Mar 2025

1 commit
Add support for gamma() and lgamma() functions.
26 Mar 2025Dean Rasheedmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #151 of 5 reviewers saw the final version

Compared with #15, the commit only clarifies the dlgamma() comment about pole errors in src/backend/utils/adt/float.c and bumps src/include/catalog/catversion.h, apart from rebase metadata. The clarification was requested in #16 and confirmed committed in #17; executable logic, documentation, and tests match.

  • Stepan Neretinsaw an earlier versiontrailerIn #4, he reviewed #1, said it looked good, and requested more informative range errors; substantial error-handling changes followed.
  • Peter Eisentrautsaw an earlier versiontrailerIn #7, he tested removing a NaN check and questioned special-value handling, overflow semantics, and thread safety in #1, before subsequent logic revisions.
  • Tom Lanesaw an earlier versiontrailerIn #9 and #10, he reviewed the rationale behind #8, advocated relying on POSIX behavior, and explained why ignoring signgam avoids the thread-safety concern; later versions changed error semantics.
  • Dmitry Kovalsaw an earlier versiontrailerIn #13, he questioned the inconsistent treatment of zero and near-zero inputs in #12, prompting the later error-policy revision in #15.
  • Alexandra Wangsaw the final versiontrailerIn #16, she approved and tested #15, reported libm behavior on Ubuntu, macOS, and Windows, and requested the pole-error comment clarification included in the commit.

Credits across 1 commit

Author
Dean Rasheed1
Reviewer
Alexandra Wang1
Dmitry Koval1
Peter Eisentraut1
Stepan Neretin1
Tom Lane1
Committer
Dean Rasheed1