Eventual-Inc/Daft - CodSpeed

Daft

Blog Docs Changelog

Performance History

Latest Results

Merge remote-tracking branch 'upstream/main' into codex-sql-read-parquet-ignore-corrupt-files

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

1 hour ago

Merge remote-tracking branch 'upstream/main' into codex-sql-read-parquet-ignore-corrupt-files

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

3 hours ago

Merge remote-tracking branch 'upstream/main' into codex-sql-read-parquet-ignore-corrupt-files

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

4 hours ago

chore(temporal): address Greptile review on unix extractors Sort the new from .datetime imports alphabetically to match the __all__ list. Add Expression method forms (unix_seconds, unix_millis, unix_micros, unix_timestamp, to_unix_timestamp, weekday) so the new functions are reachable via method-chaining like every other temporal extractor.

BABTUNA:feat/temporal-unix-extractors

6 hours ago

Merge remote-tracking branch 'upstream/main' into codex-sql-read-parquet-ignore-corrupt-files

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

6 hours ago

fix(scan): populate per-column sizes when splitting parquet by row groups Follow-up to #6542. The size estimator uses per-column uncompressed sizes from the Parquet footer (TableMetadata.column_sizes) when available, but those were only captured for the first file in a glob. split_by_row_groups already reads each file's full footer at planning time, so populate column_sizes there too — giving accurate, projection-aware estimates for every split task at zero extra I/O. This also fixes a latent over-estimate: previously a split of the first file reused the whole-file column_sizes and only updated `length`, so each split estimated the entire file (N splits => ~Nx the real size). We now recompute per-column sizes for exactly the row groups in each split, and attach an accurate row count for every file (not just the first). Adds a Ray regression test asserting the total estimated scan bytes tracks the files' uncompressed size, rather than being multiplied by the number of splits or blown up by parquet_inflation_factor. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

varun/scan-split-column-sizes

6 hours ago

fix typo in src/daft-distributed/src/pipeline_node/shuffles/backends/mod.rs Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

Liusixuuu:fix/into-partitions-flight-shuffle-panic

6 hours ago

Merge remote-tracking branch 'upstream/main' into codex-sql-read-parquet-ignore-corrupt-files

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

7 hours ago

Latest Branches

0%

feat(sql): support read_parquet ignore_corrupt_files#7133

2 hours ago

49207d1

jackylee-ch:codex-sql-read-parquet-ignore-corrupt-files

0%

feat(temporal): add Spark-style unix extractors and weekday#6920

1 month ago

30716e9

BABTUNA:feat/temporal-unix-extractors

0%

fix(scan): populate per-column sizes when splitting parquet by row groups#7155

7 hours ago

3e3a3b9

varun/scan-split-column-sizes

© 2026 CodSpeed Technology

Home Terms Privacy Docs