Eventual-Inc/Daft - CodSpeed

Daft

Blog Docs Changelog

Performance History

Latest Results

fix(parquet): Use chunk size for ArrayReader buffer size (#7149) ## Changes Made Addressing a comment from https://github.com/Eventual-Inc/Daft/pull/7141 regarding the buffer size that's passed to the Parquet Arrow array readers. Instead of using the size of the row group, we should use the chunk size that we already have laying around.

main

2 hours ago

address greptile

slade/fix-batch-sizing

2 hours ago

chore: upgrade arrow-rs from 57.1.0 to 59.0.0 (#7141) ## Summary - Upgrade all arrow-rs workspace dependencies from 57.1.0 to 59.0.0 - Update Cargo.lock with surgical arrow/parquet-only dependency resolution - Fix daft-parquet API compatibility with arrow-rs 59.0.0: - Add batch_size parameter to PrimitiveArrayReader::new, NullArrayReader::new, make_byte_view_array_reader, make_byte_array_reader, and make_fixed_len_byte_array_reader calls - Update LogicalType::Timestamp and LogicalType::Decimal pattern matching from struct variants to tuple variants (wrapping TimestampType and DecimalType) - Update daft-sketch serde_arrow feature from arrow-57 to arrow-59 - Add arrow-59 feature to daft-ext extension SDK for FFI compatibility ## Motivation arrow-rs 59.0.0 includes an architecture-aware size assertion for the parquet-variant Variant enum, which fixes a compile-time failure on s390x (IBM Z). The previous assertion expected 80 bytes on all 64-bit platforms, but s390x produces 72 bytes due to different alignment/padding rules on big-endian architectures. ## Test plan - [ ] CI passes on all platforms - [ ] Existing arrow-dependent tests (daft-sketch, daft-parquet, daft-ext) pass - [ ] s390x cross-compilation succeeds without parquet-variant patching Signed-off-by: Mike DePaulo <mdepaulo@redhat.com> Signed-off-by: Mike DePaulo <mikedep333@redhat.com>

main

3 hours ago

fix(iceberg): pass ignore_corrupt_files through table reads

jackylee-ch:codex-iceberg-table-ignore-corrupt-files

13 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

19 hours ago

fix type hint and style.

refactor--embed_text-public-api-to-delegate-expression-building-to-providers

1 day ago

perf(grouped-agg): bump NUM_SHARDS_PER_MORSEL to 8 (empirically tuned)

BABTUNA:perf/sharded-grouped-agg

2 days ago

fix(flotilla): honor explicit num_cpus=0 in autoscaler bundles aggregate_ray_bundles forced CPU >= 1 (via .max(1) on individual bundles and a fixed CPU:1 on every GPU bundle), so a task that explicitly sets num_cpus=0 still requested a CPU — breaking "explicit num_cpus passes through unchanged" and over-requesting CPU for GPU-only / memory-only workloads. Drop the .max(1); give GPU bundles CPU only when the packed tasks actually need it (gpu_cpu_sum > 0); and omit the CPU key from the Ray bundle dict when it is zero. Add a test for num_cpus=0 GPU-only and memory-only tasks.

XiaoHongbo-Hope:fix/min-cpu-per-task-wiring

2 days ago

Latest Branches

0%

fix(parquet): Use chunk size for ArrayReader buffer size#7149

3 hours ago

0a740cd

slade/fix-batch-sizing

0%

fix(iceberg): pass ignore_corrupt_files through table reads#7147

15 hours ago

e4ce31d

jackylee-ch:codex-iceberg-table-ignore-corrupt-files

0%

feat(sql): support read_parquet ignore_corrupt_files#7133

5 days ago

c19a075

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

© 2026 CodSpeed Technology

Home Terms Privacy Docs