[SCM] PostGIS branch master updated. 3.7.0beta1-175-g08f9832fb3

[git at osgeo.org]

This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "PostGIS".

The branch, master has been updated
       via  08f9832fb36e066a57e53f285cf33914e9061aaf (commit)
       via  9067ee7ba05ea014c12f40582715339bd11043f2 (commit)
      from  7bc10724ea488d17b005ef88d7135c9ce34d0558 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
commit 08f9832fb36e066a57e53f285cf33914e9061aaf
Merge: 7bc10724ea 9067ee7ba0
Author: Darafei Praliaskouski <komzpa at gmail.com>
Date:   Sun Aug 2 01:00:36 2026 -0700

    Merge pull request 'ci: stop ignoring failures in the usan-clang tier' (!645) from Komzpa/postgis:ci/usan-clang-cu-misc-20260801 into master
    
    `qa/usan-clang` is muted, so its result cannot fail a pipeline. A check that cannot fail is not a
    check; this removes the mute so the tier speaks for itself.
    
    Reviewed-on: https://gitea.osgeo.org/postgis/postgis/pulls/645


commit 9067ee7ba05ea014c12f40582715339bd11043f2
Author: Darafei Praliaskouski <me at komzpa.net>
Date:   Sat Aug 1 04:21:27 2026 +0400

    ci: stop ignoring failures in the usan-clang QA step
    
    The step was muted with this reason:
    
        # Current master trips this sanitizer in liblwgeom/cunit/cu_misc.c;
        # keep the signal visible without blocking unrelated pull requests.
    
    Both halves have stopped being true. The cu_misc.c finding was fixed by
    a73d06ef7 and the recipe exits 0 on fresh master, so there is nothing left to
    shield anyone from. And the mute is not protecting unrelated pull requests: it
    hides failures in the pull requests that cause them.
    
    Across the 200 most recent pipelines the step ran 140 times to completion --
    133 success, 7 failure -- and every one of those seven was ignored, leaving its
    pull request green. They are not sanitizer diagnostics; not one log contains a
    runtime error: line. The step runs the QA suite inside a sanitizer build, so
    what it swallowed were ordinary test failures:
    
        pipeline 6232, PR 434  topogeo_addlinestring, totopogeom          Failed: 2
        pipeline 6336, PR 429  topogeo_addlinestring_robust               Failed: 1
        pipeline 6331, PR 595  raster overview extents                    Failed: 1
    
    Each was reproduced locally and each is branch-caused: master at f0c84f8c6
    passes all three. So the step was reporting correctly and only the mute was
    discarding the report.
    
    Worth stating plainly, because anyone checking today will find those pull
    requests green: all three branches have since moved on and their current heads
    pass. The mute did not cause permanent damage. What it cost was time -- the
    authors were never shown a failure they had introduced, and each was found
    again later by someone else. That is the argument for removing it, not a claim
    that defects are sitting there now.
    
    Expect this to turn red any pull request that introduces a QA failure, which is
    the point.

diff --git a/.woodpecker/qa.yml b/.woodpecker/qa.yml
index 24fdf4e4da..1ebab3437d 100644
--- a/.woodpecker/qa.yml
+++ b/.woodpecker/qa.yml
@@ -33,9 +33,6 @@ steps:
   usan-clang:
     image: *test-image
     depends_on: autogen
-    # Current master trips this sanitizer in liblwgeom/cunit/cu_misc.c;
-    # keep the signal visible without blocking unrelated pull requests.
-    failure: ignore
     when:
       - matrix:
           QA_MODE: usan_clang

-----------------------------------------------------------------------

Summary of changes:
 .woodpecker/qa.yml | 3 ---
 1 file changed, 3 deletions(-)


hooks/post-receive
-- 
PostGIS