R> {
                let field_name = ed
                    .dest_column
                    .clone()
-                    .unwrap_or_else(|| format!("{}_embedding", &ed.source_column));
+                    .unwrap_or_else(|| format!("{}_embedding", ed.source_column));
                Ok(Field::new(
                    field_name,
                    func.dest_type()?.into_owned(),
@@ -1,7 +1,8 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

-use std::sync::PoisonError;
+use std::fmt::{self, Display, Formatter};
+use std::sync::{Arc, PoisonError};

 use arrow_schema::ArrowError;
 use datafusion_common::DataFusionError;
@@ -9,6 +10,46 @@ use snafu::Snafu;

 pub(crate) type BoxError = Box<dyn std::error::Error + Send + Sync>;

+/// Why a job failed, to whatever precision the backend provides.
+///
+/// A job run in this process carries the error it failed with in [`Self::source`].
+/// A job run remotely carries whatever the server reported, which older servers
+/// do not report at all. Every field is absent rather than invented when the
+/// backend does not supply it.
+#[derive(Debug, Clone, Default)]
+pub struct JobFailure {
+    /// The stage the job was in, when known.
+    pub phase: Option<String>,
+    /// A human-readable reason, when known.
+    pub message: Option<String>,
+    /// Whether a retry could clear the failure, when known.
+    pub retryable: Option<bool>,
+    /// The error the job failed with, when it ran in this process.
+    pub source: Option<Arc<Error>>,
+}
+
+impl JobFailure {
+    /// A failure whose only known detail is the error that caused it.
+    pub(crate) fn from_source(source: Arc<Error>) -> Self {
+        Self {
+            message: Some(source.to_string()),
+            source: Some(source),
+            ..Default::default()
+        }
+    }
+}
+
+impl Display for JobFailure {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        match (&self.message, &self.phase) {
+            (Some(message), Some(phase)) => write!(f, ": {message} (in {phase})"),
+            (Some(message), None) => write!(f, ": {message}"),
+            (None, Some(phase)) => write!(f, " in {phase}"),
+            (None, None) => Ok(()),
+        }
+    }
+}
+
 #[derive(Debug, Snafu)]
 #[snafu(visibility(pub(crate)))]
 pub enum Error {
@@ -18,6 +59,10 @@ pub enum Error {
    InvalidInput { message: String },
    #[snafu(display("Table '{name}' was not found"))]
    TableNotFound { name: String, source: BoxError },
+    #[snafu(display(
+        "Table '{name}' exists but could not be loaded (it may be corrupt or incomplete): {source}"
+    ))]
+    TableCorrupted { name: String, source: BoxError },
    #[snafu(display("Database '{name}' was not found"))]
    DatabaseNotFound { name: String },
    #[snafu(display("Database '{name}' already exists."))]
@@ -40,6 +85,13 @@ pub enum Error {
    Runtime { message: String },
    #[snafu(display("Timeout error: {message}"))]
    Timeout { message: String },
+    #[snafu(display("Job{} failed{failure}", job_id.as_ref().map(|id| format!(" {id}")).unwrap_or_default()))]
+    JobFailed {
+        job_id: Option<String>,
+        failure: JobFailure,
+    },
+    #[snafu(display("Job{} was cancelled", job_id.as_ref().map(|id| format!(" {id}")).unwrap_or_default()))]
+    JobCancelled { job_id: Option<String> },

    // 3rd party / external errors
    #[snafu(display("object_store error: {source}"))]
@@ -121,6 +173,9 @@ impl From<lance::Error> for Error {
        match source {
            lance::Error::Wrapped { error, .. } => Self::from_box_error(error),
            lance::Error::External { source } => Self::from_box_error(source),
+            lance::Error::InvalidInput { source, .. } => Self::InvalidInput {
+                message: source.to_string(),
+            },
            _ => Self::Lance { source },
        }
    }
@@ -156,7 +156,7 @@ mod tests {
        use datafusion_common::ScalarValue;
        let expr = col("data").eq(lit(ScalarValue::Binary(Some(vec![0xca, 0xfe]))));
        let sql = expr_to_sql_string(&expr).unwrap();
-        assert_eq!(sql, "(data = X'CAFE')");
+        assert_eq!(sql, "(`data` = X'CAFE')");
    }

    #[test]
@@ -166,7 +166,7 @@ mod tests {
        let int_expr = col("id").gt(lit(5i64));
        let combined = bin_expr.and(int_expr);
        let sql = expr_to_sql_string(&combined).unwrap();
-        assert_eq!(sql, "((data = X'01') AND (id > 5))");
+        assert_eq!(sql, "((`data` = X'01') AND (id > 5))");
    }

    #[test]
@@ -184,7 +184,7 @@ mod tests {
        // serialized correctly (regression test for placeholder rewrite path).
        let expr = contains(col("data"), lit(ScalarValue::Binary(Some(vec![0xff]))));
        let sql = expr_to_sql_string(&expr).unwrap();
-        assert_eq!(sql, "contains(data, X'FF')");
+        assert_eq!(sql, "contains(`data`, X'FF')");
    }

    #[test]
@@ -195,7 +195,7 @@ mod tests {
            .eq(lit(ScalarValue::Binary(Some(vec![0xab, 0xcd]))))
            .not();
        let sql = expr_to_sql_string(&expr).unwrap();
-        assert_eq!(sql, "NOT (data = X'ABCD')");
+        assert_eq!(sql, "NOT (`data` = X'ABCD')");
    }

    #[test]
@@ -205,6 +205,122 @@ mod tests {
        assert!(sql.contains("IN"), "expected IN in: {}", sql);
    }

+    #[test]
+    fn test_empty_is_in() {
+        let expr = is_in(col("id"), vec![]);
+        assert_eq!(expr_to_sql_string(&expr).unwrap(), "false");
+    }
+
+    #[test]
+    fn test_empty_is_in_discards_binary_children() {
+        use datafusion_common::ScalarValue;
+
+        let expr = is_in(
+            col("payload").eq(lit(ScalarValue::Binary(Some(vec![0x01])))),
+            vec![],
+        );
+        assert_eq!(expr_to_sql_string(&expr).unwrap(), "false");
+    }
+
+    #[test]
+    fn test_keyword_identifier() {
+        let expr = col("null").eq(lit(1i64));
+        assert_eq!(expr_to_sql_string(&expr).unwrap(), "(`null` = 1)");
+    }
+
+    #[test]
+    fn test_decimal_literal_preserves_type() {
+        use datafusion_common::ScalarValue;
+
+        let expr = col("val").lt(lit(ScalarValue::Decimal128(
+            Some(1_234_567_890_123_456_790),
+            19,
+            18,
+        )));
+        let sql = expr_to_sql_string(&expr).unwrap();
+        assert_eq!(
+            sql,
+            "(val < arrow_cast('1.234567890123456790', 'Decimal128(19, 18)'))"
+        );
+    }
+
+    #[test]
+    fn test_non_finite_float_literal_preserves_type() {
+        let expr = col("x").lt(lit(f64::INFINITY));
+        assert_eq!(
+            expr_to_sql_string(&expr).unwrap(),
+            "(x < arrow_cast('inf', 'Float64'))"
+        );
+    }
+
+    #[test]
+    fn test_cast_uses_arrow_type_name() {
+        let string = expr_cast(col("x"), DataType::Utf8);
+        assert_eq!(
+            expr_to_sql_string(&string).unwrap(),
+            "arrow_cast(x, 'Utf8')"
+        );
+
+        let int32 = expr_cast(col("x"), DataType::Int32);
+        assert_eq!(
+            expr_to_sql_string(&int32).unwrap(),
+            "arrow_cast(x, 'Int32')"
+        );
+
+        let expr = expr_cast(col("x"), DataType::Float16).lt(lit(2.0));
+        assert_eq!(
+            expr_to_sql_string(&expr).unwrap(),
+            "(arrow_cast(x, 'Float16') < 2.0)"
+        );
+
+        let decimal = expr_cast(lit("2.00"), DataType::Decimal256(40, 2));
+        assert_eq!(
+            expr_to_sql_string(&decimal).unwrap(),
+            "arrow_cast('2.00', 'Decimal256(40, 2)')"
+        );
+    }
+
+    #[test]
+    fn test_binary_placeholder_does_not_rewrite_user_string() {
+        use datafusion_common::ScalarValue;
+
+        let marker = "__lancedb_binary_placeholder_0__";
+        let expr = col("payload")
+            .eq(lit(ScalarValue::Binary(Some(vec![0x01]))))
+            .or(col("text").eq(lit(marker)));
+        assert_eq!(
+            expr_to_sql_string(&expr).unwrap(),
+            "((payload = X'01') OR (`text` = '__lancedb_binary_placeholder_0__'))"
+        );
+    }
+
+    #[test]
+    fn test_binary_binding_skips_quoted_identifiers() {
+        use datafusion_common::ScalarValue;
+
+        let expr = col("payload")
+            .eq(lit(ScalarValue::Binary(Some(vec![0x01]))))
+            .and(col("odd'name").eq(lit(1i64)))
+            .and(col("odd`'name").eq(lit(2i64)));
+        assert_eq!(
+            expr_to_sql_string(&expr).unwrap(),
+            "(((payload = X'01') AND (`odd'name` = 1)) AND (`odd``'name` = 2))"
+        );
+    }
+
+    #[test]
+    fn test_binary_placeholder_collision_search_is_linear() {
+        use datafusion_common::ScalarValue;
+
+        let collision_shaped = format!("__lancedb_binary_placeholder_0__{}", "_".repeat(64_000));
+        let expr = col("payload")
+            .eq(lit(ScalarValue::Binary(Some(vec![0x01]))))
+            .and(col("text").eq(lit(collision_shaped.clone())));
+        let sql = expr_to_sql_string(&expr).unwrap();
+        assert!(sql.contains("X'01'"));
+        assert!(sql.contains(&format!("'{collision_shaped}'")));
+    }
+
    #[test]
    fn test_multiple_binary_literals() {
        use datafusion_common::ScalarValue;
@@ -1,9 +1,19 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

+use std::collections::{HashMap, HashSet};
+
+use arrow_array::types::{
+    Decimal32Type, Decimal64Type, Decimal128Type, Decimal256Type, DecimalType,
+};
+use arrow_schema::DataType;
 use datafusion_common::ScalarValue;
 use datafusion_common::tree_node::{Transformed, TreeNode, TreeNodeRecursion};
 use datafusion_expr::Expr;
+use datafusion_functions::core::expr_fn::{
+    arrow_cast as datafusion_arrow_cast, arrow_try_cast as datafusion_arrow_try_cast,
+};
+use datafusion_sql::sqlparser::keywords::ALL_KEYWORDS;
 use datafusion_sql::unparser::{self, dialect::Dialect};

 /// Unparser dialect that matches the quoting style expected by the Lance SQL
@@ -21,11 +31,13 @@ struct LanceSqlDialect;

 impl Dialect for LanceSqlDialect {
    fn identifier_quote_style(&self, identifier: &str) -> Option<char> {
-        let needs_quote = identifier.chars().any(|c| c.is_ascii_uppercase())
-            || !identifier
-                .chars()
-                .enumerate()
-                .all(|(i, c)| c == '_' || c.is_ascii_alphabetic() || (i > 0 && c.is_ascii_digit()));
+        let identifier_upper = identifier.to_ascii_uppercase();
+        let needs_quote =
+            (identifier_upper != "ID" && ALL_KEYWORDS.contains(&identifier_upper.as_str()))
+                || identifier.chars().any(|c| c.is_ascii_uppercase())
+                || !identifier.chars().enumerate().all(|(i, c)| {
+                    c == '_' || c.is_ascii_alphabetic() || (i > 0 && c.is_ascii_digit())
+                });
        if needs_quote { Some('`') } else { None }
    }
 }
@@ -39,24 +51,128 @@ fn bytes_to_hex_sql(bytes: &[u8]) -> String {
    format!("X'{hex}'")
 }

-/// Returns true if *expr* contains a `Binary` or `LargeBinary` scalar literal
-/// anywhere in its subtree.  DataFusion's SQL unparser cannot serialize those
-/// variants, so we route such expressions through a placeholder-substitution
-/// path that emits SQL `X'...'` byte-string literals.
-fn has_binary_literal(expr: &Expr) -> bool {
-    let mut found = false;
+fn string_literals(expr: &Expr) -> HashSet<String> {
+    let mut literals = HashSet::new();
    let _ = expr.apply(&mut |e: &Expr| {
-        if matches!(
-            e,
-            Expr::Literal(ScalarValue::Binary(_) | ScalarValue::LargeBinary(_), _)
-        ) {
-            found = true;
-            Ok(TreeNodeRecursion::Stop)
-        } else {
-            Ok(TreeNodeRecursion::Continue)
+        if let Expr::Literal(
+            ScalarValue::Utf8(Some(value))
+            | ScalarValue::LargeUtf8(Some(value))
+            | ScalarValue::Utf8View(Some(value)),
+            _,
+        ) = e
+        {
+            literals.insert(value.clone());
        }
+        Ok(TreeNodeRecursion::Continue)
    });
-    found
+    literals
+}
+
+fn typed_string_literal(value: String, data_type: DataType) -> Expr {
+    datafusion_arrow_cast(
+        Expr::Literal(ScalarValue::Utf8(Some(value)), None),
+        Expr::Literal(ScalarValue::Utf8(Some(data_type.to_string())), None),
+    )
+}
+
+fn next_binary_placeholder(user_strings: &HashSet<String>, next_id: &mut usize) -> String {
+    loop {
+        let placeholder = format!("{BINARY_PLACEHOLDER_PREFIX}{}__", *next_id);
+        *next_id += 1;
+        if !user_strings.contains(&placeholder) {
+            return placeholder;
+        }
+    }
+}
+
+fn bind_binary_literals(
+    sql: &str,
+    mut bindings: HashMap<String, Vec<u8>>,
+) -> crate::Result<String> {
+    let bytes = sql.as_bytes();
+    let mut output = Vec::with_capacity(bytes.len());
+    let mut index = 0;
+
+    // Walk SQL string tokens once. Placeholders are plain, unescaped string
+    // literals, so this remains linear even when user strings are large or
+    // deliberately resemble the placeholder prefix.
+    while index < bytes.len() {
+        if bytes[index] == b'`' {
+            let identifier_start = index;
+            index += 1;
+            let mut identifier_end = None;
+            while index < bytes.len() {
+                if bytes[index] == b'`' {
+                    if index + 1 < bytes.len() && bytes[index + 1] == b'`' {
+                        index += 2;
+                    } else {
+                        index += 1;
+                        identifier_end = Some(index);
+                        break;
+                    }
+                } else {
+                    index += 1;
+                }
+            }
+
+            let Some(identifier_end) = identifier_end else {
+                return Err(crate::Error::InvalidInput {
+                    message: "unterminated identifier while binding binary literal".to_string(),
+                });
+            };
+            output.extend_from_slice(&bytes[identifier_start..identifier_end]);
+            continue;
+        }
+
+        if bytes[index] != b'\'' {
+            output.push(bytes[index]);
+            index += 1;
+            continue;
+        }
+
+        let literal_start = index;
+        index += 1;
+        let content_start = index;
+        let mut escaped = false;
+        let mut content_end = None;
+        while index < bytes.len() {
+            if bytes[index] == b'\'' {
+                if index + 1 < bytes.len() && bytes[index + 1] == b'\'' {
+                    escaped = true;
+                    index += 2;
+                } else {
+                    content_end = Some(index);
+                    index += 1;
+                    break;
+                }
+            } else {
+                index += 1;
+            }
+        }
+
+        let Some(content_end) = content_end else {
+            return Err(crate::Error::InvalidInput {
+                message: "unterminated string while binding binary literal".to_string(),
+            });
+        };
+
+        let placeholder = &sql[content_start..content_end];
+        if !escaped && let Some(value) = bindings.remove(placeholder) {
+            output.extend_from_slice(bytes_to_hex_sql(&value).as_bytes());
+        } else {
+            output.extend_from_slice(&bytes[literal_start..index]);
+        }
+    }
+
+    if !bindings.is_empty() {
+        return Err(crate::Error::InvalidInput {
+            message: "failed to bind binary literal while serializing expression".to_string(),
+        });
+    }
+
+    String::from_utf8(output).map_err(|e| crate::Error::InvalidInput {
+        message: format!("failed to bind binary literal: {e}"),
+    })
 }

 fn run_unparser(expr: &Expr) -> crate::Result<String> {
@@ -69,25 +185,37 @@ fn run_unparser(expr: &Expr) -> crate::Result<String> {
 }

 pub fn expr_to_sql_string(expr: &Expr) -> crate::Result<String> {
-    // Fast path: no binary literals — DataFusion's unparser handles everything.
-    if !has_binary_literal(expr) {
-        return run_unparser(expr);
-    }
-
-    // Slow path: DataFusion's unparser cannot serialize `Binary`/`LargeBinary`
-    // scalars, so we rewrite each one to a unique string-literal placeholder,
-    // let the unparser do the rest of the work, then substitute the SQL
-    // `X'...'` byte-string literal back in.  This keeps the operator/function
-    // serialization logic centralized in DataFusion and works for every
-    // expression node type the unparser supports.
-    let mut bindings: Vec<Vec<u8>> = Vec::new();
+    // DataFusion's unparser needs a few adaptations before its SQL can be
+    // reparsed by Lance without changing the typed expression's semantics:
+    //
+    // * decimal literals need an explicit cast to preserve precision and scale;
+    // * casts need exact Arrow type names rather than SQL type aliases;
+    // * an empty IN list is valid in DataFusion but invalid SQL;
+    // * binary literals are unsupported by the unparser and need placeholders.
+    // Eliminate empty membership expressions before visiting their children.
+    // Otherwise a discarded binary child could leave behind a stale binding.
    let rewritten = expr
        .clone()
+        .transform(|e: Expr| match e {
+            Expr::InList(in_list) if in_list.list.is_empty() => Ok(Transformed::yes(
+                Expr::Literal(ScalarValue::Boolean(Some(in_list.negated)), None),
+            )),
+            other => Ok(Transformed::no(other)),
+        })
+        .map_err(|e| crate::Error::InvalidInput {
+            message: format!("failed to rewrite expression: {e}"),
+        })?
+        .data;
+
+    let user_strings = string_literals(&rewritten);
+    let mut next_placeholder_id = 0;
+    let mut binary_bindings = HashMap::new();
+    let rewritten = rewritten
        .transform(|e: Expr| match e {
            Expr::Literal(ScalarValue::Binary(Some(bytes)), m)
            | Expr::Literal(ScalarValue::LargeBinary(Some(bytes)), m) => {
-                let placeholder = format!("{}{}__", BINARY_PLACEHOLDER_PREFIX, bindings.len());
-                bindings.push(bytes);
+                let placeholder = next_binary_placeholder(&user_strings, &mut next_placeholder_id);
+                binary_bindings.insert(placeholder.clone(), bytes);
                Ok(Transformed::yes(Expr::Literal(
                    ScalarValue::Utf8(Some(placeholder)),
                    m,
@@ -97,6 +225,57 @@ pub fn expr_to_sql_string(expr: &Expr) -> crate::Result<String> {
            | Expr::Literal(ScalarValue::LargeBinary(None), m) => {
                Ok(Transformed::yes(Expr::Literal(ScalarValue::Null, m)))
            }
+            Expr::Literal(ScalarValue::Decimal32(Some(value), precision, scale), _m) => {
+                let value = Decimal32Type::format_decimal(value, precision, scale);
+                Ok(Transformed::yes(typed_string_literal(
+                    value,
+                    DataType::Decimal32(precision, scale),
+                )))
+            }
+            Expr::Literal(ScalarValue::Decimal64(Some(value), precision, scale), _m) => {
+                let value = Decimal64Type::format_decimal(value, precision, scale);
+                Ok(Transformed::yes(typed_string_literal(
+                    value,
+                    DataType::Decimal64(precision, scale),
+                )))
+            }
+            Expr::Literal(ScalarValue::Decimal128(Some(value), precision, scale), _m) => {
+                let value = Decimal128Type::format_decimal(value, precision, scale);
+                Ok(Transformed::yes(typed_string_literal(
+                    value,
+                    DataType::Decimal128(precision, scale),
+                )))
+            }
+            Expr::Literal(ScalarValue::Decimal256(Some(value), precision, scale), _m) => {
+                let value = Decimal256Type::format_decimal(value, precision, scale);
+                Ok(Transformed::yes(typed_string_literal(
+                    value,
+                    DataType::Decimal256(precision, scale),
+                )))
+            }
+            Expr::Literal(ScalarValue::Float16(Some(value)), _m) if !value.is_finite() => Ok(
+                Transformed::yes(typed_string_literal(value.to_string(), DataType::Float16)),
+            ),
+            Expr::Literal(ScalarValue::Float32(Some(value)), _m) if !value.is_finite() => Ok(
+                Transformed::yes(typed_string_literal(value.to_string(), DataType::Float32)),
+            ),
+            Expr::Literal(ScalarValue::Float64(Some(value)), _m) if !value.is_finite() => Ok(
+                Transformed::yes(typed_string_literal(value.to_string(), DataType::Float64)),
+            ),
+            Expr::Cast(cast) => Ok(Transformed::yes(datafusion_arrow_cast(
+                *cast.expr,
+                Expr::Literal(
+                    ScalarValue::Utf8(Some(cast.field.data_type().to_string())),
+                    None,
+                ),
+            ))),
+            Expr::TryCast(cast) => Ok(Transformed::yes(datafusion_arrow_try_cast(
+                *cast.expr,
+                Expr::Literal(
+                    ScalarValue::Utf8(Some(cast.field.data_type().to_string())),
+                    None,
+                ),
+            ))),
            other => Ok(Transformed::no(other)),
        })
        .map_err(|e| crate::Error::InvalidInput {
@@ -104,12 +283,10 @@ pub fn expr_to_sql_string(expr: &Expr) -> crate::Result<String> {
        })?
        .data;

-    let mut sql = run_unparser(&rewritten)?;
-    for (i, bytes) in bindings.iter().enumerate() {
-        // The unparser quotes string literals with single quotes, so the
-        // placeholder appears as `'__lancedb_binary_placeholder_<i>__'`.
-        let quoted = format!("'{}{}__'", BINARY_PLACEHOLDER_PREFIX, i);
-        sql = sql.replace(&quoted, &bytes_to_hex_sql(bytes));
+    let sql = run_unparser(&rewritten)?;
+    if binary_bindings.is_empty() {
+        Ok(sql)
+    } else {
+        bind_binary_literals(&sql, binary_bindings)
    }
-    Ok(sql)
 }
@@ -10,7 +10,7 @@ use std::time::Duration;
 use vector::IvfFlatIndexBuilder;

 use crate::index::vector::IvfRqIndexBuilder;
-use crate::{DistanceType, Error, Result, table::BaseTable};
+use crate::{DistanceType, Error, Result, job::Job, table::BaseTable};

 use self::{
    scalar::{BTreeIndexBuilder, BitmapIndexBuilder, FmIndexBuilder, LabelListIndexBuilder},
@@ -305,6 +305,14 @@ impl IndexBuilder {
    pub async fn execute(self) -> Result<()> {
        self.parent.clone().create_index(self).await
    }
+
+    /// Creates the index, returning a [`Job`] tracking the operation.
+    ///
+    /// The job may already be complete when returned, and callers must not
+    /// assume the index exists until [`Job::wait`] resolves.
+    pub async fn execute_async(self) -> Result<Job> {
+        self.parent.clone().create_index_async(self).await
+    }
 }

 #[derive(Debug, Clone, PartialEq, Deserialize)]
@@ -132,9 +132,14 @@ impl ObjectStore for MirroringObjectStore {
        if to.primary_only() {
            self.primary.copy_opts(from, to, options).await
        } else {
-            self.secondary.copy_opts(from, to, options.clone()).await?;
-            self.primary.copy_opts(from, to, options).await?;
-            Ok(())
+            // The secondary store can be process-local and less durable than the
+            // primary, so a source written by another process may not exist here
+            // or may be evicted before the copy begins.
+            match self.secondary.copy_opts(from, to, options.clone()).await {
+                Ok(()) | Err(Error::NotFound { .. }) => {}
+                Err(err) => return Err(err),
+            }
+            self.primary.copy_opts(from, to, options).await
        }
    }
 }
@@ -192,7 +197,8 @@ mod test {
    use futures::TryStreamExt;
    use lance::{dataset::WriteParams, io::ObjectStoreParams};
    use lance_testing::datagen::{BatchGenerator, IncrementingInt32, RandomVector};
-    use object_store::local::LocalFileSystem;
+    use object_store::{local::LocalFileSystem, memory::InMemory};
+    use std::time::Duration;
    use tempfile;

    use crate::{
@@ -201,6 +207,139 @@ mod test {
        table::WriteOptions,
    };

+    #[derive(Debug)]
+    struct EvictBeforeCopyStore {
+        inner: Arc<dyn ObjectStore>,
+    }
+
+    impl std::fmt::Display for EvictBeforeCopyStore {
+        fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+            write!(f, "EvictBeforeCopyStore")
+        }
+    }
+
+    #[async_trait]
+    impl ObjectStore for EvictBeforeCopyStore {
+        async fn put_opts(
+            &self,
+            location: &Path,
+            payload: PutPayload,
+            options: PutOptions,
+        ) -> Result<PutResult> {
+            self.inner.put_opts(location, payload, options).await
+        }
+
+        async fn put_multipart_opts(
+            &self,
+            location: &Path,
+            options: PutMultipartOptions,
+        ) -> Result<Box<dyn MultipartUpload>> {
+            self.inner.put_multipart_opts(location, options).await
+        }
+
+        async fn get_opts(&self, location: &Path, options: GetOptions) -> Result<GetResult> {
+            self.inner.get_opts(location, options).await
+        }
+
+        fn delete_stream(
+            &self,
+            locations: BoxStream<'static, Result<Path>>,
+        ) -> BoxStream<'static, Result<Path>> {
+            self.inner.delete_stream(locations)
+        }
+
+        fn list(&self, prefix: Option<&Path>) -> BoxStream<'static, Result<ObjectMeta>> {
+            self.inner.list(prefix)
+        }
+
+        async fn list_with_delimiter(&self, prefix: Option<&Path>) -> Result<ListResult> {
+            self.inner.list_with_delimiter(prefix).await
+        }
+
+        async fn copy_opts(&self, from: &Path, to: &Path, options: CopyOptions) -> Result<()> {
+            self.inner.delete(from).await?;
+            self.inner.copy_opts(from, to, options).await
+        }
+    }
+
+    #[tokio::test]
+    async fn test_copy_when_source_is_missing_from_secondary() {
+        let primary_dir = tempfile::tempdir().unwrap();
+        let secondary_dir = tempfile::tempdir().unwrap();
+        let primary: Arc<dyn ObjectStore> =
+            Arc::new(LocalFileSystem::new_with_prefix(primary_dir.path()).unwrap());
+        let secondary: Arc<dyn ObjectStore> =
+            Arc::new(LocalFileSystem::new_with_prefix(secondary_dir.path()).unwrap());
+        let store = MirroringObjectStore {
+            primary: primary.clone(),
+            secondary: secondary.clone(),
+        };
+        let staging = Path::from("_versions/1.manifest-staging");
+        let finalized = Path::from("_versions/1.manifest");
+
+        primary
+            .put(&staging, "manifest contents".into())
+            .await
+            .unwrap();
+
+        tokio::time::timeout(Duration::from_secs(5), store.copy(&staging, &finalized))
+            .await
+            .expect("copy should not hang when the secondary source is missing")
+            .unwrap();
+
+        let copied = primary
+            .get(&finalized)
+            .await
+            .unwrap()
+            .bytes()
+            .await
+            .unwrap();
+        assert_eq!(copied, "manifest contents");
+        assert!(matches!(
+            secondary.head(&finalized).await,
+            Err(Error::NotFound { .. })
+        ));
+    }
+
+    #[tokio::test]
+    async fn test_copy_when_secondary_source_disappears_after_head() {
+        let primary: Arc<dyn ObjectStore> = Arc::new(InMemory::new());
+        let secondary_inner: Arc<dyn ObjectStore> = Arc::new(InMemory::new());
+        let secondary: Arc<dyn ObjectStore> = Arc::new(EvictBeforeCopyStore {
+            inner: secondary_inner.clone(),
+        });
+        let store = MirroringObjectStore {
+            primary: primary.clone(),
+            secondary,
+        };
+        let staging = Path::from("_versions/1.manifest-staging");
+        let finalized = Path::from("_versions/1.manifest");
+
+        primary
+            .put(&staging, "manifest contents".into())
+            .await
+            .unwrap();
+        secondary_inner
+            .put(&staging, "manifest contents".into())
+            .await
+            .unwrap();
+
+        store.copy(&staging, &finalized).await.unwrap();
+
+        let copied = primary
+            .get(&finalized)
+            .await
+            .unwrap()
+            .bytes()
+            .await
+            .unwrap();
+        assert_eq!(copied, "manifest contents");
+        assert!(matches!(
+            secondary_inner.head(&finalized).await,
+            Err(Error::NotFound { .. })
+        ));
+    }
+
    // This test is ignored because lance 3.0 introduced LocalWriter optimization
    // that bypasses the object store wrapper for local writes. The mirroring feature
    // still works for remote/cloud storage, but can't be tested with local storage.
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Handles to operations a server may run asynchronously.
+
+use std::sync::Arc;
+
+use async_trait::async_trait;
+use tokio::sync::watch;
+use tokio::task::{AbortHandle, JoinHandle};
+
+use crate::error::{Error, JobFailure, Result};
+
+/// Backend-specific tracking for an asynchronous operation.
+#[async_trait]
+pub(crate) trait JobHandle: Send + Sync {
+    /// Server-assigned id, when the backend has one.
+    fn id(&self) -> Option<&str> {
+        None
+    }
+    async fn status(&self) -> Result<String>;
+    async fn wait(&self) -> Result<()>;
+    async fn cancel(&self) -> Result<()>;
+}
+
+/// A handle to an operation that may still be running.
+///
+/// The operation may already be complete when the handle is created.
+pub struct Job {
+    handle: Option<Box<dyn JobHandle>>,
+}
+
+impl std::fmt::Debug for Job {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("Job")
+            .field("id", &self.id())
+            .field("done", &self.handle.is_none())
+            .finish()
+    }
+}
+
+impl Job {
+    /// A job whose operation finished before the handle was created.
+    pub(crate) fn new_done() -> Self {
+        Self { handle: None }
+    }
+
+    pub(crate) fn new(handle: Box<dyn JobHandle>) -> Self {
+        Self {
+            handle: Some(handle),
+        }
+    }
+
+    /// A job running as a task in this process.
+    pub(crate) fn spawned(task: JoinHandle<Result<()>>) -> Self {
+        Self::new(Box::new(SpawnedJob::new(task)))
+    }
+
+    /// Identifies the operation on the server that is running it.
+    ///
+    /// Returned for correlating with server logs or the jobs API. Operations
+    /// that run in this process have no server id and return `None`. The
+    /// value is opaque: parsing it or storing it to resume the job later is
+    /// not supported.
+    pub fn id(&self) -> Option<&str> {
+        self.handle.as_ref().and_then(|handle| handle.id())
+    }
+
+    /// The operation's current lifecycle state: "running", "finished",
+    /// "failed", or "cancelled".
+    ///
+    /// A point snapshot; unlike [`Job::wait`] it does not block, raise on a
+    /// terminal failure state, or retry. States a newer server reports that
+    /// this client version does not know pass through as-is.
+    pub async fn status(&self) -> Result<String> {
+        match &self.handle {
+            None => Ok("finished".to_string()),
+            Some(handle) => handle.status().await,
+        }
+    }
+
+    /// Waits until the operation reaches a terminal state.
+    ///
+    /// Returns [`crate::Error::JobFailed`] if the operation failed and
+    /// [`crate::Error::JobCancelled`] if it was cancelled.
+    pub async fn wait(&self) -> Result<()> {
+        match &self.handle {
+            None => Ok(()),
+            Some(handle) => handle.wait().await,
+        }
+    }
+
+    /// Requests cancellation of the operation.
+    ///
+    /// Cancelling an operation that already finished is a no-op.
+    pub async fn cancel(&self) -> Result<()> {
+        match &self.handle {
+            None => Ok(()),
+            Some(handle) => handle.cancel().await,
+        }
+    }
+}
+
+/// How an in-process operation ended. Cloneable so every waiter can be given
+/// the outcome; [`Error`] is not, so failures share one behind an [`Arc`].
+#[derive(Clone)]
+enum Outcome {
+    Succeeded,
+    Failed(Arc<Error>),
+    Cancelled,
+}
+
+impl Outcome {
+    fn into_result(self) -> Result<()> {
+        match self {
+            Self::Succeeded => Ok(()),
+            Self::Failed(source) => Err(Error::JobFailed {
+                job_id: None,
+                failure: JobFailure::from_source(source),
+            }),
+            Self::Cancelled => Err(Error::JobCancelled { job_id: None }),
+        }
+    }
+}
+
+/// Tracks an operation running as a task in this process. A second task
+/// watches the first so that aborting it still produces an outcome, and so
+/// that every caller of `wait` observes the same one.
+struct SpawnedJob {
+    outcome: watch::Receiver<Option<Outcome>>,
+    abort: AbortHandle,
+}
+
+impl SpawnedJob {
+    fn new(task: JoinHandle<Result<()>>) -> Self {
+        let abort = task.abort_handle();
+        let (tx, outcome) = watch::channel(None);
+        tokio::spawn(async move {
+            let outcome = match task.await {
+                Ok(Ok(())) => Outcome::Succeeded,
+                Ok(Err(err)) => Outcome::Failed(Arc::new(err)),
+                Err(err) if err.is_cancelled() => Outcome::Cancelled,
+                Err(err) => Outcome::Failed(Arc::new(Error::Runtime {
+                    message: format!("index job task failed: {err}"),
+                })),
+            };
+            let _ = tx.send(Some(outcome));
+        });
+        Self { outcome, abort }
+    }
+}
+
+#[async_trait]
+impl JobHandle for SpawnedJob {
+    async fn status(&self) -> Result<String> {
+        let label = match &*self.outcome.borrow() {
+            None => "running",
+            Some(Outcome::Succeeded) => "finished",
+            Some(Outcome::Failed(_)) => "failed",
+            Some(Outcome::Cancelled) => "cancelled",
+        };
+        Ok(label.to_string())
+    }
+
+    async fn wait(&self) -> Result<()> {
+        let mut outcome = self.outcome.clone();
+        let settled = outcome
+            .wait_for(|outcome| outcome.is_some())
+            .await
+            .map_err(|_| Error::Runtime {
+                message: "index job outcome was dropped before it completed".to_string(),
+            })?
+            .clone()
+            .expect("wait_for returns once an outcome is set");
+        settled.into_result()
+    }
+
+    async fn cancel(&self) -> Result<()> {
+        self.abort.abort();
+        Ok(())
+    }
+}
@@ -184,6 +184,7 @@ pub mod expr;
 pub mod index;
 pub mod io;
 pub mod ipc;
+pub mod job;
 #[cfg(feature = "metrics-otel")]
 pub mod metrics_otel;
 #[cfg(feature = "polars")]
@@ -203,7 +204,8 @@ use serde::{Deserialize, Serialize};

 pub use blob::{BlobRangeRequest, blob, is_blob};
 pub use connection::{ConnectNamespaceBuilder, Connection};
-pub use error::{Error, Result};
+pub use error::{Error, JobFailure, Result};
+pub use job::Job;
 use lance_index::vector::ApproxMode as LanceApproxMode;
 use lance_linalg::distance::DistanceType as LanceDistanceType;
 /// Re-export of the [`metrics`](https://docs.rs/metrics) crate facade. Enable
@@ -1661,14 +1661,8 @@ mod tests {

    #[tokio::test]
    async fn test_setters_getters() {
-        // TODO: Switch back to memory://foo after https://github.com/lancedb/lancedb/issues/1051
-        // is fixed
-        let tmp_dir = tempdir().unwrap();
-        let dataset_path = tmp_dir.path().join("test.lance");
-        let uri = dataset_path.to_str().unwrap();
-
        let batches = make_test_batches();
-        let conn = connect(uri).execute().await.unwrap();
+        let conn = connect("memory://foo").execute().await.unwrap();
        let table = conn
            .create_table("my_table", batches)
            .execute()
@@ -1763,14 +1757,8 @@ mod tests {

    #[tokio::test]
    async fn test_execute() {
-        // TODO: Switch back to memory://foo after https://github.com/lancedb/lancedb/issues/1051
-        // is fixed
-        let tmp_dir = tempdir().unwrap();
-        let dataset_path = tmp_dir.path().join("test.lance");
-        let uri = dataset_path.to_str().unwrap();
-
        let batches = make_non_empty_batches();
-        let conn = connect(uri).execute().await.unwrap();
+        let conn = connect("memory://foo").execute().await.unwrap();
        let table = conn
            .create_table("my_table", batches)
            .execute()
@@ -1889,14 +1877,8 @@ mod tests {

    #[tokio::test]
    async fn test_select_with_transform() {
-        // TODO: Switch back to memory://foo after https://github.com/lancedb/lancedb/issues/1051
-        // is fixed
-        let tmp_dir = tempdir().unwrap();
-        let dataset_path = tmp_dir.path().join("test.lance");
-        let uri = dataset_path.to_str().unwrap();
-
        let batches = make_non_empty_batches();
-        let conn = connect(uri).execute().await.unwrap();
+        let conn = connect("memory://foo").execute().await.unwrap();
        let table = conn
            .create_table("my_table", batches)
            .execute()
@@ -1993,15 +1975,9 @@ mod tests {

    #[tokio::test]
    async fn test_execute_no_vector() {
-        // TODO: Switch back to memory://foo after https://github.com/lancedb/lancedb/issues/1051
-        // is fixed
-        let tmp_dir = tempdir().unwrap();
-        let dataset_path = tmp_dir.path().join("test.lance");
-        let uri = dataset_path.to_str().unwrap();
-
        // test that it's ok to not specify a query vector (just filter / limit)
        let batches = make_non_empty_batches();
-        let conn = connect(uri).execute().await.unwrap();
+        let conn = connect("memory://foo").execute().await.unwrap();
        let table = conn
            .create_table("my_table", batches)
            .execute()
@@ -8,13 +8,13 @@

 pub(crate) mod client;
 pub(crate) mod db;
+pub(crate) mod job;
 pub mod oauth;
 mod retry;
 pub(crate) mod table;
 pub(crate) mod util;

 const ARROW_STREAM_CONTENT_TYPE: &str = "application/vnd.apache.arrow.stream";
-#[cfg(test)]
 const ARROW_FILE_CONTENT_TYPE: &str = "application/vnd.apache.arrow.file";
 #[cfg(test)]
 const JSON_CONTENT_TYPE: &str = "application/json";
@@ -373,6 +373,37 @@ pub fn parse_db_url(db_url: &str) -> Result<ParsedDbUrl> {
    Ok(ParsedDbUrl { db_name, db_prefix })
 }

+fn validate_dns_hostname(hostname: &str) -> Result<()> {
+    let ascii_hostname = match url::Host::parse(hostname) {
+        Ok(url::Host::Domain(hostname)) => hostname,
+        Ok(_) => {
+            return Err(Error::InvalidInput {
+                message: "LanceDB Cloud database URI or region produced a non-DNS hostname"
+                    .to_string(),
+            });
+        }
+        Err(err) => {
+            return Err(Error::InvalidInput {
+                message: format!(
+                    "LanceDB Cloud database URI or region produced an invalid hostname: {err}"
+                ),
+            });
+        }
+    };
+
+    if ascii_hostname.len() > 253
+        || ascii_hostname
+            .split('.')
+            .any(|label| label.is_empty() || label.len() > 63)
+    {
+        return Err(Error::InvalidInput {
+            message: "LanceDB Cloud database URI or region produced an invalid hostname: DNS labels must contain 1 to 63 bytes and the full hostname must not exceed 253 bytes".to_string(),
+        });
+    }
+
+    Ok(())
+}
+
 impl RestfulLanceDbClient<Sender> {
    fn get_timeout(passed: Option<Duration>, env_var: &str) -> Result<Option<Duration>> {
        if let Some(passed) = passed {
@@ -480,7 +511,11 @@ impl RestfulLanceDbClient<Sender> {

        let host = match host_override {
            Some(host_override) => host_override,
-            None => format!("https://{}.{}.api.lancedb.com", parsed_url.db_name, region),
+            None => {
+                let hostname = format!("{}.{}.api.lancedb.com", parsed_url.db_name, region);
+                validate_dns_hostname(&hostname)?;
+                format!("https://{hostname}")
+            }
        };
        debug!("Created client for host: {}", host);
        let retry_config = client_config.retry_config.clone().try_into()?;
@@ -706,7 +741,7 @@ impl<S: HttpSend> RestfulLanceDbClient<S> {
            .err_to_http(request_id.clone())?;
        debug!(
            "Received response for request_id={}: {:?}",
-            request_id, &response
+            request_id, response
        );
        Ok((request_id, response))
    }
@@ -768,7 +803,7 @@ impl<S: HttpSend> RestfulLanceDbClient<S> {
                Ok((status, response)) if status.is_success() => {
                    debug!(
                        "Received response for request_id={}: {:?}",
-                        retry_counter.request_id, &response
+                        retry_counter.request_id, response
                    );
                    return Ok((retry_counter.request_id, response));
                }
@@ -1157,6 +1192,29 @@ mod tests {
        assert_eq!(headers.get("x-api-key").unwrap(), "api-key");
    }

+    #[test]
+    fn test_rejects_invalid_cloud_dns_hostname() {
+        let invalid_database_names = ["a".repeat(64), "invalid..database".to_string()];
+
+        for db_name in invalid_database_names {
+            let parsed_url = parse_db_url(&format!("db://{db_name}")).unwrap();
+            let error = RestfulLanceDbClient::<Sender>::try_new(
+                &parsed_url,
+                "us-east-1",
+                None,
+                HeaderMap::new(),
+                ClientConfig::default(),
+                None,
+            )
+            .unwrap_err();
+
+            assert!(
+                matches!(error, Error::InvalidInput { ref message } if message.contains("DNS labels must contain 1 to 63 bytes")),
+                "unexpected error: {error}"
+            );
+        }
+    }
+
    // Test implementation of HeaderProvider
    #[derive(Debug, Clone)]
    struct TestHeaderProvider {
@@ -20,7 +20,7 @@ use lance_namespace::models::{
 use crate::Error;
 use crate::database::{
    CloneTableRequest, CreateTableMode, CreateTableRequest, Database, DatabaseOptions,
-    OpenTableRequest, ReadConsistency, TableNamesRequest,
+    JobDescription, JobInfo, OpenTableRequest, ReadConsistency, TableNamesRequest,
 };
 use crate::error::Result;
 use crate::remote::util::stream_as_body;
@@ -79,6 +79,10 @@ impl ServerVersion {
    pub fn support_multipart_write(&self) -> bool {
        self.0 >= semver::Version::new(0, 4, 0)
    }
+
+    pub fn support_blobs(&self) -> bool {
+        self.0 >= semver::Version::new(0, 5, 0)
+    }
 }

 pub const OPT_REMOTE_PREFIX: &str = "remote_database_";
@@ -428,6 +432,73 @@ fn build_cache_key(name: &str, namespace: &[String]) -> String {
    key.iter().map(|b| format!("{:02x}", b)).collect()
 }

+#[derive(serde::Deserialize)]
+struct RemoteListJobRow {
+    job_id: String,
+    #[serde(default)]
+    table: String,
+    #[serde(default)]
+    job_type: String,
+    #[serde(default)]
+    state: String,
+    #[serde(default)]
+    created_at_millis: i64,
+}
+
+#[derive(serde::Deserialize)]
+struct RemoteListJobsResponse {
+    #[serde(default)]
+    jobs: Vec<RemoteListJobRow>,
+    #[serde(default)]
+    page_token: Option<String>,
+}
+
+/// The server's account of why a job failed. Absent from older servers,
+/// which report only the terminal state.
+#[derive(serde::Deserialize)]
+struct RemoteReportedFailure {
+    #[serde(default)]
+    phase: Option<String>,
+    #[serde(default)]
+    message: Option<String>,
+    #[serde(default)]
+    retryable: Option<bool>,
+}
+
+#[derive(serde::Deserialize)]
+struct RemoteDescribeJobResponse {
+    job_id: String,
+    #[serde(default)]
+    job_type: String,
+    job_state: String,
+    #[serde(default)]
+    creation_ms: i64,
+    #[serde(default)]
+    spec: serde_json::Value,
+    #[serde(default)]
+    failure: Option<RemoteReportedFailure>,
+}
+
+/// Server job states -> the client vocabulary ("running" / "finished" /
+/// "failed" / "cancelled"). Covers both the describe enum (IN_PROGRESS /
+/// DONE / FAILED / CANCELLED) and the registry's lowercase list-row states
+/// (in_progress / succeeded / failed / canceled / timed_out). States this
+/// client version does not know (e.g. created, queued) pass through as-is.
+fn job_state_to_client(state: &str) -> String {
+    match state {
+        "IN_PROGRESS" | "in_progress" => "running",
+        "DONE" | "done" | "succeeded" => "finished",
+        "FAILED" | "failed" | "TIMED_OUT" | "timed_out" => "failed",
+        "CANCELLED" | "cancelled" | "canceled" => "cancelled",
+        other => other,
+    }
+    .to_string()
+}
+
+/// Bound on `list_jobs` page walking; a warning is logged when the listing
+/// is truncated at this many pages.
+const MAX_LIST_JOBS_PAGES: usize = 100;
+
 #[async_trait]
 impl<S: HttpSend> Database for RemoteDatabase<S> {
    fn uri(&self) -> &str {
@@ -441,6 +512,108 @@ impl<S: HttpSend> Database for RemoteDatabase<S> {
        })
    }

+    fn job(&self, job_id: &str) -> Result<crate::job::Job> {
+        Ok(crate::job::Job::new(Box::new(super::job::RemoteJob::new(
+            self.client.clone(),
+            job_id.to_string(),
+        ))))
+    }
+
+    async fn list_jobs(&self) -> Result<Vec<JobInfo>> {
+        let mut out = Vec::new();
+        let mut page_token: Option<String> = None;
+        for page in 0..MAX_LIST_JOBS_PAGES {
+            let mut body = serde_json::json!({});
+            if let Some(token) = &page_token {
+                body["page_token"] = serde_json::Value::String(token.clone());
+            }
+            let req = self.client.post("/v1/jobs/list").json(&body);
+            let (request_id, rsp) = self.client.send(req).await?;
+            let rsp = self.client.check_response(&request_id, rsp).await?;
+            let body: RemoteListJobsResponse = rsp.json().await.err_to_http(request_id)?;
+            out.extend(body.jobs.into_iter().map(|row| JobInfo {
+                job_id: row.job_id,
+                table: row.table,
+                job_type: row.job_type,
+                state: job_state_to_client(&row.state),
+                created_at_millis: row.created_at_millis,
+            }));
+            page_token = body.page_token;
+            if page_token.is_none() {
+                break;
+            }
+            if page + 1 == MAX_LIST_JOBS_PAGES {
+                log::warn!(
+                    "list_jobs truncated after {} pages ({} jobs)",
+                    MAX_LIST_JOBS_PAGES,
+                    out.len()
+                );
+            }
+        }
+        Ok(out)
+    }
+
+    async fn get_job(&self, job_id: &str) -> Result<Option<JobDescription>> {
+        let req = self
+            .client
+            .post("/v1/jobs/describe")
+            .json(&serde_json::json!({ "job_id": job_id }));
+        let (request_id, rsp) = self.client.send(req).await?;
+        let rsp = match self.client.check_response(&request_id, rsp).await {
+            Ok(rsp) => rsp,
+            Err(Error::Http {
+                status_code: Some(StatusCode::NOT_FOUND),
+                ..
+            }) => return Ok(None),
+            Err(err) => return Err(err),
+        };
+        let body: RemoteDescribeJobResponse = rsp.json().await.err_to_http(request_id)?;
+        Ok(Some(JobDescription {
+            job_id: body.job_id,
+            job_type: body.job_type,
+            state: job_state_to_client(&body.job_state),
+            creation_ms: body.creation_ms,
+            spec: body.spec,
+            failure: body.failure.map(|reported| crate::error::JobFailure {
+                phase: reported.phase,
+                message: reported.message,
+                retryable: reported.retryable,
+                source: None,
+            }),
+        }))
+    }
+
+    async fn cancel_job(&self, job_id: &str) -> Result<bool> {
+        let req = self
+            .client
+            .post("/v1/jobs/cancel")
+            .json(&serde_json::json!({ "job_id": job_id }));
+        let (request_id, rsp) = self.client.send(req).await?;
+        match self.client.check_response(&request_id, rsp).await {
+            Ok(_) => Ok(true),
+            Err(Error::Http {
+                status_code: Some(StatusCode::NOT_FOUND),
+                ..
+            }) => Ok(false),
+            Err(err) => Err(err),
+        }
+    }
+
+    async fn job_history(&self, job_id: Option<&str>) -> Result<Vec<arrow_array::RecordBatch>> {
+        let mut body = serde_json::json!({});
+        if let Some(job_id) = job_id {
+            body["job_id"] = serde_json::Value::String(job_id.to_string());
+        }
+        let req = self.client.post("/v1/jobs/query_events").json(&body);
+        let (request_id, rsp) = self.client.send(req).await?;
+        let rsp = self.client.check_response(&request_id, rsp).await?;
+        let bytes = rsp.bytes().await.err_to_http(request_id)?;
+        let reader = arrow_ipc::reader::StreamReader::try_new(std::io::Cursor::new(bytes), None)?;
+        reader
+            .collect::<std::result::Result<Vec<_>, _>>()
+            .map_err(Into::into)
+    }
+
    async fn table_names(&self, request: TableNamesRequest) -> Result<Vec<String>> {
        let mut req = if !request.namespace_path.is_empty() {
            let namespace_id =
@@ -661,6 +834,7 @@ impl<S: HttpSend> Database for RemoteDatabase<S> {
                RemoteTable::<S>::handle_table_not_found(&request.name, rsp, &request_id).await?;
            let rsp = self.client.check_response(&request_id, rsp).await?;
            let version = parse_server_version(&request_id, &rsp)?;
+            let describe_body = rsp.text().await.ok();
            let table_identifier = build_table_identifier(
                &request.name,
                &request.namespace_path,
@@ -673,6 +847,12 @@ impl<S: HttpSend> Database for RemoteDatabase<S> {
                table_identifier,
                version,
            ));
+            // This describe already carries the schema, so hand it to the table
+            // instead of making the first schema read fetch it again. A version or
+            // branch pin applied after this invalidates the cache.
+            if let Some(body) = &describe_body {
+                table.seed_schema(body);
+            }
            let cache_key = build_cache_key(&request.name, &request.namespace_path);
            self.table_cache.insert(cache_key, table.clone()).await;
            Ok(table)
@@ -923,6 +1103,7 @@ impl From<StorageOptions> for RemoteOptions {
 mod tests {
    use super::{NamespaceHeaderProviderContext, build_cache_key};
    use std::collections::HashMap;
+    use std::sync::atomic::{AtomicUsize, Ordering};
    use std::sync::{Arc, OnceLock};

    use arrow_array::{Int32Array, RecordBatch};
@@ -1073,6 +1254,46 @@ mod tests {
        assert_eq!(table.name(), "table1");
    }

+    #[tokio::test]
+    async fn test_open_table_seeds_the_schema_from_its_describe() {
+        let describe_calls = Arc::new(AtomicUsize::new(0));
+        let counted = describe_calls.clone();
+        let conn = Connection::new_with_handler(move |request| {
+            assert_eq!(request.url().path(), "/v1/table/table1/describe/");
+            counted.fetch_add(1, Ordering::SeqCst);
+            http::Response::builder()
+                .status(200)
+                .body(
+                    r#"{"version": 1, "schema": {"fields": [
+                        {"name": "id", "type": {"type": "int64"}, "nullable": false}
+                    ]}}"#
+                        .to_string(),
+                )
+                .unwrap()
+        });
+
+        let table = conn.open_table("table1").execute().await.unwrap();
+        let schema = table.schema().await.unwrap();
+
+        assert_eq!(schema.field(0).name(), "id");
+        assert_eq!(describe_calls.load(Ordering::SeqCst), 1);
+    }
+
+    #[tokio::test]
+    async fn test_open_table_survives_a_describe_body_it_cannot_parse() {
+        let conn = Connection::new_with_handler(|request| {
+            assert_eq!(request.url().path(), "/v1/table/table1/describe/");
+            http::Response::builder()
+                .status(200)
+                .body(r#"{"table": "table1"}"#.to_string())
+                .unwrap()
+        });
+
+        let table = conn.open_table("table1").execute().await.unwrap();
+
+        assert_eq!(table.name(), "table1");
+    }
+
    #[tokio::test]
    async fn test_open_table_branch_and_version() {
        let conn = Connection::new_with_handler(|request| {
@@ -2042,4 +2263,165 @@ mod tests {
            assert!(list_response.tables.contains(&"table3".to_string()));
        }
    }
+
+    #[tokio::test]
+    async fn test_list_jobs_paginates() {
+        let page = Arc::new(AtomicUsize::new(0));
+        let conn = Connection::new_with_handler(move |request| {
+            assert_eq!(request.method(), &reqwest::Method::POST);
+            assert_eq!(request.url().path(), "/v1/jobs/list");
+            let body: serde_json::Value =
+                serde_json::from_slice(request.body().unwrap().as_bytes().unwrap()).unwrap();
+            match page.fetch_add(1, Ordering::SeqCst) {
+                0 => {
+                    assert!(body.get("page_token").is_none());
+                    http::Response::builder()
+                        .status(200)
+                        .body(
+                            r#"{"jobs": [{"job_id": "job-1", "table": "t1", "job_type": "create_index", "state": "in_progress", "created_at_millis": 1000}], "page_token": "next"}"#,
+                        )
+                        .unwrap()
+                }
+                _ => {
+                    assert_eq!(body["page_token"], "next");
+                    http::Response::builder()
+                        .status(200)
+                        .body(
+                            r#"{"jobs": [{"job_id": "job-2", "table": "t2", "job_type": "create_index", "state": "succeeded", "created_at_millis": 2000}, {"job_id": "job-3", "table": "t3", "job_type": "create_index", "state": "timed_out", "created_at_millis": 3000}]}"#,
+                        )
+                        .unwrap()
+                }
+            }
+        });
+        let jobs = conn.list_jobs().await.unwrap();
+        assert_eq!(jobs.len(), 3);
+        assert_eq!(jobs[0].job_id, "job-1");
+        assert_eq!(jobs[0].table, "t1");
+        assert_eq!(jobs[0].state, "running");
+        assert_eq!(jobs[1].job_id, "job-2");
+        assert_eq!(jobs[1].state, "finished");
+        assert_eq!(jobs[1].created_at_millis, 2000);
+        assert_eq!(jobs[2].job_id, "job-3");
+        assert_eq!(jobs[2].state, "failed");
+    }
+
+    #[tokio::test]
+    async fn test_get_job() {
+        let conn = Connection::new_with_handler(|request| {
+            assert_eq!(request.method(), &reqwest::Method::POST);
+            assert_eq!(request.url().path(), "/v1/jobs/describe");
+            let body: serde_json::Value =
+                serde_json::from_slice(request.body().unwrap().as_bytes().unwrap()).unwrap();
+            assert_eq!(body["job_id"], "job-1");
+            http::Response::builder()
+                .status(200)
+                .body(
+                    r#"{"job_id": "job-1", "job_type": "create_index", "job_state": "FAILED", "creation_ms": 1000, "spec": {"column": "vec"}, "failure": {"phase": "execute", "message": "worker died", "retryable": true}}"#,
+                )
+                .unwrap()
+        });
+        let job = conn.get_job("job-1").await.unwrap().unwrap();
+        assert_eq!(job.job_id, "job-1");
+        assert_eq!(job.job_type, "create_index");
+        assert_eq!(job.state, "failed");
+        assert_eq!(job.creation_ms, 1000);
+        assert_eq!(job.spec["column"], "vec");
+        let failure = job.failure.unwrap();
+        assert_eq!(failure.phase.as_deref(), Some("execute"));
+        assert_eq!(failure.message.as_deref(), Some("worker died"));
+        assert_eq!(failure.retryable, Some(true));
+    }
+
+    #[tokio::test]
+    async fn test_get_job_missing_is_none() {
+        let conn = Connection::new_with_handler(|_| {
+            http::Response::builder()
+                .status(404)
+                .body("no such job")
+                .unwrap()
+        });
+        assert!(conn.get_job("nope").await.unwrap().is_none());
+    }
+
+    #[tokio::test]
+    async fn test_cancel_job() {
+        let conn = Connection::new_with_handler(|request| {
+            assert_eq!(request.url().path(), "/v1/jobs/cancel");
+            http::Response::builder()
+                .status(200)
+                .body(r#"{"job_id": "job-1"}"#)
+                .unwrap()
+        });
+        assert!(conn.cancel_job("job-1").await.unwrap());
+
+        let conn = Connection::new_with_handler(|_| {
+            http::Response::builder()
+                .status(404)
+                .body("no such job")
+                .unwrap()
+        });
+        assert!(!conn.cancel_job("nope").await.unwrap());
+    }
+
+    #[tokio::test]
+    async fn test_job_history_parses_arrow_stream() {
+        let schema = Arc::new(Schema::new(vec![Field::new(
+            "state",
+            DataType::Utf8,
+            false,
+        )]));
+        let batch = RecordBatch::try_new(
+            schema.clone(),
+            vec![Arc::new(arrow_array::StringArray::from(vec![
+                "created", "done",
+            ]))],
+        )
+        .unwrap();
+        let mut body = Vec::new();
+        {
+            let mut writer = arrow_ipc::writer::StreamWriter::try_new(&mut body, &schema).unwrap();
+            writer.write(&batch).unwrap();
+            writer.finish().unwrap();
+        }
+        let conn = Connection::new_with_handler(move |request| {
+            assert_eq!(request.url().path(), "/v1/jobs/query_events");
+            let req_body: serde_json::Value =
+                serde_json::from_slice(request.body().unwrap().as_bytes().unwrap()).unwrap();
+            assert_eq!(req_body["job_id"], "job-1");
+            http::Response::builder()
+                .status(200)
+                .body(body.clone())
+                .unwrap()
+        });
+        let batches = conn.job_history(Some("job-1")).await.unwrap();
+        assert_eq!(batches.len(), 1);
+        assert_eq!(batches[0].num_rows(), 2);
+    }
+
+    #[tokio::test]
+    async fn test_conn_job_waits_to_done() {
+        let polls = Arc::new(AtomicUsize::new(0));
+        let polls_ref = polls.clone();
+        let conn = Connection::new_with_handler(move |request| {
+            assert_eq!(request.url().path(), "/v1/jobs/describe");
+            let state = if polls_ref.fetch_add(1, Ordering::SeqCst) == 0 {
+                "IN_PROGRESS"
+            } else {
+                "DONE"
+            };
+            http::Response::builder()
+                .status(200)
+                .body(format!(
+                    r#"{{"job_id": "job-1", "job_type": "create_index", "job_state": "{}", "creation_ms": 1}}"#,
+                    state
+                ))
+                .unwrap()
+        });
+        let job = conn.job("job-1").unwrap();
+        assert_eq!(job.id(), Some("job-1"));
+        assert_eq!(job.status().await.unwrap(), "running");
+        job.wait().await.unwrap();
+        assert_eq!(job.status().await.unwrap(), "finished");
+        assert!(polls.load(Ordering::SeqCst) >= 3);
+    }
 }
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Tracking for server-side jobs through the `/v1/jobs` API.
+
+use std::time::Duration;
+
+use async_trait::async_trait;
+use tokio::time::sleep;
+
+use serde::{Deserialize, Deserializer};
+
+use crate::error::{Error, JobFailure, Result};
+use crate::job::JobHandle;
+use crate::remote::client::{HttpSend, RequestResultExt, RestfulLanceDbClient};
+
+/// Delay before the second job-state poll; doubles up to [`MAX_POLL_INTERVAL`].
+const INITIAL_POLL_INTERVAL: Duration = Duration::from_millis(200);
+const MAX_POLL_INTERVAL: Duration = Duration::from_secs(5);
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+enum JobState {
+    InProgress,
+    Cancelled,
+    Failed,
+    Done,
+    /// A state this client version does not know; treated as still running
+    /// and reported as-is if the job never settles.
+    Other(String),
+}
+
+impl<'de> Deserialize<'de> for JobState {
+    fn deserialize<D: Deserializer<'de>>(deserializer: D) -> std::result::Result<Self, D::Error> {
+        Ok(Self::from(String::deserialize(deserializer)?.as_str()))
+    }
+}
+
+impl JobState {
+    /// The client vocabulary label for this state.
+    fn client_label(&self) -> String {
+        match self {
+            Self::InProgress => "running".to_string(),
+            Self::Done => "finished".to_string(),
+            Self::Failed => "failed".to_string(),
+            Self::Cancelled => "cancelled".to_string(),
+            Self::Other(state) => state.clone(),
+        }
+    }
+}
+
+impl From<&str> for JobState {
+    fn from(state: &str) -> Self {
+        match state {
+            "IN_PROGRESS" => Self::InProgress,
+            "CANCELLED" => Self::Cancelled,
+            // The server reports a timed-out job as FAILED on describe;
+            // accept the raw registry state too in case a future server
+            // stops folding it.
+            "FAILED" | "TIMED_OUT" => Self::Failed,
+            "DONE" => Self::Done,
+            other => Self::Other(other.to_string()),
+        }
+    }
+}
+
+/// The server's account of why a job failed. Absent from older servers, which
+/// report only the terminal state.
+#[derive(Deserialize)]
+struct ReportedFailure {
+    #[serde(default)]
+    phase: Option<String>,
+    #[serde(default)]
+    message: Option<String>,
+    #[serde(default)]
+    retryable: Option<bool>,
+}
+
+#[derive(Deserialize)]
+struct DescribeJobResponse {
+    job_state: JobState,
+    #[serde(default)]
+    failure: Option<ReportedFailure>,
+}
+
+pub struct RemoteJob<S: HttpSend> {
+    client: RestfulLanceDbClient<S>,
+    job_id: String,
+}
+
+impl<S: HttpSend> RemoteJob<S> {
+    pub fn new(client: RestfulLanceDbClient<S>, job_id: String) -> Self {
+        Self { client, job_id }
+    }
+
+    /// One `/v1/jobs/describe` round trip.
+    async fn describe(&self) -> Result<DescribeJobResponse> {
+        let request = self
+            .client
+            .post("/v1/jobs/describe")
+            .json(&serde_json::json!({ "job_id": self.job_id }));
+        let (request_id, response) = self.client.send(request).await?;
+        let response = self.client.check_response(&request_id, response).await?;
+        let body = response.text().await.err_to_http(request_id.clone())?;
+        let description: DescribeJobResponse =
+            serde_json::from_str(&body).map_err(|e| Error::Http {
+                source: format!("failed to parse job description: {}", e).into(),
+                request_id,
+                status_code: None,
+            })?;
+        Ok(description)
+    }
+}
+
+#[async_trait]
+impl<S: HttpSend> JobHandle for RemoteJob<S> {
+    fn id(&self) -> Option<&str> {
+        Some(&self.job_id)
+    }
+
+    async fn status(&self) -> Result<String> {
+        Ok(self.describe().await?.job_state.client_label())
+    }
+
+    async fn wait(&self) -> Result<()> {
+        let mut interval = INITIAL_POLL_INTERVAL;
+        loop {
+            let description = self.describe().await?;
+            match description.job_state {
+                JobState::Done => return Ok(()),
+                JobState::Failed => {
+                    return Err(Error::JobFailed {
+                        job_id: Some(self.job_id.clone()),
+                        failure: description
+                            .failure
+                            .map(|reported| JobFailure {
+                                phase: reported.phase,
+                                message: reported.message,
+                                retryable: reported.retryable,
+                                source: None,
+                            })
+                            .unwrap_or_default(),
+                    });
+                }
+                JobState::Cancelled => {
+                    return Err(Error::JobCancelled {
+                        job_id: Some(self.job_id.clone()),
+                    });
+                }
+                JobState::InProgress => {}
+                JobState::Other(ref state) => {
+                    log::debug!("job {} is in unrecognized state {state}", self.job_id)
+                }
+            }
+            sleep(interval).await;
+            interval = (interval * 2).min(MAX_POLL_INTERVAL);
+        }
+    }
+
+    async fn cancel(&self) -> Result<()> {
+        let request = self
+            .client
+            .post("/v1/jobs/cancel")
+            .json(&serde_json::json!({ "job_id": self.job_id }));
+        let (request_id, response) = self.client.send(request).await?;
+        self.client
+            .check_response(&request_id, response)
+            .await
+            .map(|_| ())
+    }
+}
@@ -1,7 +1,8 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

-//! DataFusion ExecutionPlan for inserting data into remote LanceDB tables.
+//! DataFusion ExecutionPlan for streaming writes (add / merge_insert) to
+//! remote LanceDB tables.

 use std::sync::{Arc, Mutex};
 use std::time::{Duration, Instant};
@@ -23,28 +24,57 @@ use lance::io::exec::utils::InstrumentedRecordBatchStreamAdapter;
 use crate::Error;
 use crate::remote::ARROW_STREAM_CONTENT_TYPE;
 use crate::remote::client::{HttpSend, RestfulLanceDbClient, Sender};
-use crate::remote::table::RemoteTable;
-use crate::table::AddResult;
+use crate::remote::table::{MergeInsertRequest, REQUEST_TIMEOUT_HEADER, RemoteTable};
 use crate::table::datafusion::insert::COUNT_SCHEMA;
 use crate::table::write_progress::WriteProgressTracker;
+use crate::table::{AddResult, MergeResult};

-/// ExecutionPlan for inserting data into a remote LanceDB table.
+/// The write operation a [`RemoteWriteExec`] performs. Both variants share the
+/// same Arrow-IPC streaming body and error side-channel; only the target
+/// endpoint, query parameters, and parsed result type differ.
+#[derive(Debug, Clone)]
+pub(crate) enum WriteOp {
+    /// `add`: stream to `/v1/table/{id}/insert/`, optionally overwriting.
+    Insert { overwrite: bool },
+    /// `merge_insert`: stream to `/v1/table/{id}/merge_insert/` with the merge
+    /// parameters carried as query params. Multipart is not supported for this
+    /// operation (the server has no multipart merge_insert endpoint), so an
+    /// `upload_id` combined with this op is a programming error.
+    MergeInsert {
+        query: MergeInsertRequest,
+        timeout: Option<Duration>,
+    },
+}
+
+/// The parsed server response for a completed write, discriminated by the
+/// operation that produced it.
+#[derive(Debug, Clone)]
+pub(crate) enum WriteResult {
+    Add(AddResult),
+    Merge(MergeResult),
+}
+
+/// ExecutionPlan for streaming a write (add or merge_insert) to a remote
+/// LanceDB table.
 ///
-/// Streams data as Arrow IPC to `/v1/table/{id}/insert/` endpoint.
+/// Streams data as Arrow IPC to the endpoint selected by [`WriteOp`]. Both
+/// operations reuse the same error side-channel so an input stream error (e.g.
+/// NaN rejection) surfaces with its original message rather than the masked
+/// HTTP error Hyper produces when a request body stream fails under HTTP2.
 ///
 /// When `upload_id` is set, inserts are staged as part of a multipart write
 /// session and the plan supports multiple partitions for parallel uploads.
 /// Without `upload_id`, the plan requires a single partition and commits
-/// immediately.
+/// immediately. Multipart applies to `add` only.
 #[derive(Debug)]
-pub struct RemoteInsertExec<S: HttpSend = Sender> {
+pub struct RemoteWriteExec<S: HttpSend = Sender> {
    table_name: String,
    identifier: String,
    client: RestfulLanceDbClient<S>,
    input: Arc<dyn ExecutionPlan>,
-    overwrite: bool,
+    op: WriteOp,
    properties: Arc<PlanProperties>,
-    add_result: Arc<Mutex<Option<AddResult>>>,
+    result: Arc<Mutex<Option<WriteResult>>>,
    metrics: ExecutionPlanMetricsSet,
    upload_id: Option<String>,
    tracker: Option<Arc<WriteProgressTracker>>,
@@ -61,27 +91,28 @@ pub struct RemoteInsertExec<S: HttpSend = Sender> {
    max_request_duration: Option<Duration>,
 }

-impl<S: HttpSend + 'static> RemoteInsertExec<S> {
-    /// Create a new single-partition RemoteInsertExec.
+impl<S: HttpSend + 'static> RemoteWriteExec<S> {
+    /// Create a new single-partition RemoteWriteExec.
    pub fn new(
        table_name: String,
        identifier: String,
        client: RestfulLanceDbClient<S>,
        input: Arc<dyn ExecutionPlan>,
-        overwrite: bool,
+        op: WriteOp,
        tracker: Option<Arc<WriteProgressTracker>>,
        branch: Option<String>,
    ) -> Self {
        Self::new_inner(
-            table_name, identifier, client, input, overwrite, None, tracker, branch, None, None,
+            table_name, identifier, client, input, op, None, tracker, branch, None, None,
        )
    }

-    /// Create a multi-partition RemoteInsertExec for use with multipart writes.
+    /// Create a multi-partition RemoteWriteExec for use with multipart writes.
    ///
    /// Each partition's insert is staged under the given `upload_id` without
    /// committing. The caller is responsible for calling the complete (or abort)
-    /// endpoint after all partitions finish.
+    /// endpoint after all partitions finish. Multipart is insert-only, so the
+    /// op is fixed to [`WriteOp::Insert`].
    #[allow(clippy::too_many_arguments)]
    pub fn new_multipart(
        table_name: String,
@@ -100,7 +131,7 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
            identifier,
            client,
            input,
-            overwrite,
+            WriteOp::Insert { overwrite },
            Some(upload_id),
            tracker,
            branch,
@@ -115,7 +146,7 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
        identifier: String,
        client: RestfulLanceDbClient<S>,
        input: Arc<dyn ExecutionPlan>,
-        overwrite: bool,
+        op: WriteOp,
        upload_id: Option<String>,
        tracker: Option<Arc<WriteProgressTracker>>,
        branch: Option<String>,
@@ -140,9 +171,9 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
            identifier,
            client,
            input,
-            overwrite,
+            op,
            properties: Arc::new(properties),
-            add_result: Arc::new(Mutex::new(None)),
+            result: Arc::new(Mutex::new(None)),
            metrics: ExecutionPlanMetricsSet::new(),
            upload_id,
            tracker,
@@ -152,14 +183,30 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
        }
    }

-    /// Get the add result after execution.
-    // TODO: this will be used when we wire this up to Table::add().
-    #[allow(dead_code)]
+    /// Get the add result after execution, if this exec ran an insert.
    pub fn add_result(&self) -> Option<AddResult> {
-        self.add_result
+        match self
+            .result
            .lock()
            .unwrap_or_else(|e| e.into_inner())
            .clone()
+        {
+            Some(WriteResult::Add(r)) => Some(r),
+            _ => None,
+        }
+    }
+
+    /// Get the merge result after execution, if this exec ran a merge_insert.
+    pub fn merge_result(&self) -> Option<MergeResult> {
+        match self
+            .result
+            .lock()
+            .unwrap_or_else(|e| e.into_inner())
+            .clone()
+        {
+            Some(WriteResult::Merge(r)) => Some(r),
+            _ => None,
+        }
    }

    /// Stream the input into an HTTP body as an Arrow IPC stream, capturing any
@@ -464,24 +511,24 @@ impl<S: HttpSend + 'static> PartRequestCtx<'_, S> {
    }
 }

-impl<S: HttpSend + 'static> DisplayAs for RemoteInsertExec<S> {
+impl<S: HttpSend + 'static> DisplayAs for RemoteWriteExec<S> {
    fn fmt_as(&self, t: DisplayFormatType, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match t {
            DisplayFormatType::Default | DisplayFormatType::Verbose => {
-                write!(
-                    f,
-                    "RemoteInsertExec: table={}, overwrite={}",
-                    self.table_name, self.overwrite
-                )
+                write!(f, "RemoteWriteExec: table={}, op=", self.table_name)?;
+                match &self.op {
+                    WriteOp::Insert { overwrite } => write!(f, "insert, overwrite={}", overwrite),
+                    WriteOp::MergeInsert { .. } => write!(f, "merge_insert"),
+                }
            }
            DisplayFormatType::TreeRender => {
-                write!(f, "RemoteInsertExec")
+                write!(f, "RemoteWriteExec")
            }
        }
    }
 }

-impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
+impl<S: HttpSend + 'static> ExecutionPlan for RemoteWriteExec<S> {
    fn name(&self) -> &str {
        Self::static_name()
    }
@@ -516,15 +563,18 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
    ) -> DataFusionResult<Arc<dyn ExecutionPlan>> {
        if children.len() != 1 {
            return Err(DataFusionError::Internal(
-                "RemoteInsertExec requires exactly one child".to_string(),
+                "RemoteWriteExec requires exactly one child".to_string(),
            ));
        }
+        // Building a fresh exec (with a new, empty `result`) is what makes the
+        // outer rescannable retry loop work: `reset_state()` clears the captured
+        // result so a re-execution starts clean.
        Ok(Arc::new(Self::new_inner(
            self.table_name.clone(),
            self.identifier.clone(),
            self.client.clone(),
            children[0].clone(),
-            self.overwrite,
+            self.op.clone(),
            self.upload_id.clone(),
            self.tracker.clone(),
            self.branch.clone(),
@@ -540,11 +590,19 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
    ) -> DataFusionResult<SendableRecordBatchStream> {
        if self.upload_id.is_none() && partition != 0 {
            return Err(DataFusionError::Internal(
-                "RemoteInsertExec only supports single partition execution without upload_id"
+                "RemoteWriteExec only supports single partition execution without upload_id"
                    .to_string(),
            ));
        }

+        // Multipart is insert-only: the server has no multipart merge_insert
+        // endpoint, so a merge_insert with an upload_id is a programming error.
+        if self.upload_id.is_some() && matches!(self.op, WriteOp::MergeInsert { .. }) {
+            return Err(DataFusionError::Internal(
+                "merge_insert does not support multipart (upload_id) writes".to_string(),
+            ));
+        }
+
        let input_stream = self.input.execute(partition, context)?;
        let input_schema = input_stream.schema();
        let input_stream: SendableRecordBatchStream =
@@ -556,8 +614,8 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
            ));
        let client = self.client.clone();
        let identifier = self.identifier.clone();
-        let overwrite = self.overwrite;
-        let add_result = self.add_result.clone();
+        let op = self.op.clone();
+        let result_slot = self.result.clone();
        let table_name = self.table_name.clone();
        let upload_id = self.upload_id.clone();
        let tracker = self.tracker.clone();
@@ -568,10 +626,12 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
        let stream = futures::stream::once(async move {
            // Multipart writes with a byte budget split the partition into
            // several bounded, still-streamed requests so no single request
-            // stays open long enough to hit the client read timeout.
+            // stays open long enough to hit the client read timeout. This path
+            // is insert-only (guarded above).
            if let (Some(upload_id), Some(max_bytes)) =
                (upload_id.as_deref(), max_bytes_per_request)
            {
+                let overwrite = matches!(op, WriteOp::Insert { overwrite: true });
                let ctx = PartRequestCtx {
                    client: &client,
                    identifier: &identifier,
@@ -592,16 +652,36 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
                )?);
            }

-            let mut request = client
-                .post(&format!("/v1/table/{}/insert/", identifier))
-                .header(CONTENT_TYPE, ARROW_STREAM_CONTENT_TYPE);
+            // Build the request for the selected operation. Both endpoints take
+            // an Arrow-IPC streaming body and reuse the same error side-channel.
+            let mut request = match &op {
+                WriteOp::Insert { overwrite } => {
+                    let mut request = client
+                        .post(&format!("/v1/table/{}/insert/", identifier))
+                        .header(CONTENT_TYPE, ARROW_STREAM_CONTENT_TYPE);
+                    if *overwrite {
+                        request = request.query(&[("mode", "overwrite")]);
+                    }
+                    if let Some(ref uid) = upload_id {
+                        request = request.query(&[("upload_id", uid.as_str())]);
+                    }
+                    request
+                }
+                WriteOp::MergeInsert { query, timeout } => {
+                    let mut request = client
+                        .post(&format!("/v1/table/{}/merge_insert/", identifier))
+                        .query(query)
+                        .header(CONTENT_TYPE, ARROW_STREAM_CONTENT_TYPE);
+                    if let Some(timeout) = timeout {
+                        // (If it doesn't fit into u64, it's not worth sending anyways.)
+                        if let Ok(timeout_ms) = u64::try_from(timeout.as_millis()) {
+                            request = request.header(REQUEST_TIMEOUT_HEADER, timeout_ms);
+                        }
+                    }
+                    request
+                }
+            };

-            if overwrite {
-                request = request.query(&[("mode", "overwrite")]);
-            }
-            if let Some(ref uid) = upload_id {
-                request = request.query(&[("upload_id", uid.as_str())]);
-            }
            if let Some(ref b) = branch {
                request = request.query(&[("branch", b.as_str())]);
            }
@@ -635,6 +715,8 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {

            // If the request failed due to an input stream error, surface the
            // original error (e.g. NaN rejection) instead of the HTTP error.
+            // This is the crux of the #2339 fix: Hyper silently swallows body
+            // stream errors under HTTP2, so we recover the original here.
            if let Ok(stream_err) = error_rx.try_recv() {
                return Err(stream_err);
            }
@@ -642,7 +724,7 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
            let (request_id, response) = result?;

            // For multipart writes, the staging response is not the final
-            // version. Only parse AddResult for non-multipart inserts.
+            // version. Only parse the result for non-multipart writes.
            if upload_id.is_none() {
                let body_text = response.text().await.map_err(|e| {
                    DataFusionError::External(Box::new(Error::Http {
@@ -652,21 +734,44 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
                    }))
                })?;

-                let parsed_result = if body_text.trim().is_empty() {
-                    // Backward compatible with old servers
-                    AddResult { version: 0 }
-                } else {
-                    serde_json::from_str(&body_text).map_err(|e| {
-                        DataFusionError::External(Box::new(Error::Http {
-                            source: format!("Failed to parse add response: {}", e).into(),
-                            request_id: request_id.clone(),
-                            status_code: None,
-                        }))
-                    })?
+                let parsed_result = match &op {
+                    WriteOp::Insert { .. } => {
+                        let add = if body_text.trim().is_empty() {
+                            // Backward compatible with old servers
+                            AddResult { version: 0 }
+                        } else {
+                            serde_json::from_str(&body_text).map_err(|e| {
+                                DataFusionError::External(Box::new(Error::Http {
+                                    source: format!("Failed to parse add response: {}", e).into(),
+                                    request_id: request_id.clone(),
+                                    status_code: None,
+                                }))
+                            })?
+                        };
+                        WriteResult::Add(add)
+                    }
+                    WriteOp::MergeInsert { .. } => {
+                        let merge = if body_text.trim().is_empty() {
+                            // Backward compatible with old servers
+                            MergeResult::default()
+                        } else {
+                            serde_json::from_str(&body_text).map_err(|e| {
+                                DataFusionError::External(Box::new(Error::Http {
+                                    source: format!("Failed to parse merge_insert response: {}", e)
+                                        .into(),
+                                    request_id: request_id.clone(),
+                                    status_code: None,
+                                }))
+                            })?
+                        };
+                        WriteResult::Merge(merge)
+                    }
                };

-                let mut res_lock = add_result.lock().map_err(|_| {
-                    DataFusionError::Execution("Failed to acquire lock for add_result".to_string())
+                let mut res_lock = result_slot.lock().map_err(|_| {
+                    DataFusionError::Execution(
+                        "Failed to acquire lock for write result".to_string(),
+                    )
                })?;
                *res_lock = Some(parsed_result);
            } else {
@@ -680,7 +785,7 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
                })?;
            }

-            // Return a single batch with count 0 (actual count is tracked in add_result)
+            // Return a single batch with count 0 (actual count is tracked in result)
            let count_array: ArrayRef = Arc::new(UInt64Array::from(vec![0u64]));
            let batch = RecordBatch::try_new(COUNT_SCHEMA.clone(), vec![count_array])?;
            Ok::<_, DataFusionError>(batch)
@@ -711,9 +816,11 @@ mod tests {
    use std::sync::atomic::{AtomicUsize, Ordering};
    use std::sync::{Arc, Mutex};

-    use super::RemoteInsertExec;
+    use super::RemoteWriteExec;
+    use super::WriteOp;
    use crate::Table;
    use crate::remote::ARROW_STREAM_CONTENT_TYPE;
+    use crate::remote::table::MergeInsertRequest;
    use crate::table::datafusion::BaseTableAdapter;

    fn schema_json() -> &'static str {
@@ -1028,7 +1135,7 @@ mod tests {
        let input = input_plan_from_batches(schema, batches).await;

        // A 1-byte budget forces every batch into its own part.
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1068,7 +1175,7 @@ mod tests {

        // A large byte budget and no time limit keep the whole partition in a
        // single part.
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1109,7 +1216,7 @@ mod tests {
        // A large byte budget but a tiny duration budget: writing and sending
        // one batch already takes longer than the limit, so each batch is cut
        // into its own part on the time check rather than the byte check.
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1144,7 +1251,7 @@ mod tests {
        // write relies on another partition having data to commit.
        let input = input_plan_from_batches(schema, vec![]).await;

-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1184,7 +1291,7 @@ mod tests {
        let input = input_plan_from_batches(schema, batches).await;

        // A 1-byte budget forces every batch into its own part.
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1233,7 +1340,7 @@ mod tests {
        ];
        let input = input_plan_from_partitions(schema, partitions).await;

-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1267,7 +1374,7 @@ mod tests {
        // A large byte budget keeps the good batch and the following error in
        // the same part, exercising the mid-part abort path.
        let input: Arc<dyn ExecutionPlan> = Arc::new(ErroringExec::new());
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -1297,6 +1404,66 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn test_merge_insert_input_error_surfaces_original() {
+        // Regression test for #2339 on the single-request merge_insert path.
+        // When the input stream errors mid-body, Hyper masks it under HTTP2 as a
+        // generic "stream error sent by user" message. The error side-channel
+        // must recover and surface the original DataFusion error instead.
+        use futures::StreamExt;
+
+        let client = crate::remote::client::test_utils::client_with_handler(|request| {
+            assert_eq!(request.url().path(), "/v1/table/my_table/merge_insert/");
+            http::Response::builder()
+                .status(200)
+                .body(
+                    r#"{"version": 2, "num_updated_rows": 0, "num_inserted_rows": 0, "num_deleted_rows": 0}"#
+                        .to_string(),
+                )
+                .unwrap()
+        });
+
+        let query = MergeInsertRequest {
+            on: "id".to_string(),
+            when_matched_update_all: false,
+            when_matched_update_all_filt: None,
+            when_not_matched_insert_all: false,
+            when_not_matched_by_source_delete: false,
+            when_not_matched_by_source_delete_filt: None,
+            use_index: true,
+            use_lsm: None,
+        };
+
+        let input: Arc<dyn ExecutionPlan> = Arc::new(ErroringExec::new());
+        let exec = RemoteWriteExec::new(
+            "my_table".to_string(),
+            "my_table".to_string(),
+            client,
+            input,
+            WriteOp::MergeInsert {
+                query,
+                timeout: None,
+            },
+            None,
+            None,
+        );
+
+        let mut stream = exec.execute(0, Arc::new(TaskContext::default())).unwrap();
+        let mut err = None;
+        while let Some(item) = stream.next().await {
+            if let Err(e) = item {
+                err = Some(e);
+                break;
+            }
+        }
+
+        let err = err.expect("expected the input stream error to surface");
+        assert!(
+            err.to_string().contains("boom"),
+            "expected original input error, got: {err}"
+        );
+    }
+
    #[tokio::test]
    async fn test_multipart_records_progress_within_a_part() {
        use crate::table::write_progress::{ProgressCallback, WriteProgress, WriteProgressTracker};
@@ -1327,7 +1494,7 @@ mod tests {
        // A large byte budget keeps all three batches in one part; smooth
        // progress therefore requires bytes to be reported per chunk rather than
        // once when the part completes.
-        let exec = RemoteInsertExec::new_multipart(
+        let exec = RemoteWriteExec::new_multipart(
            "my_table".to_string(),
            "my_table".to_string(),
            client,
@@ -3,6 +3,7 @@

 //! LanceDB Table APIs

+use crate::blob::BlobFile;
 use arrow_array::{LargeBinaryArray, RecordBatch, RecordBatchReader};
 use arrow_schema::{Schema, SchemaRef};
 use async_trait::async_trait;
@@ -12,7 +13,6 @@ use datafusion_physical_plan::ExecutionPlan;
 use datafusion_physical_plan::display::DisplayableExecutionPlan;
 use futures::StreamExt;
 use futures::stream::FuturesUnordered;
-use lance::dataset::BlobFile;
 pub use lance::dataset::ColumnAlteration;
 pub use lance::dataset::NewColumnTransform;
 pub use lance::dataset::ReadParams;
@@ -50,12 +50,14 @@ use crate::DistanceType;
 use crate::blob::BlobRangeRequest;
 use crate::data::scannable::{PeekedScannable, Scannable, estimate_write_partitions};
 use crate::database::Database;
+use crate::database::listing::LANCE_FILE_EXTENSION;
 use crate::database::read_freshness::TableFreshness;
 use crate::embeddings::{EmbeddingDefinition, EmbeddingRegistry, MemoryRegistry};
 use crate::error::{Error, Result};
 use crate::index::IndexStatistics;
 use crate::index::{Index, IndexBuilder};
 use crate::index::{IndexConfig, IndexStatisticsImpl, IndexType};
+use crate::job::Job;
 use crate::query::{IntoQueryVector, Query, QueryExecutionOptions, TakeQuery, VectorQuery};
 use crate::table::datafusion::insert::InsertExec;
 use crate::utils::{PatchReadParam, PatchWriteParam, resolve_arrow_field_path};
@@ -63,6 +65,7 @@ use crate::utils::{PatchReadParam, PatchWriteParam, resolve_arrow_field_path};
 use self::dataset::DatasetConsistencyWrapper;
 use self::merge::MergeInsertBuilder;

+pub mod add_columns;
 mod add_data;
 pub mod branch_merge;
 mod create_index;
@@ -77,6 +80,7 @@ pub mod schema_evolution;
 pub mod update;
 pub mod write_progress;
 use crate::index::waiter::wait_for_index;
+pub use add_columns::AddColumnsBuilder;
 #[cfg(feature = "remote")]
 pub(crate) use add_data::PreprocessingOutput;
 pub use add_data::{AddDataBuilder, AddDataMode, AddResult, NaNVectorBehavior};
@@ -146,6 +150,55 @@ pub(crate) fn map_namespace_lance_error(err: lance::Error, table_name: &str) ->
    }
 }

+/// Map a `lance::Error::DatasetNotFound` for the table at `uri` into a `lancedb::Error`.
+///
+/// Lance reports "there is nothing at this location" and "there is a table directory
+/// here but nothing loadable inside it" with the same error. Only the first is a
+/// `TableNotFound`: a `<name>.lance` directory left behind by an interrupted drop and
+/// re-create is still reported by `Connection::table_names`, so callers need to be able
+/// to tell "never existed" from "exists but is broken".
+///
+/// See <https://github.com/lancedb/lancedb/issues/3127>.
+async fn map_dataset_not_found(
+    uri: &str,
+    name: &str,
+    params: ReadParams,
+    err: lance::Error,
+) -> Error {
+    let name = name.to_string();
+    let source = Box::new(err);
+    if table_dir_exists(uri, params).await.unwrap_or(false) {
+        Error::TableCorrupted { name, source }
+    } else {
+        Error::TableNotFound { name, source }
+    }
+}
+
+/// Whether a table directory is present at `uri`, even though no dataset could be
+/// loaded from it.
+///
+/// This looks for a `<name>.lance` entry in the parent directory, which is exactly what
+/// `ListingDatabase::table_names` lists, so the two APIs agree on whether a table is
+/// present. Probing `uri` itself would not work: object stores have no empty
+/// directories to probe, and on a local filesystem the interesting case is precisely an
+/// empty directory.
+async fn table_dir_exists(uri: &str, params: ReadParams) -> Result<bool> {
+    let (object_store, path, _) = DatasetBuilder::from_uri(uri)
+        .with_read_params(params)
+        .build_object_store()
+        .await?;
+    // Only `*.lance` entries are ever reported as tables, so nothing else can produce
+    // the list-then-open mismatch this guards against.
+    if path.extension() != Some(LANCE_FILE_EXTENSION) {
+        return Ok(false);
+    }
+    let (Some(parent), Some(dir_name)) = (path.parent(), path.filename()) else {
+        return Ok(false);
+    };
+    let entries = object_store.read_dir(parent).await?;
+    Ok(entries.iter().any(|entry| entry.as_str() == dir_name))
+}
+
 /// Defines the type of column
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub enum ColumnKind {
@@ -563,6 +616,9 @@ pub trait BaseTable: std::fmt::Display + std::fmt::Debug + Send + Sync {
    async fn update(&self, update: UpdateBuilder) -> Result<UpdateResult>;
    /// Create an index on the provided column(s).
    async fn create_index(&self, index: IndexBuilder) -> Result<()>;
+
+    /// Starts index creation, returning a handle to the resulting job.
+    async fn create_index_async(&self, index: IndexBuilder) -> Result<Job>;
    /// List the indices on the table.
    async fn list_indices(&self) -> Result<Vec<IndexConfig>>;
    /// Drop an index from the table.
@@ -1566,12 +1622,8 @@ impl Table {
    }

    /// Add new columns to the table, providing values to fill in.
-    pub async fn add_columns(
-        &self,
-        transforms: NewColumnTransform,
-        read_columns: Option<Vec<String>>,
-    ) -> Result<AddColumnsResult> {
-        self.inner.add_columns(transforms, read_columns).await
+    pub fn add_columns(&self) -> AddColumnsBuilder {
+        AddColumnsBuilder::new(self.inner.clone())
    }

    /// Change a column's name or nullability.
@@ -2240,6 +2292,8 @@ impl NativeTable {
            None => false,
        };

+        // Kept so that a `DatasetNotFound` can be re-checked against storage below.
+        let recovery_params = params.clone();
        let mut builder = DatasetBuilder::from_uri(uri).with_read_params(params);

        // Set up commit handler when managed_versioning is enabled
@@ -2255,13 +2309,13 @@ impl NativeTable {
            builder = builder.with_commit_handler(commit_handler);
        }

-        let dataset = builder.load().await.map_err(|e| match e {
-            lance::Error::DatasetNotFound { .. } => Error::TableNotFound {
-                name: name.to_string(),
-                source: Box::new(e),
-            },
-            e => e.into(),
-        })?;
+        let dataset = match builder.load().await {
+            Ok(dataset) => dataset,
+            Err(e @ lance::Error::DatasetNotFound { .. }) => {
+                return Err(map_dataset_not_found(uri, name, recovery_params, e).await);
+            }
+            Err(e) => return Err(e.into()),
+        };

        let dataset = DatasetConsistencyWrapper::new_latest(dataset, read_consistency_interval);
        let id = Self::build_id(&namespace, name);
@@ -3009,29 +3063,18 @@ impl BaseTable for NativeTable {
    }

    async fn create_index(&self, opts: IndexBuilder) -> Result<()> {
-        if opts.columns.len() != 1 {
-            return Err(Error::Schema {
-                message: "Multi-column (composite) indices are not yet supported".to_string(),
-            });
-        }
-        self.dataset.ensure_mutable()?;
-        let mut dataset = (*self.dataset.get().await?).clone();
-        let (column, field) = Self::resolve_index_field(dataset.schema(), &opts.columns[0])?;
+        let prepared = self.prepare_index(&opts).await?;
+        self.build_index(opts, prepared).await
+    }

-        let lance_idx_params = self.make_index_params(&field, opts.index.clone()).await?;
-        let index_type = self.get_index_type_for_field(&field, &opts.index);
-        let columns = [column.as_str()];
-        let mut builder = dataset
-            .create_index_builder(&columns, index_type, lance_idx_params.as_ref())
-            .train(opts.train)
-            .replace(opts.replace);
-
-        if let Some(name) = opts.name {
-            builder = builder.name(name);
-        }
-        builder.await?;
-        self.dataset.update(dataset);
-        Ok(())
+    async fn create_index_async(&self, opts: IndexBuilder) -> Result<Job> {
+        // Prepare before spawning so bad input is reported by this call rather
+        // than only by the job.
+        let prepared = self.prepare_index(&opts).await?;
+        let table = self.clone();
+        Ok(Job::spawned(tokio::spawn(async move {
+            table.build_index(opts, prepared).await
+        })))
    }

    async fn drop_index(&self, index_name: &str) -> Result<()> {
@@ -3584,6 +3627,103 @@ mod tests {
        assert!(matches!(table.unwrap_err(), Error::TableNotFound { .. }));
    }

+    #[tokio::test]
+    async fn test_open_not_found_missing_lance_dir() {
+        let tmp_dir = tempdir().unwrap();
+        let dataset_path = tmp_dir.path().join("test.lance");
+
+        let err = NativeTable::open(dataset_path.to_str().unwrap())
+            .await
+            .unwrap_err();
+        assert!(
+            matches!(&err, Error::TableNotFound { name, .. } if name == "test"),
+            "got {err:?}"
+        );
+    }
+
+    /// Write a table and then break it, leaving the `<name>.lance` directory in place.
+    ///
+    /// `remove_all` reproduces an interrupted drop + re-create (the directory is left
+    /// empty); otherwise only the manifests are removed, leaving the data files behind.
+    async fn write_then_corrupt_table(dir: &std::path::Path, remove_all: bool) -> String {
+        let dataset_path = dir.join("test.lance");
+        let uri = dataset_path.to_str().unwrap().to_string();
+
+        let batch = make_test_batches();
+        let reader = RecordBatchIterator::new(vec![Ok(batch.clone())], batch.schema());
+        Dataset::write(reader, &uri, None).await.unwrap();
+
+        if remove_all {
+            for entry in std::fs::read_dir(&dataset_path).unwrap() {
+                let entry = entry.unwrap();
+                if entry.file_type().unwrap().is_dir() {
+                    std::fs::remove_dir_all(entry.path()).unwrap();
+                } else {
+                    std::fs::remove_file(entry.path()).unwrap();
+                }
+            }
+            assert_eq!(std::fs::read_dir(&dataset_path).unwrap().count(), 0);
+        } else {
+            let versions = dataset_path.join("_versions");
+            assert!(versions.is_dir(), "expected manifests under {versions:?}");
+            std::fs::remove_dir_all(&versions).unwrap();
+            assert!(std::fs::read_dir(&dataset_path).unwrap().count() > 0);
+        }
+
+        uri
+    }
+
+    #[tokio::test]
+    async fn test_open_corrupt_empty_dir() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = write_then_corrupt_table(tmp_dir.path(), true).await;
+
+        let err = NativeTable::open(&uri).await.unwrap_err();
+        assert!(
+            matches!(&err, Error::TableCorrupted { name, .. } if name == "test"),
+            "got {err:?}"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_open_corrupt_missing_manifest() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = write_then_corrupt_table(tmp_dir.path(), false).await;
+
+        let err = NativeTable::open(&uri).await.unwrap_err();
+        assert!(
+            matches!(&err, Error::TableCorrupted { name, .. } if name == "test"),
+            "got {err:?}"
+        );
+    }
+
+    /// A table listed by `table_names()` must not be reported as missing by
+    /// `open_table()`. See <https://github.com/lancedb/lancedb/issues/3127>.
+    #[tokio::test]
+    async fn test_open_table_corrupt_is_still_listed() {
+        let tmp_dir = tempdir().unwrap();
+        let db = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+
+        write_then_corrupt_table(tmp_dir.path(), true).await;
+
+        assert_eq!(
+            db.table_names().execute().await.unwrap(),
+            vec!["test".to_string()]
+        );
+        let err = db.open_table("test").execute().await.unwrap_err();
+        assert!(
+            matches!(&err, Error::TableCorrupted { name, .. } if name == "test"),
+            "got {err:?}"
+        );
+        assert!(
+            err.to_string().contains("exists but could not be loaded"),
+            "got {err}"
+        );
+    }
+
    #[test]
    #[cfg(not(windows))]
    fn test_object_store_path() {
@@ -4591,7 +4731,7 @@ mod tests {
                .set_lsm_write_spec(LsmWriteSpec::bucket("id", bad))
                .await
                .expect_err("should reject");
-            assert!(matches!(err, Error::Lance { .. }), "got {:?}", err);
+            assert!(matches!(err, Error::InvalidInput { .. }), "got {:?}", err);
        }

        // Happy path: install spec; verify MemWAL details record it.
@@ -0,0 +1,161 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Builder for adding columns to a table.
+
+use std::sync::Arc;
+
+use lance::dataset::NewColumnTransform;
+
+use super::BaseTable;
+use super::schema_evolution::AddColumnsResult;
+use crate::{Error, Result};
+
+/// Adds columns to a table. See [`Table::add_columns`](super::Table::add_columns).
+pub struct AddColumnsBuilder {
+    parent: Arc<dyn BaseTable>,
+    transform: Option<NewColumnTransform>,
+    read_columns: Option<Vec<String>>,
+}
+
+impl std::fmt::Debug for AddColumnsBuilder {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("AddColumnsBuilder")
+            .field("parent", &self.parent)
+            .field("has_transform", &self.transform.is_some())
+            .field("read_columns", &self.read_columns)
+            .finish()
+    }
+}
+
+impl AddColumnsBuilder {
+    pub(crate) fn new(parent: Arc<dyn BaseTable>) -> Self {
+        Self {
+            parent,
+            transform: None,
+            read_columns: None,
+        }
+    }
+
+    /// Set how the new columns' values are produced. Required.
+    pub fn transform(mut self, transform: NewColumnTransform) -> Self {
+        self.transform = Some(transform);
+        self
+    }
+
+    /// Limit which existing columns a [`NewColumnTransform::BatchUDF`] mapper
+    /// receives. Every other transform determines what it reads, so setting
+    /// this alongside one is an error rather than a silent no-op.
+    pub fn read_columns(mut self, columns: impl IntoIterator<Item = impl Into<String>>) -> Self {
+        self.read_columns = Some(columns.into_iter().map(Into::into).collect());
+        self
+    }
+
+    /// Add the columns.
+    pub async fn execute(self) -> Result<AddColumnsResult> {
+        let Self {
+            parent,
+            transform,
+            read_columns,
+        } = self;
+
+        let Some(transform) = transform else {
+            return Err(Error::InvalidInput {
+                message: "add_columns requires a transform".into(),
+            });
+        };
+
+        if read_columns.is_some() && !matches!(transform, NewColumnTransform::BatchUDF(_)) {
+            return Err(Error::InvalidInput {
+                message: "read_columns applies only to a BatchUDF transform; \
+                          every other transform determines what it reads"
+                    .into(),
+            });
+        }
+
+        parent.add_columns(transform, read_columns).await
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::sync::Arc;
+
+    use arrow_array::{Int32Array, RecordBatch, record_batch};
+    use arrow_schema::{DataType, Field, Schema};
+    use lance::dataset::{BatchUDF, NewColumnTransform};
+
+    use crate::Table;
+    use crate::connect;
+
+    async fn table_with_two_columns(name: &str) -> Table {
+        let conn = connect("memory://").execute().await.unwrap();
+        let batch = record_batch!(("x", Int32, [1, 2, 3]), ("y", Int32, [10, 20, 30])).unwrap();
+        conn.create_table(name, batch).execute().await.unwrap()
+    }
+
+    #[tokio::test]
+    async fn test_requires_a_transform() {
+        let table = table_with_two_columns("no_transform").await;
+        let err = table.add_columns().execute().await.unwrap_err();
+        assert!(
+            err.to_string().contains("requires a transform"),
+            "got: {err}"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_read_columns_with_sql_expressions_is_rejected() {
+        let table = table_with_two_columns("read_cols_sql").await;
+        let err = table
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "doubled".into(),
+                "x * 2".into(),
+            )]))
+            .read_columns(["x"])
+            .execute()
+            .await
+            .unwrap_err();
+        assert!(err.to_string().contains("BatchUDF"), "got: {err}");
+
+        let schema = table.schema().await.unwrap();
+        assert!(
+            schema.field_with_name("doubled").is_err(),
+            "a rejected call must not commit"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_read_columns_limits_what_a_batch_udf_sees() {
+        let table = table_with_two_columns("read_cols_udf").await;
+
+        let output_schema = Arc::new(Schema::new(vec![Field::new("sum", DataType::Int32, true)]));
+        let mapper_schema = output_schema.clone();
+        let udf = BatchUDF {
+            mapper: Box::new(move |batch: &RecordBatch| {
+                assert!(batch.column_by_name("x").is_some());
+                assert!(batch.column_by_name("y").is_none(), "y was not requested");
+                let x = batch["x"].as_any().downcast_ref::<Int32Array>().unwrap();
+                let doubled: Int32Array = x.iter().map(|v| v.map(|v| v * 2)).collect();
+                Ok(RecordBatch::try_new(
+                    mapper_schema.clone(),
+                    vec![Arc::new(doubled)],
+                )?)
+            }),
+            output_schema,
+            result_checkpoint: None,
+        };
+
+        table
+            .add_columns()
+            .transform(NewColumnTransform::BatchUDF(udf))
+            .read_columns(["x"])
+            .execute()
+            .await
+            .unwrap();
+
+        let schema = table.schema().await.unwrap();
+        assert!(schema.field_with_name("sum").is_ok());
+    }
+}
@@ -576,10 +576,12 @@ mod tests {

        // Add a new physical column AFTER the embedding column.
        table
-            .add_columns(
-                NewColumnTransform::SqlExpressions(vec![("score".into(), "42.0".into())]),
-                None,
-            )
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "score".into(),
+                "42.0".into(),
+            )]))
+            .execute()
            .await
            .unwrap();

@@ -683,7 +685,9 @@ mod tests {
            true,
        )]));
        table
-            .add_columns(NewColumnTransform::AllNulls(nested_schema), None)
+            .add_columns()
+            .transform(NewColumnTransform::AllNulls(nested_schema))
+            .execute()
            .await
            .unwrap();

@@ -21,6 +21,9 @@ use lance_index::vector::pq::PQBuildParams;
 use lance_index::vector::sq::builder::SQBuildParams;

 use crate::error::{Error, Result};
+
+/// Resolved column, index parameters and index type for one build.
+pub(super) type PreparedIndex = (String, Box<dyn lance::index::IndexParams>, IndexType);
 use crate::index::Index;
 use crate::index::vector::{VectorIndex, suggested_num_sub_vectors};
 use crate::utils::{
@@ -105,6 +108,47 @@ impl NativeTable {
        }
    }

+    /// Resolves the target column and index parameters, erroring on input the
+    /// build would reject.
+    pub(super) async fn prepare_index(
+        &self,
+        opts: &crate::index::IndexBuilder,
+    ) -> Result<PreparedIndex> {
+        if opts.columns.len() != 1 {
+            return Err(Error::Schema {
+                message: "Multi-column (composite) indices are not yet supported".to_string(),
+            });
+        }
+        self.dataset.ensure_mutable()?;
+        let dataset = self.dataset.get().await?;
+        let (column, field) = Self::resolve_index_field(dataset.schema(), &opts.columns[0])?;
+        let params = self.make_index_params(&field, opts.index.clone()).await?;
+        let index_type = self.get_index_type_for_field(&field, &opts.index);
+        Ok((column, params, index_type))
+    }
+
+    /// Builds a prepared index and publishes the new dataset version.
+    pub(super) async fn build_index(
+        &self,
+        opts: crate::index::IndexBuilder,
+        prepared: PreparedIndex,
+    ) -> Result<()> {
+        let (column, lance_idx_params, index_type) = prepared;
+        let mut dataset = (*self.dataset.get().await?).clone();
+        let columns = [column.as_str()];
+        let mut builder = dataset
+            .create_index_builder(&columns, index_type, lance_idx_params.as_ref())
+            .train(opts.train)
+            .replace(opts.replace);
+
+        if let Some(name) = opts.name {
+            builder = builder.name(name);
+        }
+        builder.await?;
+        self.dataset.update(dataset);
+        Ok(())
+    }
+
    pub(super) fn resolve_index_field(
        schema: &lance_core::datatypes::Schema,
        column: &str,
@@ -475,6 +519,208 @@ mod tests {
        assert_eq!(table.list_indices().await.unwrap().len(), 0);
    }

+    #[tokio::test]
+    async fn test_execute_async_job_waits_for_local_build() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+
+        let job = table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .execute_async()
+            .await
+            .unwrap();
+        // Local jobs run in this process and have no server id.
+        assert_eq!(job.id(), None);
+        // The build runs as a task, so the index need not exist yet; it must
+        // once the job resolves.
+        job.wait().await.unwrap();
+        assert_eq!(table.list_indices().await.unwrap().len(), 1);
+        // Cancelling a finished job is a no-op.
+        job.cancel().await.unwrap();
+    }
+
+    /// Concurrent waiters, and a wait issued after the job settled, all
+    /// succeed once the build does.
+    #[tokio::test]
+    async fn test_execute_async_job_reports_success_to_every_waiter() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+
+        let job = Arc::new(
+            table
+                .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+                .execute_async()
+                .await
+                .unwrap(),
+        );
+        let waiters = (0..4)
+            .map(|_| {
+                let job = job.clone();
+                tokio::spawn(async move { job.wait().await })
+            })
+            .collect::<Vec<_>>();
+        for waiter in waiters {
+            waiter.await.unwrap().unwrap();
+        }
+        // A wait after the job settled still reports the same outcome.
+        job.wait().await.unwrap();
+        assert_eq!(table.list_indices().await.unwrap().len(), 1);
+    }
+
+    /// Every waiter sees a failure, not just the first: a waiter that missed
+    /// the outcome would be told the job succeeded.
+    #[tokio::test]
+    async fn test_execute_async_job_reports_failure_to_every_waiter() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+        table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .execute()
+            .await
+            .unwrap();
+
+        // Rebuilding the same index without replace fails once the build
+        // starts, so the failure reaches the job rather than execute_async.
+        let job = Arc::new(
+            table
+                .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+                .replace(false)
+                .execute_async()
+                .await
+                .unwrap(),
+        );
+        let waiters = (0..3)
+            .map(|_| {
+                let job = job.clone();
+                tokio::spawn(async move { job.wait().await })
+            })
+            .collect::<Vec<_>>();
+        for waiter in waiters {
+            waiter
+                .await
+                .unwrap()
+                .expect_err("every waiter must see the failure");
+        }
+        job.wait().await.expect_err("a later wait still fails");
+    }
+
+    /// A local failure keeps the error it failed with, so a caller can match on
+    /// the original variant rather than parse a message.
+    #[tokio::test]
+    async fn test_execute_async_failure_keeps_the_source_error() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+        table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .execute()
+            .await
+            .unwrap();
+
+        let job = table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .replace(false)
+            .execute_async()
+            .await
+            .unwrap();
+
+        let crate::Error::JobFailed { failure, .. } = job.wait().await.unwrap_err() else {
+            panic!("a failed job reports JobFailed");
+        };
+        let source = failure.source.expect("a local failure carries its error");
+        assert_eq!(
+            failure.message.as_deref(),
+            Some(source.to_string()).as_deref()
+        );
+        // Nothing local can report these, so they must be absent rather than invented.
+        assert!(failure.phase.is_none());
+        assert!(failure.retryable.is_none());
+    }
+
+    /// Every waiter sees the cancellation, including ones that were already
+    /// waiting when the cancel landed.
+    #[tokio::test]
+    async fn test_execute_async_job_reports_cancellation_to_every_waiter() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+
+        let job = Arc::new(
+            table
+                .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+                .execute_async()
+                .await
+                .unwrap(),
+        );
+        // Cancel before yielding, so the build cannot have started and the
+        // outcome is always the cancellation.
+        job.cancel().await.unwrap();
+
+        let waiters = (0..2)
+            .map(|_| {
+                let job = job.clone();
+                tokio::spawn(async move { job.wait().await })
+            })
+            .collect::<Vec<_>>();
+        for waiter in waiters {
+            match waiter.await.unwrap() {
+                Err(crate::Error::JobCancelled { .. }) => {}
+                other => panic!("expected the cancellation, got {other:?}"),
+            }
+        }
+        match job.wait().await {
+            Err(crate::Error::JobCancelled { .. }) => {}
+            other => panic!("expected the cancellation, got {other:?}"),
+        }
+    }
+
+    #[tokio::test]
+    async fn test_execute_async_job_cancel_stops_local_build() {
+        let tmp_dir = tempdir().unwrap();
+        let conn = connect(tmp_dir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let batch = record_batch!(("id", Int32, (0..512).collect::<Vec<_>>())).unwrap();
+        let table = conn.create_table("t", batch).execute().await.unwrap();
+
+        let job = table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .execute_async()
+            .await
+            .unwrap();
+        job.cancel().await.unwrap();
+        match job.wait().await {
+            Err(crate::Error::JobCancelled { .. }) => {}
+            // The build may finish before the abort lands.
+            Ok(()) => {}
+            other => panic!("unexpected job outcome: {other:?}"),
+        }
+    }
+
    #[tokio::test]
    async fn test_ivf_pq_uses_default_partition_size_for_num_partitions() {
        use crate::index::vector::IvfPqIndexBuilder;
@@ -315,7 +315,10 @@ pub(crate) async fn execute_merge_insert(

 #[cfg(test)]
 mod tests {
-    use arrow_array::{Int32Array, RecordBatch, RecordBatchIterator, RecordBatchReader};
+    use arrow_array::builder::FixedSizeBinaryBuilder;
+    use arrow_array::{
+        Int32Array, RecordBatch, RecordBatchIterator, RecordBatchReader, StringArray, UInt64Array,
+    };
    use arrow_schema::{DataType, Field, Schema};
    use std::sync::Arc;

@@ -337,6 +340,42 @@ mod tests {
        Box::new(RecordBatchIterator::new(vec![Ok(batch)], schema))
    }

+    fn fixed_size_binary_merge_batch(
+        id_range: std::ops::Range<u64>,
+        price: u64,
+    ) -> Box<dyn RecordBatchReader + Send> {
+        let ids = id_range.collect::<Vec<_>>();
+        let mut id_builder = FixedSizeBinaryBuilder::new(16);
+        for id in &ids {
+            let mut bytes = [0; 16];
+            bytes[..8].copy_from_slice(&id.to_le_bytes());
+            id_builder.append_value(bytes).unwrap();
+        }
+
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("id", DataType::FixedSizeBinary(16), false),
+            Field::new("id_as_int", DataType::UInt64, false),
+            Field::new("name", DataType::Utf8, false),
+            Field::new("market", DataType::Utf8, false),
+        ]));
+        let batch = RecordBatch::try_new(
+            schema.clone(),
+            vec![
+                Arc::new(id_builder.finish()),
+                Arc::new(UInt64Array::from_iter_values(ids.iter().copied())),
+                Arc::new(StringArray::from_iter_values(
+                    ids.iter().map(|id| format!("name{id}")),
+                )),
+                Arc::new(StringArray::from_iter_values(std::iter::repeat_n(
+                    format!("market_{price}"),
+                    ids.len(),
+                ))),
+            ],
+        )
+        .unwrap();
+        Box::new(RecordBatchIterator::new(vec![Ok(batch)], schema))
+    }
+
    #[tokio::test]
    async fn test_merge_insert() {
        let conn = connect("memory://").execute().await.unwrap();
@@ -388,6 +427,36 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn test_merge_insert_fixed_size_binary_non_nullable() {
+        // Regression test for #2869: an unrelated FixedSizeBinary column used to corrupt the
+        // outer join that implements when_not_matched_by_source_delete.
+        let conn = connect("memory://").execute().await.unwrap();
+        let table = conn
+            .create_table(
+                "fixed_size_binary_merge",
+                fixed_size_binary_merge_batch(0..256, 100),
+            )
+            .execute()
+            .await
+            .unwrap();
+
+        let mut merge_insert = table.merge_insert(&["id_as_int"]);
+        merge_insert
+            .when_matched_update_all(None)
+            .when_not_matched_insert_all()
+            .when_not_matched_by_source_delete(None);
+        let result = merge_insert
+            .execute(fixed_size_binary_merge_batch(100..356, 200))
+            .await
+            .unwrap();
+
+        assert_eq!(result.num_updated_rows, 156);
+        assert_eq!(result.num_inserted_rows, 100);
+        assert_eq!(result.num_deleted_rows, 100);
+        assert_eq!(table.count_rows(None).await.unwrap(), 256);
+    }
+
    #[tokio::test]
    async fn test_merge_insert_use_index() {
        let conn = connect("memory://").execute().await.unwrap();
@@ -214,12 +214,17 @@ pub(crate) async fn execute_optimize(

 #[cfg(test)]
 mod tests {
-    use arrow_array::{Int32Array, RecordBatch, StringArray};
+    use arrow_array::{
+        Array, FixedSizeListArray, Float32Array, Int32Array, RecordBatch, StringArray,
+    };
    use arrow_schema::{DataType, Field, Schema};
+    use lance_arrow::FixedSizeListArrayExt;
    use rstest::rstest;
    use std::sync::Arc;

    use crate::connect;
+    use crate::database::listing::OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS;
+    use crate::index::vector::IvfRqIndexBuilder;
    use crate::index::{Index, scalar::BTreeIndexBuilder};
    use crate::query::ExecutableQuery;
    use crate::table::{CompactionOptions, OptimizeAction, OptimizeStats};
@@ -304,6 +309,96 @@ mod tests {
        assert_eq!(all_values, expected);
    }

+    #[tokio::test]
+    async fn test_compact_with_concurrent_add() {
+        const NUM_FRAGMENTS: usize = 5;
+        const ROWS_PER_FRAGMENT: i32 = 300;
+
+        let tmpdir = tempfile::tempdir().unwrap();
+        let conn = connect(tmpdir.path().to_str().unwrap())
+            .execute()
+            .await
+            .unwrap();
+        let schema = Arc::new(Schema::new(vec![Field::new("id", DataType::Int32, false)]));
+        let batch = RecordBatch::try_new(
+            schema,
+            vec![Arc::new(Int32Array::from_iter_values(0..ROWS_PER_FRAGMENT))],
+        )
+        .unwrap();
+
+        let table = conn
+            .create_table("test_concurrent_compact", batch.clone())
+            .execute()
+            .await
+            .unwrap();
+        table
+            .create_index(&["id"], Index::BTree(BTreeIndexBuilder::default()))
+            .execute()
+            .await
+            .unwrap();
+        for _ in 0..NUM_FRAGMENTS {
+            table.add(batch.clone()).execute().await.unwrap();
+        }
+
+        // Use separate handles so the two writes actually overlap, as they can
+        // when different Node connections operate on the same S3 table.
+        let compact_table = conn
+            .open_table("test_concurrent_compact")
+            .execute()
+            .await
+            .unwrap();
+        let append_table = conn
+            .open_table("test_concurrent_compact")
+            .execute()
+            .await
+            .unwrap();
+        let compact_task = tokio::spawn(async move {
+            compact_table
+                .optimize(OptimizeAction::Compact {
+                    options: CompactionOptions {
+                        target_rows_per_fragment: 1_000,
+                        ..Default::default()
+                    },
+                    remap_options: None,
+                })
+                .await
+        });
+        tokio::task::yield_now().await;
+        for _ in 0..NUM_FRAGMENTS {
+            append_table.add(batch.clone()).execute().await.unwrap();
+        }
+        compact_task.await.unwrap().unwrap();
+
+        let table = conn
+            .open_table("test_concurrent_compact")
+            .execute()
+            .await
+            .unwrap();
+        let dataset = table.dataset().unwrap().get().await.unwrap();
+        let fragment_ids = dataset
+            .get_fragments()
+            .iter()
+            .map(|fragment| fragment.id())
+            .collect::<Vec<_>>();
+        assert!(fragment_ids.windows(2).all(|ids| ids[0] < ids[1]));
+
+        // A second compaction exposed the original out-of-order row-id bug.
+        table
+            .optimize(OptimizeAction::Compact {
+                options: CompactionOptions {
+                    target_rows_per_fragment: 1_000,
+                    ..Default::default()
+                },
+                remap_options: None,
+            })
+            .await
+            .unwrap();
+        assert_eq!(
+            table.count_rows(None).await.unwrap(),
+            ROWS_PER_FRAGMENT as usize * (NUM_FRAGMENTS * 2 + 1)
+        );
+    }
+
    #[tokio::test]
    async fn test_optimize_prune_versions() {
        let conn = connect("memory://").execute().await.unwrap();
@@ -442,6 +537,58 @@ mod tests {
        assert_eq!(final_row_count, 200);
    }

+    #[tokio::test]
+    async fn test_optimize_vector_index_after_delete_with_stable_row_ids() {
+        const NUM_ROWS: i32 = 400;
+        const DIMENSION: i32 = 32;
+
+        let conn = connect("memory://").execute().await.unwrap();
+        let vectors = FixedSizeListArray::try_new_from_values(
+            Float32Array::from_iter_values((0..NUM_ROWS).flat_map(|id| {
+                (0..DIMENSION).map(move |offset| ((id as f32 * 0.1) + (offset as f32 * 0.3)).sin())
+            })),
+            DIMENSION,
+        )
+        .unwrap();
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("id", DataType::Int32, false),
+            Field::new("vector", vectors.data_type().clone(), false),
+        ]));
+        let batch = RecordBatch::try_new(
+            schema,
+            vec![
+                Arc::new(Int32Array::from_iter_values(0..NUM_ROWS)),
+                Arc::new(vectors),
+            ],
+        )
+        .unwrap();
+        let table = conn
+            .create_table("test_vector_index_optimize_after_delete", batch)
+            .storage_option(OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, "true")
+            .execute()
+            .await
+            .unwrap();
+
+        table
+            .create_index(
+                &["vector"],
+                Index::IvfRq(IvfRqIndexBuilder::default().num_partitions(4)),
+            )
+            .execute()
+            .await
+            .unwrap();
+        table.delete("id % 3 = 0").await.unwrap();
+
+        // Regression test for #3330: deleted stable row IDs used to become
+        // misaligned with row addresses while joining small IVF partitions.
+        table
+            .optimize(OptimizeAction::Index(Default::default()))
+            .await
+            .unwrap();
+
+        assert_eq!(table.count_rows(None).await.unwrap(), 266);
+    }
+
    #[tokio::test]
    async fn test_optimize_all() {
        let conn = connect("memory://").execute().await.unwrap();
@@ -193,10 +193,12 @@ mod tests {

        // Add a computed column
        let result = table
-            .add_columns(
-                NewColumnTransform::SqlExpressions(vec![("doubled".into(), "id * 2".into())]),
-                None,
-            )
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "doubled".into(),
+                "id * 2".into(),
+            )]))
+            .execute()
            .await
            .unwrap();

@@ -251,13 +253,12 @@ mod tests {

        // Add multiple columns at once
        table
-            .add_columns(
-                NewColumnTransform::SqlExpressions(vec![
-                    ("y".into(), "x + 1".into()),
-                    ("z".into(), "x * x".into()),
-                ]),
-                None,
-            )
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![
+                ("y".into(), "x + 1".into()),
+                ("z".into(), "x * x".into()),
+            ]))
+            .execute()
            .await
            .unwrap();

@@ -283,10 +284,12 @@ mod tests {

        // Add a column with a constant value
        table
-            .add_columns(
-                NewColumnTransform::SqlExpressions(vec![("constant".into(), "42".into())]),
-                None,
-            )
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "constant".into(),
+                "42".into(),
+            )]))
+            .execute()
            .await
            .unwrap();

@@ -659,10 +662,12 @@ mod tests {

        // Add column increments version
        let add_result = table
-            .add_columns(
-                NewColumnTransform::SqlExpressions(vec![("c".into(), "a + b".into())]),
-                None,
-            )
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "c".into(),
+                "a + b".into(),
+            )]))
+            .execute()
            .await
            .unwrap();
        assert!(add_result.version > v1);
@@ -9,14 +9,17 @@ use arrow_array::{
 };
 use arrow_schema::{DataType, Field, Fields, Schema};
 use futures::TryStreamExt;
+use lance::Dataset;
 use lance_encoding::version::LanceFileVersion;
 use lancedb::{
    Connection, Error, Result, Table,
    blob::{BlobRangeRequest, blob},
    connect, connect_namespace,
-    database::listing::OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS,
+    database::listing::{
+        ListingDatabaseOptions, NewTableConfig, OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS,
+    },
    query::{ExecutableQuery, QueryBase},
-    table::{AddDataMode, CompactionOptions, OptimizeAction},
+    table::{AddDataMode, CompactionOptions, OptimizeAction, OptimizeStats},
 };
 use tempfile::tempdir;

@@ -647,7 +650,7 @@ async fn fetch_blob_ranges_validates_requests() -> Result<()> {
        .await
        .unwrap_err();
    assert!(matches!(&err, Error::InvalidInput { .. }), "got {err:?}");
-    assert!(err.to_string().contains("row ids"));
+    assert!(err.to_string().contains("row IDs"));
    Ok(())
 }

@@ -680,7 +683,7 @@ async fn fetch_blobs_out_of_range_id_errors_without_panic() -> Result<()> {
    let table = create_inline_blob_table(&db, "t", &[1], &[Some(b"x".as_slice())]).await?;

    let err = table.fetch_blobs("image", &[u64::MAX]).await.unwrap_err();
-    assert!(err.to_string().contains("row ids"));
+    assert!(err.to_string().contains("row IDs"));
    Ok(())
 }

@@ -694,11 +697,11 @@ async fn fetch_blob_apis_reject_mixed_valid_and_missing_row_ids() -> Result<()>

    let err = table.fetch_blobs("image", &row_ids).await.unwrap_err();
    assert!(matches!(&err, Error::InvalidInput { .. }), "got {err:?}");
-    assert!(err.to_string().contains("row ids"));
+    assert!(err.to_string().contains("row IDs"));

    let err = table.fetch_blob_files("image", &row_ids).await.unwrap_err();
    assert!(matches!(&err, Error::InvalidInput { .. }), "got {err:?}");
-    assert!(err.to_string().contains("row ids"));
+    assert!(err.to_string().contains("row IDs"));

    let requests = row_ids.map(|row_id| BlobRangeRequest::new(row_id, 0, 1));
    let err = table
@@ -706,7 +709,7 @@ async fn fetch_blob_apis_reject_mixed_valid_and_missing_row_ids() -> Result<()>
        .await
        .unwrap_err();
    assert!(matches!(&err, Error::InvalidInput { .. }), "got {err:?}");
-    assert!(err.to_string().contains("row ids"));
+    assert!(err.to_string().contains("row IDs"));
    Ok(())
 }

@@ -1075,3 +1078,252 @@ async fn fetch_blob_files_aligns_across_fragments_with_nulls_and_dups() -> Resul
    }
    Ok(())
 }
+
+/// Rows exercising the null/empty interleavings from
+/// <https://github.com/lancedb/lancedb/issues/3744>: a payload, a null, a valid
+/// empty value, then payloads whose descriptors a fragment rewrite used to zero.
+fn null_empty_input_batch() -> RecordBatch {
+    let owned = [
+        Some(dedicated_blob_bytes(1)),
+        None,
+        Some(Vec::new()),
+        Some(dedicated_blob_bytes(4)),
+        Some(dedicated_blob_bytes(5)),
+        Some(dedicated_blob_bytes(6)),
+    ];
+    let payloads: Vec<Option<&[u8]>> = owned.iter().map(|payload| payload.as_deref()).collect();
+    binary_input_batch(&[1, 2, 3, 4, 5, 6], &payloads)
+}
+
+/// One `(id, Some((payload length, first byte)))` per live row, or `(id, None)`
+/// for a null blob.  Comparing lengths and first bytes keeps failure output
+/// readable where comparing whole payloads would not.
+type BlobSummary = Vec<(i64, Option<(usize, Option<u8>)>)>;
+
+/// The rows [`null_empty_input_batch`] leaves behind after `id IN (1, 4)` is
+/// deleted: a null, a valid empty value, and the two payloads that follow them.
+fn expected_null_empty_survivors() -> BlobSummary {
+    vec![
+        (2, None),
+        (3, Some((0, None))),
+        (5, Some((DEDICATED_BLOB_LEN, Some(5)))),
+        (6, Some((DEDICATED_BLOB_LEN, Some(6)))),
+    ]
+}
+
+/// `optimize()` only rewrites a fragment when lance's compaction planner selects
+/// it — here because the delete pushes the fragment past
+/// `materialize_deletions_threshold` (0.1 by default; these tests delete 2 of 6
+/// rows). Without this check, a planner or threshold change upstream would leave
+/// both regression tests green while no rewrite happened at all.
+fn assert_compacted(stats: &OptimizeStats) {
+    let metrics = stats
+        .compaction
+        .as_ref()
+        .expect("OptimizeAction::All runs compaction");
+    assert!(
+        metrics.fragments_removed >= 1,
+        "optimize() rewrote no fragment, so this test proves nothing: {metrics:?}"
+    );
+}
+
+fn summarize(rows: &[(i64, Option<Vec<u8>>)]) -> BlobSummary {
+    rows.iter()
+        .map(|(id, payload)| {
+            (
+                *id,
+                payload
+                    .as_ref()
+                    .map(|bytes| (bytes.len(), bytes.first().copied())),
+            )
+        })
+        .collect()
+}
+
+async fn sorted_id_rowid(table: &Table) -> Result<Vec<(i64, u64)>> {
+    let mut pairs = collect_id_rowid(table).await?;
+    pairs.sort_by_key(|(id, _)| *id);
+    Ok(pairs)
+}
+
+/// `{position, size}` descriptors of a legacy v1 blob column, keyed by `id`.
+async fn v1_blob_descriptors(table: &Table) -> Result<Vec<(i64, Option<(u64, u64)>)>> {
+    let batches = table
+        .query()
+        .execute()
+        .await?
+        .try_collect::<Vec<_>>()
+        .await?;
+    let batch = arrow_select::concat::concat_batches(&batches[0].schema(), &batches).unwrap();
+    let ids = batch
+        .column_by_name("id")
+        .unwrap()
+        .as_any()
+        .downcast_ref::<Int64Array>()
+        .unwrap();
+    let descriptors = batch
+        .column_by_name("image")
+        .unwrap()
+        .as_any()
+        .downcast_ref::<StructArray>()
+        .expect("v1 blob column reads back as a descriptor struct");
+    let position = descriptors
+        .column_by_name("position")
+        .unwrap()
+        .as_any()
+        .downcast_ref::<UInt64Array>()
+        .unwrap();
+    let size = descriptors
+        .column_by_name("size")
+        .unwrap()
+        .as_any()
+        .downcast_ref::<UInt64Array>()
+        .unwrap();
+    let mut rows: Vec<(i64, Option<(u64, u64)>)> = (0..batch.num_rows())
+        .map(|row| {
+            let descriptor =
+                (!descriptors.is_null(row)).then(|| (position.value(row), size.value(row)));
+            (ids.value(row), descriptor)
+        })
+        .collect();
+    rows.sort_by_key(|(id, _)| *id);
+    Ok(rows)
+}
+
+/// Payload bytes of every live row of a legacy v1 blob column, keyed by `id`.
+/// [`Table::fetch_blobs`] rejects v1 columns, so read them through lance.
+async fn v1_blob_payloads(dataset_uri: &str, table: &Table) -> Result<Vec<(i64, Option<Vec<u8>>)>> {
+    let pairs = sorted_id_rowid(table).await?;
+    let row_ids: Vec<u64> = pairs.iter().map(|(_, row_id)| *row_id).collect();
+    let dataset = Arc::new(Dataset::open(dataset_uri).await?);
+    let files = dataset.take_blobs(&row_ids, "image").await?;
+    assert_eq!(
+        files.len(),
+        pairs.len(),
+        "take_blobs returned {} handles for {} live rows",
+        files.len(),
+        pairs.len()
+    );
+    let mut rows = Vec::with_capacity(pairs.len());
+    for ((id, _), file) in pairs.iter().zip(files) {
+        let payload = match file {
+            Some(file) => Some(file.read().await?.to_vec()),
+            None => None,
+        };
+        rows.push((*id, payload));
+    }
+    Ok(rows)
+}
+
+/// Length and first byte of every live blob v2 value, keyed by `id`.
+async fn blob_v2_values(table: &Table) -> Result<BlobSummary> {
+    let pairs = sorted_id_rowid(table).await?;
+    let row_ids: Vec<u64> = pairs.iter().map(|(_, row_id)| *row_id).collect();
+    let bytes = table.fetch_blobs("image", &row_ids).await?;
+    Ok(pairs
+        .iter()
+        .enumerate()
+        .map(|(slot, (id, _))| {
+            let value = (!bytes.is_null(slot))
+                .then(|| (bytes.value(slot).len(), bytes.value(slot).first().copied()));
+            (*id, value)
+        })
+        .collect())
+}
+
+/// Regression test for [#3744]: on storage 2.0 (legacy v1 descriptors),
+/// compaction rewrote every payload following a null or empty value in the same
+/// fragment as `{position: 0, size: 0}`, so the payload bytes read back as `b""`
+/// and the new fragment no longer referenced them at all.
+///
+/// [#3744]: https://github.com/lancedb/lancedb/issues/3744
+#[tokio::test]
+async fn optimize_preserves_v1_blob_payloads_with_null_and_empty() -> Result<()> {
+    let tmp = tempdir().unwrap();
+    let db_uri = tmp.path().to_str().unwrap().to_string();
+    let db = connect(&db_uri)
+        .database_options(&ListingDatabaseOptions {
+            new_table_config: NewTableConfig {
+                data_storage_version: Some(LanceFileVersion::V2_0),
+                ..Default::default()
+            },
+            ..Default::default()
+        })
+        .execute()
+        .await?;
+    let legacy = Field::new("image", DataType::LargeBinary, true).with_metadata(
+        std::collections::HashMap::from([("lance-encoding:blob".to_string(), "true".to_string())]),
+    );
+    let schema = Arc::new(Schema::new(vec![
+        Field::new("id", DataType::Int64, false),
+        legacy,
+    ]));
+    let table = db.create_empty_table("t", schema).execute().await?;
+    table.add(null_empty_input_batch()).execute().await?;
+    assert_eq!(
+        storage_format_version(&table).await,
+        LanceFileVersion::V2_0.resolve(),
+        "v1 blob descriptors only exist below storage 2.2"
+    );
+    let dataset_uri = table.uri().await?;
+
+    // Any rewrite triggers it; deleting rows is the shape from the issue.
+    table.delete("id IN (1, 4)").await?;
+    let descriptors_before = v1_blob_descriptors(&table).await?;
+    let before = v1_blob_payloads(&dataset_uri, &table).await?;
+    assert_eq!(
+        summarize(&before),
+        expected_null_empty_survivors(),
+        "test setup no longer produces the null/empty/payload mix"
+    );
+
+    let stats = table.optimize(OptimizeAction::All).await?;
+    assert_compacted(&stats);
+
+    let descriptors_after = v1_blob_descriptors(&table).await?;
+    let after = v1_blob_payloads(&dataset_uri, &table).await?;
+    assert_eq!(
+        summarize(&after),
+        summarize(&before),
+        "optimize() lost blob payloads; descriptors before={descriptors_before:?} after={descriptors_after:?}"
+    );
+    assert!(after == before, "optimize() changed blob payload bytes");
+    Ok(())
+}
+
+/// Regression test for the blob v2 half of [#3744]: compaction rewrote a valid
+/// empty value as null, destroying the null-vs-empty distinction.
+///
+/// [#3744]: https://github.com/lancedb/lancedb/issues/3744
+#[tokio::test]
+async fn optimize_preserves_blob_v2_null_and_empty_distinction() -> Result<()> {
+    let tmp = tempdir().unwrap();
+    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
+    let table = db
+        .create_empty_table("t", blob_table_schema())
+        .execute()
+        .await?;
+    table.add(null_empty_input_batch()).execute().await?;
+    assert!(
+        storage_format_version(&table).await >= LanceFileVersion::V2_2,
+        "blob v2 columns require storage >= 2.2"
+    );
+
+    table.delete("id IN (1, 4)").await?;
+    let before = blob_v2_values(&table).await?;
+    assert_eq!(
+        before,
+        expected_null_empty_survivors(),
+        "test setup no longer produces the null/empty/payload mix"
+    );
+
+    let stats = table.optimize(OptimizeAction::All).await?;
+    assert_compacted(&stats);
+
+    assert_eq!(
+        blob_v2_values(&table).await?,
+        before,
+        "optimize() changed blob v2 values"
+    );
+    Ok(())
+}