Oct 2026

1 commit
ssl: Don't disable SSL for all hosts when one SNI host fails to load
26 Oct 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19rc2
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #871 of 1 reviewer saw the final version

Takes the first of three patches in #87, preserving failed host entries and rejecting only their handshakes. Differences are test formatting and adaptation to the existing two-argument init_host_context(); the two startup-rejection patches are absent.

  • Daniel Gustafssonsaw the final versiontrailerHe acknowledged the report in #88 and, as committer, reviewed the committed version.

Oct 2026

5 commits
Delay creation of SSL_CTX structure to allow cleanup
6 Oct 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19rc1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #812 of 2 reviewers saw the final version

Takes the fifth patch from #81, delaying init_host_context() until the host is reachable by cleanup. The only changed call arguments adapt to the existing function signature; the attachment explicitly credits Zsolt Parragi.

  • Daniel Gustafssonsaw the final versiontrailerHe explained and posted Zsolt's allocation-order fix in #81 and reviewed its final form as committer.
  • Rui Zhaosaw the final versiontrailerHe explicitly reviewed all five patches and reported no further problems in #82.
Copy SSL GUCs into the hosts context
6 Oct 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19rc1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #812 of 2 reviewers saw the final version

Takes the fourth patch from #81: all added and removed lines copying SSL GUC strings with pstrdup() match. The attachment credits Zsolt Parragi, whose off-list contribution Daniel describes in the submission.

  • Daniel Gustafssonsaw the final versiontrailerHe explained the dangling-GUC-pointer problem and posted the fix in #81, then committed it.
  • Rui Zhaosaw the final versiontrailerHis explicit review of all five patches in #82 includes this GUC-lifetime fix.
Clear SSL_hosts when destroying TLS
6 Oct 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19rc1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #811 of 2 reviewers saw the final version

Takes the third patch from #81 unchanged: seven lines in be_tls_destroy() delete SSL_hosts_memcxt and clear its pointers; the constituent patch and commit have identical stable patch IDs.

  • Zsolt Parragino review foundtrailerAlthough credited in the attachment and commit, his visible reviews in #77 and #79 concern the reload-state fix, with no specific review of this cleanup shown.
  • Rui Zhaosaw the final versiontrailerHe reviewed all five patches, including TLS destruction cleanup, in #82.
Use safe_psql in SSL test setup to avoid silent failure
6 Oct 2026Daniel Gustafssonmasterunreleased
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #811 of 2 reviewers saw the final version

Takes the second patch from #81 unchanged, replacing ten psql() calls with safe_psql() in src/test/ssl/t/SSL/Server.pm; stable patch IDs match.

  • Zsolt Parragino review foundtrailerThe attachment credits him, but the supplied discussion contains no identifiable review by him of the safe_psql change in #81.
  • Rui Zhaosaw the final versiontrailerHe explicitly included all five patches in his favorable review in #82.
Keep current ssl_sni setting on SSL config reload failure
6 Oct 2026Daniel GustafssonmasterunreleasedREL_19_STABLE19rc1
LLM analysis, 97% suremodified before commitlight changesfrom patchset #841 of 3 reviewers saw the final version

Combines the reload-state fix from #81 with the latest test add-on #84, with only comment/test formatting changes to that combined work. The commit credits Anthropic OSS program and co-authors Daniel Gustafsson and Rui Zhao; the external author has no message in this thread, and Noah identifies its contribution in #74.

  • Zsolt Parragisaw an earlier versiontrailerHe found the null-dereference and rollback issues in #77 and the initialization/passphrase regression in #79, before the subsequent local logic corrections.
  • Jacob Championsaw an earlier versiontrailerHis review in #80 challenged the initialization-time state checks and led to their correction in #81.
  • Rui Zhaosaw the final versiontrailerHe reviewed and tested the corrected series in #82, then supplied the additional regression tests finalized in #84.

May 2026

1 commit
Declare load_hosts() as returning HostsFileLoadResult.
4 May 2026Tom Lanemaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surelanded as postedfrom patchset #691 of 1 reviewer saw the final version

The four-file typedef relocation and load_hosts() return-type cleanup exactly match #69, including an identical stable patch ID.

  • Daniel Gustafssonsaw the final versionLLM analysis, 100% sureHe explicitly raised no objection to the attached fix and discussed the declaration placement in #71, before the commit.

Mar 2026

3 commits
ssl: Skip passphrase reload tests in EXEC_BACKEND builds
18 Mar 2026Daniel Gustafssonmaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #551 of 1 reviewer saw the final version

Includes only the 004_sni.pl EXEC_BACKEND skip from #55, with an expanded skip message and formatting. The proposed fe-secure-openssl.c handshake-error change was omitted.

  • Jacob Championsaw the final versionLLM analysis, 99% sureHe diagnosed the landed tests' missing EXEC_BACKEND condition and proposed the precise skip correction in #54, before the fix was committed.
ssl: Serverside SNI support for libpq
18 Mar 2026Daniel Gustafssonmaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #522 of 6 reviewers saw the final version

Takes the complete SNI feature constituent of the two-patch series in #52 unchanged, with an identical stable patch ID; the independent client-CA tests landed separately.

  • Jacob Championsaw the final versiontrailerHe investigated fallback and CA isolation throughout the thread, co-developed the redesign in #45, and supplied the final CA/CRL/hook tests incorporated in #50.
  • Zsolt Parragisaw the final versiontrailerHe found parser and TLS-alert bugs in #48, then broadly approved the patch with minor defensive and warning refinements in #51.
  • Chao Lisaw an earlier versiontrailerHis reviews in #19 and #24 covered const correctness, memory ownership, configuration and documentation before the substantial redesigns.
  • Dewei Daisaw an earlier versiontrailerHe identified uninstall cleanup, duplicated context destruction and wording issues in #26, before the configuration and SSL switching redesigns.
  • Cary Huangsaw an earlier versiontrailerHis initial review in #2 requested per-host passphrase commands and questioned strict-mode handling of clients without SNI.
  • Heikki Linnakangassaw an earlier versiontrailerHe proposed explicit wildcard/non-SNI entries in #28 and supported the boolean enable switch in #38, before the final SSL implementation rewrite.
ssl: Add tests for client CA
18 Mar 2026Daniel Gustafssonmaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% surepart of a posted patchsetlight changesfrom patchset #52

Takes the client-CA test constituent of #52 unchanged, with an identical stable patch ID. Its author credit identifies Jacob Champion, whose original CA tests appear in #13, with Daniel Gustafsson as co-author.

Credits across 10 commits

Author
Daniel Gustafsson2 LLM4
Zsolt Parragi3
Anthropic OSS program1
Jacob Champion1
Tom LaneLLM1
Reviewer
Rui Zhao5
Zsolt Parragi4
Daniel Gustafsson3
Jacob Champion2
Cary Huang1
Chao Li1
Dewei Dai1
Heikki Linnakangas1
Reported by
Jacob Champion2
Anthropic OSS program1
Co author
Daniel Gustafsson2
Jacob Champion1
Rui Zhao1
Committer
Daniel Gustafsson9
Tom Lane1

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