rust/lancedb
1
0
Fork 0
mirror of https://github.com/lancedb/lancedb.git synced 2026-06-03 04:10:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: remove primary key constraint from MemWAL bucket sharding (#3435)

Browse Source 
## Summary

- Bump lance dependency from `v7.0.0-beta.13` to `v7.0.0-rc.1`
- Remove PK constraint from `LsmWriteSpec::Bucket` docs and
`Table::set_lsm_write_spec` docs
- Remove test assertions that expected rejection when no PK is set or
when bucket column != PK

Closes https://github.com/lance-format/lance/issues/6917
This commit is contained in:
Jack Ye
2026-05-26 17:35:28 -07:00
committed by GitHub
parent 7dba793629
commit a7d9f2e99d
6 changed files with 82 additions and 108 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
rust/lancedb/src/table.rs
View File

@@ -312,17 +312,15 @@ pub use self::merge::MergeResult;
/// date) and [`LsmWriteSpec::with_writer_config_defaults`] (default
/// `ShardWriter` configuration recorded in the MemWAL index).
///
/// All variants require the table to have an unenforced primary key.
///
/// 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.
#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
pub enum LsmWriteSpec {
/// Hash-bucket sharding by the unenforced primary key column.
/// Hash-bucket sharding by a scalar column.
///
/// `column` must equal the table's currently-set single-column
/// unenforced primary key. `num_buckets` must be in `[1, 1024]`.
/// `column` must be a non-nested column with a supported scalar type.
/// `num_buckets` must be in `[1, 1024]`.
/// Iceberg-compatible Murmur3-x86-32 (seed 0) is used so each row's
/// `bucket(column, num_buckets)` value is stable across processes.
Bucket {
@@ -1360,21 +1358,15 @@ impl Table {
///
/// [`LsmWriteSpec`] chooses one of three sharding strategies:
///
/// - [`LsmWriteSpec::bucket`] — hash-bucket writes by the single-column
/// unenforced primary key.
/// - [`LsmWriteSpec::bucket`] — hash-bucket writes by a scalar column.
/// - [`LsmWriteSpec::identity`] — shard by the raw value of a scalar column.
/// - [`LsmWriteSpec::unsharded`] — route every write to a single shard.
///
/// All variants require the table to have an unenforced primary key
/// ([`Table::set_unenforced_primary_key`]); bucket sharding additionally
/// requires it to be the single column being bucketed.
///
/// # Example
///
/// ```
/// # use lancedb::table::{LsmWriteSpec, Table};
/// # async fn example(table: &Table) -> Result<(), Box<dyn std::error::Error>> {
/// table.set_unenforced_primary_key(["id"]).await?;
/// table
/// .set_lsm_write_spec(
/// LsmWriteSpec::bucket("id", 16).with_maintained_indexes(["id_idx"]),
@@ -4661,21 +4653,6 @@ mod tests {
.unwrap();
let table = conn.create_table("t", reader).execute().await.unwrap();
// Reject when no PK is set.
let err = table
.set_lsm_write_spec(LsmWriteSpec::bucket("id", 4))
.await
.expect_err("should reject without PK");
assert!(matches!(err, Error::Lance { .. }), "got {:?}", err);
// Set PK, then a mismatched column on the spec must be rejected.
table.set_unenforced_primary_key(["id"]).await.unwrap();
let err = table
.set_lsm_write_spec(LsmWriteSpec::bucket("name", 4))
.await
.expect_err("should reject column != PK");
assert!(matches!(err, Error::Lance { .. }), "got {:?}", err);
// Reject num_buckets out of range.
for bad in [0u32, 1025] {
let err = table
@@ -4741,9 +4718,6 @@ mod tests {
.unwrap();
let table = conn.create_table("t", reader).execute().await.unwrap();
// Lance's MemWAL still requires *some* unenforced primary key on
// the dataset; Unsharded just skips the per-row hashing step.
table.set_unenforced_primary_key(["id"]).await.unwrap();
table
.set_lsm_write_spec(LsmWriteSpec::unsharded())
.await
@@ -4790,7 +4764,6 @@ mod tests {
.unwrap();
let table = conn.create_table("t", reader).execute().await.unwrap();
table.set_unenforced_primary_key(["id"]).await.unwrap();
table
.set_lsm_write_spec(
LsmWriteSpec::identity("region")
@@ -4846,7 +4819,6 @@ mod tests {
table.unset_lsm_write_spec().await.unwrap_err();
// Install a spec, then unset it.
table.set_unenforced_primary_key(["id"]).await.unwrap();
table
.set_lsm_write_spec(LsmWriteSpec::bucket("id", 4))
.await