Issues

docs: add more engines to user guide

Task

lance-format/lance#10536 · opened

on Oct 30, 2026

bug(docs): headings are confusing on integrations pages

Bug

lance-format/lance#10535 · opened

on Oct 30, 2026

perf: page faults on fresh decode buffers cost as much CPU as decoding large columns

lance-format/lance#10529 · opened

on Oct 30, 2026

perf: take_rows opens a scheduler per fragment, so its I/O concurrency is fragments x LANCE_IO_THREADS

lance-format/lance#10528 · opened

on Oct 30, 2026

perf: zstd level 1 beats the default level 3 on the per-value path for large text values

lance-format/lance#10527 · opened

on Oct 30, 2026

perf: writer emits one page per inserted array for large per-value values (multi-GB pages, single-threaded compression)

lance-format/lance#10526 · opened

on Oct 30, 2026

RTree index build and optimize_indices fail on a geoarrow.point column with a CRS: "field bbox does not match the file schema"

lance-format/lance#10523 · opened

on Oct 30, 2026

delta(begin_version, end_version) reads rows from the handle's version instead of end_version

lance-format/lance#10517 · opened

on Oct 30, 2026

delta().get_inserted_rows() and get_updated_rows() return no rows on a dataset without stable row ids

lance-format/lance#10516 · opened

on Oct 30, 2026

create_scalar_index commits over a concurrent drop_columns, and every filtered read then fails

lance-format/lance#10515 · opened

on Oct 30, 2026

LanceDataset.merge picks an arbitrary right-side row for a duplicate key, and the pick changes from run to run

lance-format/lance#10514 · opened

on Oct 30, 2026

Write data, deletion and index files with If-None-Match: *, so a store can refuse overwriting a committed file

lance-format/lance#10513 · opened

on Oct 30, 2026

1

drop_columns and column renames commit over a concurrent schema or field metadata update and discard it

lance-format/lance#10510 · opened

on Oct 30, 2026

Column groups: keep chosen columns in their own data files, across writes and compaction

lance-format/lance#10509 · opened

on Oct 30, 2026

A BTREE filter misses rows after compact_files() follows a deferred remap

lance-format/lance#10506 · opened

on Oct 30, 2026

A concurrent create_branch that fails still writes its version into the branch

lance-format/lance#10503 · opened

on Oct 30, 2026

cleanup_old_versions deletes a version tagged while it runs

lance-format/lance#10499 · opened

on Oct 30, 2026

An append concurrent with an alter_columns cast commits with that column NULL

lance-format/lance#10498 · opened

on Oct 30, 2026

bug: FTS pagination repeats and omits matches with explicit column ordering

lance-format/lance#10496 · opened

on Oct 30, 2026

What should asof=t return once the version current at t is cleaned up?

lance-format/lance#10493 · opened

on Oct 30, 2026

A scalar index on the key makes merge_insert drop source rows with a NULL key

lance-format/lance#10491 · opened

on Oct 30, 2026

cleanup_old_versions(retain_versions=1) panics on a handle to a deleted branch

lance-format/lance#10490 · opened

on Oct 30, 2026

DirectoryNamespace drop_table(["a$b"]) deletes the table ["a", "b"]

lance-format/lance#10487 · opened

on Oct 30, 2026

branches.delete("a") leaves branch a readable and writable while "a/b" exists

lance-format/lance#10484 · opened

on Oct 30, 2026

feature: carry clustering_id and metadata through Transaction V2 fragment and index actions

lance-format/lance#10472 · opened

on Oct 29, 2026

Navigation Menu

Repository navigation

Issues

docs: add more engines to user guide

bug(docs): headings are confusing on integrations pages

perf: page faults on fresh decode buffers cost as much CPU as decoding large columns

perf: take_rows opens a scheduler per fragment, so its I/O concurrency is fragments x LANCE_IO_THREADS

perf: zstd level 1 beats the default level 3 on the per-value path for large text values

perf: writer emits one page per inserted array for large per-value values (multi-GB pages, single-threaded compression)

RTree index build and optimize_indices fail on a geoarrow.point column with a CRS: "field bbox does not match the file schema"

delta(begin_version, end_version) reads rows from the handle's version instead of end_version

delta().get_inserted_rows() and get_updated_rows() return no rows on a dataset without stable row ids

create_scalar_index commits over a concurrent drop_columns, and every filtered read then fails

LanceDataset.merge picks an arbitrary right-side row for a duplicate key, and the pick changes from run to run

Write data, deletion and index files with If-None-Match: *, so a store can refuse overwriting a committed file

drop_columns and column renames commit over a concurrent schema or field metadata update and discard it

Column groups: keep chosen columns in their own data files, across writes and compaction

A BTREE filter misses rows after compact_files() follows a deferred remap

A concurrent create_branch that fails still writes its version into the branch

cleanup_old_versions deletes a version tagged while it runs

An append concurrent with an alter_columns cast commits with that column NULL

bug: FTS pagination repeats and omits matches with explicit column ordering

What should asof=t return once the version current at t is cleaned up?

A scalar index on the key makes merge_insert drop source rows with a NULL key

cleanup_old_versions(retain_versions=1) panics on a handle to a deleted branch

DirectoryNamespace drop_table(["a$b"]) deletes the table ["a", "b"]

branches.delete("a") leaves branch a readable and writable while "a/b" exists

feature: carry clustering_id and metadata through Transaction V2 fragment and index actions