Eventual-Inc/Daft - CodSpeed

Daft

Blog Docs Changelog

Performance History

Latest Results

fix: Rename `daft.read_mcap` parameter `use_legacy_types` to `output_format` (#7625) ## Changes Made I want to rename the keyword parameter `use_legacy_types` to `output_format` in `daft.read_mcap`. This shouldn't be a breaking change since we haven't released the old parameter yet, and it allows us to potentially add a new parameter in the future such as a `output_format='decoded'`. --------- Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

main

42 minutes ago

everettVT/bump-webpki-roots

50 minutes ago

everettVT/bump-webpki-roots

2 hours ago

everettVT/bump-webpki-roots

2 hours ago

upgrade aws-lc-rs

everettVT/bump-webpki-roots

3 hours ago

fix: Bump webpki-roots to 1.0.10 for eMudhra CA compromise Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

everettVT/bump-webpki-roots

17 hours ago

fix(iceberg): honor write.parquet page-size, dict-size and page-row-limit properties

YuangGao:fix/iceberg-write-page-properties

21 hours ago

fix(core): compute decimal arithmetic with i256 and checked ops Bounding inferred precision at 38 lets previously rejected expressions plan, but evaluation still used unchecked i128 kernels. Two values of scale 20 square their raw representation to 10^40 when multiplied even though the answer fits, and 9e37 + 9e37 exceeds i128, so evaluation could panic or wrap to a wrong-signed value. Multiply and divide now compute through i256 with checked operations; add and subtract use checked i128. A result that does not fit the inferred Decimal128(38, s) becomes null instead of panicking or wrapping. Division inference also keeps at least the divisor's scale. The shared cast path casts the divisor to the result type, so a bounded scale below the divisor's would truncate its fraction and turn 1 / 0.5 into a division by zero. Integer digits give way instead; quotients that no longer fit decode to null through the checked kernel. Adds value-level regression tests through the Series cast + kernel path for large-scale intermediates, fractional divisors, and overflow.

hello-peter-tang:fix/7601-decimal-precision-overflow

2 days ago

Latest Branches

0%

fix: Bump webpki-roots to 1.0.10 for eMudhra CA compromise#7628

1 hour ago

3e82c8a

everettVT/bump-webpki-roots

0%

fix(iceberg): honor write.parquet page-size, dict-size and page-row-limit properties#7627

21 hours ago

307b0d9

YuangGao:fix/iceberg-write-page-properties

0%

fix(core): bound decimal arithmetic precision at 38 instead of failing to plan#7603

2 days ago

245e248

hello-peter-tang:fix/7601-decimal-precision-overflow

© 2026 CodSpeed Technology

Home Terms Privacy Docs