Jul 2024

4 commits
Require memory barrier support.
30 Jul 2024Thomas Munromaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1Also in: Build with LTO / -flto on macOS (not this thread)
LLM analysis, 100% surelanded as postedfrom patchset #231 of 1 reviewer saw the final version

The diff exactly matches the fourth patch in #23, removing pg_spinlock_barrier and memory-barrier emulation from atomics.c, fallback.h and generic.h while requiring pg_memory_barrier_impl in atomics.h. This supersedes the from_discussion attribution in thread #49720: that thread supplied the proposal, but its June attachments precede this matching July attachment, which explicitly incorporates that discussion.

  • Heikki Linnakangassaw the final versiontrailerHeikki approved the revised series in #24, and its memory-barrier patch landed without code changes.
Require compiler barrier support.
30 Jul 2024Thomas Munromaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1Also in: Build with LTO / -flto on macOS (not this thread)
LLM analysis, 100% surelanded as postedfrom patchset #231 of 1 reviewer saw the final version

The diff exactly matches the third patch in #23, removing pg_extern_compiler_barrier from src/backend/port/atomics.c and requiring compiler-barrier support in src/include/port/atomics.h. This supersedes the from_discussion attribution in thread #49720: that thread supplied the rationale, but its June attachments precede this matching July attachment, which explicitly incorporates that discussion.

  • Heikki Linnakangassaw the final versiontrailerHeikki approved the revised series in #24, and its compiler-barrier patch landed without code changes.
Remove --disable-atomics, require 32 bit atomics.
30 Jul 2024Thomas Munromaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #231 of 3 reviewers saw the final version

The diff matches the second patch in #23, removing --disable-atomics and flag/uint32 simulation while retaining the uint64 fallback.

  • Heikki Linnakangassaw the final versiontrailerHeikki approved the final attached series in #24.
  • Tom Lanesaw an earlier versiontrailerTom endorsed removing --disable-atomics in #22, before #23 changed the compilation check from recognizing a compiler to requiring an actual uint32 implementation.
  • Andres Freundsaw an earlier versiontrailerAndres proposed dropping flag/32-bit emulation while retaining 64-bit fallback in #10 and explained compare-exchange sufficiency in #12, before this series was attached.
Remove --disable-spinlocks.
30 Jul 2024Thomas Munromaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #232 of 3 reviewers saw the final version

The diff matches the first patch in #23, removing --disable-spinlocks, semaphore-based spinlock emulation, associated configuration, and regression-test branches.

  • Heikki Linnakangassaw the final versiontrailerHeikki approved the final attached series in #24.
  • Tom Lanesaw the final versiontrailerTom endorsed removing --disable-spinlocks in #22, and subsequent changes to this component were comment cleanup and rebasing.
  • Andres Freundsaw an earlier versiontrailerAndres advocated requiring spinlocks and explained the growing performance cost of semaphore emulation in #10 and #12, before an implementation was attached.

Jul 2024

1 commit
Remove support for HPPA (a/k/a PA-RISC) architecture.
1 Jul 2024Tom Lanemaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surelanded as postedfrom patchset #14 of 4 reviewers saw the final version

The HPPA removal matches #1 after rebasing for the copyright year and installation documentation; the omitted cpluspluscheck hunk targets a file already absent from the commit's parent.

  • Michael Paquiersaw the final versionLLM analysis, 100% sureMichael explicitly approved the C changes and asked whether config.guess/config.sub needed updating in #2.
  • Noah Mischsaw the final versionLLM analysis, 100% sureNoah challenged the removal rationale in #4, accepted the atomics-cleanup justification in #13, and explicitly supported removal in #19.
  • Thomas Munrosaw the final versionLLM analysis, 94% sureThomas explicitly supported the proposed HPPA removal with a +1 in #6.
  • Andres Freundsaw the final versionLLM analysis, 94% sureAndres supported removal by identifying HPPA's atomics-emulation burden in #8 and explaining the maintenance costs in #15.

Credits across 5 commits

Author
Thomas MunroLLM4
Tom LaneLLM1
Reviewer
Heikki Linnakangas4
Andres Freund2
Tom Lane2
Committer
Thomas Munro4
Tom Lane1

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