Snowflake-Labs/pg_lake v3.5.0

v3.5.0

Repository: Snowflake-Labs/pg_lake

Tag: v3.5.0

Published: 2026-09-03T17:18:21Z

Prerelease: no

Release notes:

What's Changed

• Push down string_agg to pgduck_server by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/440
• Fix/40 autovacuum logging content filter by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/444
• Add sinh, cosh, tanh pushdown support by @sfc-gh-amzheng in https://github.com/Snowflake-Labs/pg_lake/pull/438
• Fix oid type mapping: BIGINT in Parquet, 'long' in Iceberg metadata by @sfc-gh-amzheng in https://github.com/Snowflake-Labs/pg_lake/pull/443
• Eagerly remove orphaned write-through cache staging files by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/445
• Add CODEOWNERS file by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/452
• Update pytest_all.yml by @sfc-gh-nsaliaris in https://github.com/Snowflake-Labs/pg_lake/pull/453
• Revert CDWRegionTest environments; skip cross-region tests when creds absent by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/458
• fix: disallow copy unbounded numerics in nested types by @sfc-gh-abozkurt in https://github.com/Snowflake-Labs/pg_lake/pull/448
• Bump extension versions to 3.5 -- let's not forget these! by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/450
• Fix memory leaks during Iceberg manifest processing by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/454
• duckdb_pglake: bind size-clamp on multi-dimensional arrays by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/461
• Back off base worker restarts with an exponential delay after failures by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/447
• Disable the CSV sniffer when reading our own CSVs by column map by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/463
• pg_extension_base: keep default_version at 3.5 instead of bumping to 3.6 by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/464
• Fix flaky test_sequences_serial_and_generated_columns[rest] by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/466
• pg_lake_engine: precompute per-column DuckDB serialization choice by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/476
• pg_lake_engine: hoist element container check out of array output loop by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/479
• pgduck_server: fix typo and stray newlines in socket error messages by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/469
• pgduck_server: free the logical type alias in blob_to_text by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/472
• pg_lake_iceberg: fix off-by-one overflow on partition field names by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/467
• pgduck_server: free partial output buffer on nested conversion errors by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/473
• pgduck_server: stop leaking the per-connection cancellation token by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/471
• pg_lake_table: reset errno before parsing column_stats_mode length by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/468
• pg_lake_engine: don't leak the PGresult when COPY stats parsing throws by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/474
• pgduck_server: free query result when sending CopyDone fails by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/470
• Scale Iceberg TRUNCATE metadata cleanup by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/457
• pg_lake_table: hoist loop-invariant lookups out of data file pruning by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/477
• Document file descriptor limits for partitioned writes by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/487
• pgduck: never re-send a COPY (data-file write) on send failure by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/485
• Support iceberg schema type change by @sfc-gh-abozkurt in https://github.com/Snowflake-Labs/pg_lake/pull/223
• pg_extension_base: fix flaky oneshot-worker tests (poll instead of fixed sleep) by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/475
• Remove redundant pre-commit stage from pgindent hook by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/491
• pg_lake_table: mark fully matched data files with a linear merge by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/478
• pg_lake_engine: cap the CSV read buffer DuckDB allocates by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/492
• Fix data race on PgLakeQueryListener active query state by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/390
• pg_lake_iceberg: leave dropped attributes out of Iceberg structs by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/501
• pg_extension_base: find control files via extension_control_path by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/496
• pg_lake_iceberg: skip dropped attributes in struct value serde by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/504
• pg_lake_iceberg: read object store catalog.json without the file cache by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/480
• pg_lake_iceberg: one-sided primitives for the stored Iceberg representation of a type by @sfc-gh-okalaci in https://github.com/Snowflake-Labs/pg_lake/pull/495
• test_common: bump Apache Polaris REST catalog to 1.7.0 by @sfc-gh-agedemenli in https://github.com/Snowflake-Labs/pg_lake/pull/508
• pg_lake_table: bound memory used by the pre-commit data file diff by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/511
• pg_lake_table: fix preload order for the test hideable extension by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/520
• pg_extension_base: wait for a deregistered base worker to exit by @sfc-gh-dachristensen in https://github.com/Snowflake-Labs/pg_lake/pull/505
• Make Azure append blob writes safe to retry by @sfc-gh-mslot in https://github.com/Snowflake-Labs/pg_lake/pull/514
• Fix nested timetz reads from Iceberg by @marknefedov in https://github.com/Snowflake-Labs/pg_lake/pull/482
• pg_extension_base: let attached workers inherit caller writability by @sfc-gh-mslot in...