Nov 2019

1 commit
Use memcpy instead of a byte loop in pglz_decompress
28 Nov 2019Tomas Vondramaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #634 of 4 reviewers saw the final version

The commit preserves the executable decompression changes from #63 in src/common/pg_lzcompress.c, reformats the overlap explanation and drops the unrelated pglz_maximum_compressed_size whitespace hunk. Tomas explicitly confirms these edits in #66.

  • Michael Paquiersaw the final versiontrailerMichael evaluated the benchmark conclusion and identified a comment typo in the unrestricted alternative in #59, after which only comment and formatting changes followed.
  • Konstantin Knizhniksaw the final versiontrailerKonstantin checked the memcpy algorithm's correctness and benchmarked competing implementations in #11; the committed unrestricted algorithm retains that approach with light loop-bound, expression and comment changes.
  • Telssaw the final versiontrailerTels inspected both alternatives and proposed batching literal-byte copies in #49, while the reviewed match-copy implementation reached the commit with only light subsequent changes.
  • Tomas Vondrasaw the final versioncommittercommitted the patch

Credits across 1 commit

Author
Andrey Borodin1
Reviewer
Konstantin Knizhnik1
Michael Paquier1
Tels1
Tomas Vondracommitter1
Committer
Tomas Vondra1