Jan 2019

5 commits
Remove _configthreadlocale() calls in ecpg test suite.
24 Jan 2019Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surewritten from the discussion, no matching patch

Implements Tom Lane's proposal in #33 to remove application-side _configthreadlocale() calls now that ecpglib handles them internally. The commit changes five thread test programs and their generated expected C files; none of the attached patchsets changes tests.

Remove infinite-loop hazards in ecpg test suite.
24 Jan 2019Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+6 more
LLM analysis, 100% surewritten from the discussion, no matching patch

Following Andrew Dunstan's hang report in #28 and Tom Lane's loop analysis in #32, this adds 100-iteration limits to nine ecpg test programs and updates their expected C files. No attached patchset contains these safeguards.

Blind attempt to fix _configthreadlocale() failures on MinGW.
24 Jan 2019Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+6 more
LLM analysis, 100% surewritten from the discussion, no matching patch0 of 1 reviewer saw the final version

Implements the MinGW fallback proposed in #31, allowing execution when _configthreadlocale() returns -1, with the push confirmed in #32. It also moves thread-locale restoration outside the oldlocale check in execute.c, addressing the patch defect identified in #25; these changes have no matching attachment.

  • Tsunakawa, Takayukisaw an earlier versionLLM analysis, 90% sureIn #25, he examined #14 and identified missing thread-locale restoration after initialization failure, which this commit fixes alongside additional MinGW error-handling changes.
Second try at fixing ecpglib thread-safety problem.
21 Jan 2019Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+1 more
LLM analysis, 100% surewritten from the discussion, no matching patch

Responds to the MinGW build failure reported in #18 and implements Tom Lane's configure-probe decision in #22. It adds HAVE__CONFIGTHREADLOCALE detection and replaces WIN32 guards in ecpglib; none of the attached patchsets contains this follow-up.

Avoid thread-safety problem in ecpglib.
21 Jan 2019Tom Lanemaster12beta1REL_19_STABLE12beta1REL_18_STABLE12beta1REL_17_STABLE12beta1REL_16_STABLE12beta1REL_15_STABLE12beta1REL_14_STABLE12beta1REL_13_STABLE12beta1+6 more
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #141 of 2 reviewers saw the final version

Matches #14 in the configuration files, statement fields, and execute.c logic, with a comment adjustment and an additional locale-handling conversion in descriptor.c that Tom Lane explains in #16. The commit credits Michael Meskes and Tom Lane; Michael supplied the original implementation in #2.

  • Tsunakawa, Takayukisaw an earlier versionLLM analysis, 98% sureHe proposed the Windows implementation after #10 in #11 and explicitly reviewed both platform paths of #14 in #15, before the additional descriptor.c conversion.
  • Tom Lanesaw the final versioncommittercommitted the patch

Credits across 5 commits

Author
Michael MeskesLLM1
Reviewer
Tom Lanecommitter1
Committer
Tom Lane5

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