docs: state what a computed column promises

The declaration API shipped without a runnable example, and none of the three binding docs said when values appear, what happens to them when an input changes, which schema operations a declaration blocks, or that the feature is local-only. Those are the questions a caller has to answer before using it. Adds Rust doctests on both entry points and the same semantics to the Python and TypeScript parameter docs, plus a worked Python example. Also adds the abstract refresh_column that both concrete Python tables already implemented, so the surface is declared in one place and the cross-references resolve.
fix: make computed columns explicitly local-only
2026-08-28 00:48:40 +00:00 · 2026-08-12 21:01:11 -07:00 · 2026-08-12 20:00:18 -07:00 · 2026-08-12 20:00:18 -07:00 · 2026-08-12 20:00:18 -07:00 · 2026-08-12 20:00:18 -07:00
61 changed files with 3412 additions and 849 deletions
@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.37.1"
+current_version = "0.37.1-beta.1"
 parse = """(?x)
    (?P<major>0|[1-9]\\d*)\\.
    (?P<minor>0|[1-9]\\d*)\\.
@@ -36,7 +36,9 @@ jobs:
    permissions:
      contents: read
    outputs:
+      checker_outcome: ${{ steps.lychee.outcome }}
      exit_code: ${{ steps.lychee.outputs.exit_code }}
+      status: ${{ steps.validate.outputs.status }}
    steps:
      - name: Checkout
        uses: actions/checkout@v6
@@ -50,6 +52,7 @@ jobs:

      - name: Check links
        id: lychee
+        continue-on-error: true
        uses: lycheeverse/lychee-action@e7477775783ea5526144ba13e8db5eec57747ce8 # v2.9.0
        with:
          # Restricted to http(s) on purpose. Much of docs/src is generated
@@ -68,38 +71,50 @@ jobs:
          format: json
          output: ./lychee/out.json
          jobSummary: false
-          # The report, not a red build, is the signal for broken links. The
-          # validation step below still fails the run if the check itself
-          # breaks.
+          # The report issue, not a red workflow run, is the signal for link
+          # findings and checker failures alike.
          fail: false

      - name: Validate report
+        id: validate
        # lychee does not reserve exit code 2 for broken links: its CLI
        # parser also exits 2 on an invalid option, before any link was
        # checked or any report written. Only a parseable report whose
-        # counts agree with the exit code counts as a link verdict; anything
-        # else fails here, and the report job below is skipped entirely, so
-        # the tracking issue is never touched. Exit 2 covers timeouts as
-        # well as errors, and a timed-out host is exactly the transient
-        # unavailability this report exists to surface, so both count as
-        # findings. Requiring total > 0 also catches a glob that silently
-        # stopped matching any file.
-        if: steps.lychee.outputs.exit_code == 0 || steps.lychee.outputs.exit_code == 2
+        # counts agree with a completed exit code (0 or 2) counts as a link
+        # verdict. Everything else becomes a checker-error report instead of
+        # failing the workflow. Exit 2 covers timeouts as well as errors, and a
+        # timed-out host is exactly the transient unavailability this report
+        # exists to surface, so both count as findings. Requiring total > 0
+        # also catches a glob that silently stopped matching any file.
+        if: always()
        env:
+          CHECKER_OUTCOME: ${{ steps.lychee.outcome }}
          EXIT_CODE: ${{ steps.lychee.outputs.exit_code }}
        run: |
-          jq -e --argjson code "$EXIT_CODE" '
-            (.total > 0) and
-            (if $code == 0
-             then .errors == 0 and .timeouts == 0
-               and (.error_map | length == 0) and (.timeout_map | length == 0)
-             else (.errors + .timeouts) > 0
-               and ((.error_map | length) + (.timeout_map | length)) > 0
-             end)
-          ' ./lychee/out.json
+          status=checker-error
+          if [[ "$CHECKER_OUTCOME" == success ]] &&
+             [[ "$EXIT_CODE" == 0 || "$EXIT_CODE" == 2 ]] &&
+             jq -e --argjson code "$EXIT_CODE" '
+               (.total > 0) and
+               (if $code == 0
+                then .errors == 0 and .timeouts == 0
+                  and (.error_map | length == 0) and (.timeout_map | length == 0)
+                else (.errors + .timeouts) > 0
+                  and ((.error_map | length) + (.timeout_map | length)) > 0
+                end)
+             ' ./lychee/out.json
+          then
+            if [[ "$EXIT_CODE" == 0 ]]; then
+              status=healthy
+            else
+              status=findings
+            fi
+          fi
+          echo "status=$status" >> "$GITHUB_OUTPUT"
+          echo "Validated link check as $status"

      - name: Upload report
-        if: steps.lychee.outputs.exit_code == 2
+        if: steps.validate.outputs.status == 'findings'
        uses: actions/upload-artifact@v7
        with:
          name: link-report
@@ -115,26 +130,11 @@ jobs:
    permissions:
      issues: write
    env:
+      CHECKER_OUTCOME: ${{ needs.scan.outputs.checker_outcome }}
      EXIT_CODE: ${{ needs.scan.outputs.exit_code }}
+      STATUS: ${{ needs.scan.outputs.status }}
      GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
    steps:
-      - name: Classify checker result
-        # lychee exits 0 when every link resolves and 2 when links fail,
-        # both already cross-checked against the report by the scan job's
-        # validation step. Anything else (1 runtime, 3 bad config) means the
-        # check never produced a link verdict, which must surface as a failed
-        # run rather than be published as "broken documentation links".
-        run: |
-          case "$EXIT_CODE" in
-            0|2)
-              echo "lychee exit code $EXIT_CODE"
-              ;;
-            *)
-              echo "::error::lychee exited with '$EXIT_CODE': the link check did not complete. Leaving the report issue untouched."
-              exit 1
-              ;;
-          esac
-
      - name: Find existing report issue
        id: report
        # Matched on title alone, and through search rather than a listing:
@@ -144,7 +144,7 @@ jobs:
        # Closed issues are included because a healthy run closes the report:
        # an open-only lookup would forget that identity and the next failing
        # run would open a duplicate. The oldest match stays the canonical
-        # report and is reopened below when links break again.
+        # report and is reopened below when a problem recurs.
        run: |
          match=$(gh issue list --repo "$GITHUB_REPOSITORY" --state all \
            --search "in:title \"$REPORT_TITLE\" author:app/github-actions" \
@@ -154,14 +154,14 @@ jobs:
          echo "state=$(jq -r '.state // empty' <<<"$match")" >> "$GITHUB_OUTPUT"

      - name: Download report
-        if: env.EXIT_CODE == 2
+        if: env.STATUS == 'findings'
        uses: actions/download-artifact@v8
        with:
          name: link-report
          path: ./lychee

      - name: Compose report
-        if: env.EXIT_CODE == 2
+        if: env.STATUS == 'findings'
        run: |
          run_url="$GITHUB_SERVER_URL/$GITHUB_REPOSITORY/actions/runs/$GITHUB_RUN_ID"
          {
@@ -185,22 +185,41 @@ jobs:
            ' ./lychee/out.json
          } > ./lychee/issue.md

+      - name: Compose checker error report
+        if: env.STATUS == 'checker-error'
+        run: |
+          mkdir -p ./lychee
+          run_url="$GITHUB_SERVER_URL/$GITHUB_REPOSITORY/actions/runs/$GITHUB_RUN_ID"
+          {
+            echo "The documentation link check did not complete in [the latest run]($run_url)."
+            echo
+            echo "This issue is rewritten by every scheduled run and closed automatically once a trustworthy run finds that all links resolve."
+            echo
+            echo "The checker did not produce a trustworthy link verdict. Treat the previous result, if any, as stale until a later run completes."
+            echo
+            echo "* Action outcome: \`$CHECKER_OUTCOME\`"
+            echo "* Exit code: \`${EXIT_CODE:-not reported}\`"
+            echo "* Verdict validation: \`failed\`"
+          } > ./lychee/issue.md
+
      - name: Reopen report issue
        # A healthy run closes the report, and the issue action below only
        # rewrites the body of whatever number it is given. Without an
-        # explicit reopen, the 2 -> 0 -> 2 sequence would keep rewriting a
-        # closed issue while links are broken. A CLOSED state implies the
-        # lookup found a canonical issue, so no separate emptiness check.
-        if: env.EXIT_CODE == 2 && steps.report.outputs.state == 'CLOSED'
+        # explicit reopen, a later finding or checker error would rewrite a
+        # closed issue. A CLOSED state implies the lookup found a canonical
+        # issue, so no separate emptiness check.
+        if: >-
+          env.STATUS != 'healthy' &&
+          steps.report.outputs.state == 'CLOSED'
        env:
          ISSUE_NUMBER: ${{ steps.report.outputs.number }}
        run: |
          run_url="$GITHUB_SERVER_URL/$GITHUB_REPOSITORY/actions/runs/$GITHUB_RUN_ID"
          gh issue reopen "$ISSUE_NUMBER" --repo "$GITHUB_REPOSITORY" \
-            --comment "Broken documentation links found again in [the latest run]($run_url)."
+            --comment "The documentation link checker reported a problem again in [the latest run]($run_url)."

-      - name: Report broken links
-        if: env.EXIT_CODE == 2
+      - name: Report link-check problem
+        if: env.STATUS != 'healthy'
        uses: peter-evans/create-issue-from-file@fca9117c27cdc29c6c4db3b86c48e4115a786710 # v6.0.0
        with:
          # Empty on the first failing run, which creates the issue; afterwards
@@ -213,7 +232,9 @@ jobs:
      - name: Close report issue once links are healthy
        # An OPEN state implies the lookup found a canonical issue; a report
        # that is already closed needs nothing.
-        if: env.EXIT_CODE == 0 && steps.report.outputs.state == 'OPEN'
+        if: >-
+          env.STATUS == 'healthy' &&
+          steps.report.outputs.state == 'OPEN'
        env:
          ISSUE_NUMBER: ${{ steps.report.outputs.number }}
        run: |
@@ -1,8 +1,7 @@
 name: Create release commit

-# This workflow increments the version, updates lockfiles, tags the final
-# commit, and pushes it. All SDKs share a single version, so one tag releases
-# all of them.
+# This workflow increments the version, tags it, and pushes it. All SDKs share
+# a single version, so one tag releases all of them.
 # When a tag is pushed, another workflow is triggered that creates a GH release
 # and uploads the binaries. This workflow is only for creating the tag.

@@ -64,7 +63,6 @@ jobs:
          pip install bump-my-version PyGithub packaging
          bash ci/bump_version.sh ${{ inputs.type }} ${{ inputs.bump-minor }}
          bash ci/update_lockfiles.sh --amend
-          bash ci/create_release_tag.sh
      - name: Push new version tag
        if: ${{ !inputs.dry_run }}
        uses: ad-m/github-push-action@881a6320fdb16eb5318c5054f31c218aec2b324c # v1.3.0
@@ -4,16 +4,6 @@ on:
  push:
    tags:
      - 'v*'
-  workflow_dispatch:
-    inputs:
-      release_tag:
-        description: Stable release tag to recover (for example, v0.37.1)
-        required: true
-        type: string
-      source_run_id:
-        description: Failed tag workflow run containing the completed non-Windows wheels
-        required: true
-        type: string
  pull_request:
    # This should trigger a dry run (we skip the final publish step)
    paths:
@@ -39,7 +29,6 @@ concurrency:
 jobs:
  linux:
    name: Python ${{ matrix.config.package_name }} ${{ matrix.config.platform }} manylinux${{ matrix.config.manylinux }}
-    if: github.event_name != 'workflow_dispatch'
    timeout-minutes: 60
    strategy:
      matrix:
@@ -95,7 +84,6 @@ jobs:
          path: target/wheels/*.whl
          if-no-files-found: error
  mac:
-    if: github.event_name != 'workflow_dispatch'
    timeout-minutes: 90
    runs-on: ${{ matrix.config.runner }}
    strategy:
@@ -125,8 +113,7 @@ jobs:
          path: target/wheels/lancedb-*.whl
          if-no-files-found: error
  windows:
-    if: github.event_name != 'workflow_dispatch'
-    timeout-minutes: 120
+    timeout-minutes: 90
    runs-on: windows-latest
    env:
      # link.exe is single-threaded and the long pole on Windows builds. Use
@@ -158,32 +145,6 @@ jobs:
          name: wheels-windows
          path: target/wheels/lancedb-*.whl
          if-no-files-found: error
-  recover-windows:
-    name: Recover Windows wheel
-    if: github.event_name == 'workflow_dispatch'
-    timeout-minutes: 120
-    runs-on: windows-latest
-    env:
-      CARGO_TARGET_X86_64_PC_WINDOWS_MSVC_LINKER: rust-lld
-    steps:
-      - uses: actions/checkout@v6
-        with:
-          ref: ${{ inputs.release_tag }}
-          fetch-depth: 0
-          lfs: true
-      - name: Set up Python
-        uses: actions/setup-python@v6
-        with:
-          python-version: "3.13"
-      - uses: ./.github/workflows/build_windows_wheel
-        with:
-          python-minor-version: 10
-          args: "--release --strip"
-      - uses: actions/upload-artifact@v7
-        with:
-          name: wheels-windows-recovery
-          path: target/wheels/lancedb-*.whl
-          if-no-files-found: error
  publish:
    name: Publish wheels
    if: startsWith(github.ref, 'refs/tags/v')
@@ -204,13 +165,11 @@ jobs:
        run: ls -la target/wheels
      - name: Choose repo
        id: choose_repo
-        env:
-          RELEASE_REF: ${{ github.ref }}
        run: |
-          if [[ "$RELEASE_REF" == *beta* ]]; then
-            echo "repo=fury" >> "$GITHUB_OUTPUT"
+          if [[ ${{ github.ref }} == *beta* ]]; then
+            echo "repo=fury" >> $GITHUB_OUTPUT
          else
-            echo "repo=pypi" >> "$GITHUB_OUTPUT"
+            echo "repo=pypi" >> $GITHUB_OUTPUT
          fi
      - name: Publish to Fury
        if: steps.choose_repo.outputs.repo == 'fury'
@@ -237,71 +196,6 @@ jobs:
        uses: pypa/gh-action-pypi-publish@release/v1
        with:
          packages-dir: target/wheels/
-  recover-publish:
-    name: Recover PyPI publish
-    if: github.event_name == 'workflow_dispatch'
-    needs: [recover-windows]
-    runs-on: ubuntu-latest
-    permissions:
-      actions: read
-      id-token: write
-      contents: read
-    steps:
-      - uses: actions/checkout@v6
-        with:
-          ref: ${{ inputs.release_tag }}
-          fetch-depth: 0
-      - name: Verify recovery source
-        env:
-          GH_TOKEN: ${{ github.token }}
-          RELEASE_TAG: ${{ inputs.release_tag }}
-          SOURCE_RUN_ID: ${{ inputs.source_run_id }}
-        run: |
-          if [[ "$RELEASE_TAG" != v* || "$RELEASE_TAG" == *beta* ]]; then
-            echo "Recovery only supports stable v* release tags" >&2
-            exit 1
-          fi
-          TAG_SHA=$(git rev-parse HEAD)
-          RUN_SHA=$(gh api "/repos/${{ github.repository }}/actions/runs/$SOURCE_RUN_ID" --jq .head_sha)
-          if [[ "$TAG_SHA" != "$RUN_SHA" ]]; then
-            echo "Source run $SOURCE_RUN_ID ($RUN_SHA) does not match $RELEASE_TAG ($TAG_SHA)" >&2
-            exit 1
-          fi
-      - name: Download Linux wheel artifacts
-        uses: actions/download-artifact@v8
-        with:
-          github-token: ${{ github.token }}
-          repository: ${{ github.repository }}
-          run-id: ${{ inputs.source_run_id }}
-          pattern: wheels-linux-*
-          path: target/wheels
-          merge-multiple: true
-      - name: Download macOS wheel artifacts
-        uses: actions/download-artifact@v8
-        with:
-          github-token: ${{ github.token }}
-          repository: ${{ github.repository }}
-          run-id: ${{ inputs.source_run_id }}
-          pattern: wheels-mac-*
-          path: target/wheels
-          merge-multiple: true
-      - name: Download recovered Windows wheel
-        uses: actions/download-artifact@v8
-        with:
-          name: wheels-windows-recovery
-          path: target/wheels
-      - name: Validate recovered wheels
-        run: |
-          find target/wheels -maxdepth 1 -type f -name '*.whl' -print
-          WHEEL_COUNT=$(find target/wheels -maxdepth 1 -type f -name '*.whl' | wc -l)
-          if [[ "$WHEEL_COUNT" -ne 5 ]]; then
-            echo "Expected 5 wheels, found $WHEEL_COUNT" >&2
-            exit 1
-          fi
-      - name: Publish to PyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: target/wheels/
  report-failure:
    name: Report Workflow Failure
    runs-on: ubuntu-latest
@@ -775,7 +775,7 @@ dependencies = [
 "http 0.2.12",
 "http 1.5.0",
 "http-body 1.1.0",
- "lru",
+ "lru 0.16.4",
 "percent-encoding",
 "regex-lite",
 "sha2 0.11.0",
@@ -1970,15 +1970,6 @@ dependencies = [
 "spin 0.10.1",
 ]

-[[package]]
-name = "crc32c"
-version = "0.6.8"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3a47af21622d091a8f0fb295b88bc886ac74efcc613efc19f5d0b21de5c89e47"
-dependencies = [
- "rustc_version",
-]
-
 [[package]]
 name = "crc32fast"
 version = "1.5.0"
@@ -3450,6 +3441,12 @@ dependencies = [
 "percent-encoding",
 ]

+[[package]]
+name = "frostem"
+version = "1.20260804.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "82eb03a32a1d50555353c85a7b9d3279a6f1e91af9890b789acdf544ed57c8d7"
+
 [[package]]
 name = "fs_extra"
 version = "1.3.0"
@@ -3458,9 +3455,8 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"

 [[package]]
 name = "fsst"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d0981ce90521824089f3cd68a48b1c4c89e9ad96d0eb74f58d6e550a3d604545"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "rand 0.9.5",
@@ -3817,14 +3813,23 @@ dependencies = [

 [[package]]
 name = "goosefs-sdk"
-version = "0.1.5"
+version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9ae079b88ffe7772d12cfc5c40a5a324babb357893d95b5e3a22ae857f236c5f"
+checksum = "e1ea4eee6dcbc31b25ab4fd577adc55b677d2bed3aa3016c44c58fbe1b2298a5"
 dependencies = [
+ "arc-swap",
 "async-trait",
 "bytes",
 "dashmap",
+ "fastrand",
+ "futures",
 "hostname",
+ "io-uring",
+ "itoa",
+ "libc",
+ "lru 0.18.2",
+ "memmap2 0.9.10",
+ "moka",
 "prost",
 "prost-types",
 "rand 0.9.5",
@@ -3837,6 +3842,7 @@ dependencies = [
 "tonic-prost",
 "tracing",
 "uuid",
+ "xxhash-rust",
 ]

 [[package]]
@@ -4809,9 +4815,8 @@ checksum = "e037a2e1d8d5fdbd49b16a4ea09d5d6401c1f29eca5ff29d03d3824dba16256a"

 [[package]]
 name = "lance"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "830548f9fc92ae74b848a504282d4da06f016f2ee94b663773c2738b9cb61c42"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arc-swap",
 "arrow",
@@ -4885,9 +4890,8 @@ dependencies = [

 [[package]]
 name = "lance-arrow"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "24187f972374567bb3573cffd8728f2f4f7f06d644c3c5d4430b79d6b0b67300"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4909,8 +4913,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-scalar"
 version = "58.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "771f68b04b47f3addf781116f65061808de94b05e1e9411c23c18f32d14ebe79"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4924,20 +4927,17 @@ dependencies = [
 [[package]]
 name = "lance-arrow-stats"
 version = "58.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dd47ec33c90bf29f688fd02118e37d3a5ad5c339caa3163f89e417dc0867001f"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-schema",
- "half",
 "lance-arrow-scalar",
 ]

 [[package]]
 name = "lance-bitpacking"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "55fa50ad941e25298afb54eec107c3584f80c16db4adf7a6e4e9c4b6066f4281"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrayref",
 "crunchy",
@@ -4947,9 +4947,8 @@ dependencies = [

 [[package]]
 name = "lance-core"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "21f4fd872bfe948150a6878d983327c4f150dc1b629e94ae7677310fa6a3f35f"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4989,9 +4988,8 @@ dependencies = [

 [[package]]
 name = "lance-datafusion"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "230770734cd5f6fe1f1cb4a66750acad5c5a862c23c7f587d7def57c9f2c5f6b"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -5021,9 +5019,8 @@ dependencies = [

 [[package]]
 name = "lance-datagen"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c0c9edcea9dbf154a3baf470595bdcdb7ef6b44dcf2b5a0c28e8815c2d4837c5"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -5040,9 +5037,8 @@ dependencies = [

 [[package]]
 name = "lance-derive"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "03ac280ef94d66c2e7a4b0104e60d548f45156f114f02cd0ac57423721ec96ec"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "proc-macro2",
 "quote",
@@ -5051,9 +5047,8 @@ dependencies = [

 [[package]]
 name = "lance-encoding"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "65b345fecf1792d4147982e9ded1cf211992a442bcf73dd17a598b1c8c9b53a5"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-arith",
 "arrow-array",
@@ -5079,7 +5074,6 @@ dependencies = [
 "prost",
 "prost-build",
 "rand 0.9.5",
- "strum 0.26.3",
 "tokio",
 "tracing",
 "xxhash-rust",
@@ -5088,9 +5082,8 @@ dependencies = [

 [[package]]
 name = "lance-file"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "203400160ecd4ac6f67f6bfdb3c186f94758ff70a6ca76d4f300c327f5c14a11"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-arith",
 "arrow-array",
@@ -5121,9 +5114,8 @@ dependencies = [

 [[package]]
 name = "lance-index"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db67792e22332a7be08d35e60ce9e4c6fb7f75f70ab1566b67b462d1aac21b13"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arc-swap",
 "arrow",
@@ -5190,9 +5182,8 @@ dependencies = [

 [[package]]
 name = "lance-index-core"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "49b5700d71f43e5da2057908f3243a6bc01247b5dd12247313f789c280155aae"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-schema",
@@ -5214,9 +5205,8 @@ dependencies = [

 [[package]]
 name = "lance-io"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4b8a6872b9bff95e75e223e04011feebd990d8784878eaf5df73c86d6acebb47"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -5228,7 +5218,6 @@ dependencies = [
 "bytes",
 "chrono",
 "futures",
- "goosefs-sdk",
 "http 1.5.0",
 "io-uring",
 "lance-arrow",
@@ -5253,9 +5242,8 @@ dependencies = [

 [[package]]
 name = "lance-linalg"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5a1dd49fdf61da86650dc335fbe1f239803f9d73bc2bc5ffbc0c905dd8809892"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -5271,9 +5259,8 @@ dependencies = [

 [[package]]
 name = "lance-namespace"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1449221f599147e570bec6e09eaea6ce6b9286f632e00a27f945dd0c6b70ccd8"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "async-trait",
@@ -5285,9 +5272,8 @@ dependencies = [

 [[package]]
 name = "lance-namespace-impls"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c30bff40ab9a9a58313818ab97241cff7bab2b90b8d65039908425a065a797de"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "arrow-ipc",
@@ -5317,7 +5303,6 @@ dependencies = [
 "serde",
 "serde_json",
 "sha2 0.10.9",
- "time",
 "tokio",
 "tower",
 "tower-http 0.5.2",
@@ -5341,9 +5326,8 @@ dependencies = [

 [[package]]
 name = "lance-select"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c66c5cb12ed194c7c8e427a0e90e990cdbc7159edba2bfc7b7e8db2c90c747af"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -5358,9 +5342,8 @@ dependencies = [

 [[package]]
 name = "lance-table"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6cd2a39f2e288d1acecf887cb9cc655711fb67bcf3b24a2034158dbac7a6f60e"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -5370,6 +5353,7 @@ dependencies = [
 "async-trait",
 "aws-credential-types",
 "aws-sdk-dynamodb",
+ "blake3",
 "byteorder",
 "bytes",
 "chrono",
@@ -5399,9 +5383,8 @@ dependencies = [

 [[package]]
 name = "lance-testing"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0847352699e869a465b49bd49ecbc467bc5a3dcb168757bf39368362aad05f99"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
 "arrow-array",
 "arrow-schema",
@@ -5414,14 +5397,13 @@ dependencies = [

 [[package]]
 name = "lance-tokenizer"
-version = "10.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "254c1e9287789c8d744f93f2010a8389a5ea3fcedc0dddf7649c4117ec830c57"
+version = "11.0.0-beta.7"
+source = "git+https://github.com/lance-format/lance.git?tag=v11.0.0-beta.7#e581c49338bc83baf1ea50c5e235bd702f3fbeea"
 dependencies = [
+ "frostem",
 "icu_segmenter",
 "jieba-rs",
 "lindera",
- "rust-stemmers",
 "serde",
 "stop-words",
 "unicode-normalization",
@@ -5429,7 +5411,7 @@ dependencies = [

 [[package]]
 name = "lancedb"
-version = "0.37.1"
+version = "0.37.1-beta.1"
 dependencies = [
 "ahash",
 "anyhow",
@@ -5517,7 +5499,7 @@ dependencies = [

 [[package]]
 name = "lancedb-nodejs"
-version = "0.37.1"
+version = "0.37.1-beta.1"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -5542,7 +5524,7 @@ dependencies = [

 [[package]]
 name = "lancedb-python"
-version = "0.37.1"
+version = "0.37.1-beta.1"
 dependencies = [
 "arrow",
 "async-trait",
@@ -5699,7 +5681,7 @@ dependencies = [
 "serde",
 "serde_json",
 "serde_yaml_ng",
- "strum 0.28.0",
+ "strum",
 "strum_macros 0.28.0",
 "unicode-blocks",
 "unicode-normalization",
@@ -5729,7 +5711,7 @@ dependencies = [
 "rkyv",
 "serde",
 "serde_json",
- "strum 0.28.0",
+ "strum",
 "strum_macros 0.28.0",
 "thiserror 2.0.18",
 ]
@@ -5801,6 +5783,15 @@ dependencies = [
 "hashbrown 0.16.1",
 ]

+[[package]]
+name = "lru"
+version = "0.18.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5d2f2f9b4ba7e6b24d95e7e899329d35be83bcded72c8540cdd5368932d1d90a"
+dependencies = [
+ "hashbrown 0.17.1",
+]
+
 [[package]]
 name = "lru-slab"
 version = "0.1.2"
@@ -6438,9 +6429,9 @@ dependencies = [

 [[package]]
 name = "object_store_opendal"
-version = "0.57.0"
+version = "0.58.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0eb12a624a41fce745838d0ef3701ff6c47797c13cd18ad3612fd2a3134fdbd8"
+checksum = "88f165780495c17aa3ce86846600504198c3fffd99073521552751c2430fa6ac"
 dependencies = [
 "async-trait",
 "bytes",
@@ -6501,12 +6492,13 @@ checksum = "d6790f58c7ff633d8771f42965289203411a5e5c68388703c06e14f24770b41e"

 [[package]]
 name = "opendal"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "96c9c85ce253ff87225e7669979d877a20c98a06604ec9d6dd5f4473e08f1ae1"
+checksum = "4f20562cc7447fcc915fc5c23df305a412ea80a733c9f2fd9e2d267e2815be6d"
 dependencies = [
 "ctor 1.0.12",
 "opendal-core",
+ "opendal-http-transport-reqwest",
 "opendal-layer-concurrent-limit",
 "opendal-layer-logging",
 "opendal-layer-retry",
@@ -6523,24 +6515,22 @@ dependencies = [

 [[package]]
 name = "opendal-core"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c4f8607c90e2c963a91467f50fb49fbc7fb3d573f88cea219ca59ccd3740b309"
+checksum = "ec75551ff4cf3e57da98979f6a937aaa9ddb3915bf68cc17d03df733be6646ed"
 dependencies = [
 "anyhow",
- "base64 0.22.1",
+ "base64 0.23.1",
 "bytes",
 "futures",
 "http 1.5.0",
- "http-body 1.1.0",
 "jiff",
 "log",
 "md-5 0.11.0",
 "mea",
 "percent-encoding",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-core",
- "reqwest 0.13.4",
 "serde",
 "serde_json",
 "tokio",
@@ -6550,10 +6540,24 @@ dependencies = [
 ]

 [[package]]
-name = "opendal-layer-concurrent-limit"
-version = "0.57.0"
+name = "opendal-http-transport-reqwest"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0d6f81ba6960e3fae1882f253b114b21d7e444e1534f209c7737a79f6243eb6f"
+checksum = "ad4d4f19c3ce01126a30611f8e544eaa217104a278c889ac17c9374fe4f9e4ef"
+dependencies = [
+ "bytes",
+ "futures",
+ "http 1.5.0",
+ "http-body 1.1.0",
+ "opendal-core",
+ "reqwest 0.13.4",
+]
+
+[[package]]
+name = "opendal-layer-concurrent-limit"
+version = "0.58.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "249ac5b0aa5a7a6c3737342d10456067937f9c9a6f3f02544271f7908ab91081"
 dependencies = [
 "futures",
 "http 1.5.0",
@@ -6563,9 +6567,9 @@ dependencies = [

 [[package]]
 name = "opendal-layer-logging"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "58ada45c6d81d1aa4c9305d0c7d4bc317c59c85866a0908a2d75a7a978aa5ee2"
+checksum = "5c75411ab00f77851ff086b686c1e9ca8175ac18c15afa2cb75b9036436cb06c"
 dependencies = [
 "log",
 "opendal-core",
@@ -6573,9 +6577,9 @@ dependencies = [

 [[package]]
 name = "opendal-layer-retry"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7b2a25a718afb81fad81cb9a0580a1cb989221fa2317f888c6a37f8dad408eb7"
+checksum = "80b7738bd5f233ad8da39af9b9316b9b7a4eaddd91e8e32a1e19b7030688121d"
 dependencies = [
 "backon",
 "log",
@@ -6584,9 +6588,9 @@ dependencies = [

 [[package]]
 name = "opendal-layer-timeout"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1e91f731724c213af81e9d03517859c8fc47b4578e64ad61ae4f099f10fe36e3"
+checksum = "a704141924500f3803c05ed871b53305d2a2f11cb5ef20160c3ee688a1857f66"
 dependencies = [
 "opendal-core",
 "tokio",
@@ -6594,17 +6598,17 @@ dependencies = [

 [[package]]
 name = "opendal-service-azblob"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0030644366ef5d8cbe3a4a5822bf99a4aafddc1666e9d24b44d158d9062fc76a"
+checksum = "b3310fbbb48f111c6f590473c2cd15e1b7f8e384444b0d4e328f0464c864d767"
 dependencies = [
- "base64 0.22.1",
+ "base64 0.23.1",
 "bytes",
 "http 1.5.0",
 "log",
 "opendal-core",
 "opendal-service-azure-common",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-azure-storage",
 "reqsign-core",
 "reqsign-file-read-tokio",
@@ -6615,17 +6619,18 @@ dependencies = [

 [[package]]
 name = "opendal-service-azdls"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6dea4908d490143a9b0b7f7a790e139ff829b06a023f670455ed3d44f664b361"
+checksum = "2e3c406729935fe214ce574d68681a1ff7e0b322548f14094912bdbfe50e5c53"
 dependencies = [
- "base64 0.22.1",
+ "base64 0.23.1",
 "bytes",
 "http 1.5.0",
 "log",
+ "mea",
 "opendal-core",
 "opendal-service-azure-common",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-azure-storage",
 "reqsign-core",
 "reqsign-file-read-tokio",
@@ -6635,9 +6640,9 @@ dependencies = [

 [[package]]
 name = "opendal-service-azure-common"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9b489f13c42e69d69bdd72952b634356ec43a7881a20259b38b540fcecdf4051"
+checksum = "7348c88edf15af435b7be930077746b569fac5e738c1bf6a363b675e7317c9df"
 dependencies = [
 "http 1.5.0",
 "opendal-core",
@@ -6645,15 +6650,15 @@ dependencies = [

 [[package]]
 name = "opendal-service-cos"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "aa8cafe9729213375c7331019b0cb756ad3e1aff7f45cd32c45eae91ebde8901"
+checksum = "d533d4582105d269c8aebeee5f0e8bcf960f41b8aab6197df7012254d9f39bf0"
 dependencies = [
 "bytes",
 "http 1.5.0",
 "log",
 "opendal-core",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-core",
 "reqsign-file-read-tokio",
 "reqsign-tencent-cos",
@@ -6662,9 +6667,9 @@ dependencies = [

 [[package]]
 name = "opendal-service-gcs"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "48de101aac565ed06af4b47903c24eafd249075553ec1fb18256751c45148d47"
+checksum = "007f3fba63c21e516c956b891e96ff9892d8175662bfb781cdada9d3766a11e6"
 dependencies = [
 "async-trait",
 "bytes",
@@ -6672,7 +6677,7 @@ dependencies = [
 "log",
 "opendal-core",
 "percent-encoding",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-core",
 "reqsign-file-read-tokio",
 "reqsign-google",
@@ -6683,9 +6688,9 @@ dependencies = [

 [[package]]
 name = "opendal-service-goosefs"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "69e43048bde419947ba826fbdc2f134d6c03f44ebf48bd33a03b72f9fc45fcb4"
+checksum = "60871e6386f04d831e6a5bdbc032af4a91aeba49963252d0ef456a2cf36a9b78"
 dependencies = [
 "bytes",
 "goosefs-sdk",
@@ -6697,9 +6702,9 @@ dependencies = [

 [[package]]
 name = "opendal-service-hf"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c4922661976a1d40794a2adfbdb888cc3c23097690f825a92f773af38908a848"
+checksum = "b41fd41eb7ed03c5e66cefda61e8e117808ffd2908f2916737cb020a6beb02c7"
 dependencies = [
 "bytes",
 "hf-xet",
@@ -6707,22 +6712,21 @@ dependencies = [
 "log",
 "opendal-core",
 "percent-encoding",
- "reqwest 0.13.4",
 "serde",
 "serde_json",
 ]

 [[package]]
 name = "opendal-service-oss"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "328fa55e8888cbdfe00826bfea2a79042422b720e8369e9e021e46121dea5ace"
+checksum = "cd528ec2d49c5ca69e674ffed7b3e0686fb9cfcfea0596870de381467fda4f1b"
 dependencies = [
 "bytes",
 "http 1.5.0",
 "log",
 "opendal-core",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-aliyun-oss",
 "reqsign-core",
 "reqsign-file-read-tokio",
@@ -6731,18 +6735,18 @@ dependencies = [

 [[package]]
 name = "opendal-service-s3"
-version = "0.57.0"
+version = "0.58.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "313d46c9f5ae70bca26b7c3e3fbb9b639292625f28af73aa016f47e788af9deb"
+checksum = "58e80cdf192d7eff05feed747894d64f81905ac4eaf132edf7ea270abdd2d663"
 dependencies = [
- "base64 0.22.1",
+ "base64 0.23.1",
 "bytes",
- "crc32c",
+ "crc-fast",
 "http 1.5.0",
 "log",
 "md-5 0.11.0",
 "opendal-core",
- "quick-xml 0.39.4",
+ "quick-xml 0.41.0",
 "reqsign-aws-v4",
 "reqsign-core",
 "reqsign-file-read-tokio",
@@ -8675,16 +8679,6 @@ dependencies = [
 "ordered-multimap",
 ]

-[[package]]
-name = "rust-stemmers"
-version = "1.2.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e46a2036019fdb888131db7a4c847a1063a7493f971ed94ea82c67eada63ca54"
-dependencies = [
- "serde",
- "serde_derive",
-]
-
 [[package]]
 name = "rustc-demangle"
 version = "0.1.27"
@@ -9558,15 +9552,6 @@ version = "0.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7da8b5736845d9f2fcb837ea5d9e2628564b3b043a70948a3f0b778838c5fb4f"

-[[package]]
-name = "strum"
-version = "0.26.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8fec0f0aef304996cf250b31b5a10dee7980c85da9d759361292b8bca5a18f06"
-dependencies = [
- "strum_macros 0.26.4",
-]
-
 [[package]]
 name = "strum"
 version = "0.28.0"
@@ -9589,19 +9574,6 @@ dependencies = [
 "syn 2.0.117",
 ]

-[[package]]
-name = "strum_macros"
-version = "0.26.4"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4c6bee85a5a24955dc440386795aa378cd9cf82acd5f764469152d2270e581be"
-dependencies = [
- "heck 0.5.0",
- "proc-macro2",
- "quote",
- "rustversion",
- "syn 2.0.117",
-]
-
 [[package]]
 name = "strum_macros"
 version = "0.28.0"
@@ -13,20 +13,20 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"

 [workspace.dependencies]
-lance = { "version" = "=10.0.0", default-features = false }
-lance-core = "=10.0.0"
-lance-datagen = "=10.0.0"
-lance-file = "=10.0.0"
-lance-io = { "version" = "=10.0.0", default-features = false }
-lance-index = "=10.0.0"
-lance-linalg = "=10.0.0"
-lance-namespace = "=10.0.0"
-lance-namespace-impls = { "version" = "=10.0.0", default-features = false }
-lance-table = "=10.0.0"
-lance-testing = "=10.0.0"
-lance-datafusion = "=10.0.0"
-lance-encoding = "=10.0.0"
-lance-arrow = "=10.0.0"
+lance = { "version" = "=11.0.0-beta.7", default-features = false, "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=11.0.0-beta.7", default-features = false, "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=11.0.0-beta.7", default-features = false, "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=11.0.0-beta.7", "tag" = "v11.0.0-beta.7", "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 }
@@ -6,11 +6,16 @@ HEAD_SHA=$(git rev-parse HEAD)

 readonly TAG_PREFIX="v"
 readonly SELF_DIR=$(cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
-readonly BUMP_ARGS="--no-tag"

 PREV_TAG=$(git tag --sort='version:refname' | grep ^$TAG_PREFIX | python $SELF_DIR/semver_sort.py $TAG_PREFIX | tail -n 1)
 echo "Found previous tag $PREV_TAG"

+# Initially, we don't want to tag if we are doing stable, because we will bump
+# again later. See comment at end for why.
+if [[ "$RELEASE_TYPE" == 'stable' ]]; then
+  BUMP_ARGS="--no-tag"
+fi
+
 # If last is stable and not bumping minor
 if [[ $PREV_TAG != *beta* ]]; then
    if [[ "$BUMP_MINOR" != "false" ]]; then
@@ -34,12 +39,12 @@ fi
 # a stable version, bump the pre-release level ("pre_l") to make it stable.
 if [[ $RELEASE_TYPE == 'stable' ]]; then
  # X.Y.Z-beta.N -> X.Y.Z
-  bump-my-version bump -vv $BUMP_ARGS pre_l
+  bump-my-version bump -vv pre_l
 fi

 # Validate that we have incremented version appropriately for breaking changes
-NEW_VERSION=$(python -c 'import tomllib; print(tomllib.load(open(".bumpversion.toml", "rb"))["tool"]["bumpversion"]["current_version"])')
-NEW_TAG="$TAG_PREFIX$NEW_VERSION"
+NEW_TAG=$(git describe --tags --exact-match HEAD)
+NEW_VERSION=$(echo $NEW_TAG | sed "s/^$TAG_PREFIX//")
 LAST_STABLE_RELEASE=$(git tag --sort='version:refname' | grep ^$TAG_PREFIX | grep -v beta | grep -vF "$NEW_TAG" | python $SELF_DIR/semver_sort.py $TAG_PREFIX | tail -n 1)
 LAST_STABLE_VERSION=$(echo $LAST_STABLE_RELEASE | sed "s/^$TAG_PREFIX//")

@@ -1,21 +0,0 @@
-#!/usr/bin/env bash
-set -euo pipefail
-
-RELEASE_VERSION=$(python -c 'import tomllib; print(tomllib.load(open(".bumpversion.toml", "rb"))["tool"]["bumpversion"]["current_version"])')
-RELEASE_TAG="v${RELEASE_VERSION}"
-
-if git rev-parse --quiet --verify "refs/tags/${RELEASE_TAG}" >/dev/null; then
-  echo "Release tag ${RELEASE_TAG} already exists" >&2
-  exit 1
-fi
-
-git tag --annotate "$RELEASE_TAG" --message "Release ${RELEASE_TAG}"
-
-HEAD_SHA=$(git rev-parse HEAD)
-TAG_SHA=$(git rev-parse "refs/tags/${RELEASE_TAG}^{}")
-if [[ "$TAG_SHA" != "$HEAD_SHA" ]]; then
-  echo "Release tag ${RELEASE_TAG} points to ${TAG_SHA}, expected ${HEAD_SHA}" >&2
-  exit 1
-fi
-
-echo "Created ${RELEASE_TAG} at ${HEAD_SHA}"
@@ -1,72 +0,0 @@
-import tempfile
-import unittest
-from pathlib import Path
-
-from validate_stable_lance import validate
-
-
-class ValidateStableLanceTest(unittest.TestCase):
-    def write_fixture(
-        self,
-        root: Path,
-        *,
-        rust: str = "=10.0.0",
-        python: str = "10.0.0",
-        java: str = "10.0.0",
-    ) -> None:
-        (root / "python").mkdir()
-        (root / "java").mkdir()
-        (root / "Cargo.toml").write_text(
-            f'[workspace.dependencies]\nlance = "{rust}"\nlance-core = "{rust}"\n'
-        )
-        (root / "python" / "pyproject.toml").write_text(
-            '[project.optional-dependencies]\ntests = ["pylance==' + python + '"]\n'
-        )
-        (root / "java" / "pom.xml").write_text(
-            "<project><properties><lance-core.version>"
-            + java
-            + "</lance-core.version></properties></project>"
-        )
-
-    def test_accepts_matching_stable_versions(self) -> None:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            root = Path(temp_dir)
-            self.write_fixture(root)
-            self.assertEqual(validate(root), "10.0.0")
-
-    def test_rejects_prerelease(self) -> None:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            root = Path(temp_dir)
-            self.write_fixture(root, python="10.0.0rc1")
-            with self.assertRaisesRegex(ValueError, "not stable"):
-                validate(root)
-
-    def test_rejects_sdk_mismatch(self) -> None:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            root = Path(temp_dir)
-            self.write_fixture(root, java="9.0.0")
-            with self.assertRaisesRegex(ValueError, "do not match across SDKs"):
-                validate(root)
-
-    def test_rejects_non_exact_rust_dependency(self) -> None:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            root = Path(temp_dir)
-            self.write_fixture(root, rust="10.0.0")
-            with self.assertRaisesRegex(ValueError, "not exact"):
-                validate(root)
-
-    def test_rejects_unpublished_rust_source(self) -> None:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            root = Path(temp_dir)
-            self.write_fixture(root)
-            (root / "Cargo.toml").write_text(
-                "[workspace.dependencies]\n"
-                'lance = { version = "=10.0.0", git = "https://example.com/lance", '
-                'tag = "v10.0.0" }\n'
-            )
-            with self.assertRaisesRegex(ValueError, "unpublished source fields"):
-                validate(root)
-
-
-if __name__ == "__main__":
-    unittest.main()
@@ -1,112 +1,34 @@
-#!/usr/bin/env python3
-"""Validate that every SDK uses the same published stable Lance release."""
-
-from __future__ import annotations
-
-import re
-import xml.etree.ElementTree as ET
-from pathlib import Path
-
 import tomllib

-STABLE_VERSION = re.compile(r"[0-9]+\.[0-9]+\.[0-9]+")
+found_preview_lance = False

+with open("Cargo.toml", "rb") as f:
+    cargo_data = tomllib.load(f)

-def _stable_version(raw: str, *, dependency: str, exact: bool = False) -> str:
-    value = raw.strip()
-    if exact and not value.startswith("="):
-        raise ValueError(f"Dependency '{dependency}' is not exact: {raw}")
-    value = value.removeprefix("=").removeprefix("v")
-    if STABLE_VERSION.fullmatch(value) is None:
-        raise ValueError(f"Dependency '{dependency}' is not stable: {raw}")
-    return value
+    for name, dep in cargo_data["workspace"]["dependencies"].items():
+        if name == "lance" or name.startswith("lance-"):
+            if isinstance(dep, str):
+                version = dep
+            elif isinstance(dep, dict):
+                # Version doesn't have the beta tag in it, so we instead look
+                # at the git tag.
+                version = dep.get('tag', dep.get('version'))
+            else:
+                raise ValueError("Unexpected type for dependency: " + str(dep))

+            if "beta" in version:
+                found_preview_lance = True
+                print(f"Dependency '{name}' is a preview version: {version}")

-def rust_lance_version(repo_root: Path) -> str:
-    with (repo_root / "Cargo.toml").open("rb") as cargo_file:
-        dependencies = tomllib.load(cargo_file)["workspace"]["dependencies"]
+with open("python/pyproject.toml", "rb") as f:
+    py_proj_data = tomllib.load(f)

-    versions: dict[str, str] = {}
-    for name, dependency in dependencies.items():
-        if name != "lance" and not name.startswith("lance-"):
-            continue
+    for dep in py_proj_data["project"]["dependencies"]:
+        if dep.startswith("pylance"):
+            if "b" in dep:
+                found_preview_lance = True
+                print(f"Dependency '{dep}' is a preview version")
+            break  # Only one pylance dependency

-        if isinstance(dependency, str):
-            raw_version = dependency
-        elif isinstance(dependency, dict):
-            forbidden_sources = [
-                source
-                for source in ("git", "path", "branch", "rev", "tag")
-                if source in dependency
-            ]
-            if forbidden_sources:
-                joined = ", ".join(forbidden_sources)
-                raise ValueError(
-                    f"Dependency '{name}' uses unpublished source fields: {joined}"
-                )
-            raw_version = dependency.get("version")
-            if raw_version is None:
-                raise ValueError(f"Dependency '{name}' has no version")
-        else:
-            raise TypeError(f"Dependency '{name}' has an unexpected definition")
-
-        versions[name] = _stable_version(raw_version, dependency=name, exact=True)
-
-    if not versions:
-        raise ValueError("No Rust Lance dependencies found")
-    unique_versions = set(versions.values())
-    if len(unique_versions) != 1:
-        details = ", ".join(f"{name}={version}" for name, version in versions.items())
-        raise ValueError(f"Rust Lance dependency versions do not match: {details}")
-    return unique_versions.pop()
-
-
-def python_lance_version(repo_root: Path) -> str:
-    with (repo_root / "python" / "pyproject.toml").open("rb") as pyproject_file:
-        pyproject = tomllib.load(pyproject_file)
-
-    requirements = pyproject["project"]["optional-dependencies"]["tests"]
-    pylance_requirements = [
-        requirement for requirement in requirements if requirement.startswith("pylance")
-    ]
-    if len(pylance_requirements) != 1:
-        raise ValueError(
-            "Expected exactly one pylance requirement in the Python test dependencies"
-        )
-
-    requirement = pylance_requirements[0]
-    prefix = "pylance=="
-    if not requirement.startswith(prefix):
-        raise ValueError(f"Python test dependency is not exact: {requirement}")
-    return _stable_version(requirement[len(prefix) :], dependency="pylance")
-
-
-def java_lance_version(repo_root: Path) -> str:
-    pom_root = ET.parse(repo_root / "java" / "pom.xml").getroot()
-    versions = [
-        element.text
-        for element in pom_root.iter()
-        if element.tag.rsplit("}", 1)[-1] == "lance-core.version"
-    ]
-    if len(versions) != 1 or versions[0] is None:
-        raise ValueError("Expected exactly one Java lance-core.version property")
-    return _stable_version(versions[0], dependency="Java lance-core")
-
-
-def validate(repo_root: Path) -> str:
-    versions = {
-        "Rust": rust_lance_version(repo_root),
-        "Python": python_lance_version(repo_root),
-        "Java": java_lance_version(repo_root),
-    }
-    if len(set(versions.values())) != 1:
-        details = ", ".join(f"{sdk}={version}" for sdk, version in versions.items())
-        raise ValueError(
-            f"Lance dependency versions do not match across SDKs: {details}"
-        )
-    return versions["Rust"]
-
-
-if __name__ == "__main__":
-    version = validate(Path(__file__).resolve().parents[1])
-    print(f"Validated published stable Lance v{version} across Rust, Python, and Java")
+if found_preview_lance:
+    raise ValueError("Found preview version of Lance in dependencies")
@@ -101,6 +101,13 @@ ignore = [
    # https://rustsec.org/advisories/RUSTSEC-2026-0195
    { id = "RUSTSEC-2026-0194", reason = "transitive via inferno/lance/opendal; XML from trusted cloud endpoints, not attacker-controlled" },
    { id = "RUSTSEC-2026-0195", reason = "transitive via inferno/lance/opendal; XML from trusted cloud endpoints, not attacker-controlled" },
+    # smartstring: unmaintained — the repository was archived by its author on
+    # 2026-05-03. Not a vulnerability. Reached only transitively through polars
+    # (polars-core/-io/-ops/-time/-utils); nothing in LanceDB depends on it directly.
+    # The advisory states no safe upgrade is available: upstream recommends
+    # compact_str/smol_str, so clearing this requires polars to migrate.
+    # https://rustsec.org/advisories/RUSTSEC-2026-0249
+    { id = "RUSTSEC-2026-0249", reason = "smartstring unmaintained via polars; no fixed upstream release" },
 ]

 # ---------------------------------------------------------------------------
@@ -14,7 +14,7 @@ Add the following dependency to your `pom.xml`:
 <dependency>
    <groupId>com.lancedb</groupId>
    <artifactId>lancedb-core</artifactId>
-    <version>0.37.1</version>
+    <version>0.37.1-beta.1</version>
 </dependency>
 ```

@@ -69,14 +69,33 @@ abstract addColumns(newColumnTransforms): Promise<AddColumnsResult>

 Add new columns with defined values.

+The `{ computed }` form stores the expression rather than evaluating it
+now: the column is committed with no values, and rows get them from
+[Table#refreshColumn](Table.md#refreshcolumn). Declaring one therefore costs the same on a
+large table as on an empty one.
+
+A refresh does not revisit rows it has already filled, so mutating an
+input leaves the value computed at fill time; recomputing means dropping
+the column and declaring it again. While a declaration reads a column,
+that column cannot be renamed, retyped or dropped.
+
+Computed columns are local-only: LanceDB Cloud and Enterprise reject a
+declaration.
+
 #### Parameters

-* **newColumnTransforms**: `Field`&lt;`any`&gt; \| `Field`&lt;`any`&gt;[] \| `Schema`&lt;`any`&gt; \| [`AddColumnsSql`](../interfaces/AddColumnsSql.md)[]
+* **newColumnTransforms**:
+    \| `Field`&lt;`any`&gt;
+    \| `Field`&lt;`any`&gt;[]
+    \| `Schema`&lt;`any`&gt;
+    \| [`AddColumnsSql`](../interfaces/AddColumnsSql.md)[]
+    \| `object`
    Either:
    - An array of objects with column names and SQL expressions to calculate values
    - A single Arrow Field defining one column with its data type (column will be initialized with null values)
    - An array of Arrow Fields defining columns with their data types (columns will be initialized with null values)
    - An Arrow Schema defining columns with their data types (columns will be initialized with null values)
+    - `{ computed }`, declaring columns defined by a SQL expression whose type and inputs are derived from it

 #### Returns

@@ -85,6 +104,13 @@ Add new columns with defined values.
 A promise that resolves to an object
 containing the new version number of the table after adding the columns.

+#### Example
+
+```ts
+await table.addColumns({ computed: [{ name: "doubled", valueSql: "x * 2" }] });
+const { rowsFilled } = await table.refreshColumn("doubled");
+```
+
 ***

 ### alterColumns()
@@ -431,9 +457,10 @@ Read the [LsmWriteSpec](../interfaces/LsmWriteSpec.md) currently installed on th

 Resolves to `undefined` when the MemWAL LSM write path is not enabled (no
 spec has been set, or it was removed with [Table#unsetLsmWriteSpec](Table.md#unsetlsmwritespec)).
-The returned spec — including its `maintainedIndexes` and
-`writerConfigDefaults` — mirrors what was passed to
-[Table#setLsmWriteSpec](Table.md#setlsmwritespec).
+The returned spec mirrors what was passed to
+[Table#setLsmWriteSpec](Table.md#setlsmwritespec), except that `maintainedIndexes` always
+reports the concrete list resolved when the spec was set — `undefined`
+never round-trips.

 #### Returns

@@ -717,6 +744,32 @@ for await (const batch of table.query()) {

 ***

+### refreshColumn()
+
+```ts
+abstract refreshColumn(column): Promise<RefreshColumnResult>
+```
+
+Fill the rows of a computed column that hold no value yet.
+
+Rows appended since the last refresh are filled by the next one; rows
+already filled are left as they are, so the call is idempotent and does
+not observe a mutated input. Local tables only.
+
+#### Parameters
+
+* **column**: `string`
+    The name of the computed column to fill.
+
+#### Returns
+
+`Promise`&lt;[`RefreshColumnResult`](../interfaces/RefreshColumnResult.md)&gt;
+
+A promise that resolves to the
+number of rows filled and the new version number of the table.
+
+***
+
 ### restore()

 ```ts
@@ -806,6 +859,11 @@ All variants require the table to have an unenforced primary key
 ([Table#setUnenforcedPrimaryKey](Table.md#setunenforcedprimarykey)); bucket sharding additionally
 requires it to be the single column being bucketed.

+Omitting `maintainedIndexes` maintains every index on the table, resolved
+here, failing if one cannot be maintained — name them to install anyway.
+Naming them pins an exact set, and a still-building index is rejected
+rather than quietly omitted.
+
 #### Parameters

 * **spec**: [`LsmWriteSpec`](../interfaces/LsmWriteSpec.md)
@@ -105,6 +105,7 @@
 - [OptimizeOptions](interfaces/OptimizeOptions.md)
 - [OptimizeStats](interfaces/OptimizeStats.md)
 - [QueryExecutionOptions](interfaces/QueryExecutionOptions.md)
+- [RefreshColumnResult](interfaces/RefreshColumnResult.md)
 - [RemovalStats](interfaces/RemovalStats.md)
 - [RenameTableOptions](interfaces/RenameTableOptions.md)
 - [RestNamespaceConfig](interfaces/RestNamespaceConfig.md)
@@ -34,7 +34,9 @@ Bucket and identity variants: the sharding column.
 optional maintainedIndexes: string[];
 ```

-Names of indexes the MemWAL should keep up to date during writes.
+Indexes the MemWAL keeps up to date. Omit to maintain every supported
+index, resolved on install — a snapshot, so indexes created later are not
+maintained. Pass `[]` for none.

 ***

@@ -0,0 +1,23 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / RefreshColumnResult
+
+# Interface: RefreshColumnResult
+
+## Properties
+
+### rowsFilled
+
+```ts
+rowsFilled: number;
+```
+
+***
+
+### version
+
+```ts
+version: number;
+```
@@ -8,7 +8,7 @@
    <parent>
      <groupId>com.lancedb</groupId>
      <artifactId>lancedb-parent</artifactId>
-      <version>0.37.1-final.0</version>
+      <version>0.37.1-beta.1</version>
      <relativePath>../pom.xml</relativePath>
    </parent>

@@ -6,7 +6,7 @@

    <groupId>com.lancedb</groupId>
    <artifactId>lancedb-parent</artifactId>
-    <version>0.37.1-final.0</version>
+    <version>0.37.1-beta.1</version>
    <packaging>pom</packaging>
    <name>${project.artifactId}</name>
    <description>LanceDB Java SDK Parent POM</description>
@@ -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</lance-core.version>
+        <lance-core.version>11.0.0-beta.6</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>
@@ -1,7 +1,7 @@
 [package]
 name = "lancedb-nodejs"
 edition.workspace = true
-version = "0.37.1"
+version = "0.37.1-beta.1"
 publish = false
 license.workspace = true
 description.workspace = true
@@ -3340,3 +3340,45 @@ describe("LSM merge insert", () => {
    await expect(table.query().useLsm(true).toArray()).rejects.toThrow();
  });
 });
+
+describe("computed columns", () => {
+  let tmpDir: tmp.DirResult;
+  beforeEach(() => {
+    tmpDir = tmp.dirSync({ unsafeCleanup: true });
+  });
+  afterEach(() => tmpDir.removeCallback());
+
+  it("declares a column and fills it on refresh", async () => {
+    const db = await connect(tmpDir.name);
+    const table = await db.createTable("computed", [{ x: 1 }, { x: 2 }]);
+
+    await table.addColumns({
+      computed: [{ name: "doubled", valueSql: "x * 2" }],
+    });
+    let rows = await table.query().toArray();
+    expect(rows.map((r) => r.doubled)).toEqual([null, null]);
+
+    const result = await table.refreshColumn("doubled");
+    expect(result.rowsFilled).toBe(2);
+
+    rows = await table.query().toArray();
+    expect(rows.map((r) => r.doubled).sort()).toEqual([2, 4]);
+  });
+
+  it("fills rows added since the last refresh", async () => {
+    const db = await connect(tmpDir.name);
+    const table = await db.createTable("computed_append", [{ x: 1 }]);
+
+    await table.addColumns({
+      computed: [{ name: "doubled", valueSql: "x * 2" }],
+    });
+    await table.refreshColumn("doubled");
+    await table.add([{ x: 5 }]);
+
+    const result = await table.refreshColumn("doubled");
+    expect(result.rowsFilled).toBe(1);
+
+    const rows = await table.query().toArray();
+    expect(rows.map((r) => r.doubled).sort()).toEqual([10, 2]);
+  });
+});
@@ -50,6 +50,7 @@ export {
  MergeResult,
  AddResult,
  AddColumnsResult,
+  RefreshColumnResult,
  AlterColumnsResult,
  UpdateFieldMetadataResult,
  DeleteResult,
@@ -33,6 +33,7 @@ import {
  Job,
  Branches as NativeBranches,
  OptimizeStats,
+  RefreshColumnResult,
  TableStatistics,
  Tags,
  UpdateFieldMetadataResult,
@@ -197,7 +198,11 @@ export interface LsmWriteSpec {
  column?: string;
  /** Bucket variant: the number of buckets, in `[1, 1024]`. */
  numBuckets?: number;
-  /** Names of indexes the MemWAL should keep up to date during writes. */
+  /**
+   * Indexes the MemWAL keeps up to date. Omit to maintain every supported
+   * index, resolved on install — a snapshot, so indexes created later are not
+   * maintained. Pass `[]` for none.
+   */
  maintainedIndexes?: string[];
  /** Default `ShardWriter` configuration recorded in the MemWAL index. */
  writerConfigDefaults?: Record<string, string>;
@@ -521,18 +526,54 @@ export abstract class Table {
  abstract vectorSearch(vector: IntoVector | MultiVector): VectorQuery;
  /**
   * Add new columns with defined values.
+   *
+   * The `{ computed }` form stores the expression rather than evaluating it
+   * now: the column is committed with no values, and rows get them from
+   * {@link Table#refreshColumn}. Declaring one therefore costs the same on a
+   * large table as on an empty one.
+   *
+   * A refresh does not revisit rows it has already filled, so mutating an
+   * input leaves the value computed at fill time; recomputing means dropping
+   * the column and declaring it again. While a declaration reads a column,
+   * that column cannot be renamed, retyped or dropped.
+   *
+   * Computed columns are local-only: LanceDB Cloud and Enterprise reject a
+   * declaration.
   * @param {AddColumnsSql[] | Field | Field[] | Schema} newColumnTransforms Either:
   *   - An array of objects with column names and SQL expressions to calculate values
   *   - A single Arrow Field defining one column with its data type (column will be initialized with null values)
   *   - An array of Arrow Fields defining columns with their data types (columns will be initialized with null values)
   *   - An Arrow Schema defining columns with their data types (columns will be initialized with null values)
+   *   - `{ computed }`, declaring columns defined by a SQL expression whose type and inputs are derived from it
   * @returns {Promise<AddColumnsResult>} A promise that resolves to an object
   * containing the new version number of the table after adding the columns.
+   * @example
+   * ```ts
+   * await table.addColumns({ computed: [{ name: "doubled", valueSql: "x * 2" }] });
+   * const { rowsFilled } = await table.refreshColumn("doubled");
+   * ```
   */
  abstract addColumns(
-    newColumnTransforms: AddColumnsSql[] | Field | Field[] | Schema,
+    newColumnTransforms:
+      | AddColumnsSql[]
+      | Field
+      | Field[]
+      | Schema
+      | { computed: AddColumnsSql[] },
  ): Promise<AddColumnsResult>;

+  /**
+   * Fill the rows of a computed column that hold no value yet.
+   *
+   * Rows appended since the last refresh are filled by the next one; rows
+   * already filled are left as they are, so the call is idempotent and does
+   * not observe a mutated input. Local tables only.
+   * @param {string} column The name of the computed column to fill.
+   * @returns {Promise<RefreshColumnResult>} A promise that resolves to the
+   * number of rows filled and the new version number of the table.
+   */
+  abstract refreshColumn(column: string): Promise<RefreshColumnResult>;
+
  /**
   * Alter the name or nullability of columns.
   * @param {ColumnAlteration[]} columnAlterations One or more alterations to
@@ -595,6 +636,11 @@ export abstract class Table {
   * All variants require the table to have an unenforced primary key
   * ({@link Table#setUnenforcedPrimaryKey}); bucket sharding additionally
   * requires it to be the single column being bucketed.
+   *
+   * Omitting `maintainedIndexes` maintains every index on the table, resolved
+   * here, failing if one cannot be maintained — name them to install anyway.
+   * Naming them pins an exact set, and a still-building index is rejected
+   * rather than quietly omitted.
   * @param {LsmWriteSpec} spec The sharding spec to install.
   * @returns {Promise<void>}
   * @example
@@ -622,9 +668,10 @@ export abstract class Table {
   *
   * Resolves to `undefined` when the MemWAL LSM write path is not enabled (no
   * spec has been set, or it was removed with {@link Table#unsetLsmWriteSpec}).
-   * The returned spec — including its `maintainedIndexes` and
-   * `writerConfigDefaults` — mirrors what was passed to
-   * {@link Table#setLsmWriteSpec}.
+   * The returned spec mirrors what was passed to
+   * {@link Table#setLsmWriteSpec}, except that `maintainedIndexes` always
+   * reports the concrete list resolved when the spec was set — `undefined`
+   * never round-trips.
   * @returns {Promise<LsmWriteSpec | undefined>}
   */
  abstract getLsmWriteSpec(): Promise<LsmWriteSpec | undefined>;
@@ -1078,8 +1125,22 @@ export class LocalTable extends Table {
  // TODO: Support BatchUDF

  async addColumns(
-    newColumnTransforms: AddColumnsSql[] | Field | Field[] | Schema,
+    newColumnTransforms:
+      | AddColumnsSql[]
+      | Field
+      | Field[]
+      | Schema
+      | { computed: AddColumnsSql[] },
  ): Promise<AddColumnsResult> {
+    // Columns defined by an expression are declared, not materialized here.
+    if (
+      typeof newColumnTransforms === "object" &&
+      !Array.isArray(newColumnTransforms) &&
+      "computed" in newColumnTransforms
+    ) {
+      return await this.inner.addComputedColumns(newColumnTransforms.computed);
+    }
+
    // Handle single Field -> convert to array of Fields
    if (newColumnTransforms instanceof Field) {
      newColumnTransforms = [newColumnTransforms];
@@ -1114,6 +1175,10 @@ export class LocalTable extends Table {
    throw new Error("Invalid input type for addColumns");
  }

+  async refreshColumn(column: string): Promise<RefreshColumnResult> {
+    return await this.inner.refreshColumn(column);
+  }
+
  async alterColumns(
    columnAlterations: ColumnAlteration[],
  ): Promise<AlterColumnsResult> {
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",
@@ -1,6 +1,6 @@
 {
  "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.37.1",
+  "version": "0.37.1-beta.1",
  "os": [
    "win32"
  ],
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.37.1",
+	"version": "0.37.1-beta.1",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",
@@ -1,12 +1,12 @@
 {
  "name": "@lancedb/lancedb",
-  "version": "0.37.1",
+  "version": "0.37.1-beta.1",
  "lockfileVersion": 3,
  "requires": true,
  "packages": {
    "": {
      "name": "@lancedb/lancedb",
-      "version": "0.37.1",
+      "version": "0.37.1-beta.1",
      "cpu": [
        "x64",
        "arm64"
@@ -11,7 +11,7 @@
    "ann"
  ],
  "private": false,
-  "version": "0.37.1",
+  "version": "0.37.1-beta.1",
  "main": "dist/index.js",
  "exports": {
    ".": "./dist/index.js",
@@ -347,6 +347,30 @@ impl Table {
        Ok(res.into())
    }

+    #[napi(catch_unwind)]
+    pub async fn add_computed_columns(
+        &self,
+        columns: Vec<AddColumnsSql>,
+    ) -> napi::Result<AddColumnsResult> {
+        let table = self.inner_ref()?;
+        let mut builder = table.add_columns();
+        for column in columns {
+            builder = builder.computed(column.name, column.value_sql);
+        }
+        let res = builder.execute().await.default_error()?;
+        Ok(res.into())
+    }
+
+    #[napi(catch_unwind)]
+    pub async fn refresh_column(&self, column: String) -> napi::Result<RefreshColumnResult> {
+        let res = self
+            .inner_ref()?
+            .refresh_column(column)
+            .await
+            .default_error()?;
+        Ok(res.into())
+    }
+
    #[napi(catch_unwind)]
    pub async fn add_columns_with_schema(
        &self,
@@ -772,7 +796,8 @@ pub struct LsmWriteSpec {
    pub column: Option<String>,
    /// Bucket variant: the number of buckets, in `[1, 1024]`.
    pub num_buckets: Option<u32>,
-    /// Names of indexes the MemWAL should keep up to date during writes.
+    /// Indexes the MemWAL keeps up to date. Omitted resolves every
+    /// maintainable index on install; an empty array means none.
    pub maintained_indexes: Option<Vec<String>>,
    /// Default `ShardWriter` configuration recorded in the MemWAL index.
    pub writer_config_defaults: Option<HashMap<String, String>>,
@@ -782,7 +807,6 @@ impl TryFrom<LsmWriteSpec> for lancedb::table::LsmWriteSpec {
    type Error = napi::Error;

    fn try_from(value: LsmWriteSpec) -> napi::Result<Self> {
-        let maintained = value.maintained_indexes.unwrap_or_default();
        let writer_config_defaults = value.writer_config_defaults.unwrap_or_default();
        let spec = match value.spec_type.as_str() {
            "bucket" => {
@@ -809,7 +833,7 @@ impl TryFrom<LsmWriteSpec> for lancedb::table::LsmWriteSpec {
            }
        };
        Ok(spec
-            .with_maintained_indexes(maintained)
+            .with_maintained_indexes(value.maintained_indexes)
            .with_writer_config_defaults(writer_config_defaults))
    }
 }
@@ -827,7 +851,7 @@ impl From<lancedb::table::LsmWriteSpec> for LsmWriteSpec {
                spec_type: "bucket".to_string(),
                column: Some(column),
                num_buckets: Some(num_buckets),
-                maintained_indexes: Some(maintained_indexes),
+                maintained_indexes,
                writer_config_defaults: Some(writer_config_defaults),
            },
            Native::Identity {
@@ -838,7 +862,7 @@ impl From<lancedb::table::LsmWriteSpec> for LsmWriteSpec {
                spec_type: "identity".to_string(),
                column: Some(column),
                num_buckets: None,
-                maintained_indexes: Some(maintained_indexes),
+                maintained_indexes,
                writer_config_defaults: Some(writer_config_defaults),
            },
            Native::Unsharded {
@@ -848,7 +872,7 @@ impl From<lancedb::table::LsmWriteSpec> for LsmWriteSpec {
                spec_type: "unsharded".to_string(),
                column: None,
                num_buckets: None,
-                maintained_indexes: Some(maintained_indexes),
+                maintained_indexes,
                writer_config_defaults: Some(writer_config_defaults),
            },
        }
@@ -1196,6 +1220,21 @@ pub struct AddColumnsResult {
    pub version: i64,
 }

+#[napi(object)]
+pub struct RefreshColumnResult {
+    pub rows_filled: i64,
+    pub version: i64,
+}
+
+impl From<lancedb::table::RefreshColumnResult> for RefreshColumnResult {
+    fn from(value: lancedb::table::RefreshColumnResult) -> Self {
+        Self {
+            rows_filled: value.rows_filled as i64,
+            version: value.version as i64,
+        }
+    }
+}
+
 impl From<lancedb::table::AddColumnsResult> for AddColumnsResult {
    fn from(value: lancedb::table::AddColumnsResult) -> Self {
        Self {
@@ -1,6 +1,6 @@
 [package]
 name = "lancedb-python"
-version = "0.37.1"
+version = "0.37.1-beta.1"
 publish = false
 edition.workspace = true
 description = "Python bindings for LanceDB"
@@ -63,7 +63,7 @@ tests = [
    "polars>=0.19, <=1.32.3",
    "pyarrow<25",
    "pyarrow-stubs>=16.0",
-    "pylance==10.0.0",
+    "pylance==9.0.0rc1",
    "requests>=2.31.0",
    "datafusion>=54,<55",
    "opentelemetry-sdk>=1.30.0",
@@ -335,6 +335,10 @@ class Table:
    ) -> list[FtsToken]: ...
    async def delete(self, filter: Union[str, PyExpr]) -> DeleteResult: ...
    async def add_columns(self, columns: list[tuple[str, str]]) -> AddColumnsResult: ...
+    async def add_computed_columns(
+        self, columns: list[tuple[str, str]]
+    ) -> AddColumnsResult: ...
+    async def refresh_column(self, column: str) -> RefreshColumnResult: ...
    async def add_columns_with_schema(self, schema: pa.Schema) -> AddColumnsResult: ...
    async def alter_columns(
        self, columns: list[dict[str, Any]]
@@ -653,9 +657,10 @@ class LsmWriteSpec:
    def identity(column: str) -> "LsmWriteSpec": ...
    @staticmethod
    def unsharded() -> "LsmWriteSpec": ...
-    def with_maintained_indexes(self, indexes: List[str]) -> "LsmWriteSpec":
-        """Return a copy of this spec asking the MemWAL to keep the named
-        indexes up to date as rows are appended."""
+    def with_maintained_indexes(self, indexes: Optional[List[str]]) -> "LsmWriteSpec":
+        """Set which indexes the MemWAL keeps up to date. None resolves every
+        index on the table at install, failing if one cannot be maintained;
+        a list is verbatim, empty means none."""
        ...
    def with_writer_config_defaults(self, defaults: Dict[str, str]) -> "LsmWriteSpec":
        """Return a copy of this spec recording the given default
@@ -670,13 +675,19 @@ class LsmWriteSpec:
    @property
    def num_buckets(self) -> Optional[int]: ...
    @property
-    def maintained_indexes(self) -> List[str]: ...
+    def maintained_indexes(self) -> Optional[List[str]]:
+        """Indexes the MemWAL keeps up to date, or None for every supported one."""
+        ...
    @property
    def writer_config_defaults(self) -> Dict[str, str]: ...

 class AddColumnsResult:
    version: int

+class RefreshColumnResult:
+    rows_filled: int
+    version: int
+
 class AlterColumnsResult:
    version: int

@@ -958,9 +958,21 @@ class RemoteTable(Table):
    def count_rows(self, filter: Optional[str] = None) -> int:
        return LOOP.run(self._table.count_rows(filter))

-    def add_columns(self, transforms: Dict[str, str]) -> AddColumnsResult:
+    def add_columns(
+        self,
+        transforms: Dict[str, str] | None = None,
+        *,
+        computed: Dict[str, str] | None = None,
+    ) -> AddColumnsResult:
+        if computed:
+            raise NotImplementedError(
+                "computed columns are supported only on local tables"
+            )
        return LOOP.run(self._table.add_columns(transforms))

+    def refresh_column(self, column: str):
+        raise NotImplementedError("computed columns are supported only on local tables")
+
    def alter_columns(
        self, *alterations: Iterable[Dict[str, str]]
    ) -> AlterColumnsResult:
@@ -11,6 +11,11 @@ Provides StreamingDataset, a PyTorch IterableDataset that guarantees:
 - **Resumability**: state_dict / load_state_dict capture per-split consumption
  counts so training can resume from an exact mid-epoch position even when the
  distributed topology changes between runs.
+
+Transform failures on bad rows (e.g. nulls or NaNs from incomplete data) can
+be tolerated with ``on_transform_error="skip"``; see the parameter
+documentation on StreamingDataset for how this interacts with the guarantees
+above.
 """

 import ctypes
@@ -22,7 +27,7 @@ import time
 from collections import deque
 from concurrent.futures import ThreadPoolExecutor
 from multiprocessing import RawArray
-from typing import Any, Callable, Iterator, Optional
+from typing import Any, Callable, Iterator, Optional, Union

 from torch.utils.data import IterableDataset, get_worker_info

@@ -127,6 +132,49 @@ class StreamingDataset(IterableDataset):
        Maximum number of transforms to run concurrently.  Must be greater
        than zero.  When ``None`` (the default), uses ``os.cpu_count()`` or 1
        when the CPU count is unavailable.
+    on_transform_error:
+        What to do when the transform raises an exception:
+
+        - ``"raise"`` (the default): the exception propagates and iteration
+          aborts.
+        - ``"skip"``: the failing rows are dropped and iteration continues.
+        - ``"warn"``: like ``"skip"``, but a warning is logged for each
+          failing batch.
+        - a callable ``handler(exc) -> bool``: called with the exception;
+          return ``True`` to skip the failing rows or ``False`` to re-raise.
+          Useful to skip only expected error types (compatible with
+          ``webdataset.handlers`` style handlers).
+
+        When a batch fails, the transform is re-invoked on each single-row
+        slice of the batch so that only the rows that actually fail are
+        dropped.  Transforms should therefore be deterministic and accept
+        batches of any size (including one row).  Skipped rows are counted in
+        ``rows_skipped``.
+
+        Skipping weakens the elastic-determinism guarantee at the end of the
+        epoch: splits that lose more rows than others run dry earlier, and
+        each rank's iterator ends at the last cycle where every split *it
+        owns* still has a row.  Because bad rows are not distributed evenly
+        across splits, this means one rank's iterator can yield noticeably
+        fewer or more steps than another rank's *in the same run* — there is
+        no cross-rank coordination that stops every rank at the same global
+        step.  This is generally safe for asynchronous or single-rank use,
+        but synchronous distributed training (e.g. ranks that call
+        ``all_reduce`` every step) can hang or deadlock if one rank's
+        iterator is exhausted while others are still stepping; callers doing
+        synchronous multi-rank training with ``on_transform_error != "raise"``
+        are responsible for their own cross-rank stopping mechanism (e.g.
+        broadcasting a stop signal on ``StopIteration``).  The final few
+        global steps can also differ across topologies (bounded by the skew
+        in bad-row counts across splits).  The sequence of samples yielded
+        from each split remains deterministic.  Mid-epoch
+        checkpoints remain exact provided the transform fails
+        deterministically; in multi-rank training each rank must save its
+        own ``state_dict`` and the states must be combined with
+        ``merge_state_dicts`` before resuming on a different topology.
+        Prefer the ``filter`` parameter when bad rows can be expressed as a
+        SQL predicate (e.g. ``"col IS NOT NULL"``) — filtering happens before
+        splits are built, so every guarantee is fully preserved.
    worker_info_override:
        If set, used in place of ``torch.utils.data.get_worker_info()`` to
        determine the DataLoader worker assignment.  Intended for unit tests
@@ -152,6 +200,7 @@ class StreamingDataset(IterableDataset):
        filter: Optional[str] = None,
        transform: Optional[Callable] = None,
        transform_parallelism: Optional[int] = None,
+        on_transform_error: Union[str, Callable[[Exception], bool]] = "raise",
        connection_factory: Optional[Callable[[str], Any]] = None,
        worker_info_override=None,
    ):
@@ -167,6 +216,13 @@ class StreamingDataset(IterableDataset):
            )
        if transform_parallelism is not None and transform_parallelism <= 0:
            raise ValueError("transform_parallelism must be greater than 0")
+        if on_transform_error not in ("raise", "skip", "warn") and not callable(
+            on_transform_error
+        ):
+            raise ValueError(
+                "on_transform_error must be 'raise', 'skip', 'warn', or a "
+                f"callable, got {on_transform_error!r}"
+            )

        self._table = table
        self._num_splits = num_splits
@@ -182,6 +238,7 @@ class StreamingDataset(IterableDataset):
        self._filter = filter
        self._transform = transform
        self._transform_parallelism = transform_parallelism
+        self._on_transform_error = on_transform_error
        self._connection_factory = connection_factory
        self._worker_info_override = worker_info_override

@@ -199,19 +256,28 @@ class StreamingDataset(IterableDataset):
        # in the main process.  RawArray is picklable via the forkserver
        # reduction protocol so it survives the dataset pickle round-trip.
        # Layout: [unscanned_rows, raw_rows, cooked_rows, consumed_rows,
-        #          bytes_loaded, fetch_time_us, transform_time_us]
-        self._worker_stats: RawArray = RawArray(ctypes.c_int64, 7)
+        #          bytes_loaded, fetch_time_us, transform_time_us,
+        #          rows_skipped]
+        self._worker_stats: RawArray = RawArray(ctypes.c_int64, 8)

        # Cumulative bytes of Arrow buffer data fetched across all iterations.
        self._bytes_loaded: int = 0
        # Cumulative seconds spent in LanceDB I/O and in transform functions.
        self._fetch_time: float = 0.0
        self._transform_time: float = 0.0
+        # Cumulative rows dropped by on_transform_error across all iterations.
+        self._rows_skipped: int = 0

        # Number of samples each split has already been consumed.  At global
        # step boundaries all splits have consumed this many samples, so a
        # single scalar captures the topology-independent checkpoint state.
        self._resume_offset: int = 0
+        # Permutation position each split has consumed through, keyed by
+        # global split index.  Equal to _resume_offset for every split unless
+        # on_transform_error skipped rows, in which case skipped positions
+        # push the watermark of the affected splits further ahead.  Splits
+        # this instance has never iterated have no entry.
+        self._resume_positions: dict[int, int] = {}

        # Build the permutation table once, deterministically.
        builder = permutation_builder(table)
@@ -275,6 +341,7 @@ class StreamingDataset(IterableDataset):
        # Set identity transform on each Permutation so __getitems__ returns
        # the raw RecordBatch.  Stage 2 applies the real transform.
        permutations: list[Permutation] = []
+        initial_positions: list[int] = []
        for split_idx in my_splits:
            perm = Permutation.from_tables(
                self._table, self._perm_table, split=split_idx
@@ -282,14 +349,20 @@ class StreamingDataset(IterableDataset):
            if self._columns is not None:
                perm = perm.select_columns(self._columns)
            perm = perm.with_transform(lambda batch: batch)
-            if self._resume_offset > 0:
-                perm = perm.with_skip(self._resume_offset)
+            start_pos = self._resume_positions.get(split_idx, self._resume_offset)
+            if start_pos > 0:
+                perm = perm.with_skip(start_pos)
+            initial_positions.append(start_pos)
            permutations.append(perm)

        n = len(permutations)
        split_sizes = [perm.num_rows for perm in permutations]
        initial_offset = self._resume_offset
        local_consumed = [0] * n
+        # Permutation position each split has consumed through (absolute,
+        # i.e. counted from the start of the unskipped split).  Runs ahead of
+        # initial + local_consumed when rows are skipped.
+        pos_consumed = list(initial_positions)

        batch_size = self._read_batch_size
        max_prefetch = self._prefetch_batches
@@ -302,12 +375,14 @@ class StreamingDataset(IterableDataset):
            self._transform if self._transform is not None else Transforms.arrow2python
        )

-        # Per-split pipeline state.
+        # Per-split pipeline state.  Batches are paired with the absolute
+        # permutation position of their first row so that skipped rows can be
+        # accounted for in pos_consumed.
        fetch_head = [0] * n
-        io_pending = [deque() for _ in range(n)]  # Future[RecordBatch]
-        raw_batches = [deque() for _ in range(n)]  # RecordBatch — fetched, awaiting tx
-        tx_pending = [deque() for _ in range(n)]  # Future[list[Any]]
-        cooked = [deque() for _ in range(n)]  # rows ready to yield
+        io_pending = [deque() for _ in range(n)]  # (abs_start, Future[RecordBatch])
+        raw_batches = [deque() for _ in range(n)]  # (abs_start, RecordBatch)
+        tx_pending = [deque() for _ in range(n)]  # Future[list[(abs_pos, row)]]
+        cooked = [deque() for _ in range(n)]  # (abs_pos, row) ready to yield

        # Limit simultaneous transforms to transform_workers across all splits.
        tx_semaphore = threading.Semaphore(transform_workers)
@@ -330,7 +405,8 @@ class StreamingDataset(IterableDataset):
            fetch_head[i] += fetch
            perm_i = permutations[i]
            indices = list(range(start, start + fetch))
-            io_pending[i].append(io_pool.submit(_io_call, perm_i, indices))
+            abs_start = initial_positions[i] + start
+            io_pending[i].append((abs_start, io_pool.submit(_io_call, perm_i, indices)))

        def _fill_io(i: int) -> None:
            while len(io_pending[i]) < max_prefetch and fetch_head[i] < split_sizes[i]:
@@ -338,15 +414,72 @@ class StreamingDataset(IterableDataset):

        def _drain_io(i: int) -> None:
            """Move completed I/O futures into raw_batches non-blockingly."""
-            while io_pending[i] and io_pending[i][0].done():
-                raw_batches[i].append(io_pending[i].popleft().result())
+            while io_pending[i] and io_pending[i][0][1].done():
+                abs_start, fut = io_pending[i].popleft()
+                raw_batches[i].append((abs_start, fut.result()))

        # ── Stage 2 helpers ───────────────────────────────────────────────────

-        def _tx_call_guarded(batch):
+        on_error = self._on_transform_error
+
+        def _should_skip(exc: Exception) -> bool:
+            if on_error == "raise":
+                return False
+            if callable(on_error):
+                return bool(on_error(exc))
+            return True  # "skip" or "warn"
+
+        def _check_row_count(rows: list, num_rows: int) -> None:
+            if len(rows) != num_rows:
+                raise ValueError(
+                    f"transform returned {len(rows)} rows for a batch of "
+                    f"{num_rows}; transforms must return exactly one output "
+                    "row per input row.  To drop bad rows, raise inside the "
+                    "transform and pass on_transform_error='skip'."
+                )
+
+        def _transform_isolated(abs_start, batch, batch_exc):
+            """Re-run the transform on single-row slices, dropping failures."""
+            out = []
+            skipped = 0
+            first_exc = None
+            for j in range(batch.num_rows):
+                try:
+                    rows = list(final_transform(batch.slice(j, 1)))
+                except Exception as exc:
+                    if not _should_skip(exc):
+                        raise
+                    skipped += 1
+                    if first_exc is None:
+                        first_exc = exc
+                    continue
+                _check_row_count(rows, 1)
+                out.append((abs_start + j, rows[0]))
+            self._rows_skipped += skipped
+            if skipped and on_error == "warn":
+                logger.warning(
+                    "Skipped %d of %d rows whose transform failed (first error: %r)",
+                    skipped,
+                    batch.num_rows,
+                    first_exc if first_exc is not None else batch_exc,
+                )
+            return out
+
+        def _transform_batch(abs_start, batch):
+            """Apply the transform, returning [(abs_pos, row), ...]."""
+            try:
+                rows = list(final_transform(batch))
+            except Exception as exc:
+                if not _should_skip(exc):
+                    raise
+                return _transform_isolated(abs_start, batch, exc)
+            _check_row_count(rows, batch.num_rows)
+            return [(abs_start + j, row) for j, row in enumerate(rows)]
+
+        def _tx_call_guarded(abs_start, batch):
            try:
                t0 = time.perf_counter()
-                result = final_transform(batch)
+                result = _transform_batch(abs_start, batch)
                self._transform_time += time.perf_counter() - t0
                return result
            finally:
@@ -355,8 +488,8 @@ class StreamingDataset(IterableDataset):
        def _try_submit_tx(i: int) -> None:
            """Submit transforms for raw_batches[i] up to available capacity."""
            while raw_batches[i] and tx_semaphore.acquire(blocking=False):
-                batch = raw_batches[i].popleft()
-                tx_pending[i].append(tx_pool.submit(_tx_call_guarded, batch))
+                abs_start, batch = raw_batches[i].popleft()
+                tx_pending[i].append(tx_pool.submit(_tx_call_guarded, abs_start, batch))

        def _drain_tx(i: int) -> None:
            """Move completed transform futures into cooked non-blockingly."""
@@ -384,11 +517,14 @@ class StreamingDataset(IterableDataset):
                    # Acquire a transform slot (may block briefly if all
                    # transform_workers are busy with other splits).
                    tx_semaphore.acquire()
-                    batch = raw_batches[i].popleft()
-                    tx_pending[i].append(tx_pool.submit(_tx_call_guarded, batch))
+                    abs_start, batch = raw_batches[i].popleft()
+                    tx_pending[i].append(
+                        tx_pool.submit(_tx_call_guarded, abs_start, batch)
+                    )
                elif io_pending[i]:
                    # Block on the oldest in-flight I/O fetch.
-                    raw_batches[i].append(io_pending[i].popleft().result())
+                    abs_start, fut = io_pending[i].popleft()
+                    raw_batches[i].append((abs_start, fut.result()))
                    _advance(i)
                else:
                    break  # split exhausted
@@ -407,15 +543,28 @@ class StreamingDataset(IterableDataset):
                        _fill_io(i)

                    while True:
-                        # Stop when any split is exhausted (all exhaust
-                        # simultaneously: equal split sizes + round-robin).
-                        if any(local_consumed[i] >= split_sizes[i] for i in range(n)):
+                        # A cycle only runs if every split can still produce a
+                        # row.  Without skips all splits exhaust simultaneously
+                        # (equal split sizes + round-robin); when
+                        # on_transform_error drops rows a split can run dry
+                        # early, ending the epoch at the last complete cycle.
+                        # This check only sees splits owned by this rank/worker
+                        # (my_splits) — there is no cross-rank coordination, so
+                        # a different rank with fewer skipped rows keeps going;
+                        # see the on_transform_error docstring.
+                        exhausted = False
+                        for i in range(n):
+                            _ensure_cooked(i)
+                            if not cooked[i]:
+                                exhausted = True
+                                break
+                        if exhausted:
                            break

                        for i in range(n):
-                            _ensure_cooked(i)
-                            row = cooked[i].popleft()
+                            pos, row = cooked[i].popleft()
                            local_consumed[i] += 1
+                            pos_consumed[i] = pos + 1
                            _advance(i)

                            # After the last split in each cycle: update the
@@ -424,21 +573,39 @@ class StreamingDataset(IterableDataset):
                            # even when __iter__ runs in a worker process.
                            if i == n - 1:
                                self._resume_offset = initial_offset + local_consumed[i]
+                                for j, split_idx in enumerate(my_splits):
+                                    self._resume_positions[split_idx] = pos_consumed[j]
                                ws = self._worker_stats
                                ws[0] = sum(
                                    split_sizes[j] - fetch_head[j] for j in range(n)
                                )
                                ws[1] = sum(
-                                    batch.num_rows for q in raw_batches for batch in q
+                                    batch.num_rows
+                                    for q in raw_batches
+                                    for _, batch in q
                                )
                                ws[2] = sum(len(q) for q in cooked)
                                ws[3] = sum(local_consumed)
                                ws[4] = self._bytes_loaded
                                ws[5] = int(self._fetch_time * 1_000_000)
                                ws[6] = int(self._transform_time * 1_000_000)
+                                ws[7] = self._rows_skipped

                            yield row
                finally:
+                    # Final stats flush: the per-cycle write above never runs
+                    # when iteration ends mid-cycle (e.g. a split whose rows
+                    # were all skipped before completing a single cycle), so
+                    # counters like rows_skipped would otherwise be stale.
+                    ws = self._worker_stats
+                    ws[0] = sum(split_sizes[j] - fetch_head[j] for j in range(n))
+                    ws[1] = 0  # queue-depth properties document 0 when idle
+                    ws[2] = 0
+                    ws[3] = sum(local_consumed)
+                    ws[4] = self._bytes_loaded
+                    ws[5] = int(self._fetch_time * 1_000_000)
+                    ws[6] = int(self._transform_time * 1_000_000)
+                    ws[7] = self._rows_skipped
                    self._raw_batches_ref = None
                    self._cooked_ref = None
                    self._fetch_head_ref = None
@@ -492,7 +659,7 @@ class StreamingDataset(IterableDataset):
        batches.  Returns 0 when not iterating.
        """
        if self._raw_batches_ref is not None:
-            return sum(batch.num_rows for q in self._raw_batches_ref for batch in q)
+            return sum(batch.num_rows for q in self._raw_batches_ref for _, batch in q)
        return int(self._worker_stats[1])

    @property
@@ -522,6 +689,19 @@ class StreamingDataset(IterableDataset):
            )
        return int(self._worker_stats[0])

+    @property
+    def rows_skipped(self) -> int:
+        """Number of rows dropped because their transform raised an exception.
+
+        Only ever non-zero when ``on_transform_error`` is set to ``"skip"``,
+        ``"warn"``, or a callable that returned ``True``.  Accumulates across
+        multiple iterations of the same dataset instance and is never reset
+        automatically.
+        """
+        if self._raw_batches_ref is not None:
+            return self._rows_skipped
+        return int(self._worker_stats[7])
+
    @property
    def consumed_rows(self) -> int:
        """Number of rows already yielded to the caller across all splits.
@@ -587,12 +767,27 @@ class StreamingDataset(IterableDataset):
        every split has been consumed the same number of times (by the
        round-robin design), so the per-split count is a single uniform value
        that is identical across all ranks and DataLoader workers.
+
+        ``positions_consumed_per_split`` records how far into each split's
+        permutation iteration has advanced.  It only differs from
+        ``samples_consumed_per_split`` when ``on_transform_error`` skipped
+        rows, in which case entries are exact for the splits this instance
+        iterated and a lower bound (the sample count) for splits owned by
+        other ranks or workers.  Combine the state dicts from all ranks with
+        [merge_state_dicts][lancedb.streaming.StreamingDataset.merge_state_dicts]
+        to recover the exact value for every split before resuming on a
+        different topology.
        """
+        positions = [
+            self._resume_positions.get(split, self._resume_offset)
+            for split in range(self._num_splits)
+        ]
        return {
            "shuffle_seed": self._shuffle_seed,
            "num_splits": self._num_splits,
            "epoch": self._epoch,
            "samples_consumed_per_split": [self._resume_offset] * self._num_splits,
+            "positions_consumed_per_split": positions,
        }

    def load_state_dict(self, state: dict) -> None:
@@ -618,3 +813,96 @@ class StreamingDataset(IterableDataset):
            self._resume_offset = consumed[0] if consumed else 0
        else:
            self._resume_offset = int(consumed)
+        # Older checkpoints predate positions_consumed_per_split; without
+        # skipped rows positions equal sample counts, so falling back to
+        # _resume_offset (the .get default in __iter__) is exact.
+        positions = state.get("positions_consumed_per_split")
+        if positions is None:
+            self._resume_positions = {}
+        else:
+            self._resume_positions = {
+                split: int(pos) for split, pos in enumerate(positions)
+            }
+
+    @staticmethod
+    def merge_state_dicts(states: list[dict]) -> dict:
+        """Merge state dicts saved by different ranks into one exact state.
+
+        Only needed when ``on_transform_error`` skips rows in multi-rank
+        training: each rank then knows the exact permutation position only for
+        its own splits, and records a lower bound for the rest.  Because
+        exactly one rank owns each split, the elementwise maximum across all
+        ranks' ``positions_consumed_per_split`` recovers the exact position of
+        every split.  Without skipped rows every rank's state is already
+        identical and merging is a no-op.
+
+        Raises ``ValueError`` if the states are empty or were not produced by
+        the same run (mismatched seed, split count, epoch, or sample counts).
+
+        The merge is always all-to-all and topology-agnostic: collect the
+        ``state_dict()`` from every rank of the *previous* run into one list,
+        merge that whole list, and hand the identical merged result to every
+        rank of the *next* run — regardless of whether the rank count grew,
+        shrank, or stayed the same. There is no pairwise or subset merging
+        step, because each split's exact position is only known to whichever
+        rank owned that split, and the elementwise maximum needs every rank's
+        contribution to be correct.
+
+        For example, checkpointing 8 ranks and resuming on 4 (the same
+        pattern applies when growing, e.g. 4 ranks resuming on 8)::
+
+            states = [ds.state_dict() for ds in previous_run_datasets]  # 8
+            merged = StreamingDataset.merge_state_dicts(states)
+            for ds in resumed_datasets:  # now only 4 ranks
+                ds.load_state_dict(merged)  # same dict on every rank
+
+        The rank count on either side never affects the merge itself, since
+        ``merge_state_dicts`` only cares about the list of states it is
+        given.  Each split's position is recovered by elementwise maximum;
+        here rank 0 owned split 0 (and skipped two rows there) while rank 1
+        owned split 1 (and skipped one row):
+
+        >>> rank0 = {
+        ...     "shuffle_seed": 0, "num_splits": 2, "epoch": 0,
+        ...     "samples_consumed_per_split": [3, 3],
+        ...     "positions_consumed_per_split": [5, 3],
+        ... }
+        >>> rank1 = {
+        ...     "shuffle_seed": 0, "num_splits": 2, "epoch": 0,
+        ...     "samples_consumed_per_split": [3, 3],
+        ...     "positions_consumed_per_split": [3, 4],
+        ... }
+        >>> merged = StreamingDataset.merge_state_dicts([rank0, rank1])
+        >>> merged["positions_consumed_per_split"]
+        [5, 4]
+        """
+        if not states:
+            raise ValueError("merge_state_dicts requires at least one state dict")
+        first = states[0]
+        for state in states[1:]:
+            for key in ("shuffle_seed", "num_splits", "epoch"):
+                if state[key] != first[key]:
+                    raise ValueError(
+                        f"{key} mismatch across state dicts: "
+                        f"{state[key]} != {first[key]}"
+                    )
+            if (
+                state["samples_consumed_per_split"]
+                != first["samples_consumed_per_split"]
+            ):
+                raise ValueError(
+                    "samples_consumed_per_split mismatch across state dicts; "
+                    "state_dict() must be called at the same global step "
+                    "boundary on every rank"
+                )
+        merged = dict(first)
+        all_positions = [
+            state.get(
+                "positions_consumed_per_split", state["samples_consumed_per_split"]
+            )
+            for state in states
+        ]
+        merged["positions_consumed_per_split"] = [
+            max(per_split) for per_split in zip(*all_positions)
+        ]
+        return merged
@@ -176,6 +176,7 @@ if TYPE_CHECKING:
        CompactionStats,
        Tag,
        AddColumnsResult,
+        RefreshColumnResult,
        AddResult,
        AlterColumnsResult,
        UpdateFieldMetadataResult,
@@ -1916,7 +1917,14 @@ class Table(ABC):

    @abstractmethod
    def add_columns(
-        self, transforms: Dict[str, str] | pa.Field | List[pa.Field] | pa.Schema
+        self,
+        transforms: Dict[str, str]
+        | pa.Field
+        | List[pa.Field]
+        | pa.Schema
+        | None = None,
+        *,
+        computed: Dict[str, str] | None = None,
    ):
        """
        Add new columns with defined values.
@@ -1930,11 +1938,68 @@ class Table(ABC):
            Alternatively, a pyarrow Field or Schema can be provided to add
            new columns with the specified data types. The new columns will
            be initialized with null values.
+        computed: Dict[str, str], optional
+            A map of column name to a SQL expression defining the column. The
+            column's type and inputs are derived from the expression, so no
+            data type is supplied.
+
+            Unlike ``transforms``, the expression is stored rather than
+            evaluated now: the column is committed with no values, and rows get
+            them from [`refresh_column`][lancedb.table.Table.refresh_column].
+            Declaring one therefore costs the same on a large table as on an
+            empty one.
+
+            A refresh does not revisit rows it has already filled, so mutating
+            an input leaves the value computed at fill time; recomputing means
+            dropping the column and declaring it again. While a declaration
+            reads a column, that column cannot be renamed, retyped or dropped.
+
+            Local tables only; LanceDB Cloud and Enterprise raise
+            ``NotImplementedError``. Cannot be combined with ``transforms``.

        Returns
        -------
        AddColumnsResult
            version: the new version number of the table after adding columns.
+
+        Examples
+        --------
+        >>> import lancedb
+        >>> db = lancedb.connect("./.lancedb")
+        >>> table = db.create_table("computed_demo", [{"x": 1}, {"x": 2}])
+        >>> table.add_columns(computed={"doubled": "x * 2"})
+        AddColumnsResult(version=2)
+        >>> table.refresh_column("doubled")
+        RefreshColumnResult(rows_filled=2, version=3)
+        >>> table.to_arrow().sort_by("x").to_pandas()
+           x  doubled
+        0  1        2
+        1  2        4
+        """
+
+    @abstractmethod
+    def refresh_column(self, column: str) -> "RefreshColumnResult":
+        """
+        Fill the rows of a computed column that hold no value yet.
+
+        Declared with ``add_columns(computed=...)``, a column starts empty and
+        gets its values here. Rows appended since the last refresh are filled
+        by the next one; rows already filled are left as they are, so the call
+        is idempotent and does not observe a mutated input.
+
+        Local tables only; LanceDB Cloud and Enterprise raise
+        ``NotImplementedError``.
+
+        Parameters
+        ----------
+        column: str
+            The name of the computed column to fill.
+
+        Returns
+        -------
+        RefreshColumnResult
+            rows_filled: the number of rows given a value.
+            version: the new version number of the table.
        """

    @abstractmethod
@@ -3939,9 +4004,21 @@ class LanceTable(Table):
        return LOOP.run(self._table.index_stats(index_name))

    def add_columns(
-        self, transforms: Dict[str, str] | pa.field | List[pa.field] | pa.Schema
+        self,
+        transforms: Dict[str, str]
+        | pa.field
+        | List[pa.field]
+        | pa.Schema
+        | None = None,
+        *,
+        computed: Dict[str, str] | None = None,
    ) -> AddColumnsResult:
-        return LOOP.run(self._table.add_columns(transforms))
+        return LOOP.run(self._table.add_columns(transforms, computed=computed))
+
+    def refresh_column(self, column: str) -> "RefreshColumnResult":
+        """Fill a computed column's unfilled rows. See
+        [`AsyncTable.refresh_column`][lancedb.AsyncTable.refresh_column]."""
+        return LOOP.run(self._table.refresh_column(column))

    def alter_columns(
        self, *alterations: Iterable[Dict[str, str]]
@@ -4676,6 +4753,13 @@ class AsyncTable:
        via [`set_unenforced_primary_key`]; bucket sharding additionally
        requires it to be the single column being bucketed.

+        By default the MemWAL maintains every index on the table, resolved
+        here — a snapshot, so an index created afterwards needs the spec unset
+        and set again. This fails if one cannot be maintained; name the set
+        with ``with_maintained_indexes`` to install anyway. That pins an exact
+        set (a still-building index is rejected, not omitted); ``[]`` maintains
+        none.
+
        Parameters
        ----------
        spec : LsmWriteSpec
@@ -4702,9 +4786,9 @@ class AsyncTable:

        Returns ``None`` when the MemWAL LSM write path is not enabled (no
        spec has been set, or it was removed with `unset_lsm_write_spec`).
-        The returned spec — including its ``maintained_indexes`` and
-        ``writer_config_defaults`` — mirrors what was passed to
-        `set_lsm_write_spec`.
+        The returned spec mirrors what was passed to `set_lsm_write_spec`,
+        except that ``maintained_indexes`` always reports the concrete list
+        resolved when the spec was set — ``None`` never round-trips.
        """
        return await self._inner.get_lsm_write_spec()

@@ -5849,7 +5933,14 @@ class AsyncTable:
        return await self._inner.update(updates_sql, where)

    async def add_columns(
-        self, transforms: dict[str, str] | pa.field | List[pa.field] | pa.Schema
+        self,
+        transforms: dict[str, str]
+        | pa.field
+        | List[pa.field]
+        | pa.Schema
+        | None = None,
+        *,
+        computed: dict[str, str] | None = None,
    ) -> AddColumnsResult:
        """
        Add new columns with defined values.
@@ -5862,6 +5953,21 @@ class AsyncTable:
            each row in the table, and can reference existing columns.
            Alternatively, you can pass a pyarrow field or schema to add
            new columns with NULLs.
+        computed: Dict[str, str], optional
+            A map of column name to a SQL expression defining the column. The
+            column's type and inputs are derived from the expression.
+
+            Unlike ``transforms``, the expression is stored rather than
+            evaluated now: the column is committed with no values, and rows get
+            them from
+            [`refresh_column`][lancedb.table.AsyncTable.refresh_column].
+
+            A refresh does not revisit rows it has already filled, so mutating
+            an input leaves the value computed at fill time. While a
+            declaration reads a column, that column cannot be renamed, retyped
+            or dropped.
+
+            Local tables only. Cannot be combined with ``transforms``.

        Returns
        -------
@@ -5875,11 +5981,43 @@ class AsyncTable:
            {isinstance(f, pa.Field) for f in transforms}
        ):
            transforms = pa.schema(transforms)
+        if computed:
+            if transforms:
+                raise ValueError(
+                    "add_columns cannot take both transforms and computed columns"
+                )
+            return await self._inner.add_computed_columns(list(computed.items()))
+        if transforms is None:
+            raise ValueError("add_columns requires transforms or computed columns")
        if isinstance(transforms, pa.Schema):
            return await self._inner.add_columns_with_schema(transforms)
        else:
            return await self._inner.add_columns(list(transforms.items()))

+    async def refresh_column(self, column: str) -> RefreshColumnResult:
+        """
+        Fill the rows of a computed column that hold no value yet.
+
+        Declared with ``add_columns(computed=...)``, a column starts empty and
+        gets its values here. Rows appended since the last refresh are filled
+        by the next one; rows already filled are left as they are, so the call
+        is idempotent and does not observe a mutated input.
+
+        Local tables only; LanceDB Cloud and Enterprise raise
+        ``NotImplementedError``.
+
+        Parameters
+        ----------
+        column: str
+            The name of the computed column to fill.
+
+        Returns
+        -------
+        RefreshColumnResult
+            The number of rows filled and the new version of the table.
+        """
+        return await self._inner.refresh_column(column)
+
    async def alter_columns(
        self, *alterations: Iterable[dict[str, Any]]
    ) -> AlterColumnsResult:
@@ -1456,6 +1456,408 @@ def test_shuffle_clump_size_yields_all_rows(lance_table):
    )


+# ---------------------------------------------------------------------------
+# on_transform_error tests
+# ---------------------------------------------------------------------------
+
+
+class BadRowError(ValueError):
+    """Raised by the failing transforms below when a batch contains a bad id."""
+
+
+def _failing_transform(bad_ids: set):
+    """A transform that raises BadRowError whenever the batch has a bad id.
+
+    Raises on the full batch and on any single-row slice containing a bad id,
+    so per-row isolation drops exactly the bad rows.
+    """
+
+    def transform(batch: pa.RecordBatch) -> list:
+        ids = batch.column("id").to_pylist()
+        bad = sorted(set(ids) & bad_ids)
+        if bad:
+            raise BadRowError(f"bad ids in batch: {bad}")
+        return [{"id": i} for i in ids]
+
+    return transform
+
+
+def _sequential_split_members(table) -> list[list[int]]:
+    """Return each split's ids in yield order for shuffle=False.
+
+    With a single rank and no workers the round-robin yields one row per split
+    per cycle, so item k of a clean run belongs to split k % NUM_SPLITS.
+    """
+    ds = StreamingDataset(table, num_splits=NUM_SPLITS, shuffle=False)
+    members: list[list[int]] = [[] for _ in range(NUM_SPLITS)]
+    for k, row in enumerate(ds):
+        members[k % NUM_SPLITS].append(row["id"])
+    return members
+
+
+def test_on_transform_error_default_raises(lance_table):
+    """By default a transform exception propagates and aborts iteration."""
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle_seed=SHUFFLE_SEED,
+        transform=_failing_transform({7}),
+    )
+    with pytest.raises(BadRowError):
+        list(ds)
+
+
+def test_on_transform_error_invalid_value(lance_table):
+    with pytest.raises(ValueError, match="on_transform_error"):
+        StreamingDataset(lance_table, num_splits=NUM_SPLITS, on_transform_error="bogus")
+
+
+def test_on_transform_error_skip_drops_bad_rows(lance_table):
+    """With one bad row per split, 'skip' yields every good row exactly once
+    and counts the dropped rows in rows_skipped."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = {members[i][4] for i in range(NUM_SPLITS)}
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="skip",
+    )
+    assert ds.rows_skipped == 0
+
+    ids = [row["id"] for row in ds]
+
+    assert sorted(ids) == sorted(set(range(NUM_ROWS)) - bad_ids)
+    assert ds.rows_skipped == NUM_SPLITS
+
+
+def test_on_transform_error_skip_uneven_ends_at_last_complete_cycle(lance_table):
+    """When one split loses more rows than the others, the epoch ends at the
+    last cycle where every split still has a row — no crash, no bad rows, and
+    every step remains one sample per split."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = set(members[0][:3])  # all 3 bad rows in split 0
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="skip",
+    )
+    items = [row["id"] for row in ds]
+
+    rows_per_split = NUM_ROWS // NUM_SPLITS
+    expected_cycles = rows_per_split - len(bad_ids)
+    assert len(items) == expected_cycles * NUM_SPLITS
+    assert len(set(items)) == len(items), "duplicate samples yielded"
+    assert not set(items) & bad_ids, "a bad row was yielded"
+    # Split 0 contributed exactly its surviving rows, in order, one per cycle.
+    survivors = [i for i in members[0] if i not in bad_ids]
+    assert items[0::NUM_SPLITS] == survivors[:expected_cycles]
+
+
+def test_on_transform_error_warn_logs(lance_table, caplog):
+    """'warn' skips like 'skip' but logs a warning for the failing batch."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = {members[i][3] for i in range(NUM_SPLITS)}
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="warn",
+    )
+    with caplog.at_level(logging.WARNING, logger="lancedb.streaming"):
+        items = list(ds)
+
+    assert len(items) == NUM_ROWS - NUM_SPLITS
+    assert ds.rows_skipped == NUM_SPLITS
+    assert "Skipped" in caplog.text
+    assert "BadRowError" in caplog.text
+
+
+def test_on_transform_error_callable_selective(lance_table):
+    """A callable handler can skip expected errors and re-raise the rest."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = {members[i][0] for i in range(NUM_SPLITS)}
+
+    handled: list[Exception] = []
+
+    def handler(exc: Exception) -> bool:
+        handled.append(exc)
+        return isinstance(exc, BadRowError)
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error=handler,
+    )
+    items = list(ds)
+    assert len(items) == NUM_ROWS - NUM_SPLITS
+    assert handled and all(isinstance(exc, BadRowError) for exc in handled)
+
+    def broken_transform(batch: pa.RecordBatch) -> list:
+        raise TypeError("boom")
+
+    ds2 = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=broken_transform,
+        on_transform_error=handler,
+    )
+    with pytest.raises(TypeError, match="boom"):
+        list(ds2)
+
+
+def test_transform_wrong_row_count_raises(lance_table):
+    """A transform that returns the wrong number of rows is an error even with
+    on_transform_error='skip' — silent shrinkage would corrupt accounting."""
+
+    def drops_rows(batch: pa.RecordBatch) -> list:
+        return batch.column("id").to_pylist()[:-1]
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle_seed=SHUFFLE_SEED,
+        transform=drops_rows,
+        on_transform_error="skip",
+    )
+    with pytest.raises(ValueError, match="one output row per input row"):
+        list(ds)
+
+
+def test_skip_deterministic_across_runs(lance_table):
+    """With a fixed seed, skipping produces the identical sample sequence on
+    every run — skips are data-dependent, not run-dependent."""
+    bad_ids = {5, 17, 46}
+
+    def run() -> tuple[list[int], int]:
+        ds = StreamingDataset(
+            lance_table,
+            num_splits=NUM_SPLITS,
+            shuffle_seed=SHUFFLE_SEED,
+            transform=_failing_transform(bad_ids),
+            on_transform_error="skip",
+        )
+        return [row["id"] for row in ds], ds.rows_skipped
+
+    ids_a, skipped_a = run()
+    ids_b, skipped_b = run()
+    assert ids_a == ids_b
+    assert skipped_a == skipped_b
+    assert not set(ids_a) & bad_ids
+
+
+def test_skip_elastic_det_across_world_sizes(lance_table):
+    """With equal bad-row counts per split, skipping preserves the full
+    elastic-determinism guarantee: identical global batches at every step for
+    every compatible world_size."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = {members[i][6] for i in range(NUM_SPLITS)}
+
+    def collect(world_size: int) -> list[frozenset[int]]:
+        micro = GLOBAL_BATCH_SIZE // world_size
+        iters = [
+            iter(
+                StreamingDataset(
+                    lance_table,
+                    num_splits=NUM_SPLITS,
+                    shuffle=False,
+                    rank=rank,
+                    world_size=world_size,
+                    transform=_failing_transform(bad_ids),
+                    on_transform_error="skip",
+                )
+            )
+            for rank in range(world_size)
+        ]
+        _STOP = object()
+        batches: list[frozenset[int]] = []
+        while True:
+            step_samples: set[int] = set()
+            exhausted = 0
+            for it in iters:
+                for _ in range(micro):
+                    val = next(it, _STOP)
+                    if val is _STOP:
+                        exhausted += 1
+                        break
+                    step_samples.add(val["id"])
+            if exhausted == len(iters):
+                break
+            assert exhausted == 0, (
+                "Rank iterators exhausted at different steps despite equal "
+                "bad-row counts per split"
+            )
+            batches.append(frozenset(step_samples))
+        return batches
+
+    reference = collect(1)
+    assert len(reference) == NUM_ROWS // NUM_SPLITS - 1
+    for ws in (2, 3, 4):
+        assert collect(ws) == reference, f"world_size={ws} diverged"
+
+
+def test_resumability_with_skips_same_topology(lance_table):
+    """Checkpointing mid-epoch with skipped rows resumes exactly: no sample
+    repeated, no sample lost, skipped rows stay skipped."""
+    members = _sequential_split_members(lance_table)
+    # Uneven skips: positions diverge across splits (2 bad in split 0, 1 in
+    # split 5), which only a position-based checkpoint can resume exactly.
+    bad_ids = {members[0][2], members[0][3], members[5][7]}
+    kwargs = dict(
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="skip",
+    )
+
+    reference = [row["id"] for row in StreamingDataset(lance_table, **kwargs)]
+    rows_per_split = NUM_ROWS // NUM_SPLITS
+    assert len(reference) == (rows_per_split - 2) * NUM_SPLITS
+
+    steps = 3
+    ds = StreamingDataset(lance_table, **kwargs)
+    it = iter(ds)
+    consumed = [next(it)["id"] for _ in range(steps * NUM_SPLITS)]
+    checkpoint = ds.state_dict()
+    it.close()
+
+    # Split 0 skipped positions 2 and 3 within its first 3 yields; split 5's
+    # bad row is beyond the checkpoint.  Everything else is at 3 = the sample
+    # count.
+    positions = checkpoint["positions_consumed_per_split"]
+    assert positions[0] == 5
+    assert positions[1:] == [3] * (NUM_SPLITS - 1)
+    assert checkpoint["samples_consumed_per_split"] == [3] * NUM_SPLITS
+
+    ds2 = StreamingDataset(lance_table, **kwargs)
+    ds2.load_state_dict(checkpoint)
+    resumed = [row["id"] for row in ds2]
+
+    assert consumed == reference[: steps * NUM_SPLITS]
+    assert resumed == reference[steps * NUM_SPLITS :]
+
+
+def test_resumability_with_skips_elastic_merge(lance_table):
+    """Elastic resume with skips: each rank's checkpoint knows exact positions
+    only for its own splits; merge_state_dicts recovers the global state, and
+    a run on a different world_size continues exactly."""
+    members = _sequential_split_members(lance_table)
+    # Bad rows early in split 0 (rank 0) and split 6 (rank 1 of a ws=2 run) so
+    # both ranks' position vectors diverge before the checkpoint.
+    bad_ids = {members[0][0], members[0][2], members[6][1]}
+    kwargs = dict(
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="skip",
+    )
+
+    reference = [row["id"] for row in StreamingDataset(lance_table, **kwargs)]
+
+    steps = 3
+    world_size = 2
+    micro = GLOBAL_BATCH_SIZE // world_size
+    datasets = [
+        StreamingDataset(lance_table, rank=rank, world_size=world_size, **kwargs)
+        for rank in range(world_size)
+    ]
+    iters = [iter(ds) for ds in datasets]
+    seen: list[frozenset[int]] = []
+    for _ in range(steps):
+        step_samples = set()
+        for it in iters:
+            for _ in range(micro):
+                step_samples.add(next(it)["id"])
+        seen.append(frozenset(step_samples))
+    states = [ds.state_dict() for ds in datasets]
+    for it in iters:
+        it.close()
+
+    merged = StreamingDataset.merge_state_dicts(states)
+    expected_positions = [3] * NUM_SPLITS
+    expected_positions[0] = 5  # skipped positions 0 and 2
+    expected_positions[6] = 4  # skipped position 1
+    assert merged["positions_consumed_per_split"] == expected_positions
+
+    # The first 3 global batches match the world_size=1 reference.
+    ref_batches = [
+        frozenset(reference[s * NUM_SPLITS : (s + 1) * NUM_SPLITS])
+        for s in range(len(reference) // NUM_SPLITS)
+    ]
+    assert seen == ref_batches[:steps]
+
+    # Resume on world_size=1 from the merged state.
+    ds_resume = StreamingDataset(lance_table, **kwargs)
+    ds_resume.load_state_dict(merged)
+    resumed = [row["id"] for row in ds_resume]
+    assert resumed == reference[steps * NUM_SPLITS :]
+
+
+def test_rows_skipped_flushed_when_split_entirely_bad(lance_table):
+    """A split whose rows all fail never completes a cycle, so the epoch ends
+    immediately — but rows_skipped must still report the drops after the
+    iterator exits (the shared-memory counter is flushed on exhaustion)."""
+    members = _sequential_split_members(lance_table)
+    bad_ids = set(members[0])  # every row of split 0 is bad
+
+    ds = StreamingDataset(
+        lance_table,
+        num_splits=NUM_SPLITS,
+        shuffle=False,
+        transform=_failing_transform(bad_ids),
+        on_transform_error="skip",
+    )
+    assert list(ds) == []
+    assert ds.rows_skipped == len(bad_ids)
+
+
+def test_merge_state_dicts_validates_consistency(lance_table):
+    ds = StreamingDataset(lance_table, num_splits=NUM_SPLITS, shuffle_seed=SHUFFLE_SEED)
+    state = ds.state_dict()
+    other = dict(state, shuffle_seed=SHUFFLE_SEED + 1)
+    with pytest.raises(ValueError, match="shuffle_seed mismatch"):
+        StreamingDataset.merge_state_dicts([state, other])
+    with pytest.raises(ValueError, match="at least one"):
+        StreamingDataset.merge_state_dicts([])
+
+
+def test_load_state_dict_without_positions_key(lance_table):
+    """Checkpoints from before positions_consumed_per_split existed still
+    resume exactly (positions equal sample counts when nothing is skipped)."""
+    reference = [
+        row["id"]
+        for row in StreamingDataset(
+            lance_table, num_splits=NUM_SPLITS, shuffle_seed=SHUFFLE_SEED
+        )
+    ]
+
+    steps = 4
+    ds = StreamingDataset(lance_table, num_splits=NUM_SPLITS, shuffle_seed=SHUFFLE_SEED)
+    it = iter(ds)
+    for _ in range(steps * NUM_SPLITS):
+        next(it)
+    checkpoint = ds.state_dict()
+    it.close()
+    del checkpoint["positions_consumed_per_split"]
+
+    ds2 = StreamingDataset(
+        lance_table, num_splits=NUM_SPLITS, shuffle_seed=SHUFFLE_SEED
+    )
+    ds2.load_state_dict(checkpoint)
+    resumed = [row["id"] for row in ds2]
+    assert resumed == reference[steps * NUM_SPLITS :]
+
+
 def test_num_splits_defaults_to_world_size(lance_table):
    """Omitting num_splits gives world_size splits (one per rank)."""
    ds = StreamingDataset(
@@ -83,7 +83,9 @@ def test_lsm_write_spec_repr():
    assert s.spec_type == "bucket"
    assert s.column == "id"
    assert s.num_buckets == 4
-    assert s.maintained_indexes == []
+    # A fresh spec defers its maintained set to install time.
+    assert s.maintained_indexes is None
+    assert s.with_maintained_indexes([]).maintained_indexes == []
    assert "bucket" in repr(s)
    assert "id" in repr(s)
    assert "4" in repr(s)
@@ -169,18 +171,23 @@ def test_get_lsm_write_spec(tmp_path):
    table.unset_lsm_write_spec()
    assert table.get_lsm_write_spec() is None

-    # Identity round-trips (column recovered from the schema).
+    # Identity round-trips (column recovered from the schema). Leaving the
+    # maintained set to be inferred picks up the index on the table, so the
+    # spec reads back naming it rather than as "infer".
    table.set_lsm_write_spec(LsmWriteSpec.identity("id"))
    spec = table.get_lsm_write_spec()
    assert spec.spec_type == "identity"
    assert spec.column == "id"
+    assert spec.maintained_indexes == [idx_name]
    table.unset_lsm_write_spec()

-    # Unsharded round-trips (no routing column).
-    table.set_lsm_write_spec(LsmWriteSpec.unsharded())
+    # Unsharded round-trips (no routing column). Opting out is distinct from
+    # the inferred default.
+    table.set_lsm_write_spec(LsmWriteSpec.unsharded().with_maintained_indexes([]))
    spec = table.get_lsm_write_spec()
    assert spec.spec_type == "unsharded"
    assert spec.column is None
+    assert spec.maintained_indexes == []


@pytest.mark.asyncio
@@ -544,7 +544,7 @@ def test_lsm_read_fts_unmaintained_index_errors(tmp_path):
    table.create_index("text", config=FTS())
    # No maintained indexes: the active memtable FTS arm cannot serve un-compacted
    # docs, so the search would silently omit them — reject instead.
-    table.set_lsm_write_spec(LsmWriteSpec.unsharded())
+    table.set_lsm_write_spec(LsmWriteSpec.unsharded().with_maintained_indexes([]))
    with pytest.raises(Exception, match="maintained"):
        table.search("fox", query_type="fts", fts_columns="text").to_arrow()

@@ -631,7 +631,7 @@ def test_lsm_read_vector_unmaintained_index_errors(tmp_path):
    )
    # Spec with NO maintained indexes: the base vector index's catch-up is untracked,
    # so the scanner rejects rather than risk dropping compacted-but-unindexed rows.
-    table.set_lsm_write_spec(LsmWriteSpec.unsharded())
+    table.set_lsm_write_spec(LsmWriteSpec.unsharded().with_maintained_indexes([]))
    with pytest.raises(Exception, match="maintained"):
        table.search([1.0] * VECTOR_DIM).to_arrow()

@@ -3854,3 +3854,37 @@ async def test_async_search_runs_embedding_on_dedicated_executor(
    assert all(name.startswith("lancedb-embedding") for name in captured_threads), (
        f"embedding ran off the dedicated executor: {captured_threads}"
    )
+
+
+def test_computed_column_declare_and_refresh(tmp_path):
+    db = lancedb.connect(tmp_path)
+    table = db.create_table("computed", [{"x": 1}, {"x": 2}])
+
+    table.add_columns(computed={"doubled": "x * 2"})
+    assert table.to_arrow()["doubled"].to_pylist() == [None, None]
+
+    result = table.refresh_column("doubled")
+    assert result.rows_filled == 2
+    assert sorted(table.to_arrow()["doubled"].to_pylist()) == [2, 4]
+
+    table.add([{"x": 5}])
+    assert table.refresh_column("doubled").rows_filled == 1
+    assert sorted(table.to_arrow()["doubled"].to_pylist()) == [2, 4, 10]
+
+
+def test_computed_column_rejects_transforms_and_computed_together(tmp_path):
+    db = lancedb.connect(tmp_path)
+    table = db.create_table("computed_mixed", [{"x": 1}])
+    with pytest.raises(ValueError):
+        table.add_columns({"a": "x + 1"}, computed={"b": "x * 2"})
+
+
+@pytest.mark.asyncio
+async def test_computed_column_async(tmp_path):
+    db = await lancedb.connect_async(tmp_path)
+    table = await db.create_table("computed_async", [{"x": 3}])
+
+    await table.add_columns(computed={"tripled": "x * 3"})
+    await table.refresh_column("tripled")
+
+    assert (await table.to_arrow())["tripled"].to_pylist() == [9]
@@ -289,7 +289,7 @@ struct IvfHnswFlatParams {
    target_partition_size: Option<u32>,
 }

-#[pyclass(get_all)]
+#[pyclass(module = "lancedb._lancedb", get_all)]
 /// A description of an index currently configured on a column
 pub struct IndexConfig {
    /// The type of the index
@@ -16,7 +16,8 @@ use query::{FTSQuery, HybridQuery, Query, VectorQuery};
 use session::Session;
 use table::{
    AddColumnsResult, AddResult, AlterColumnsResult, DeleteResult, DropColumnsResult, FtsToken,
-    LsmWriteSpec, MergeResult, PyBlobFile, Table, UpdateFieldMetadataResult, UpdateResult,
+    LsmWriteSpec, MergeResult, PyBlobFile, RefreshColumnResult, Table, UpdateFieldMetadataResult,
+    UpdateResult,
 };

 pub mod arrow;
@@ -57,6 +58,7 @@ pub fn _lancedb(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
    m.add_class::<VectorQuery>()?;
    m.add_class::<RecordBatchStream>()?;
    m.add_class::<AddColumnsResult>()?;
+    m.add_class::<RefreshColumnResult>()?;
    m.add_class::<AlterColumnsResult>()?;
    m.add_class::<UpdateFieldMetadataResult>()?;
    m.add_class::<AddResult>()?;
@@ -11,7 +11,7 @@ use pyo3::{PyResult, pyclass, pymethods};
 /// Sessions allow you to configure cache sizes for index and metadata caches,
 /// which can significantly impact memory use and performance. They can
 /// also be re-used across multiple connections to share the same cache state.
-#[pyclass(from_py_object)]
+#[pyclass(module = "lancedb._lancedb", from_py_object)]
 #[derive(Clone)]
 pub struct Session {
    pub(crate) inner: Arc<LanceSession>,
@@ -246,12 +246,22 @@ impl From<lancedb::table::MergeResult> for MergeResult {
    }
 }

+/// Render for `__repr__`, so the default reads as Python's `None` rather than
+/// Rust's `Some([..])`.
+fn fmt_maintained(maintained: &Option<Vec<String>>) -> String {
+    match maintained {
+        Some(names) => format!("{:?}", names),
+        None => "None".to_string(),
+    }
+}
+
 /// Specification selecting Lance's MemWAL LSM-style write path for
 /// `merge_insert`.
 ///
 /// Constructed via the `bucket(...)`, `identity(...)`, or `unsharded()`
 /// classmethods, then optionally chain `with_maintained_indexes(...)` and
-/// `with_writer_config_defaults(...)`.
+/// `with_writer_config_defaults(...)`. A fresh spec maintains every index the
+/// MemWAL supports, resolved on install.
 #[pyclass(from_py_object)]
 #[derive(Clone, Debug)]
 pub struct LsmWriteSpec {
@@ -291,11 +301,11 @@ impl LsmWriteSpec {
        }
    }

-    /// Replace the list of indexes the MemWAL should keep up to date as
-    /// rows are appended. Each name must reference an index that
-    /// already exists on the table at the time `set_lsm_write_spec`
-    /// is called.
-    pub fn with_maintained_indexes(&self, indexes: Vec<String>) -> Self {
+    /// Set which indexes the MemWAL maintains. `None` (the default)
+    /// resolves every supported index on install; a list is verbatim,
+    /// and an empty list maintains nothing.
+    #[pyo3(signature = (indexes))]
+    pub fn with_maintained_indexes(&self, indexes: Option<Vec<String>>) -> Self {
        Self {
            inner: self.inner.clone().with_maintained_indexes(indexes),
        }
@@ -317,23 +327,29 @@ impl LsmWriteSpec {
                maintained_indexes,
                writer_config_defaults,
            } => format!(
-                "LsmWriteSpec.bucket(column={:?}, num_buckets={}, maintained_indexes={:?}, writer_config_defaults={:?})",
-                column, num_buckets, maintained_indexes, writer_config_defaults,
+                "LsmWriteSpec.bucket(column={:?}, num_buckets={}, maintained_indexes={}, writer_config_defaults={:?})",
+                column,
+                num_buckets,
+                fmt_maintained(maintained_indexes),
+                writer_config_defaults,
            ),
            lancedb::table::LsmWriteSpec::Identity {
                column,
                maintained_indexes,
                writer_config_defaults,
            } => format!(
-                "LsmWriteSpec.identity(column={:?}, maintained_indexes={:?}, writer_config_defaults={:?})",
-                column, maintained_indexes, writer_config_defaults,
+                "LsmWriteSpec.identity(column={:?}, maintained_indexes={}, writer_config_defaults={:?})",
+                column,
+                fmt_maintained(maintained_indexes),
+                writer_config_defaults,
            ),
            lancedb::table::LsmWriteSpec::Unsharded {
                maintained_indexes,
                writer_config_defaults,
            } => format!(
-                "LsmWriteSpec.unsharded(maintained_indexes={:?}, writer_config_defaults={:?})",
-                maintained_indexes, writer_config_defaults,
+                "LsmWriteSpec.unsharded(maintained_indexes={}, writer_config_defaults={:?})",
+                fmt_maintained(maintained_indexes),
+                writer_config_defaults,
            ),
        }
    }
@@ -368,10 +384,10 @@ impl LsmWriteSpec {
        }
    }

-    /// Names of indexes the MemWAL should keep up to date during writes.
+    /// Indexes the MemWAL keeps up to date, or `None` for every supported one.
    #[getter]
-    pub fn maintained_indexes(&self) -> Vec<String> {
-        self.inner.maintained_indexes().to_vec()
+    pub fn maintained_indexes(&self) -> Option<Vec<String>> {
+        self.inner.maintained_indexes().map(<[String]>::to_vec)
    }

    /// Default `ShardWriter` configuration recorded by this spec.
@@ -399,6 +415,32 @@ pub struct AddColumnsResult {
    pub version: u64,
 }

+#[pyclass(get_all, from_py_object)]
+#[derive(Clone, Debug)]
+pub struct RefreshColumnResult {
+    pub rows_filled: u64,
+    pub version: u64,
+}
+
+#[pymethods]
+impl RefreshColumnResult {
+    pub fn __repr__(&self) -> String {
+        format!(
+            "RefreshColumnResult(rows_filled={}, version={})",
+            self.rows_filled, self.version
+        )
+    }
+}
+
+impl From<lancedb::table::RefreshColumnResult> for RefreshColumnResult {
+    fn from(result: lancedb::table::RefreshColumnResult) -> Self {
+        Self {
+            rows_filled: result.rows_filled,
+            version: result.version,
+        }
+    }
+}
+
 #[pymethods]
 impl AddColumnsResult {
    pub fn __repr__(&self) -> String {
@@ -563,7 +605,7 @@ impl PyBlobFile {
    }
 }

-#[pyclass(get_all, from_py_object)]
+#[pyclass(module = "lancedb._lancedb", get_all, from_py_object)]
 #[derive(Clone, Debug)]
 pub struct FtsToken {
    pub text: String,
@@ -1494,6 +1536,29 @@ impl Table {
        })
    }

+    pub fn add_computed_columns(
+        self_: PyRef<'_, Self>,
+        columns: Vec<(String, String)>,
+    ) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.inner_ref()?.clone();
+        future_into_py(self_.py(), async move {
+            let mut builder = inner.add_columns();
+            for (name, expression) in columns {
+                builder = builder.computed(name, expression);
+            }
+            let result = builder.execute().await.infer_error()?;
+            Ok(AddColumnsResult::from(result))
+        })
+    }
+
+    pub fn refresh_column(self_: PyRef<'_, Self>, column: String) -> PyResult<Bound<'_, PyAny>> {
+        let inner = self_.inner_ref()?.clone();
+        future_into_py(self_.py(), async move {
+            let result = inner.refresh_column(column).await.infer_error()?;
+            Ok(RefreshColumnResult::from(result))
+        })
+    }
+
    pub fn add_columns_with_schema(
        self_: PyRef<'_, Self>,
        schema: PyArrowType<Schema>,
@@ -799,7 +799,7 @@ name = "cuda-bindings"
 version = "13.3.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "cuda-pathfinder" },
+    { name = "cuda-pathfinder", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 wheels = [
    { url = "https://files.pythonhosted.org/packages/a9/21/8464d133752951c154feafb3b65c297e7d80f301183d220bec4c830f1441/cuda_bindings-13.3.1-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:120fcc53d57903df529c3486962c56528cba5b7d6c57c99537320ed9922c8b86", size = 6073403, upload-time = "2026-05-29T23:11:36.22Z" },
@@ -834,37 +834,37 @@ wheels = [

 [package.optional-dependencies]
 cublas = [
-    { name = "nvidia-cublas" },
+    { name = "nvidia-cublas", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 cudart = [
-    { name = "nvidia-cuda-runtime" },
+    { name = "nvidia-cuda-runtime", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 cufft = [
-    { name = "nvidia-cufft" },
+    { name = "nvidia-cufft", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 cufile = [
-    { name = "nvidia-cufile" },
+    { name = "nvidia-cufile", marker = "sys_platform == 'linux'" },
 ]
 cupti = [
-    { name = "nvidia-cuda-cupti" },
+    { name = "nvidia-cuda-cupti", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 curand = [
-    { name = "nvidia-curand" },
+    { name = "nvidia-curand", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 cusolver = [
-    { name = "nvidia-cusolver" },
+    { name = "nvidia-cusolver", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 cusparse = [
-    { name = "nvidia-cusparse" },
+    { name = "nvidia-cusparse", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 nvjitlink = [
-    { name = "nvidia-nvjitlink" },
+    { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 nvrtc = [
-    { name = "nvidia-cuda-nvrtc" },
+    { name = "nvidia-cuda-nvrtc", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]
 nvtx = [
-    { name = "nvidia-nvtx" },
+    { name = "nvidia-nvtx", marker = "(python_full_version < '3.14' and sys_platform == 'win32') or sys_platform == 'linux'" },
 ]

 [[package]]
@@ -1023,7 +1023,7 @@ name = "exceptiongroup"
 version = "1.3.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "typing-extensions" },
+    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/50/79/66800aadf48771f6b62f7eb014e352e5d06856655206165d775e675a02c9/exceptiongroup-1.3.1.tar.gz", hash = "sha256:8b412432c6055b0b7d14c310000ae93352ed6754f70fa8f7c34141f91c4e3219", size = 30371, upload-time = "2025-11-21T23:01:54.787Z" }
 wheels = [
@@ -1440,16 +1440,16 @@ resolution-markers = [
    "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "cachetools" },
-    { name = "certifi" },
-    { name = "httpx" },
-    { name = "ibm-cos-sdk" },
-    { name = "lomond" },
-    { name = "packaging" },
-    { name = "pandas", version = "2.2.3", source = { registry = "https://pypi.org/simple" } },
-    { name = "requests" },
-    { name = "tabulate" },
-    { name = "urllib3" },
+    { name = "cachetools", marker = "python_full_version < '3.11'" },
+    { name = "certifi", marker = "python_full_version < '3.11'" },
+    { name = "httpx", marker = "python_full_version < '3.11'" },
+    { name = "ibm-cos-sdk", marker = "python_full_version < '3.11'" },
+    { name = "lomond", marker = "python_full_version < '3.11'" },
+    { name = "packaging", marker = "python_full_version < '3.11'" },
+    { name = "pandas", version = "2.2.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "requests", marker = "python_full_version < '3.11'" },
+    { name = "tabulate", marker = "python_full_version < '3.11'" },
+    { name = "urllib3", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/c7/56/2e3df38a1f13062095d7bde23c87a92f3898982993a15186b1bfecbd206f/ibm_watsonx_ai-1.3.42.tar.gz", hash = "sha256:ee5be59009004245d957ce97d1227355516df95a2640189749487614fef674ff", size = 688651, upload-time = "2025-10-01T13:35:41.527Z" }
 wheels = [
@@ -1468,17 +1468,17 @@ resolution-markers = [
    "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "cachetools" },
-    { name = "certifi" },
-    { name = "httpx" },
-    { name = "ibm-cos-sdk" },
-    { name = "lomond" },
-    { name = "packaging" },
-    { name = "pandas", version = "2.3.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.14'" },
+    { name = "cachetools", marker = "python_full_version >= '3.11'" },
+    { name = "certifi", marker = "python_full_version >= '3.11'" },
+    { name = "httpx", marker = "python_full_version >= '3.11'" },
+    { name = "ibm-cos-sdk", marker = "python_full_version >= '3.11'" },
+    { name = "lomond", marker = "python_full_version >= '3.11'" },
+    { name = "packaging", marker = "python_full_version >= '3.11'" },
+    { name = "pandas", version = "2.3.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11' and python_full_version < '3.14'" },
    { name = "pandas", version = "3.0.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
-    { name = "requests" },
-    { name = "tabulate" },
-    { name = "urllib3" },
+    { name = "requests", marker = "python_full_version >= '3.11'" },
+    { name = "tabulate", marker = "python_full_version >= '3.11'" },
+    { name = "urllib3", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/29/a3/c756b534696ab2f3f29882fdb7ca7198b7a5c94e10c0a3a327853d6d6b79/ibm_watsonx_ai-1.5.14.tar.gz", hash = "sha256:a756488bd57e87c0fc51be42dcba871143cfe0ac1e805c497c5047e1e4f13e9d", size = 735804, upload-time = "2026-06-22T12:32:43.85Z" }
 wheels = [
@@ -1554,17 +1554,17 @@ resolution-markers = [
    "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "colorama", marker = "sys_platform == 'win32'" },
-    { name = "decorator" },
-    { name = "exceptiongroup" },
-    { name = "jedi" },
-    { name = "matplotlib-inline" },
-    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit" },
-    { name = "pygments" },
-    { name = "stack-data" },
-    { name = "traitlets" },
-    { name = "typing-extensions" },
+    { name = "colorama", marker = "python_full_version < '3.11' and sys_platform == 'win32'" },
+    { name = "decorator", marker = "python_full_version < '3.11'" },
+    { name = "exceptiongroup", marker = "python_full_version < '3.11'" },
+    { name = "jedi", marker = "python_full_version < '3.11'" },
+    { name = "matplotlib-inline", marker = "python_full_version < '3.11'" },
+    { name = "pexpect", marker = "python_full_version < '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit", marker = "python_full_version < '3.11'" },
+    { name = "pygments", marker = "python_full_version < '3.11'" },
+    { name = "stack-data", marker = "python_full_version < '3.11'" },
+    { name = "traitlets", marker = "python_full_version < '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/40/18/f8598d287006885e7136451fdea0755af4ebcbfe342836f24deefaed1164/ipython-8.39.0.tar.gz", hash = "sha256:4110ae96012c379b8b6db898a07e186c40a2a1ef5d57a7fa83166047d9da7624", size = 5513971, upload-time = "2026-03-27T10:02:13.94Z" }
 wheels = [
@@ -1583,18 +1583,18 @@ resolution-markers = [
    "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "colorama", marker = "sys_platform == 'win32'" },
-    { name = "decorator" },
-    { name = "ipython-pygments-lexers" },
-    { name = "jedi" },
-    { name = "matplotlib-inline" },
-    { name = "pexpect", marker = "sys_platform != 'emscripten' and sys_platform != 'win32'" },
-    { name = "prompt-toolkit" },
-    { name = "psutil", marker = "sys_platform != 'cygwin' and sys_platform != 'emscripten'" },
-    { name = "pygments" },
-    { name = "stack-data" },
-    { name = "traitlets" },
-    { name = "typing-extensions", marker = "python_full_version < '3.12'" },
+    { name = "colorama", marker = "python_full_version >= '3.11' and sys_platform == 'win32'" },
+    { name = "decorator", marker = "python_full_version >= '3.11'" },
+    { name = "ipython-pygments-lexers", marker = "python_full_version >= '3.11'" },
+    { name = "jedi", marker = "python_full_version >= '3.11'" },
+    { name = "matplotlib-inline", marker = "python_full_version >= '3.11'" },
+    { name = "pexpect", marker = "python_full_version >= '3.11' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "prompt-toolkit", marker = "python_full_version >= '3.11'" },
+    { name = "psutil", marker = "python_full_version >= '3.11' and sys_platform != 'cygwin' and sys_platform != 'emscripten'" },
+    { name = "pygments", marker = "python_full_version >= '3.11'" },
+    { name = "stack-data", marker = "python_full_version >= '3.11'" },
+    { name = "traitlets", marker = "python_full_version >= '3.11'" },
+    { name = "typing-extensions", marker = "python_full_version == '3.11.*'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/53/59/165d3b4d75cc34add3122c4417ecb229085140ac573103c223cd01dde96f/ipython-9.15.0.tar.gz", hash = "sha256:da2819ce2aa83135257df830660b1176d986c3d2876db24df01974fa955b2756", size = 4442580, upload-time = "2026-06-26T11:03:35.913Z" }
 wheels = [
@@ -1606,7 +1606,7 @@ name = "ipython-pygments-lexers"
 version = "1.1.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "pygments" },
+    { name = "pygments", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/ef/4c/5dd1d8af08107f88c7f741ead7a40854b8ac24ddf9ae850afbcf698aa552/ipython_pygments_lexers-1.1.1.tar.gz", hash = "sha256:09c0138009e56b6854f9535736f4171d855c8c08a563a0dcd8022f78355c7e81", size = 8393, upload-time = "2025-01-17T11:24:34.505Z" }
 wheels = [
@@ -2005,7 +2005,7 @@ requires-dist = [
    { name = "pyarrow-stubs", marker = "extra == 'tests'", specifier = ">=16.0" },
    { name = "pydantic", specifier = ">=1.10" },
    { name = "pylance", marker = "extra == 'pylance'", specifier = ">=5.0.0b5" },
-    { name = "pylance", marker = "extra == 'tests'", specifier = "==10.0.0" },
+    { name = "pylance", marker = "extra == 'tests'", specifier = "==9.0.0rc1" },
    { name = "pyright", marker = "extra == 'dev'", specifier = ">=1.1.350" },
    { name = "pytest", marker = "extra == 'tests'", specifier = ">=7.0" },
    { name = "pytest-asyncio", marker = "extra == 'tests'", specifier = ">=0.21" },
@@ -2858,7 +2858,7 @@ name = "nvidia-cudnn-cu13"
 version = "9.19.0.56"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "nvidia-cublas" },
+    { name = "nvidia-cublas", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 wheels = [
    { url = "https://files.pythonhosted.org/packages/f1/84/26025437c1e6b61a707442184fa0c03d083b661adf3a3eecfd6d21677740/nvidia_cudnn_cu13-9.19.0.56-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:6ed29ffaee1176c612daf442e4dd6cfeb6a0caa43ddcbeb59da94953030b1be4", size = 433781201, upload-time = "2026-02-03T20:40:53.805Z" },
@@ -2870,7 +2870,7 @@ name = "nvidia-cufft"
 version = "12.0.0.61"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "nvidia-nvjitlink" },
+    { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 wheels = [
    { url = "https://files.pythonhosted.org/packages/8b/ae/f417a75c0259e85c1d2f83ca4e960289a5f814ed0cea74d18c353d3e989d/nvidia_cufft-12.0.0.61-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2708c852ef8cd89d1d2068bdbece0aa188813a0c934db3779b9b1faa8442e5f5", size = 214053554, upload-time = "2025-09-04T08:31:38.196Z" },
@@ -2900,9 +2900,9 @@ name = "nvidia-cusolver"
 version = "12.0.4.66"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "nvidia-cublas" },
-    { name = "nvidia-cusparse" },
-    { name = "nvidia-nvjitlink" },
+    { name = "nvidia-cublas", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
+    { name = "nvidia-cusparse", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
+    { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 wheels = [
    { url = "https://files.pythonhosted.org/packages/c8/c3/b30c9e935fc01e3da443ec0116ed1b2a009bb867f5324d3f2d7e533e776b/nvidia_cusolver-12.0.4.66-py3-none-manylinux_2_27_aarch64.whl", hash = "sha256:02c2457eaa9e39de20f880f4bd8820e6a1cfb9f9a34f820eb12a155aa5bc92d2", size = 223467760, upload-time = "2025-09-04T08:33:04.222Z" },
@@ -2914,7 +2914,7 @@ name = "nvidia-cusparse"
 version = "12.6.3.3"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "nvidia-nvjitlink" },
+    { name = "nvidia-nvjitlink", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 wheels = [
    { url = "https://files.pythonhosted.org/packages/f8/94/5c26f33738ae35276672f12615a64bd008ed5be6d1ebcb23579285d960a9/nvidia_cusparse-12.6.3.3-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:80bcc4662f23f1054ee334a15c72b8940402975e0eab63178fc7e670aa59472c", size = 162155568, upload-time = "2025-09-04T08:33:42.864Z" },
@@ -3091,10 +3091,10 @@ resolution-markers = [
    "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
-    { name = "python-dateutil" },
-    { name = "pytz" },
-    { name = "tzdata" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "python-dateutil", marker = "python_full_version < '3.11'" },
+    { name = "pytz", marker = "python_full_version < '3.11'" },
+    { name = "tzdata", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/9c/d6/9f8431bacc2e19dca897724cd097b1bb224a6ad5433784a44b587c7c13af/pandas-2.2.3.tar.gz", hash = "sha256:4f18ba62b61d7e192368b84517265a99b4d7ee8912f8708660fb4a366cc82667", size = 4399213, upload-time = "2024-09-20T13:10:04.827Z" }
 wheels = [
@@ -3143,11 +3143,11 @@ resolution-markers = [
    "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.12' or python_full_version >= '3.14'" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.11.*'" },
    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12' and python_full_version < '3.14'" },
-    { name = "python-dateutil" },
-    { name = "pytz" },
-    { name = "tzdata" },
+    { name = "python-dateutil", marker = "python_full_version >= '3.11' and python_full_version < '3.14'" },
+    { name = "pytz", marker = "python_full_version >= '3.11' and python_full_version < '3.14'" },
+    { name = "tzdata", marker = "python_full_version >= '3.11' and python_full_version < '3.14'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/33/01/d40b85317f86cf08d853a4f495195c73815fdf205eef3993821720274518/pandas-2.3.3.tar.gz", hash = "sha256:e05e1af93b977f7eafa636d043f9f94c7ee3ac81af99c13508215942e64c993b", size = 4495223, upload-time = "2025-09-29T23:34:51.853Z" }
 wheels = [
@@ -3210,9 +3210,9 @@ resolution-markers = [
    "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" } },
-    { name = "python-dateutil" },
-    { name = "tzdata", marker = "sys_platform == 'emscripten' or sys_platform == 'win32'" },
+    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.14'" },
+    { name = "python-dateutil", marker = "python_full_version >= '3.14'" },
+    { name = "tzdata", marker = "(python_full_version >= '3.14' and sys_platform == 'emscripten') or (python_full_version >= '3.14' and sys_platform == 'win32')" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/f8/87/4341c6252d1c47b08768c3d25ac487362bf403f0313ddae4a2a26c9b1b4c/pandas-3.0.3.tar.gz", hash = "sha256:696a4a00a2a2a35d4e5deb3fc946641b96c944f02230e4f76137fe35d806c4fc", size = 4651414, upload-time = "2026-05-11T18:54:29.21Z" }
 wheels = [
@@ -3320,7 +3320,7 @@ name = "pexpect"
 version = "4.9.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "ptyprocess" },
+    { name = "ptyprocess", marker = "(python_full_version < '3.14' and sys_platform == 'emscripten') or (python_full_version < '3.14' and sys_platform == 'win32') or (sys_platform != 'emscripten' and sys_platform != 'win32')" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/42/92/cc564bf6381ff43ce1f4d06852fc19a2f11d180f23dc32d9588bee2f149d/pexpect-4.9.0.tar.gz", hash = "sha256:ee7d41123f3c9911050ea2c2dac107568dc43b2d3b0c7557a33212c398ead30f", size = 166450, upload-time = "2023-11-25T09:07:26.339Z" }
 wheels = [
@@ -3912,7 +3912,7 @@ crypto = [

 [[package]]
 name = "pylance"
-version = "10.0.0"
+version = "7.0.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "lance-namespace" },
@@ -3922,12 +3922,12 @@ dependencies = [
    { name = "pyarrow" },
 ]
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/82/b2/c81de196076c4c8d768f485324a0043e113ce0950a339977d83fee9783ef/pylance-10.0.0-cp310-abi3-macosx_11_0_arm64.whl", hash = "sha256:d4bba56ae829202b7e9cdc82c92c00a4b52f03e679dc3edfad1db09d1285be2e", size = 69279797, upload-time = "2026-08-07T18:25:24.812Z" },
-    { url = "https://files.pythonhosted.org/packages/d2/e9/af671a6225740bd70c1e70bd83fa08091d628b960397cbecc477f16aaaf1/pylance-10.0.0-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:489b944827c0271e16a62b4006f8c75acc3bd7fbc381f35388a2afaaae38d438", size = 72775384, upload-time = "2026-08-07T18:31:29.686Z" },
-    { url = "https://files.pythonhosted.org/packages/5d/22/e07194195bb3bbdf062b0c31690fc92fccb2686d5e768efa1dc379a93350/pylance-10.0.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:018efe7d437d326b9049c1223458bd955e85e48509b4b0bbf8b2d3d94075dbff", size = 76632194, upload-time = "2026-08-07T18:44:29.001Z" },
-    { url = "https://files.pythonhosted.org/packages/f2/71/ed9956cf657e86a5fee5d5ddcfa7bfed0d7c2968bee39ed85d12629d8c93/pylance-10.0.0-cp310-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:254bad5d765c14db6c4eddd5323133dee3c76ff6d40c4777afe0b0e91d4c04d2", size = 72804222, upload-time = "2026-08-07T18:31:24.264Z" },
-    { url = "https://files.pythonhosted.org/packages/94/96/de449c246b2892df9d5a0af775b79a061e9897e6942bcee5c4ab6d6071f8/pylance-10.0.0-cp310-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:9f0c089e30389d9b7765a4c16f3d9325b05ed4fbf6cc2bbc9418292983472e54", size = 76602015, upload-time = "2026-08-07T18:46:27.724Z" },
-    { url = "https://files.pythonhosted.org/packages/53/81/4a5a9072b6d68c4dbb8c7b5530a381c7a6bea4ba92f9eca58ac885142722/pylance-10.0.0-cp310-abi3-win_amd64.whl", hash = "sha256:9fecf46e6835dc64b2d71767d5eab4f73787543a852432e9ba8ecf8527b3cdab", size = 82799247, upload-time = "2026-08-07T18:47:46.855Z" },
+    { url = "https://files.pythonhosted.org/packages/ac/ad/2f64921bf346e7075aef24a72595db44821724a3d89a9a92dd24e79632aa/pylance-7.0.0-cp39-abi3-macosx_11_0_arm64.whl", hash = "sha256:98422021975be76e72b1572f41b8c9abb3bee5bdc9bfa5e9ce731110a65ed4d1", size = 62134146, upload-time = "2026-05-27T21:59:37.459Z" },
+    { url = "https://files.pythonhosted.org/packages/73/1c/c5a01bee0160b55d9a98895cbd33091d038f0a0995b121ab72e629008d02/pylance-7.0.0-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4bec86ee5b6fbd8bfc493e653f0a1fba0303cfe5492b9b46fc25ab908edc7183", size = 65373684, upload-time = "2026-05-27T22:04:01.584Z" },
+    { url = "https://files.pythonhosted.org/packages/eb/da/1fe8b8f7dbfe734d76af76acc994fc360a0d0c79a4874ef69f5a72a58fe3/pylance-7.0.0-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:881491432c53184e52f8d1db8d5f872f39a03f36fb104bec77b33d379519d8b5", size = 69458555, upload-time = "2026-05-27T22:16:50.567Z" },
+    { url = "https://files.pythonhosted.org/packages/76/f0/dd505cf3fd0226ab9d94759acd713125af1d3bfacfd80bbd52e3b9f89509/pylance-7.0.0-cp39-abi3-manylinux_2_28_aarch64.whl", hash = "sha256:18453999e7fff4f76b16d6b7882c9df0628bd142ff95e2461bd7dd5ee3fe0af3", size = 65394430, upload-time = "2026-05-27T22:05:30.923Z" },
+    { url = "https://files.pythonhosted.org/packages/17/ba/2357b81034f28eb00790e258ed140289a6a887a7468ca9df6349fd186b27/pylance-7.0.0-cp39-abi3-manylinux_2_28_x86_64.whl", hash = "sha256:04a58051d408c60fe76d41a220dcaf8fea8fb6d1aa0ca78a709b60bc3cc8d19a", size = 69473470, upload-time = "2026-05-27T22:17:18.935Z" },
+    { url = "https://files.pythonhosted.org/packages/1f/ec/5c00b6303a67d787f9475141832cbdc513d674ac3dcaeef8a7b169905e65/pylance-7.0.0-cp39-abi3-win_amd64.whl", hash = "sha256:467d4864af047eaab4e1370e2f1e88e2c6f507c079874421116cb41d78bc3629", size = 74792863, upload-time = "2026-05-27T22:19:23.875Z" },
 ]

 [[package]]
@@ -4683,10 +4683,10 @@ resolution-markers = [
    "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "joblib" },
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
-    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" } },
-    { name = "threadpoolctl" },
+    { name = "joblib", marker = "python_full_version < '3.11'" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "scipy", version = "1.15.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "threadpoolctl", marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/98/c2/a7855e41c9d285dfe86dc50b250978105dce513d6e459ea66a6aeb0e1e0c/scikit_learn-1.7.2.tar.gz", hash = "sha256:20e9e49ecd130598f1ca38a1d85090e1a600147b9c02fa6f15d69cb53d968fda", size = 7193136, upload-time = "2025-09-09T08:21:29.075Z" }
 wheels = [
@@ -4734,13 +4734,13 @@ resolution-markers = [
    "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "joblib" },
-    { name = "narwhals" },
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.12'" },
+    { name = "joblib", marker = "python_full_version >= '3.11'" },
+    { name = "narwhals", marker = "python_full_version >= '3.11'" },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.11.*'" },
    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12'" },
-    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.12'" },
+    { name = "scipy", version = "1.17.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.11.*'" },
    { name = "scipy", version = "1.18.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12'" },
-    { name = "threadpoolctl" },
+    { name = "threadpoolctl", marker = "python_full_version >= '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/fa/6f/37092bdb25f712817231799fc5674d8e704066a8a70c1d2d40517e18b4ab/scikit_learn-1.9.0.tar.gz", hash = "sha256:8833266989d3a5110178a9fae30783675460724d0e1efb13b14901d2c660c557", size = 7750767, upload-time = "2026-06-02T11:54:32.706Z" }
 wheels = [
@@ -4784,7 +4784,7 @@ resolution-markers = [
    "python_full_version < '3.11'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/0f/37/6964b830433e654ec7485e45a00fc9a27cf868d622838f6b6d9c5ec0d532/scipy-1.15.3.tar.gz", hash = "sha256:eae3cf522bc7df64b42cad3925c876e1b0b6c35c1337c93e12c0f366f55b0eaf", size = 59419214, upload-time = "2025-05-08T16:13:05.955Z" }
 wheels = [
@@ -4843,7 +4843,7 @@ resolution-markers = [
    "python_full_version == '3.11.*'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" } },
+    { name = "numpy", version = "2.4.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.11.*'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/7a/97/5a3609c4f8d58b039179648e62dd220f89864f56f7357f5d4f45c29eb2cc/scipy-1.17.1.tar.gz", hash = "sha256:95d8e012d8cb8816c226aef832200b1d45109ed4464303e997c5b13122b297c0", size = 30573822, upload-time = "2026-02-23T00:26:24.851Z" }
 wheels = [
@@ -4920,7 +4920,7 @@ resolution-markers = [
    "python_full_version >= '3.12' and python_full_version < '3.14'",
 ]
 dependencies = [
-    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" } },
+    { name = "numpy", version = "2.5.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.12'" },
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/a7/25/c2700dfaf6442b4effaa91af24ebce5dc9d31bb4a69706313aae70d72cd0/scipy-1.18.0.tar.gz", hash = "sha256:67b2ad2ad54c72ca6d04975a9b2df8c3638c34ddd5b28738e94fc2b57929d378", size = 30774447, upload-time = "2026-06-19T15:01:43.456Z" }
 wheels = [
@@ -1,6 +1,6 @@
 [package]
 name = "lancedb"
-version = "0.37.1"
+version = "0.37.1-beta.1"
 edition.workspace = true
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license.workspace = true
@@ -17,7 +17,7 @@ use arrow_array::builder::LargeBinaryBuilder;
 use arrow_schema::{DataType, Field, Schema};
 use lance::dataset::{BlobRangeRequest as LanceBlobRangeRequest, Dataset, WriteParams};
 use lance_arrow::FieldExt;
-use lance_file::version::LanceFileVersion;
+use lance_file::version::{ConcreteFileVersion, LanceFileVersion};
 use lance_io::object_store::ObjectStore;
 use object_store::path::Path;

@@ -333,7 +333,10 @@ pub(crate) fn ensure_blob_storage_version(schema: &Schema, params: &mut WritePar
        .data_storage_version
        .unwrap_or(LanceFileVersion::Stable)
        .resolve();
-    if resolved < LanceFileVersion::V2_2 {
+    if matches!(
+        resolved,
+        ConcreteFileVersion::V1 | ConcreteFileVersion::V2_0 | ConcreteFileVersion::V2_1
+    ) {
        params.data_storage_version = Some(LanceFileVersion::V2_2);
    }
 }
@@ -499,7 +502,7 @@ mod tests {
        ensure_blob_storage_version(&blob_schema(), &mut params);
        assert_eq!(
            params.data_storage_version.unwrap().resolve(),
-            LanceFileVersion::V2_2
+            ConcreteFileVersion::V2_2
        );
    }

@@ -512,7 +515,7 @@ mod tests {
        ensure_blob_storage_version(&blob_schema(), &mut params);
        assert_eq!(
            params.data_storage_version.unwrap().resolve(),
-            LanceFileVersion::V2_2
+            ConcreteFileVersion::V2_2
        );
    }

@@ -438,10 +438,9 @@ mod tests {
            .await
            .unwrap()
            .data_storage_format
-            .lance_file_version()
-            .unwrap();
+            .lance_file_format();
        // Compare resolved versions since Stable/Next are aliases that resolve at storage time
-        assert_eq!(storage_format.resolve(), data_storage_version.resolve());
+        assert_eq!(storage_format, data_storage_version.resolve());
    }

    #[tokio::test]
@@ -12,7 +12,8 @@ use lance_encoding::decoder::{DecoderPlugins, FilterExpression};
 use lance_file::{
    reader::{FileReader, FileReaderOptions},
    version::ConcreteFileVersion,
-    writer::{FileWriter, FileWriterOptions},
+    versions,
+    writer::FileWriterOptions,
 };
 use lance_io::{
    ReadBatchParams,
@@ -153,13 +154,11 @@ impl Shuffler {
                    source: None,
                })?;
            let object_writer = object_store.create(&path).await?;
-            let writer = FileWriter::try_new(
+            let writer = versions::create_writer(
+                ConcreteFileVersion::V2_1,
                object_writer,
                schema.clone(),
-                FileWriterOptions {
-                    format_version: Some(ConcreteFileVersion::V2_1.into()),
-                    ..Default::default()
-                },
+                FileWriterOptions::default(),
            )?;
            file_writers.push(writer);
        }
@@ -71,6 +71,14 @@ pub enum Error {
    IndexNotFound { name: String },
    #[snafu(display("Embedding function '{name}' was not found. : {reason}"))]
    EmbeddingFunctionNotFound { name: String, reason: String },
+    #[snafu(display("Column '{name}' was not found"))]
+    ColumnNotFound { name: String },
+    #[snafu(display("Column '{name}' already exists"))]
+    ColumnAlreadyExists { name: String },
+    #[snafu(display("Column '{name}' is not a computed column"))]
+    NotAComputedColumn { name: String },
+    #[snafu(display("Invalid expression for column '{column}': {message}"))]
+    InvalidExpression { column: String, message: String },

    #[snafu(display("Table '{name}' already exists"))]
    TableAlreadyExists { name: String },
@@ -2520,9 +2520,9 @@ impl<S: HttpSend> BaseTable for RemoteTable<S> {
        self.check_mutable().await?;

        // Map the spec onto the server's request DTO. `sharding` is internally
-        // tagged on `mode` to mirror sophon's `Sharding` enum; `maintained_indexes`
-        // and `writer_config_defaults` are sent verbatim (an empty list means "no
-        // maintained indexes", not "default to all").
+        // tagged on `mode` to mirror sophon's `Sharding` enum. A null
+        // `maintained_indexes` asks the server to resolve every maintainable
+        // index at HEAD; a list is verbatim, an empty one meaning none.
        let sharding = match &spec {
            LsmWriteSpec::Bucket {
                column,
@@ -2706,6 +2706,13 @@ impl<S: HttpSend> BaseTable for RemoteTable<S> {

                Ok(result)
            }
+            // A declaration reaches here as AllNulls, which the remote protocol
+            // has no representation for.
+            NewColumnTransform::AllNulls(_) => {
+                return Err(Error::NotSupported {
+                    message: "computed columns are supported only on local tables".into(),
+                });
+            }
            _ => {
                return Err(Error::NotSupported {
                    message: "Only SQL expressions are supported for adding columns".into(),
@@ -5955,17 +5962,18 @@ mod tests {
            .await
            .unwrap();

-        // Lance 10 retains original token positions after stop-word removal.
+        // Positions are relative to the first retained token, so dropping the
+        // leading "hello" stop word does not shift the remaining tokens.
        assert_eq!(
            tokens,
            vec![
                FtsToken {
                    text: "こんにちは".to_string(),
-                    position: 1,
+                    position: 0,
                },
                FtsToken {
                    text: "世界".to_string(),
-                    position: 2,
+                    position: 1,
                },
            ]
        );
@@ -6454,6 +6462,37 @@ mod tests {
        assert_eq!(result.version, if old_server { 0 } else { 43 });
    }

+    /// Computed columns are local-only. Both halves say so here rather than
+    /// reaching the wire and failing somewhere less legible.
+    #[tokio::test]
+    async fn test_computed_columns_are_refused() {
+        let table = Table::new_with_handler("my_table", |request| -> http::Response<String> {
+            panic!("unexpected request: {}", request.url().path())
+        });
+
+        let declared = Arc::new(Schema::new(vec![Field::new(
+            "doubled",
+            DataType::Int32,
+            true,
+        )]));
+        let err = table
+            .add_columns()
+            .transform(NewColumnTransform::AllNulls(declared))
+            .execute()
+            .await
+            .unwrap_err();
+        assert!(
+            matches!(&err, Error::NotSupported { message } if message.contains("local tables")),
+            "{err:?}"
+        );
+
+        let err = table.refresh_column("doubled").await.unwrap_err();
+        assert!(
+            matches!(&err, Error::NotSupported { message } if message.contains("local tables")),
+            "{err:?}"
+        );
+    }
+
    #[tokio::test]
    async fn test_prewarm_index() {
        let table = Table::new_with_handler("my_table", |request| {
@@ -6598,7 +6637,7 @@ mod tests {
                .unwrap()
        });
        let spec = crate::table::LsmWriteSpec::unsharded()
-            .with_maintained_indexes(["id_idx"])
+            .with_maintained_indexes(vec!["id_idx".to_string()])
            .with_writer_config_defaults([("max_memtable_rows", "1000")]);
        table.set_lsm_write_spec(spec).await.unwrap();
    }
@@ -6617,7 +6656,8 @@ mod tests {
                body["sharding"],
                serde_json::json!({ "mode": "bucket", "column": "id", "num_buckets": 16 })
            );
-            assert_eq!(body["maintained_indexes"], serde_json::json!([]));
+            // An unpinned maintained set sends null: resolve server-side.
+            assert_eq!(body["maintained_indexes"], serde_json::Value::Null);
            http::Response::builder().status(200).body("{}").unwrap()
        });
        table
@@ -6626,6 +6666,23 @@ mod tests {
            .unwrap();
    }

+    /// `[]` (none) must stay distinguishable on the wire from null (all).
+    #[tokio::test]
+    async fn test_set_lsm_write_spec_no_maintained_indexes() {
+        let table = Table::new_with_handler("my_table", |request| {
+            let body = request.body().unwrap().as_bytes().unwrap();
+            let body: serde_json::Value = serde_json::from_slice(body).unwrap();
+            assert_eq!(body["maintained_indexes"], serde_json::json!([]));
+            http::Response::builder().status(200).body("{}").unwrap()
+        });
+        table
+            .set_lsm_write_spec(
+                crate::table::LsmWriteSpec::bucket("id", 16).with_maintained_indexes(Vec::new()),
+            )
+            .await
+            .unwrap();
+    }
+
    #[tokio::test]
    async fn test_set_lsm_write_spec_identity() {
        let table = Table::new_with_handler("my_table", |request| {
@@ -6700,7 +6757,7 @@ mod tests {
            } => {
                assert_eq!(column, "id");
                assert_eq!(num_buckets, 4);
-                assert_eq!(maintained_indexes, vec!["id_idx".to_string()]);
+                assert_eq!(maintained_indexes, Some(vec!["id_idx".to_string()]));
                assert_eq!(
                    writer_config_defaults
                        .get("durable_write")
@@ -69,6 +69,7 @@ pub mod add_columns;
 mod add_data;
 pub mod branch_merge;
 pub mod checkpoint;
+pub mod computed_columns;
 mod create_index;
 pub mod datafusion;
 pub(crate) mod dataset;
@@ -78,6 +79,7 @@ pub mod merge;
 pub mod optimize;
 mod primary_key;
 pub mod query;
+pub mod refresh;
 pub mod schema_evolution;
 pub mod update;
 pub mod write_progress;
@@ -91,6 +93,9 @@ pub use branch_merge::{
    MergeBranchResult, MergeBranchStatus, MergePreview, RowCountSummary,
 };
 pub use chrono::Duration;
+pub use computed_columns::{
+    ComputedColumn, ComputedColumnKind, computed_column_from_field, computed_columns,
+};
 pub use delete::DeleteResult;
 use futures::future::join_all;
 pub use lance::dataset::refs::{BranchContents, Ref, TagContents, Tags as LanceTags};
@@ -98,6 +103,7 @@ pub use lance::dataset::scanner::DatasetRecordBatchStream;
 pub use lance_index::optimize::OptimizeOptions;
 pub use lsm_stats::{BucketStats, GenerationStats, LsmStats, MemtableStats};
 pub use optimize::{CompactionOptions, OptimizeAction, OptimizeStats};
+pub use refresh::RefreshColumnResult;
 pub use schema_evolution::{
    AddColumnsResult, AlterColumnsResult, DropColumnsResult, FieldMetadataUpdate,
    UpdateFieldMetadataResult,
@@ -370,6 +376,8 @@ pub use self::merge::MergeResult;
 /// date) and [`LsmWriteSpec::with_writer_config_defaults`] (default
 /// `ShardWriter` configuration recorded in the MemWAL index).
 ///
+/// A fresh spec maintains every index on the table, resolved on install.
+///
 /// Install a spec with [`Table::set_lsm_write_spec`] and remove it with
 /// [`Table::unset_lsm_write_spec`]. The actual `merge_insert` dispatch
 /// onto the MemWAL writer is a follow-up.
@@ -384,9 +392,12 @@ pub enum LsmWriteSpec {
    Bucket {
        column: String,
        num_buckets: u32,
-        /// Names of indexes (already created on the table) that the
-        /// MemWAL should maintain in-memory as rows are appended.
-        maintained_indexes: Vec<String>,
+        /// Indexes the MemWAL maintains in-memory as rows are appended.
+        ///
+        /// `None` means every index it can maintain, resolved on install — a
+        /// snapshot, so indexes created later need the spec unset and re-set.
+        /// `Some([])` maintains nothing.
+        maintained_indexes: Option<Vec<String>>,
        /// Default `ShardWriter` configuration recorded in the MemWAL index.
        writer_config_defaults: HashMap<String, String>,
    },
@@ -396,35 +407,41 @@ pub enum LsmWriteSpec {
    /// distinct value of `column` becomes its own shard.
    Identity {
        column: String,
-        /// Names of indexes (already created on the table) that the
-        /// MemWAL should maintain in-memory as rows are appended.
-        maintained_indexes: Vec<String>,
+        /// Indexes the MemWAL maintains in-memory as rows are appended.
+        ///
+        /// `None` means every index it can maintain, resolved on install — a
+        /// snapshot, so indexes created later need the spec unset and re-set.
+        /// `Some([])` maintains nothing.
+        maintained_indexes: Option<Vec<String>>,
        /// Default `ShardWriter` configuration recorded in the MemWAL index.
        writer_config_defaults: HashMap<String, String>,
    },
    /// No sharding — every `merge_insert` call writes to a single MemWAL shard.
    Unsharded {
-        /// Names of indexes (already created on the table) that the
-        /// MemWAL should maintain in-memory as rows are appended.
-        maintained_indexes: Vec<String>,
+        /// Indexes the MemWAL maintains in-memory as rows are appended.
+        ///
+        /// `None` means every index it can maintain, resolved on install — a
+        /// snapshot, so indexes created later need the spec unset and re-set.
+        /// `Some([])` maintains nothing.
+        maintained_indexes: Option<Vec<String>>,
        /// Default `ShardWriter` configuration recorded in the MemWAL index.
        writer_config_defaults: HashMap<String, String>,
    },
 }

 impl LsmWriteSpec {
-    /// Construct a hash-bucket sharding spec with no maintained indexes.
+    /// Construct a hash-bucket sharding spec maintaining every index on the table.
    pub fn bucket(column: impl Into<String>, num_buckets: u32) -> Self {
        Self::Bucket {
            column: column.into(),
            num_buckets,
-            maintained_indexes: Vec::new(),
+            maintained_indexes: None,
            writer_config_defaults: HashMap::new(),
        }
    }

    /// Construct an identity-sharding spec (shard by the raw value of
-    /// `column`) with no maintained indexes.
+    /// `column`) maintaining every index on the table.
    ///
    /// `column` must be a deterministic function of the unenforced primary
    /// key: every row with a given primary key must always produce the same
@@ -436,28 +453,37 @@ impl LsmWriteSpec {
    pub fn identity(column: impl Into<String>) -> Self {
        Self::Identity {
            column: column.into(),
-            maintained_indexes: Vec::new(),
+            maintained_indexes: None,
            writer_config_defaults: HashMap::new(),
        }
    }

-    /// Construct an unsharded spec with no maintained indexes.
+    /// Construct an unsharded spec maintaining every index on the table.
    pub fn unsharded() -> Self {
        Self::Unsharded {
-            maintained_indexes: Vec::new(),
+            maintained_indexes: None,
            writer_config_defaults: HashMap::new(),
        }
    }

-    /// Replace the list of indexes the MemWAL should keep up to date as
-    /// rows are appended. Each name must reference an index that already
-    /// exists on the table at the time `set_lsm_write_spec` is called.
-    pub fn with_maintained_indexes<I, S>(mut self, indexes: I) -> Self
-    where
-        I: IntoIterator<Item = S>,
-        S: Into<String>,
-    {
-        let v: Vec<String> = indexes.into_iter().map(Into::into).collect();
+    /// Set which indexes the MemWAL maintains.
+    ///
+    /// `None` (the default) resolves to every index on the table at install,
+    /// failing if one cannot be maintained — name the set to install anyway. A
+    /// list is verbatim: each name must already exist and be maintainable, and
+    /// an empty list maintains nothing.
+    ///
+    /// ```
+    /// # use lancedb::table::LsmWriteSpec;
+    /// // Every index the table has when the spec is installed:
+    /// LsmWriteSpec::unsharded().with_maintained_indexes(None);
+    /// // Exactly these:
+    /// LsmWriteSpec::unsharded().with_maintained_indexes(vec!["id_idx".to_string()]);
+    /// // None at all:
+    /// LsmWriteSpec::unsharded().with_maintained_indexes(Vec::new());
+    /// ```
+    pub fn with_maintained_indexes(mut self, indexes: impl Into<Option<Vec<String>>>) -> Self {
+        let indexes = indexes.into();
        match &mut self {
            Self::Bucket {
                maintained_indexes, ..
@@ -467,7 +493,7 @@ impl LsmWriteSpec {
            }
            | Self::Unsharded {
                maintained_indexes, ..
-            } => *maintained_indexes = v,
+            } => *maintained_indexes = indexes,
        }
        self
    }
@@ -503,8 +529,9 @@ impl LsmWriteSpec {
        self
    }

-    /// Borrow the list of index names this spec asks MemWAL to maintain.
-    pub fn maintained_indexes(&self) -> &[String] {
+    /// Borrow the list of index names this spec asks MemWAL to maintain, or
+    /// `None` when it asks for every index on the table.
+    pub fn maintained_indexes(&self) -> Option<&[String]> {
        match self {
            Self::Bucket {
                maintained_indexes, ..
@@ -514,7 +541,7 @@ impl LsmWriteSpec {
            }
            | Self::Unsharded {
                maintained_indexes, ..
-            } => maintained_indexes,
+            } => maintained_indexes.as_deref(),
        }
    }

@@ -761,6 +788,14 @@ pub trait BaseTable: std::fmt::Display + std::fmt::Debug + Send + Sync {
        transforms: NewColumnTransform,
        read_columns: Option<Vec<String>>,
    ) -> Result<AddColumnsResult>;
+    /// Fill a computed column's unfilled rows.
+    ///
+    /// The default returns `NotSupported`; Lance-backed tables override it.
+    async fn refresh_column(&self, _column: &str) -> Result<RefreshColumnResult> {
+        Err(Error::NotSupported {
+            message: "computed columns are supported only on local tables".into(),
+        })
+    }
    /// Alter columns in the table.
    async fn alter_columns(&self, alterations: &[ColumnAlteration]) -> Result<AlterColumnsResult>;
    /// Drop columns from the table.
@@ -1653,6 +1688,29 @@ impl Table {
        AddColumnsBuilder::new(self.inner.clone())
    }

+    /// Fill the fragments of a computed column that hold no values yet.
+    ///
+    /// Declared with
+    /// [`AddColumnsBuilder::computed`](add_columns::AddColumnsBuilder::computed),
+    /// a column starts empty and gets its values here. Fragments appended
+    /// since the last refresh are filled by the next one; fragments already
+    /// filled are left as they are, so the call is idempotent and does not
+    /// observe a mutated input.
+    ///
+    /// Local tables only.
+    ///
+    /// ```
+    /// # use lancedb::Table;
+    /// # async fn refresh(table: &Table) -> Result<(), Box<dyn std::error::Error>> {
+    /// let result = table.refresh_column("doubled").await?;
+    /// println!("filled {} rows at version {}", result.rows_filled, result.version);
+    /// # Ok(())
+    /// # }
+    /// ```
+    pub async fn refresh_column(&self, column: impl AsRef<str>) -> Result<RefreshColumnResult> {
+        self.inner.refresh_column(column.as_ref()).await
+    }
+
    /// Change a column's name or nullability.
    pub async fn alter_columns(
        &self,
@@ -1712,7 +1770,7 @@ impl Table {
    /// # async fn example(table: &Table) -> Result<(), Box<dyn std::error::Error>> {
    /// table
    ///     .set_lsm_write_spec(
-    ///         LsmWriteSpec::bucket("id", 16).with_maintained_indexes(["id_idx"]),
+    ///         LsmWriteSpec::bucket("id", 16).with_maintained_indexes(vec!["id_idx".to_string()]),
    ///     )
    ///     .await?;
    /// # Ok(())
@@ -1734,9 +1792,10 @@ impl Table {
    ///
    /// Returns `Ok(None)` when the MemWAL LSM write path is not enabled (no
    /// spec has been set, or it was removed with [`Table::unset_lsm_write_spec`]).
-    /// The returned spec — including its [`LsmWriteSpec::maintained_indexes`] and
-    /// [`LsmWriteSpec::writer_config_defaults`] — mirrors what was passed to
-    /// [`Table::set_lsm_write_spec`].
+    /// The returned spec mirrors what was passed to
+    /// [`Table::set_lsm_write_spec`], except that
+    /// [`LsmWriteSpec::maintained_indexes`] always reports the concrete list
+    /// resolved when the spec was set — `None` never round-trips.
    ///
    /// # Example
    ///
@@ -3319,6 +3378,12 @@ impl BaseTable for NativeTable {
        Ok(result)
    }

+    async fn refresh_column(&self, column: &str) -> Result<RefreshColumnResult> {
+        let result = refresh::execute_refresh_column(self, column).await?;
+        self.bump_freshness();
+        Ok(result)
+    }
+
    async fn alter_columns(&self, alterations: &[ColumnAlteration]) -> Result<AlterColumnsResult> {
        let result = schema_evolution::execute_alter_columns(self, alterations).await?;
        self.bump_freshness();
@@ -5085,7 +5150,7 @@ mod tests {
        // Bucket spec round-trips exactly, including the routing column (recovered
        // from its field id), maintained indexes, and writer config defaults.
        let spec = LsmWriteSpec::bucket("id", 4)
-            .with_maintained_indexes([idx_name])
+            .with_maintained_indexes(vec![idx_name.clone()])
            .with_writer_config_defaults([("durable_write", "false")]);
        table.set_lsm_write_spec(spec.clone()).await.unwrap();
        assert_eq!(table.get_lsm_write_spec().await.unwrap(), Some(spec));
@@ -5095,15 +5160,125 @@ mod tests {
        assert_eq!(table.get_lsm_write_spec().await.unwrap(), None);

        // Identity sharding round-trips (column recovered from the schema).
+        // A spec left at its default maintains every index on the table, so it
+        // reads back naming the one on the table rather than as "infer".
        let spec = LsmWriteSpec::identity("region");
        table.set_lsm_write_spec(spec.clone()).await.unwrap();
-        assert_eq!(table.get_lsm_write_spec().await.unwrap(), Some(spec));
+        assert_eq!(
+            table.get_lsm_write_spec().await.unwrap(),
+            Some(spec.with_maintained_indexes(vec![idx_name.clone()]))
+        );
        table.unset_lsm_write_spec().await.unwrap();

        // Unsharded round-trips (no routing column).
        let spec = LsmWriteSpec::unsharded();
        table.set_lsm_write_spec(spec.clone()).await.unwrap();
-        assert_eq!(table.get_lsm_write_spec().await.unwrap(), Some(spec));
+        assert_eq!(
+            table.get_lsm_write_spec().await.unwrap(),
+            Some(spec.with_maintained_indexes(vec![idx_name]))
+        );
+    }
+
+    /// The maintained set defaults to every index on the table, resolved at
+    /// install. An index the memtable cannot build fails the install rather
+    /// than being dropped: maintaining it would take the table offline for
+    /// writes, dropping it would hide that from the caller.
+    #[tokio::test]
+    async fn test_set_lsm_write_spec_infers_maintained_indexes() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = tmp_dir.path().to_str().unwrap();
+
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("id", DataType::Int64, false),
+            Field::new("tag", DataType::Utf8, true),
+        ]));
+        let batch = RecordBatch::try_new(
+            schema.clone(),
+            vec![
+                Arc::new(arrow_array::Int64Array::from(vec![1, 2, 3])),
+                Arc::new(StringArray::from(vec!["a", "b", "c"])),
+            ],
+        )
+        .unwrap();
+        let reader: Box<dyn arrow_array::RecordBatchReader + Send> =
+            Box::new(RecordBatchIterator::new(vec![Ok(batch)], schema.clone()));
+        let conn = ConnectBuilder::new(uri)
+            .read_consistency_interval(Duration::from_secs(0))
+            .execute()
+            .await
+            .unwrap();
+        let table = conn.create_table("t", reader).execute().await.unwrap();
+
+        table
+            .create_index(&["id"], Index::BTree(Default::default()))
+            .name("id_btree".to_string())
+            .execute()
+            .await
+            .unwrap();
+        table
+            .create_index(&["tag"], Index::Bitmap(Default::default()))
+            .name("tag_bitmap".to_string())
+            .execute()
+            .await
+            .unwrap();
+
+        // Explicitly naming the bitmap index fails before anything commits.
+        let err = table
+            .set_lsm_write_spec(
+                LsmWriteSpec::unsharded().with_maintained_indexes(vec!["tag_bitmap".to_string()]),
+            )
+            .await
+            .unwrap_err();
+        assert!(
+            matches!(err, Error::InvalidInput { ref message } if message.contains("tag_bitmap")),
+            "expected the bitmap index to be rejected, got {err:?}"
+        );
+        assert_eq!(table.get_lsm_write_spec().await.unwrap(), None);
+
+        // The default covers every index, so the bitmap fails it too.
+        let err = table
+            .set_lsm_write_spec(LsmWriteSpec::unsharded())
+            .await
+            .unwrap_err();
+        assert!(
+            matches!(err, Error::InvalidInput { ref message }
+                if message.contains("tag_bitmap") && message.contains("maintained_indexes")),
+            "expected the inferred set to be rejected, got {err:?}"
+        );
+        assert_eq!(table.get_lsm_write_spec().await.unwrap(), None);
+
+        // Naming the maintainable subset installs.
+        table
+            .set_lsm_write_spec(
+                LsmWriteSpec::unsharded().with_maintained_indexes(vec!["id_btree".to_string()]),
+            )
+            .await
+            .unwrap();
+        assert_eq!(
+            table
+                .get_lsm_write_spec()
+                .await
+                .unwrap()
+                .unwrap()
+                .maintained_indexes(),
+            Some(["id_btree".to_string()].as_slice())
+        );
+
+        // Opting out entirely is distinct from the default.
+        table.unset_lsm_write_spec().await.unwrap();
+        table
+            .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes(Vec::new()))
+            .await
+            .unwrap();
+        assert_eq!(
+            table
+                .get_lsm_write_spec()
+                .await
+                .unwrap()
+                .unwrap()
+                .maintained_indexes(),
+            Some([].as_slice())
+        );
    }

    #[tokio::test]
@@ -5207,8 +5382,8 @@ mod tests {
    pub async fn test_stats_includes_index_and_overlay_files() {
        use lance::dataset::WriteDestination;
        use lance::dataset::transaction::{DataOverlayGroup, Operation};
-        use lance_file::version::{ConcreteFileVersion, LanceFileVersion};
-        use lance_file::writer::{FileWriter, FileWriterOptions};
+        use lance_file::version::stable_file_version;
+        use lance_file::writer::FileWriterOptions;
        use lance_io::utils::CachedFileSize;
        use lance_table::format::DataFile;
        use lance_table::format::overlay::{DataOverlayFile, OverlayCoverage};
@@ -5273,19 +5448,17 @@ mod tests {
        let fragment_id = dataset.get_fragments()[0].id() as u64;
        let foo_field_id = dataset.schema().field("foo").unwrap().id;
        let overlay_schema = dataset.schema().project_by_ids(&[foo_field_id], true);
-        let file_version = ConcreteFileVersion::from(LanceFileVersion::Stable);
+        let file_version = stable_file_version();

        let filename = "overlay.lance".to_string();
        let store = dataset.object_store(None).await.unwrap();
        let path = dataset.data_dir().child(filename.clone());
        let obj_writer = store.create(&path).await.unwrap();
-        let mut writer = FileWriter::try_new(
+        let mut writer = lance_file::versions::create_writer(
+            file_version,
            obj_writer,
            overlay_schema,
-            FileWriterOptions {
-                format_version: Some(file_version.into()),
-                ..Default::default()
-            },
+            FileWriterOptions::default(),
        )
        .unwrap();
        writer
@@ -8,6 +8,7 @@ use std::sync::Arc;
 use lance::dataset::NewColumnTransform;

 use super::BaseTable;
+use super::computed_columns;
 use super::schema_evolution::AddColumnsResult;
 use crate::{Error, Result};

@@ -15,6 +16,7 @@ use crate::{Error, Result};
 pub struct AddColumnsBuilder {
    parent: Arc<dyn BaseTable>,
    transform: Option<NewColumnTransform>,
+    computed: Vec<(String, String)>,
    read_columns: Option<Vec<String>>,
 }

@@ -23,6 +25,7 @@ impl std::fmt::Debug for AddColumnsBuilder {
        f.debug_struct("AddColumnsBuilder")
            .field("parent", &self.parent)
            .field("has_transform", &self.transform.is_some())
+            .field("computed", &self.computed)
            .field("read_columns", &self.read_columns)
            .finish()
    }
@@ -33,19 +36,57 @@ impl AddColumnsBuilder {
        Self {
            parent,
            transform: None,
+            computed: Vec::new(),
            read_columns: None,
        }
    }

-    /// Set how the new columns' values are produced. Required.
+    /// Set how the new columns' values are produced.
    pub fn transform(mut self, transform: NewColumnTransform) -> Self {
        self.transform = Some(transform);
        self
    }

+    /// Add a column defined by `expression`, evaluated by a later refresh
+    /// rather than by this commit. Its type and inputs are derived from the
+    /// expression.
+    ///
+    /// The column is committed with no values, so declaring one costs the same
+    /// on an empty table as on a large one. Rows get values from
+    /// [`Table::refresh_column`](super::Table::refresh_column), which fills
+    /// every fragment that has none -- including fragments appended since the
+    /// last refresh.
+    ///
+    /// Refresh does not revisit a fragment it has filled, so mutating an input
+    /// leaves the value computed at fill time; recomputing means dropping the
+    /// column and declaring it again. An input cannot be renamed, retyped or
+    /// dropped while a declaration reads it, since the expression names it.
+    ///
+    /// Local tables only: LanceDB Cloud and Enterprise reject a declaration
+    /// with `NotSupported`.
+    ///
+    /// ```
+    /// # use lancedb::Table;
+    /// # async fn declare(table: &Table) -> Result<(), Box<dyn std::error::Error>> {
+    /// table
+    ///     .add_columns()
+    ///     .computed("doubled", "x * 2")
+    ///     .execute()
+    ///     .await?;
+    /// let filled = table.refresh_column("doubled").await?;
+    /// println!("filled {} rows", filled.rows_filled);
+    /// # Ok(())
+    /// # }
+    /// ```
+    pub fn computed(mut self, name: impl Into<String>, expression: impl Into<String>) -> Self {
+        self.computed.push((name.into(), expression.into()));
+        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.
+    /// receives. Every other transform, and a computed column, 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
@@ -56,24 +97,43 @@ impl AddColumnsBuilder {
        let Self {
            parent,
            transform,
+            computed,
            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"
+        match (transform, computed.is_empty()) {
+            (None, true) => Err(Error::InvalidInput {
+                message: "add_columns requires a transform or a computed column".into(),
+            }),
+            // The two commit through different transforms, so one call covering
+            // both would be two commits and could half-apply.
+            (Some(_), false) => Err(Error::InvalidInput {
+                message: "add_columns cannot mix a transform with computed columns; \
+                          they cannot be added atomically in one call"
                    .into(),
-            });
+            }),
+            (Some(transform), true) => {
+                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
+            }
+            (None, false) => {
+                if read_columns.is_some() {
+                    return Err(Error::InvalidInput {
+                        message: "read_columns applies only to a BatchUDF transform; \
+                                  a computed column's inputs come from its expression"
+                            .into(),
+                    });
+                }
+                let transform = computed_columns::declare(parent.schema().await?, &computed)?;
+                parent.add_columns(transform, None).await
+            }
        }
-
-        parent.add_columns(transform, read_columns).await
    }
 }

@@ -85,8 +145,8 @@ mod tests {
    use arrow_schema::{DataType, Field, Schema};
    use lance::dataset::{BatchUDF, NewColumnTransform};

-    use crate::Table;
    use crate::connect;
+    use crate::{Error, Table};

    async fn table_with_two_columns(name: &str) -> Table {
        let conn = connect("memory://").execute().await.unwrap();
@@ -98,10 +158,7 @@ mod tests {
    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}"
-        );
+        assert!(matches!(err, Error::InvalidInput { .. }));
    }

    #[tokio::test]
@@ -117,7 +174,7 @@ mod tests {
            .execute()
            .await
            .unwrap_err();
-        assert!(err.to_string().contains("BatchUDF"), "got: {err}");
+        assert!(matches!(err, Error::InvalidInput { .. }));

        let schema = table.schema().await.unwrap();
        assert!(
@@ -126,6 +183,47 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn test_mixing_transform_and_computed_is_rejected() {
+        let table = table_with_two_columns("mixed_add").await;
+        let err = table
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "eager".into(),
+                "x * 2".into(),
+            )]))
+            .computed("lazy", "x * 3")
+            .execute()
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::InvalidInput { .. }));
+
+        let schema = table.schema().await.unwrap();
+        assert!(schema.field_with_name("eager").is_err());
+        assert!(schema.field_with_name("lazy").is_err());
+    }
+
+    #[tokio::test]
+    async fn test_read_columns_with_computed_is_rejected() {
+        let table = table_with_two_columns("read_cols_computed").await;
+        let err = table
+            .add_columns()
+            .computed("doubled", "x * 2")
+            .read_columns(["x"])
+            .execute()
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::InvalidInput { .. }));
+        assert!(
+            table
+                .schema()
+                .await
+                .unwrap()
+                .field_with_name("doubled")
+                .is_err()
+        );
+    }
+
    #[tokio::test]
    async fn test_read_columns_limits_what_a_batch_udf_sees() {
        let table = table_with_two_columns("read_cols_udf").await;
@@ -0,0 +1,705 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Computed columns.
+//!
+//! A computed column is defined by a rule rather than by values supplied at
+//! write time. Declaring one commits the column carrying that rule in field
+//! metadata but no data, so the cost does not scale with the table; a later
+//! refresh fills the rows.
+//!
+//! The rule is tagged by kind ([`ComputedColumnKind`]) because kinds differ in
+//! where the column's type and inputs come from. A SQL expression is
+//! self-describing -- both are derived from the expression, so a caller writes
+//! neither -- while a kind resolved through a registry cannot be typed without
+//! consulting it. Only SQL exists today; the tag is what lets another kind be
+//! added without a second reading of the same key.
+//!
+//! [`computed_columns`] and [`computed_column_from_field`] read declarations
+//! back off a schema.
+
+use std::collections::HashMap;
+use std::sync::Arc;
+
+use arrow_schema::{Field as ArrowField, Schema as ArrowSchema, SchemaRef};
+use lance::dataset::NewColumnTransform;
+use lance_datafusion::planner::Planner;
+
+use crate::{Error, Result};
+
+/// Field metadata key marking a column as computed. The value is `"true"`.
+pub const COMPUTED_COLUMN_META_KEY: &str = "computed_column";
+
+/// Field metadata key naming the kind of rule that defines the column.
+pub const KIND_META_KEY: &str = "computed_column.kind";
+
+/// Field metadata key holding the SQL expression that defines the column.
+pub const EXPRESSION_META_KEY: &str = "computed_column.expression";
+
+/// Field metadata key holding the column's inputs, as a JSON array of names.
+pub const INPUTS_META_KEY: &str = "computed_column.inputs";
+
+/// Value of [`KIND_META_KEY`] for a column defined by a SQL expression.
+pub const SQL_KIND: &str = "sql";
+
+/// The rule that defines a computed column's values.
+///
+/// Non-exhaustive: a kind added later is an additive change, and a caller that
+/// only handles the kinds it knows keeps compiling.
+#[derive(Debug, Clone, PartialEq, Eq)]
+#[non_exhaustive]
+pub enum ComputedColumnKind {
+    /// A SQL expression evaluated by DataFusion. It is the whole definition:
+    /// the column's type and its inputs are both derived from it.
+    Sql {
+        /// The expression.
+        expression: String,
+    },
+    /// A kind this version does not understand, written by a newer one.
+    ///
+    /// Reported rather than hidden so a caller can tell a column it cannot
+    /// refresh apart from one that was never computed. Nothing produces this.
+    Unrecognized {
+        /// The kind as it was found in the metadata.
+        kind: String,
+    },
+}
+
+/// A computed column's declaration, as read back from field metadata.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ComputedColumn {
+    /// Name of the computed column.
+    pub name: String,
+    /// The rule that defines it.
+    pub kind: ComputedColumnKind,
+    /// Columns the rule reads, recorded at declaration time.
+    ///
+    /// Outside the kind because every kind has inputs and the consumers that
+    /// use them -- refresh planning, dependency ordering -- do not care which
+    /// kind produced them. Where they come from does differ, and that is
+    /// settled at declaration: derived from a SQL expression, supplied by the
+    /// caller for a kind that cannot be parsed.
+    pub inputs: Vec<String>,
+}
+
+/// Build the field metadata recording a SQL binding.
+fn computed_column_metadata(expression: &str, inputs: &[String]) -> HashMap<String, String> {
+    HashMap::from([
+        (COMPUTED_COLUMN_META_KEY.to_string(), "true".to_string()),
+        (KIND_META_KEY.to_string(), SQL_KIND.to_string()),
+        (EXPRESSION_META_KEY.to_string(), expression.to_string()),
+        (
+            INPUTS_META_KEY.to_string(),
+            serde_json::to_string(inputs).unwrap_or_else(|_| "[]".to_string()),
+        ),
+    ])
+}
+
+/// Read a field's computed-column declaration, if it carries one.
+///
+/// A field flagged computed but carrying no kind, or a SQL one missing its
+/// expression, is not a computed column here: without the rule there is
+/// nothing to refresh from, so it is reported as absent rather than as a
+/// half-formed declaration. An unrecognized kind is different -- the rule is
+/// there and intact, this version just cannot act on it -- and comes back as
+/// [`ComputedColumnKind::Unrecognized`].
+pub fn computed_column_from_field(field: &ArrowField) -> Option<ComputedColumn> {
+    let metadata = field.metadata();
+    if metadata.get(COMPUTED_COLUMN_META_KEY).map(String::as_str) != Some("true") {
+        return None;
+    }
+    let kind = match metadata.get(KIND_META_KEY)?.as_str() {
+        SQL_KIND => ComputedColumnKind::Sql {
+            expression: metadata.get(EXPRESSION_META_KEY)?.clone(),
+        },
+        other => ComputedColumnKind::Unrecognized {
+            kind: other.to_string(),
+        },
+    };
+    let inputs = metadata
+        .get(INPUTS_META_KEY)
+        .and_then(|raw| serde_json::from_str::<Vec<String>>(raw).ok())
+        .unwrap_or_default();
+    Some(ComputedColumn {
+        name: field.name().clone(),
+        kind,
+        inputs,
+    })
+}
+
+/// Read every computed-column declaration carried by `schema`, in field order.
+///
+/// Introspection is a pure read of the schema the caller already holds, the
+/// way a SQL catalog reports a generation expression as another column of
+/// `information_schema.columns`.
+pub fn computed_columns(schema: &ArrowSchema) -> Vec<ComputedColumn> {
+    schema
+        .fields()
+        .iter()
+        .filter_map(|field| computed_column_from_field(field))
+        .collect()
+}
+
+/// Reject a schema change to a column some declaration reads.
+///
+/// A binding is SQL text naming its inputs, so renaming, retyping or dropping
+/// one leaves an expression that no longer resolves. Refusing the change keeps
+/// a declaration that survived [`plan`] evaluable for as long as it exists.
+///
+/// Paths are compared at their root: a declaration reading `metadata` is
+/// invalidated by a change to `metadata.age` just as surely.
+pub(crate) fn ensure_not_an_input(schema: &ArrowSchema, paths: &[&str]) -> Result<()> {
+    let root = |path: &str| path.split('.').next().unwrap_or(path).to_string();
+    for declaration in computed_columns(schema) {
+        for path in paths {
+            // A declaration does not read itself, so it is free to be dropped
+            // or renamed along with its binding.
+            if declaration.name == root(path) {
+                continue;
+            }
+            if declaration
+                .inputs
+                .iter()
+                .any(|input| root(input) == root(path))
+            {
+                return Err(Error::InvalidInput {
+                    message: format!(
+                        "column '{}' is read by computed column '{}'; drop that column first",
+                        path, declaration.name
+                    ),
+                });
+            }
+        }
+    }
+    Ok(())
+}
+
+/// Resolve `(name, expression)` pairs against `schema` into fields carrying
+/// their bindings.
+///
+/// Everything that can be known statically is checked here rather than at
+/// refresh time: that the expression parses, that every column it reads
+/// exists, and that the target name is free. A declaration that survives this
+/// is one a refresh can always act on.
+pub(crate) fn plan(schema: SchemaRef, columns: &[(String, String)]) -> Result<Vec<ArrowField>> {
+    if columns.is_empty() {
+        return Err(Error::InvalidInput {
+            message: "at least one computed column is required".into(),
+        });
+    }
+
+    let planner = Planner::new(schema.clone());
+    let mut fields = Vec::with_capacity(columns.len());
+    let mut declared: Vec<&str> = Vec::with_capacity(columns.len());
+
+    for (name, expression) in columns {
+        if schema.field_with_name(name).is_ok() || declared.contains(&name.as_str()) {
+            return Err(Error::ColumnAlreadyExists { name: name.clone() });
+        }
+
+        let expr = planner
+            .parse_expr(expression)
+            .and_then(|expr| planner.optimize_expr(expr))
+            .map_err(|e| Error::InvalidExpression {
+                column: name.clone(),
+                message: e.to_string(),
+            })?;
+
+        let mut inputs = Planner::column_names_in_expr(&expr);
+        inputs.sort();
+        inputs.dedup();
+
+        // Resolved here rather than left to the planner so an unknown column
+        // names itself in the error instead of surfacing as a plan failure.
+        let mut indices = Vec::with_capacity(inputs.len());
+        for input in &inputs {
+            let index = schema
+                .index_of(input)
+                .map_err(|_| Error::InvalidExpression {
+                    column: name.clone(),
+                    message: format!("unknown column '{input}'"),
+                })?;
+            indices.push(index);
+        }
+
+        // Physical expressions address columns by position, so the planner
+        // that types the expression has to be built on the projected schema
+        // the refresh will actually read.
+        let read_schema =
+            Arc::new(
+                schema
+                    .project(&indices)
+                    .map_err(|e| Error::InvalidExpression {
+                        column: name.clone(),
+                        message: e.to_string(),
+                    })?,
+            );
+        let physical = Planner::new(read_schema.clone())
+            .create_physical_expr(&expr)
+            .map_err(|e| Error::InvalidExpression {
+                column: name.clone(),
+                message: e.to_string(),
+            })?;
+        let data_type =
+            physical
+                .data_type(read_schema.as_ref())
+                .map_err(|e| Error::InvalidExpression {
+                    column: name.clone(),
+                    message: e.to_string(),
+                })?;
+
+        // Declared columns start entirely null, so nullability is a property
+        // of the declaration rather than of what the expression yields.
+        fields.push(
+            ArrowField::new(name, data_type, true)
+                .with_metadata(computed_column_metadata(expression, &inputs)),
+        );
+        declared.push(name);
+    }
+
+    Ok(fields)
+}
+
+/// Build the transform that declares `columns` against `schema`.
+///
+/// An all-null column is how a binding with no values yet is carried into a
+/// commit; that it is spelled `AllNulls` is a detail of the commit, not of the
+/// column, which is why this is internal and
+/// [`AddColumnsBuilder::computed`](super::AddColumnsBuilder::computed) is the
+/// public way in.
+pub(crate) fn declare(
+    schema: SchemaRef,
+    columns: &[(String, String)],
+) -> Result<NewColumnTransform> {
+    let fields = plan(schema, columns)?;
+    Ok(NewColumnTransform::AllNulls(Arc::new(ArrowSchema::new(
+        fields,
+    ))))
+}
+
+/// Commit a declaration of a kind this version does not produce, the way a
+/// newer lancedb would leave one behind. Shared with the refresh tests, which
+/// need the same column to check that refresh refuses it.
+#[cfg(test)]
+pub(super) async fn add_foreign_kind(table: &crate::Table, name: &str, kind: &str) {
+    use arrow_schema::DataType;
+
+    let field = ArrowField::new(name, DataType::Int32, true).with_metadata(HashMap::from([
+        (COMPUTED_COLUMN_META_KEY.to_string(), "true".to_string()),
+        (KIND_META_KEY.to_string(), kind.to_string()),
+        (INPUTS_META_KEY.to_string(), r#"["x"]"#.to_string()),
+    ]));
+    table
+        .add_columns()
+        .transform(NewColumnTransform::AllNulls(Arc::new(ArrowSchema::new(
+            vec![field],
+        ))))
+        .execute()
+        .await
+        .unwrap();
+}
+
+#[cfg(test)]
+mod tests {
+    use arrow_array::record_batch;
+    use arrow_schema::DataType;
+    use futures::TryStreamExt;
+    use lance::dataset::ColumnAlteration;
+
+    use super::*;
+    use crate::connect;
+    use crate::query::{ExecutableQuery, QueryBase, Select};
+    use crate::{Error, Table};
+
+    async fn table_with_ints(name: &str) -> Table {
+        let conn = connect("memory://").execute().await.unwrap();
+        let batch = record_batch!(("x", Int32, [1, 2, 3])).unwrap();
+        conn.create_table(name, batch).execute().await.unwrap()
+    }
+
+    /// Declare `columns` the way a caller would: plan the expressions, then
+    /// add them through the ordinary column API.
+    async fn add_computed(table: &Table, columns: &[(String, String)]) -> Result<u64> {
+        let mut builder = table.add_columns();
+        for (name, expression) in columns {
+            builder = builder.computed(name, expression);
+        }
+        Ok(builder.execute().await?.version)
+    }
+
+    async fn declared(table: &Table) -> Vec<ComputedColumn> {
+        computed_columns(table.schema().await.unwrap().as_ref())
+    }
+
+    #[tokio::test]
+    async fn test_declare_infers_type_and_inputs() {
+        let table = table_with_ints("declare_infers").await;
+        let initial = table.version().await.unwrap();
+
+        let version = add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+        assert!(version > initial);
+
+        let schema = table.schema().await.unwrap();
+        let field = schema.field_with_name("doubled").unwrap();
+        assert_eq!(field.data_type(), &DataType::Int32);
+        assert!(field.is_nullable());
+
+        assert_eq!(
+            declared(&table).await,
+            vec![ComputedColumn {
+                name: "doubled".into(),
+                kind: ComputedColumnKind::Sql {
+                    expression: "x * 2".into()
+                },
+                inputs: vec!["x".into()],
+            }]
+        );
+    }
+
+    /// The binding reaches the schema only if `AllNulls` carries per-field
+    /// metadata through the commit. The whole representation rests on it.
+    #[tokio::test]
+    async fn test_all_nulls_preserves_field_metadata() {
+        let table = table_with_ints("metadata_survives").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        let schema = table.schema().await.unwrap();
+        let metadata = schema.field_with_name("doubled").unwrap().metadata();
+        assert_eq!(
+            metadata.get(COMPUTED_COLUMN_META_KEY).map(String::as_str),
+            Some("true")
+        );
+        assert_eq!(metadata.get(KIND_META_KEY).map(String::as_str), Some("sql"));
+        assert_eq!(
+            metadata.get(EXPRESSION_META_KEY).map(String::as_str),
+            Some("x * 2")
+        );
+        assert_eq!(
+            metadata.get(INPUTS_META_KEY).map(String::as_str),
+            Some(r#"["x"]"#)
+        );
+    }
+
+    #[tokio::test]
+    async fn test_declared_column_is_all_null() {
+        let table = table_with_ints("declare_is_null").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        let batches = table
+            .query()
+            .select(Select::columns(&["doubled"]))
+            .execute()
+            .await
+            .unwrap()
+            .try_collect::<Vec<_>>()
+            .await
+            .unwrap();
+
+        let total: usize = batches.iter().map(|b| b.num_rows()).sum();
+        assert_eq!(total, 3);
+        for batch in &batches {
+            assert_eq!(batch["doubled"].null_count(), batch.num_rows());
+        }
+    }
+
+    #[tokio::test]
+    async fn test_unknown_column_fails_at_declare_time() {
+        let table = table_with_ints("unknown_input").await;
+        let err = add_computed(&table, &[("bad".into(), "missing + 1".into())])
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::InvalidExpression { column, .. } if column == "bad"));
+
+        let schema = table.schema().await.unwrap();
+        assert!(schema.field_with_name("bad").is_err());
+    }
+
+    #[tokio::test]
+    async fn test_unparsable_expression_fails_at_declare_time() {
+        let table = table_with_ints("bad_syntax").await;
+        let err = add_computed(&table, &[("bad".into(), "x *".into())])
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::InvalidExpression { column, .. } if column == "bad"));
+        assert!(
+            table
+                .schema()
+                .await
+                .unwrap()
+                .field_with_name("bad")
+                .is_err()
+        );
+    }
+
+    /// A user-defined function is an expression like any other; only its
+    /// resolution is missing. When a registry-aware planner exists this
+    /// becomes a supported declaration rather than a new API.
+    #[tokio::test]
+    async fn test_unregistered_function_is_rejected_for_now() {
+        let table = table_with_ints("udf_not_yet").await;
+        let err = add_computed(&table, &[("vec".into(), "embed(x)".into())])
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::InvalidExpression { column, .. } if column == "vec"));
+        assert!(
+            table
+                .schema()
+                .await
+                .unwrap()
+                .field_with_name("vec")
+                .is_err()
+        );
+    }
+
+    #[tokio::test]
+    async fn test_existing_column_name_is_rejected() {
+        let table = table_with_ints("name_taken").await;
+        let err = add_computed(&table, &[("x".into(), "x * 2".into())])
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::ColumnAlreadyExists { name } if name == "x"));
+        assert!(declared(&table).await.is_empty());
+    }
+
+    #[tokio::test]
+    async fn test_constant_expression_needs_no_inputs() {
+        let table = table_with_ints("constant").await;
+        add_computed(&table, &[("answer".into(), "42".into())])
+            .await
+            .unwrap();
+
+        let declared = declared(&table).await;
+        assert_eq!(declared.len(), 1);
+        assert!(declared[0].inputs.is_empty());
+    }
+
+    #[tokio::test]
+    async fn test_multiple_columns_in_one_commit() {
+        let table = table_with_ints("multi").await;
+        let initial = table.version().await.unwrap();
+
+        add_computed(
+            &table,
+            &[
+                ("plus".into(), "x + 1".into()),
+                ("squared".into(), "x * x".into()),
+            ],
+        )
+        .await
+        .unwrap();
+
+        assert_eq!(table.version().await.unwrap(), initial + 1);
+        let declared = declared(&table).await;
+        assert_eq!(declared.len(), 2);
+        assert_eq!(declared[0].name, "plus");
+        assert_eq!(declared[1].name, "squared");
+    }
+
+    #[tokio::test]
+    async fn test_duplicate_declaration_in_one_call_is_rejected() {
+        let table = table_with_ints("dupe").await;
+        let err = add_computed(
+            &table,
+            &[
+                ("dup".into(), "x + 1".into()),
+                ("dup".into(), "x + 2".into()),
+            ],
+        )
+        .await
+        .unwrap_err();
+        assert!(matches!(err, Error::ColumnAlreadyExists { name } if name == "dup"));
+        assert!(declared(&table).await.is_empty());
+    }
+
+    /// A column added by an ordinary transform is materialized, not bound, so
+    /// it carries no declaration to report.
+    #[tokio::test]
+    async fn test_ordinary_columns_are_not_reported_as_computed() {
+        let table = table_with_ints("plain").await;
+        assert!(declared(&table).await.is_empty());
+
+        table
+            .add_columns()
+            .transform(NewColumnTransform::SqlExpressions(vec![(
+                "eager".into(),
+                "x * 2".into(),
+            )]))
+            .execute()
+            .await
+            .unwrap();
+        assert!(declared(&table).await.is_empty());
+    }
+
+    /// Built-in functions type the column the same way an operator does.
+    #[tokio::test]
+    async fn test_builtin_function_inference() {
+        let conn = connect("memory://").execute().await.unwrap();
+        let batch = record_batch!(("name", Utf8, ["ada", "grace"]), ("n", Int32, [-1, 2])).unwrap();
+        let table = conn
+            .create_table("builtins", batch)
+            .execute()
+            .await
+            .unwrap();
+
+        add_computed(
+            &table,
+            &[
+                ("shout".into(), "upper(name)".into()),
+                ("width".into(), "length(name)".into()),
+                ("magnitude".into(), "abs(n)".into()),
+            ],
+        )
+        .await
+        .unwrap();
+
+        let schema = table.schema().await.unwrap();
+        assert_eq!(
+            schema.field_with_name("shout").unwrap().data_type(),
+            &DataType::Utf8
+        );
+        assert_eq!(
+            schema.field_with_name("magnitude").unwrap().data_type(),
+            &DataType::Int32
+        );
+        // length() returns a width-dependent integer type; assert it is one
+        // rather than pinning which.
+        assert!(
+            schema
+                .field_with_name("width")
+                .unwrap()
+                .data_type()
+                .is_integer()
+        );
+
+        let declared = declared(&table).await;
+        assert_eq!(declared.len(), 3);
+        assert_eq!(declared[0].inputs, vec!["name".to_string()]);
+        assert_eq!(declared[2].inputs, vec!["n".to_string()]);
+    }
+
+    /// The reason the kind is tagged: a declaration written by a newer version
+    /// has to read back as a computed column this one cannot evaluate, not as
+    /// an ordinary column. Reported as absent it would be refreshable by
+    /// nothing and redeclarable over, silently.
+    #[tokio::test]
+    async fn test_unrecognized_kind_is_reported_rather_than_hidden() {
+        let table = table_with_ints("foreign_kind").await;
+        super::add_foreign_kind(&table, "embedding", "udf").await;
+
+        assert_eq!(
+            declared(&table).await,
+            vec![ComputedColumn {
+                name: "embedding".into(),
+                kind: ComputedColumnKind::Unrecognized { kind: "udf".into() },
+                inputs: vec!["x".into()],
+            }]
+        );
+
+        let err = add_computed(&table, &[("embedding".into(), "x * 2".into())])
+            .await
+            .unwrap_err();
+        assert!(matches!(err, Error::ColumnAlreadyExists { name } if name == "embedding"));
+    }
+
+    /// A kind is what makes a declaration readable at all, so the flag alone
+    /// is half-formed in the same way a missing expression is.
+    #[test]
+    fn test_flag_without_a_kind_is_not_a_declaration() {
+        let field =
+            ArrowField::new("half", DataType::Int32, true).with_metadata(HashMap::from([(
+                COMPUTED_COLUMN_META_KEY.to_string(),
+                "true".to_string(),
+            )]));
+        assert_eq!(computed_column_from_field(&field), None);
+    }
+
+    /// A SQL declaration is its expression; without one there is nothing to
+    /// refresh from.
+    #[test]
+    fn test_sql_kind_without_an_expression_is_not_a_declaration() {
+        let field = ArrowField::new("half", DataType::Int32, true).with_metadata(HashMap::from([
+            (COMPUTED_COLUMN_META_KEY.to_string(), "true".to_string()),
+            (KIND_META_KEY.to_string(), SQL_KIND.to_string()),
+        ]));
+        assert_eq!(computed_column_from_field(&field), None);
+    }
+
+    #[tokio::test]
+    async fn test_inputs_are_deduplicated_and_sorted() {
+        let conn = connect("memory://").execute().await.unwrap();
+        let batch = record_batch!(("b", Int32, [1, 2]), ("a", Int32, [3, 4])).unwrap();
+        let table = conn.create_table("dedupe", batch).execute().await.unwrap();
+
+        add_computed(&table, &[("total".into(), "b + a + b".into())])
+            .await
+            .unwrap();
+
+        assert_eq!(
+            declared(&table).await[0].inputs,
+            vec!["a".to_string(), "b".to_string()]
+        );
+    }
+
+    #[tokio::test]
+    async fn test_dropping_an_input_is_refused() {
+        let table = table_with_ints("drop_input").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        let err = table.drop_columns(&["x"]).await.unwrap_err();
+        assert!(
+            matches!(&err, Error::InvalidInput { message } if message.contains("doubled")),
+            "{err:?}"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_renaming_an_input_is_refused() {
+        let table = table_with_ints("rename_input").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        let err = table
+            .alter_columns(&[ColumnAlteration::new("x".into()).rename("y".into())])
+            .await
+            .unwrap_err();
+        assert!(
+            matches!(&err, Error::InvalidInput { message } if message.contains("doubled")),
+            "{err:?}"
+        );
+    }
+
+    /// Nothing resolves against nullability, so it is not a rebinding.
+    #[tokio::test]
+    async fn test_altering_an_input_nullability_is_allowed() {
+        let table = table_with_ints("nullable_input").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        table
+            .alter_columns(&[ColumnAlteration::new("x".into()).set_nullable(true)])
+            .await
+            .unwrap();
+    }
+
+    /// A declaration does not read itself, so it travels with its binding.
+    #[tokio::test]
+    async fn test_dropping_the_computed_column_is_allowed() {
+        let table = table_with_ints("drop_computed").await;
+        add_computed(&table, &[("doubled".into(), "x * 2".into())])
+            .await
+            .unwrap();
+
+        table.drop_columns(&["doubled"]).await.unwrap();
+        assert!(declared(&table).await.is_empty());
+    }
+}
@@ -1161,7 +1161,7 @@ mod lsm_tests {
            .unwrap();
        let fts_index = table.list_indices().await.unwrap()[0].name.clone();
        table
-            .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes([fts_index]))
+            .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes(vec![fts_index]))
            .await
            .unwrap();

@@ -1254,7 +1254,7 @@ mod lsm_tests {
            .unwrap();
        let vec_index = table.list_indices().await.unwrap()[0].name.clone();
        table
-            .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes([vec_index]))
+            .set_lsm_write_spec(LsmWriteSpec::unsharded().with_maintained_indexes(vec![vec_index]))
            .await
            .unwrap();

@@ -29,6 +29,7 @@ use arrow_schema::{DataType, Schema as ArrowSchema, SchemaRef};
 use lance::Dataset;
 use lance::dataset::mem_wal::{
    DatasetMemWalExt, ShardWriter, ShardWriterConfig, evaluate_sharding_spec,
+    validate_maintained_indexes,
 };
 use lance::index::DatasetIndexExt;
 use lance_core::datatypes::Schema as LanceSchema;
@@ -37,8 +38,9 @@ use tokio::sync::RwLock;
 use uuid::Uuid;

 use crate::error::{Error, Result};
+use crate::index::IndexConfig;
 use crate::table::merge::{MergeInsertBuilder, MergeResult};
-use crate::table::{LsmWriteSpec, NativeTable};
+use crate::table::{BaseTable, LsmWriteSpec, NativeTable};

 /// Spec id of the sole sharding spec installed by [`set_lsm_write_spec`].
 /// Must match Lance's `InitializeMemWalBuilder` (`SHARDING_SPEC_ID`).
@@ -80,32 +82,44 @@ pub(crate) async fn set_lsm_write_spec(table: &NativeTable, spec: LsmWriteSpec)
        }
    }

+    // Before the builder borrows the dataset clone. `list_indices` merges an
+    // index's segments into one entry, so the result needs no dedup.
+    let maintained_indexes = {
+        let dataset = table.dataset.get().await?;
+        resolve_maintained_indexes(
+            &dataset,
+            &table.list_indices().await?,
+            spec.maintained_indexes(),
+        )
+        .await?
+    };
+
    let mut dataset = (*table.dataset.get().await?).clone();
    let mut builder = dataset.initialize_mem_wal();
-    let (maintained_indexes, writer_config_defaults) = match spec {
+    let writer_config_defaults = match spec {
        LsmWriteSpec::Bucket {
            column,
            num_buckets,
-            maintained_indexes,
            writer_config_defaults,
+            ..
        } => {
            builder = builder.bucket_sharding(column, num_buckets);
-            (maintained_indexes, writer_config_defaults)
+            writer_config_defaults
        }
        LsmWriteSpec::Identity {
            column,
-            maintained_indexes,
            writer_config_defaults,
+            ..
        } => {
            builder = builder.identity_sharding(column);
-            (maintained_indexes, writer_config_defaults)
+            writer_config_defaults
        }
        LsmWriteSpec::Unsharded {
-            maintained_indexes,
            writer_config_defaults,
+            ..
        } => {
            builder = builder.unsharded();
-            (maintained_indexes, writer_config_defaults)
+            writer_config_defaults
        }
    };
    builder = builder.maintained_indexes(maintained_indexes);
@@ -117,6 +131,58 @@ pub(crate) async fn set_lsm_write_spec(table: &NativeTable, spec: LsmWriteSpec)
    Ok(())
 }

+/// Resolve a spec's maintained-index selection against `indices`, as reported
+/// by [`Table::list_indices`](crate::Table::list_indices).
+///
+/// `None` means every index on the table, snapshotted now. Lance validates
+/// either selection against its shard-writer rules, so a spec that installs is
+/// one the MemWAL can open.
+///
+/// An unmaintainable index fails an inferred set rather than being dropped from
+/// it — dropping would leave the caller believing it is maintained.
+async fn resolve_maintained_indexes(
+    dataset: &Dataset,
+    indices: &[IndexConfig],
+    requested: Option<&[String]>,
+) -> Result<Vec<String>> {
+    let Some(requested) = requested else {
+        let all: Vec<String> = indices.iter().map(|index| index.name.clone()).collect();
+        validate_maintained_indexes(dataset, &all)
+            .await
+            .map_err(|source| Error::InvalidInput {
+                message: format!(
+                    "cannot maintain every index on this table: {source}. Set \
+                     maintained_indexes explicitly to choose from {}",
+                    index_name_list(indices),
+                ),
+            })?;
+        return Ok(all);
+    };
+    for name in requested {
+        if !indices.iter().any(|index| &index.name == name) {
+            return Err(Error::InvalidInput {
+                message: format!(
+                    "maintained index '{}' does not exist on this table; it has {}",
+                    name,
+                    index_name_list(indices),
+                ),
+            });
+        }
+    }
+    validate_maintained_indexes(dataset, requested).await?;
+    Ok(requested.to_vec())
+}
+
+/// Index names for an error message.
+fn index_name_list(indices: &[IndexConfig]) -> String {
+    if indices.is_empty() {
+        return "no indexes".to_string();
+    }
+    let mut names: Vec<&str> = indices.iter().map(|index| index.name.as_str()).collect();
+    names.sort_unstable();
+    format!("[{}]", names.join(", "))
+}
+
 // =============================================================================
 // unset_lsm_write_spec
 // =============================================================================
@@ -0,0 +1,523 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Filling computed columns.
+//!
+//! A row without a value gets one; a row that has one keeps it. Refresh is
+//! therefore idempotent and does not observe input mutation -- once a row is
+//! filled, changing what the expression reads leaves the stored result alone.
+//!
+//! Convergence comes from staging nothing when nothing would change, so an
+//! expression yielding null settles after one pass rather than re-selecting
+//! the same rows forever. Fragments that already cover the column and hold no
+//! nulls are skipped without evaluating it at all.
+
+use std::sync::Arc;
+
+use arrow_array::RecordBatch;
+use arrow_schema::Schema as ArrowSchema;
+use futures::{TryStreamExt, stream};
+use lance::Dataset;
+use lance::dataset::WriteDestination;
+use lance::dataset::fragment::FileFragment;
+use lance::dataset::transaction::Operation;
+use lance_core::ROW_ID;
+use lance_core::datatypes::Schema as LanceSchema;
+use serde::{Deserialize, Serialize};
+
+use super::NativeTable;
+use super::computed_columns::{ComputedColumnKind, computed_column_from_field};
+use crate::{Error, Result};
+
+/// Alias the expression is projected under, so its result and the column's
+/// current values can be read side by side.
+const COMPUTED_ALIAS: &str = "__lancedb_computed";
+
+/// The result of refreshing a computed column.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
+pub struct RefreshColumnResult {
+    /// Rows that had a value computed.
+    #[serde(default)]
+    pub rows_filled: u64,
+    /// The commit version associated with the operation.
+    #[serde(default)]
+    pub version: u64,
+}
+
+/// Internal implementation of the refresh logic.
+pub(crate) async fn execute_refresh_column(
+    table: &NativeTable,
+    column: &str,
+) -> Result<RefreshColumnResult> {
+    table.dataset.ensure_mutable()?;
+    let dataset = table.dataset.get().await?;
+
+    let expression = declared_expression(&dataset, column)?;
+    let field = dataset
+        .schema()
+        .field(column)
+        .ok_or_else(|| Error::ColumnNotFound {
+            name: column.to_string(),
+        })?;
+    // The dataset's own field, so the identity write_column checks against the
+    // manifest holds by construction.
+    let column_schema = LanceSchema {
+        fields: vec![field.clone()],
+        metadata: Default::default(),
+    };
+
+    let mut rows_filled = 0u64;
+    let mut replacements = Vec::new();
+    for fragment in fragments_to_consider(&dataset, column, field.id).await? {
+        let Some((filled, values)) =
+            fill_fragment(&dataset, &fragment, column, &expression).await?
+        else {
+            continue;
+        };
+        rows_filled += filled;
+        replacements.push(
+            fragment
+                .write_column(stream::iter(values.into_iter().map(Ok)), &column_schema)
+                .await?,
+        );
+    }
+
+    if replacements.is_empty() {
+        return Ok(RefreshColumnResult {
+            rows_filled: 0,
+            version: dataset.version().version,
+        });
+    }
+
+    let read_version = dataset.version().version;
+    let new_dataset = Dataset::commit(
+        WriteDestination::Dataset(dataset.clone()),
+        Operation::DataReplacement { replacements },
+        Some(read_version),
+        None,
+        None,
+        Arc::new(Default::default()),
+        false,
+    )
+    .await?;
+
+    let version = new_dataset.version().version;
+    table.dataset.update(new_dataset);
+    Ok(RefreshColumnResult {
+        rows_filled,
+        version,
+    })
+}
+
+/// The SQL expression `column` is declared with.
+fn declared_expression(dataset: &Dataset, column: &str) -> Result<String> {
+    let schema = ArrowSchema::from(dataset.schema());
+    let field = schema
+        .field_with_name(column)
+        .map_err(|_| Error::ColumnNotFound {
+            name: column.to_string(),
+        })?;
+    let declaration =
+        computed_column_from_field(field).ok_or_else(|| Error::NotAComputedColumn {
+            name: column.to_string(),
+        })?;
+    match declaration.kind {
+        ComputedColumnKind::Sql { expression } => Ok(expression),
+        ComputedColumnKind::Unrecognized { kind } => Err(Error::NotSupported {
+            message: format!(
+                "computed column '{column}' is defined by '{kind}', which this version of \
+                 lancedb cannot evaluate"
+            ),
+        }),
+    }
+}
+
+/// Quote `name` as a lance SQL identifier.
+///
+/// Lance's dialect delimits with backticks, so a double-quoted name would
+/// parse as a string literal rather than a column.
+fn quote_identifier(name: &str) -> String {
+    format!("`{}`", name.replace('`', "``"))
+}
+
+/// Fragments that could hold a row needing a value.
+///
+/// A fragment whose data files do not carry the field cannot hold one that
+/// does. One that carries it is asked, since a row rewrite -- an update, or a
+/// compaction folding an unfilled fragment into a filled one -- can leave
+/// nulls behind a covering file.
+async fn fragments_to_consider(
+    dataset: &Dataset,
+    column: &str,
+    field_id: i32,
+) -> Result<Vec<FileFragment>> {
+    let unfilled = format!("{} IS NULL", quote_identifier(column));
+    let mut considered = Vec::new();
+    for fragment in dataset.get_fragments() {
+        let covered = fragment
+            .metadata()
+            .files
+            .iter()
+            .any(|file| file.fields.contains(&field_id));
+        if !covered || fragment.count_rows(Some(unfilled.clone())).await? > 0 {
+            considered.push(fragment);
+        }
+    }
+    Ok(considered)
+}
+
+/// Compute one fragment's column, keeping every value it already holds.
+///
+/// `Ok(None)` when no live row gained a value, which is what keeps a refresh
+/// from restaging a fragment whose expression yields null. Deleted rows are
+/// carried through so the values line up positionally with the fragment's data
+/// files; they are never read back, but the column file has to cover them.
+async fn fill_fragment(
+    dataset: &Dataset,
+    fragment: &FileFragment,
+    column: &str,
+    expression: &str,
+) -> Result<Option<(u64, Vec<RecordBatch>)>> {
+    let mut scanner = dataset.scan();
+    scanner
+        .with_fragments(vec![fragment.metadata().clone()])
+        .with_row_id()
+        .include_deleted_rows()
+        .project_with_transform(&[
+            (column, quote_identifier(column).as_str()),
+            (COMPUTED_ALIAS, expression),
+        ])?;
+
+    let projected = Arc::new(ArrowSchema::new(vec![
+        ArrowSchema::from(dataset.schema())
+            .field_with_name(column)
+            .map_err(|_| Error::ColumnNotFound {
+                name: column.to_string(),
+            })?
+            .clone(),
+    ]));
+
+    let missing = |name: &str| Error::Runtime {
+        message: format!("refreshing {column} produced no {name} column"),
+    };
+
+    let mut filled = 0u64;
+    let mut values = Vec::new();
+    let mut batches = scanner.try_into_stream().await?;
+    while let Some(batch) = batches.try_next().await? {
+        let existing = batch
+            .column_by_name(column)
+            .ok_or_else(|| missing(column))?;
+        let computed = batch
+            .column_by_name(COMPUTED_ALIAS)
+            .ok_or_else(|| missing("expression"))?;
+        let row_ids = batch
+            .column_by_name(ROW_ID)
+            .ok_or_else(|| missing(ROW_ID))?;
+
+        // A row is filled only if it gains a value: an expression yielding null
+        // leaves it as unfilled as it was, which is what lets a refresh settle.
+        // A deleted row has a null row id; its value is written but not counted.
+        let unfilled = arrow::compute::is_null(existing.as_ref())?;
+        filled += (0..unfilled.len())
+            .filter(|i| unfilled.value(*i) && row_ids.is_valid(*i) && computed.is_valid(*i))
+            .count() as u64;
+
+        let merged = arrow_select::zip::zip(&unfilled, computed, existing)?;
+        values.push(RecordBatch::try_new(projected.clone(), vec![merged])?);
+    }
+
+    Ok((filled > 0).then_some((filled, values)))
+}
+
+#[cfg(test)]
+mod tests {
+    use arrow_array::{Int32Array, record_batch};
+    use futures::TryStreamExt;
+
+    use crate::connect;
+    use crate::query::{ExecutableQuery, QueryBase, Select};
+    use crate::{Error, Result, Table};
+
+    async fn table_with(name: &str, values: Vec<i32>) -> Table {
+        let conn = connect("memory://").execute().await.unwrap();
+        let batch = record_batch!(("x", Int32, values)).unwrap();
+        conn.create_table(name, batch).execute().await.unwrap()
+    }
+
+    async fn declare_doubled(table: &Table) -> Result<u64> {
+        Ok(table
+            .add_columns()
+            .computed("doubled", "x * 2")
+            .execute()
+            .await?
+            .version)
+    }
+
+    async fn read(table: &Table, column: &str) -> Vec<Option<i32>> {
+        let batches = table
+            .query()
+            .select(Select::columns(&[column]))
+            .execute()
+            .await
+            .unwrap()
+            .try_collect::<Vec<_>>()
+            .await
+            .unwrap();
+        let mut values: Vec<Option<i32>> = batches
+            .iter()
+            .flat_map(|batch| {
+                batch[column]
+                    .as_any()
+                    .downcast_ref::<Int32Array>()
+                    .unwrap()
+                    .iter()
+                    .collect::<Vec<_>>()
+            })
+            .collect();
+        values.sort();
+        values
+    }
+
+    async fn append(table: &Table, values: Vec<i32>) {
+        let batch = record_batch!(("x", Int32, values)).unwrap();
+        table.add(batch).execute().await.unwrap();
+    }
+
+    #[tokio::test]
+    async fn test_refresh_fills_a_declared_column() {
+        let table = table_with("refresh_fills", vec![1, 2, 3]).await;
+        let declared = declare_doubled(&table).await.unwrap();
+        assert_eq!(read(&table, "doubled").await, vec![None, None, None]);
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert!(result.version > declared);
+        assert_eq!(result.rows_filled, 3);
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(4), Some(6)]
+        );
+    }
+
+    /// Values written after the last refresh must be reachable by another one.
+    #[tokio::test]
+    async fn test_refresh_fills_rows_appended_since_the_last_refresh() {
+        let table = table_with("refresh_appended", vec![1, 2]).await;
+        declare_doubled(&table).await.unwrap();
+        table.refresh_column("doubled").await.unwrap();
+
+        append(&table, vec![5, 6]).await;
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![None, None, Some(2), Some(4)]
+        );
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(result.rows_filled, 2);
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(4), Some(10), Some(12)]
+        );
+    }
+
+    #[tokio::test]
+    async fn test_refresh_with_nothing_to_fill() {
+        let table = table_with("refresh_noop", vec![1, 2, 3]).await;
+        declare_doubled(&table).await.unwrap();
+        table.refresh_column("doubled").await.unwrap();
+
+        let again = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(again.rows_filled, 0);
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(4), Some(6)]
+        );
+    }
+
+    /// A row is filled only by gaining a value, so an expression yielding null
+    /// settles at once instead of re-selecting the same rows forever. Nothing
+    /// is staged, so the version does not move either.
+    #[tokio::test]
+    async fn test_refresh_converges_on_a_null_result() {
+        let table = table_with("refresh_null_result", vec![1, 2, 3]).await;
+        let declared = table
+            .add_columns()
+            .computed("maybe", "nullif(x, x)")
+            .execute()
+            .await
+            .unwrap()
+            .version;
+
+        let first = table.refresh_column("maybe").await.unwrap();
+        assert_eq!(first.rows_filled, 0);
+        assert_eq!(first.version, declared);
+        assert_eq!(read(&table, "maybe").await, vec![None, None, None]);
+
+        let again = table.refresh_column("maybe").await.unwrap();
+        assert_eq!(again.rows_filled, 0);
+        assert_eq!(again.version, declared);
+    }
+
+    /// The contract's boundary: a filled fragment is not revisited, so
+    /// mutating an input leaves the value computed at fill time.
+    #[tokio::test]
+    async fn test_refresh_does_not_observe_input_mutation() {
+        let table = table_with("refresh_mutation", vec![1]).await;
+        declare_doubled(&table).await.unwrap();
+        table.refresh_column("doubled").await.unwrap();
+        assert_eq!(read(&table, "doubled").await, vec![Some(2)]);
+
+        table.update().column("x", "3").execute().await.unwrap();
+
+        let again = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(again.rows_filled, 0);
+        assert_eq!(read(&table, "doubled").await, vec![Some(2)]);
+    }
+
+    /// A row rewrite before the first refresh materializes the declared
+    /// column as null behind a covering data file. Those rows are still
+    /// unfilled and a later refresh has to reach them.
+    #[tokio::test]
+    async fn test_update_before_the_first_refresh() {
+        let table = table_with("refresh_update_first", vec![1]).await;
+        declare_doubled(&table).await.unwrap();
+
+        table.update().column("x", "3").execute().await.unwrap();
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(result.rows_filled, 1);
+        assert_eq!(read(&table, "doubled").await, vec![Some(6)]);
+    }
+
+    /// The contract holds row by row, not fragment by fragment: revisiting a
+    /// fragment to fill one row must not recompute a filled row sitting beside
+    /// it, even where the input behind it has since changed.
+    #[tokio::test]
+    async fn test_refresh_does_not_recompute_a_filled_row_beside_an_unfilled_one() {
+        let table = table_with("refresh_mixed", vec![1, 2]).await;
+        declare_doubled(&table).await.unwrap();
+        table.refresh_column("doubled").await.unwrap();
+
+        append(&table, vec![5]).await;
+        table
+            .update()
+            .column("x", "100")
+            .only_if("x = 1")
+            .execute()
+            .await
+            .unwrap();
+        table
+            .optimize(crate::table::OptimizeAction::Compact {
+                options: crate::table::CompactionOptions::default(),
+                remap_options: None,
+            })
+            .await
+            .unwrap();
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(result.rows_filled, 1);
+        // 2 is the mutated row keeping the value it was filled with, not 200.
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(4), Some(10)]
+        );
+    }
+
+    /// Filling a fragment must not disturb the values it already holds, which
+    /// is what makes a compaction-mixed fragment safe to revisit.
+    #[tokio::test]
+    async fn test_refresh_preserves_already_filled_rows() {
+        let table = table_with("refresh_preserves", vec![1, 2]).await;
+        declare_doubled(&table).await.unwrap();
+        table.refresh_column("doubled").await.unwrap();
+
+        append(&table, vec![5]).await;
+        table
+            .optimize(crate::table::OptimizeAction::Compact {
+                options: crate::table::CompactionOptions::default(),
+                remap_options: None,
+            })
+            .await
+            .unwrap();
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(result.rows_filled, 1);
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(4), Some(10)]
+        );
+    }
+
+    #[tokio::test]
+    async fn test_refresh_leaves_deleted_rows_alone() {
+        let table = table_with("refresh_deleted", vec![1, 2, 3, 4]).await;
+        declare_doubled(&table).await.unwrap();
+        table.delete("x = 2").await.unwrap();
+
+        let result = table.refresh_column("doubled").await.unwrap();
+        assert_eq!(result.rows_filled, 3);
+        assert_eq!(
+            read(&table, "doubled").await,
+            vec![Some(2), Some(6), Some(8)]
+        );
+    }
+
+    #[tokio::test]
+    async fn test_refresh_a_constant_expression() {
+        let table = table_with("refresh_constant", vec![1, 2, 3]).await;
+        table
+            .add_columns()
+            .computed("answer", "42")
+            .execute()
+            .await
+            .unwrap();
+
+        let result = table.refresh_column("answer").await.unwrap();
+        assert_eq!(result.rows_filled, 3);
+    }
+
+    /// A name needing quotes reaches the evaluator intact: it is carried as a
+    /// projection alias, never spliced into SQL text.
+    #[tokio::test]
+    async fn test_refresh_a_column_whose_name_needs_quoting() {
+        let table = table_with("refresh_quoted", vec![1, 2, 3]).await;
+        table
+            .add_columns()
+            .computed("double value", "x * 2")
+            .execute()
+            .await
+            .unwrap();
+
+        let result = table.refresh_column("double value").await.unwrap();
+        assert_eq!(result.rows_filled, 3);
+        assert_eq!(
+            read(&table, "double value").await,
+            vec![Some(2), Some(4), Some(6)]
+        );
+    }
+
+    #[tokio::test]
+    async fn test_refresh_rejects_a_plain_column() {
+        let table = table_with("refresh_plain", vec![1, 2, 3]).await;
+        let err = table.refresh_column("x").await.unwrap_err();
+        assert!(matches!(err, Error::NotAComputedColumn { name } if name == "x"));
+    }
+
+    #[tokio::test]
+    async fn test_refresh_rejects_an_unknown_column() {
+        let table = table_with("refresh_missing", vec![1, 2, 3]).await;
+        let err = table.refresh_column("nope").await.unwrap_err();
+        assert!(matches!(err, Error::ColumnNotFound { name } if name == "nope"));
+    }
+
+    /// A declaration of a kind this version cannot evaluate is refused by
+    /// name, rather than mistaken for a plain column or fed to the SQL path.
+    #[tokio::test]
+    async fn test_refresh_rejects_a_kind_it_cannot_evaluate() {
+        let table = table_with("refresh_foreign", vec![1, 2, 3]).await;
+        super::super::computed_columns::add_foreign_kind(&table, "embedding", "udf").await;
+
+        let err = table.refresh_column("embedding").await.unwrap_err();
+        assert!(matches!(err, Error::NotSupported { message } if message.contains("udf")));
+    }
+}
@@ -8,11 +8,13 @@
 //! - [`alter_columns`](execute_alter_columns): Rename columns, change types, or modify nullability
 //! - [`drop_columns`](execute_drop_columns): Remove columns from the table

+use arrow_schema::Schema as ArrowSchema;
 use lance::dataset::{ColumnAlteration, NewColumnTransform};
 use serde::{Deserialize, Serialize};
 use std::collections::HashMap;

 use super::NativeTable;
+use super::computed_columns;
 use crate::Result;

 /// The result of an add columns operation.
@@ -116,6 +118,14 @@ pub(crate) async fn execute_alter_columns(
 ) -> Result<AlterColumnsResult> {
    table.dataset.ensure_mutable()?;
    let mut dataset = (*table.dataset.get().await?).clone();
+    // Nullability is not part of what an expression resolves against, so only
+    // a rename or a retype can invalidate a binding.
+    let rebinding = alterations
+        .iter()
+        .filter(|alteration| alteration.rename.is_some() || alteration.data_type.is_some())
+        .map(|alteration| alteration.path.as_str())
+        .collect::<Vec<_>>();
+    computed_columns::ensure_not_an_input(&ArrowSchema::from(dataset.schema()), &rebinding)?;
    dataset.alter_columns(alterations).await?;
    let version = dataset.version().version;
    table.dataset.update(dataset);
@@ -131,6 +141,7 @@ pub(crate) async fn execute_drop_columns(
 ) -> Result<DropColumnsResult> {
    table.dataset.ensure_mutable()?;
    let mut dataset = (*table.dataset.get().await?).clone();
+    computed_columns::ensure_not_an_input(&ArrowSchema::from(dataset.schema()), columns)?;
    dataset.drop_columns(columns).await?;
    let version = dataset.version().version;
    table.dataset.update(dataset);
@@ -10,7 +10,7 @@ use arrow_array::{
 use arrow_schema::{DataType, Field, Fields, Schema};
 use futures::TryStreamExt;
 use lance::Dataset;
-use lance_file::version::LanceFileVersion;
+use lance_file::version::{ConcreteFileVersion, LanceFileVersion};
 use lancedb::{
    Connection, Error, Result, Table,
    blob::{BlobRangeRequest, blob},
@@ -61,7 +61,7 @@ async fn create_inline_blob_table(
    Ok(table)
 }

-async fn storage_format_version(table: &Table) -> LanceFileVersion {
+async fn storage_format_version(table: &Table) -> ConcreteFileVersion {
    table
        .as_native()
        .unwrap()
@@ -69,9 +69,14 @@ async fn storage_format_version(table: &Table) -> LanceFileVersion {
        .await
        .unwrap()
        .data_storage_format
-        .lance_file_version()
-        .unwrap()
-        .resolve()
+        .lance_file_format()
+}
+
+fn supports_blob_v2(version: ConcreteFileVersion) -> bool {
+    matches!(
+        version,
+        ConcreteFileVersion::V2_2 | ConcreteFileVersion::V2_3
+    )
 }

 async fn uses_stable_row_ids(table: &Table) -> bool {
@@ -112,7 +117,7 @@ async fn declaring_blob_column_bumps_format_and_enables_stable_row_ids() -> Resu
        .execute()
        .await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(uses_stable_row_ids(&table).await);
    Ok(())
 }
@@ -127,7 +132,7 @@ async fn explicit_stable_row_id_setting_wins_over_blob_default() -> Result<()> {
        .execute()
        .await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(!uses_stable_row_ids(&table).await);
    Ok(())
 }
@@ -139,7 +144,7 @@ async fn non_blob_table_keeps_default_format_and_row_id_setting() -> Result<()>
    let schema = Arc::new(Schema::new(vec![Field::new("id", DataType::Int64, false)]));
    let table = db.create_empty_table("t", schema).execute().await?;

-    assert!(storage_format_version(&table).await < LanceFileVersion::V2_2);
+    assert!(!supports_blob_v2(storage_format_version(&table).await));
    assert!(!uses_stable_row_ids(&table).await);
    Ok(())
 }
@@ -171,7 +176,7 @@ async fn creating_with_blob_data_bumps_format() -> Result<()> {
    .unwrap();
    let table = db.create_table("t", batch).execute().await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(uses_stable_row_ids(&table).await);
    assert_eq!(table.count_rows(None).await?, 1);
    Ok(())
@@ -281,7 +286,7 @@ async fn connection_level_stable_row_id_setting_wins_over_blob_default() -> Resu
        .execute()
        .await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(!uses_stable_row_ids(&table).await);
    Ok(())
 }
@@ -297,7 +302,7 @@ async fn namespace_create_applies_blob_defaults() -> Result<()> {
        .execute()
        .await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(uses_stable_row_ids(&table).await);
    Ok(())
 }
@@ -474,7 +479,7 @@ async fn fetch_blobs_round_trips_nested_blob_column() -> Result<()> {
    let batch = RecordBatch::try_new(schema, vec![Arc::new(info_array) as ArrayRef]).unwrap();
    let table = db.create_table("t", batch).execute().await?;

-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
+    assert!(supports_blob_v2(storage_format_version(&table).await));
    assert!(uses_stable_row_ids(&table).await);

    let ids = collect_row_ids(&table).await?;
@@ -1305,7 +1310,7 @@ async fn optimize_preserves_blob_v2_null_and_empty_distinction() -> Result<()> {
        .await?;
    table.add(null_empty_input_batch()).execute().await?;
    assert!(
-        storage_format_version(&table).await >= LanceFileVersion::V2_2,
+        supports_blob_v2(storage_format_version(&table).await),
        "blob v2 columns require storage >= 2.2"
    );
Author	SHA1	Message	Date
Wyatt Alt	c463ca1503	docs: state what a computed column promises The declaration API shipped without a runnable example, and none of the three binding docs said when values appear, what happens to them when an input changes, which schema operations a declaration blocks, or that the feature is local-only. Those are the questions a caller has to answer before using it. Adds Rust doctests on both entry points and the same semantics to the Python and TypeScript parameter docs, plus a worked Python example. Also adds the abstract refresh_column that both concrete Python tables already implemented, so the surface is declared in one place and the cross-references resolve.	2026-08-12 21:01:11 -07:00
Wyatt Alt	79e9dffd06	fix: make computed columns explicitly local-only Both operations were advertised on remote tables and neither could work. A declaration reaches the wire as AllNulls, which RemoteTable::add_columns does not accept, and refresh_column fell through to the BaseTable default; the TypeScript wrapper reached the same surface. Callers got errors that named neither the feature nor the reason. The remote protocol has no representation for a stored expression, and what one should look like is not settled -- the server persists a declaration under a different vocabulary. So this states the boundary rather than guessing at a wire format: an explicit AllNulls arm, a default that names computed columns, and NotImplementedError raised in Python before the round trip.	2026-08-12 20:00:18 -07:00
Wyatt Alt	c3efc320a6	fix(rust): refuse schema changes that invalidate a computed column A declaration records the columns its expression reads, but nothing consulted them: renaming an input left an expression naming a column that no longer exists, and the failure surfaced at refresh time as a plan error rather than at the operation that caused it. Dropping or retyping an input did the same. alter_columns and drop_columns now reject a change to a column some declaration reads. Nullability is not part of what an expression resolves against, so it stays allowed. A declaration does not read itself and so travels with its own binding, and paths compare at their root, since a change to `metadata.age` invalidates an expression reading `metadata` just as surely. Binding to field ids instead would leave the expression text naming the old column, so it would need rewriting stored SQL on every rename. Refusing the operation is what a generated column does elsewhere.	2026-08-12 20:00:18 -07:00
Wyatt Alt	8d6dea6313	fix(rust): fill computed columns row by row Refresh selected rows with `{column} IS NULL` and rewrote them through an UPDATE, which made the output value double as the record of whether the row had been computed. Two consequences: an expression yielding null re-selected the same rows on every run and reported them as filled forever, and the target name was interpolated into SQL unquoted, so a column named `double value` could be declared and never refreshed. Filling is now per fragment. Each fragment that could hold an unfilled row has the expression evaluated over its physical rows and the result written as a standalone column file, published together in one DataReplacement. A row that already holds a value keeps it -- the computed and current values are merged on the is-null mask -- and a row counts as filled only when it gains a value, so a fragment where nothing would change is never staged and a null expression settles after one pass. Which fragments are worth looking at comes from the manifest first: one whose data files do not carry the field cannot hold a filled row. A fragment that does carry it is still asked, because a row rewrite -- an update, or a compaction folding an unfilled fragment into a filled one -- leaves nulls behind a covering file. That case is the reason coverage alone is not the marker; the test for it fails against a coverage-only implementation. Names now reach the evaluator through a projection alias or a backtick-quoted identifier, lance's dialect having no other way to spell one -- a double-quoted name parses as a string literal.	2026-08-12 20:00:18 -07:00
Wyatt Alt	5a2d3f39e2	chore: update lance dependency to v11.0.0-beta.7 Filling a computed column needs FileFragment::write_column, which lands in this release.	2026-08-12 20:00:18 -07:00
Wyatt Alt	219f41339d	refactor(rust): tag a computed column's definition by kind ComputedColumn carried a bare expression string, which asserts that every computed column is a SQL expression. That holds for the only kind there is, but it is the wrong shape for the next one: a column defined by a registered function cannot be typed by parsing its definition, so its type and its inputs have a different provenance than a SQL column's. A single string has nowhere to say which it is. The definition is now ComputedColumnKind, non-exhaustive so another kind is additive, and the metadata carries a matching computed_column.kind tag beside the payload. Tagging the persisted form is the point -- the Rust type stays cheap to change and field metadata does not, and a second kind distinguished only by which keys happen to be present would leave every reader sniffing the shape. An unrecognized kind reads back as Unrecognized rather than as absent. A newer version's declaration is a computed column this one cannot evaluate, not a plain column: reported as absent it would be redeclarable over and would fail refresh as "not a computed column". Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>	2026-08-12 18:48:48 -07:00
Wyatt Alt	5982eebbb3	feat(nodejs): expose computed columns and refreshColumn addComputedColumns declares columns defined by a SQL expression, and refreshColumn fills a computed column's unfilled rows: await table.addComputedColumns([{ name: "doubled", valueSql: "x * 2" }]); await table.refreshColumn("doubled");	2026-08-12 18:48:48 -07:00
Wyatt Alt	5a1c839382	feat(python): expose computed columns and refresh_column add_columns gains a computed= mapping of column name to SQL expression, and refresh_column fills a computed column's unfilled rows. Both are available on the sync and async tables: table.add_columns(computed={"doubled": "x * 2"}) table.refresh_column("doubled") transforms and computed are mutually exclusive, since they commit through different paths and could half-apply.	2026-08-12 18:48:48 -07:00
Wyatt Alt	e40e073a9d	feat(rust): fill computed columns with refresh_column Declaring a computed column stores its expression but computes nothing, so until now the column stayed null with no way to fill it. refresh_column evaluates the expression over the rows that still hold no value and commits the results: table.refresh_column("doubled").await? Rows without a value are the ones to fill, which also makes the operation idempotent and resumable after a failure: refreshing again picks up whatever did not land. A row whose expression evaluates to null is indistinguishable from an unfilled one and is recomputed, which costs work but cannot change the result. Values written after a refresh are reachable by the next one, which is the case that matters -- an expression column populated once and then appended to would otherwise read null for every later row forever.	2026-08-12 18:48:48 -07:00
Wyatt Alt	a47c22b26e	feat(rust): declare computed columns through add_columns A computed column is a column defined by a SQL expression rather than by values supplied at write time, so it is added through add_columns like any other: table.add_columns().computed("doubled", "x * 2").execute().await? Declaring does not compute. The column is committed carrying its expression in field metadata but no data, which makes declaration cost the same on a large table as on an empty one and leaves a single code path that ever produces values. A later refresh fills it. The expression is the whole definition: both the result type and the input columns are derived from it with lance-datafusion's planner, so a caller writes neither, and the two can never disagree the way a hand-declared input list can. Everything statically knowable is rejected at declare time rather than deferred: an expression that does not parse, one referencing a column that does not exist, a name already in use, and the same name declared twice in one call. The binding lives in three field metadata keys: virtual_column marks the column, virtual_column.expression holds it, and virtual_column.inputs holds the parsed inputs as a JSON array. computed_columns() reads them back off a schema, the way a SQL catalog reports a generation expression as another column of its information schema, so introspection needs no round trip. A transform and computed columns cannot be combined in one call, since they commit through different transforms and could half-apply. Only functions the query engine already knows can be named; resolving a user-defined one needs a planner aware of the function registry, which does not exist yet.	2026-08-12 18:48:48 -07:00
LanceDB Robot	6fb976cf89	chore: update lance dependency to v11.0.0-beta.6 (#3922 ) Updates the Rust workspace Lance dependencies and Java lance-core dependency to v11.0.0-beta.6. Includes compatibility updates for the new concrete Lance file-version API. Trigger: https://github.com/lance-format/lance/releases/tag/v11.0.0-beta.6 --------- Co-authored-by: XYZhan <zhaner08@hotmail.com>	2026-08-12 02:43:44 -04:00
Sravan Avvaru	a615306f39	feat(python): add on_transform_error fault tolerance to StreamingDataset (#3763 ) Closes #3704 ## Problem Transforms can fail on bad data (e.g. nulls/NaNs from incomplete user surveys). Today any transform exception aborts iteration, and there is no way to skip invalid rows during loading. ## Solution New `on_transform_error` parameter on `StreamingDataset`: - `"raise"` (default, matches current behavior and the convention in tf.data / WebDataset / Ray Data) - `"skip"` — drop the failing rows and continue - `"warn"` — like skip, plus a logged warning per failing batch - a WebDataset-style callable `handler(exc) -> bool`, so users can skip only expected error types Key design points: - Row-granular skipping: when a batch fails, the transform is re-run on single-row slices so only the rows that actually fail are dropped (avoids Ray-style whole-block loss). Skips are counted in a new `rows_skipped` property. - No crash on uneven skips: the round-robin loop now ends the epoch at the last cycle where every split still has a row, instead of hitting `IndexError` when a split runs dry early. - Exact resumability under skips: checkpoints are now position-based. `state_dict` gains `positions_consumed_per_split` (exact for owned splits), and a new `merge_state_dicts` static method combines per-rank states via elementwise max for elastic resume across topology changes. Old checkpoints without the new key still load. Positions equal sample counts when nothing is skipped, so existing behavior is unchanged. - Guardrail: transforms returning the wrong number of rows now raise a clear `ValueError` instead of silently corrupting split accounting. ### Answers to the issue's open questions - Can we do this? Yes — all transforms funnel through one guarded call in the Stage 2 pipeline. - What do other libraries do? tf.data `ignore_errors()`, WebDataset `handler=`, Ray `max_errored_blocks`; MosaicML StreamingDataset offers nothing (skipping conflicts with its determinism model). This design follows the common conventions: raise by default, opt-in skipping, count/log drops. - Error handling or pre-filtering? Both: the existing `filter=` remains the recommended tool for predictable bad data (splits are built post-filter, so all guarantees hold — now documented); `on_transform_error` covers failures not expressible as a predicate. - Impact on splits / elastic determinism? Per-split sample sequences stay deterministic (skips are data-dependent, not topology-dependent). With unequal bad-row counts across splits the last few global steps of an epoch can differ across topologies (bounded by the skew), which is documented on the parameter. With equal counts per split, full determinism is preserved — covered by a test. ## Testing 15 new tests in `test_elastic_dataloader.py` covering: default raise, invalid values, uniform and uneven skips (including epoch-end truncation), warn logging, selective callable handlers, wrong-row-count guardrail, determinism across runs and across world sizes (1/2/3/4) with skips, exact mid-epoch resume with skips on the same topology, elastic resume via `merge_state_dicts` (ws=2 → ws=1), merge validation, and backward-compat loading of old checkpoints. Note: relying on CI for the test run — my local machine OOMs during the final link of the native extension. The change itself is pure Python. --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com>	2026-08-10 09:22:06 -07:00
Xuanwo	920fc0e455	fix(python): set native module metadata (#3913 ) PyO3 defaults native extension classes to `builtins`, so mkdocstrings/Griffe could not resolve the newly documented `lancedb.Session` alias and `Deploy docs to Pages` failed on `main`. Declare the extension module for the public native types referenced by the Python API docs so Griffe resolves them through `lancedb._lancedb` and Pages can build again. Validated with the docs toolchain used by CI (`griffe==0.49.0`, `mkdocstrings==0.25.2`, and `mkdocs==1.6.1`); `PYTHONPATH=. mkdocs build` succeeds.	2026-08-10 21:40:31 +08:00
Xuanwo	5acce6782e	ci(docs): report link checker failures through issues (#3909 )	2026-08-10 15:08:36 +08:00