Dec 2018

4 commits
Use pg_strong_random() to select each server process's random seed.
29 Dec 2018Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% suremodified before commitlight changesfrom patchset #64

The postmaster.c changes match #64; the commit additionally documents early-startup constraints in src/port/pg_strong_random.c, switches its include to c.h, and makes the filename argument const.

Use a separate random seed for SQL random()/setseed() functions.
29 Dec 2018Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% suremodified before commitlight changesfrom patchset #63

The float.c implementation matches #63 exactly; the only substantive textual difference is describing pgcrypto as a 'more secure alternative' in func.sgml.

Marginal performance hacking in erand48.c.
28 Dec 2018Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 99% surewritten from the discussion, no matching patch

No attached patchset contains this erand48.c optimization; #62 proposes uint64 multiply-and-add coding inline, following Fabien COELHO's observations in #52. The commit implements that approach and credits Fabien for the idea, not for reviewing a posted patch.

Fix latent problem with pg_jrand48().
28 Dec 2018Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+3 more
LLM analysis, 100% surewritten from the discussion, no matching patch

Tom Lane identifies pg_jrand48() returning unsigned values on platforms with 64-bit long in #56; the commit adds the int32 cast and documentation in src/port/erand48.c without a matching attached patchset.

Nov 2018

2 commits
Silence compiler warning
30 Nov 2018Alvaro Herreramaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surelanded as postedfrom patchset #331 of 1 reviewer saw the final version

The postgres.c diff exactly matches #33, assigning in_sample through exceeded && ... to silence the compiler warning while preserving conditional random() evaluation.

  • Sergei Kornilovsaw the final versionLLM analysis, 100% sureAfter reporting the warning, he tested #33 with GCC 4.9.2 and confirmed a clean build in #34 before the commit.
Add log_statement_sample_rate parameter
29 Nov 2018Alvaro Herreramaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #271 of 3 reviewers saw the final version

Uses the complete feature from #27, with documentation rewording and an added if (exceeded) guard around in_sample computation in postgres.c; Alvaro Herrera explicitly describes both edits in #28.

  • David Rowleysaw an earlier versiontrailerIn #2 he identified the comparison boundary issue and proposed bypassing random() at rates zero and one, before the sampling expression was restructured and conditionally evaluated.
  • Vik Fearingsaw an earlier versiontrailerHe supplied a readability and documentation revision in #6 and marked #20 ready in #21, before the committer added conditional evaluation of in_sample.
  • Alvaro Herrerasaw the final versioncommittercommitted the patch

Credits across 6 commits

Author
Tom LaneLLM2
Adrien Nayrat1
Alvaro HerreraLLM1
Reviewer
Alvaro Herreracommitter1
David Rowley1
Vik Fearing1
Reported by
Sergei Kornilov1
Committer
Tom Lane4
Alvaro Herrera2

N LLM means N of the credits come from LLM analysis, not from commit trailers.