Den tors 7 maj 2026 kl 19:46 skrev Dan Langille <[email protected]>:
>
> Hello,
>
> I'm a port maintainer for the FreeBSD project and I'm a user of the 
> subversion project.
>
> I'm hitting a build problem with Python 3.14 and subversion-1.14.5 - the 
> problem does not appear with Python 3.12 or 3.13.
>
> I first opened a PR with FreeBSD at 
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=294183 - so far, it does 
> not seem to be a packaging issue.
>
> The problem I hit (the full build log is in the URL above) is below.
>
> Thank you.
>
> if [ "LD_LIBRARY_PATH" = "DYLD_LIBRARY_PATH" ]; then  for d in 
> /wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python/libsvn_swig_py
>  
> /wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python/../../../libsvn_*;
>  do  if [ -n "$DYLD_LIBRARY_PATH" ]; then  
> LD_LIBRARY_PATH="$LD_LIBRARY_PATH:$d/.libs";  else  
> LD_LIBRARY_PATH="$d/.libs";  fi;  done;  export LD_LIBRARY_PATH;  fi;  cd 
> /wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python;
>   /usr/local/bin/python3.14 
> /wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python/tests/run_all.py
> ..................F.........................s........................................................................F...........................................
> ======================================================================
> FAIL: test_mergeinfo_leakage__incorrect_range_t_refcounts 
> (mergeinfo.SubversionMergeinfoTestCase.test_mergeinfo_leakage__incorrect_range_t_refcounts)
> Ensure that the ref counts on svn_merge_range_t objects returned by
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File 
> "/wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python/tests/mergeinfo.py",
>  line 147, in test_mergeinfo_leakage__incorrect_range_t_refcounts
>     self.assertEqual(refcount, expected, (
>     ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^
>       "Memory leak!  Expected a ref count of %d for svn_merge_range_t "
>       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>       "object, but got %d instead (path: %s, index: %d).  Probable "
>       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>       "cause: incorrect Py_INCREF/Py_DECREF usage in libsvn_swig_py/"
>       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>       "swigutil_py.c." % (expected, refcount, path, i)))
>       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> AssertionError: 3 != 4 : Memory leak!  Expected a ref count of 4 for 
> svn_merge_range_t object, but got 3 instead (path: b'/trunk', index: 0).  
> Probable cause: incorrect Py_INCREF/Py_DECREF usage in 
> libsvn_swig_py/swigutil_py.c.
>
> ======================================================================
> FAIL: test_replay_batons_refcounts 
> (repository.SubversionRepositoryTestCase.test_replay_batons_refcounts)
> Issue SVN-4917: check ref-count of batons created and used in callbacks
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File 
> "/wrkdirs/usr/ports/devel/py-subversion/work-py314/subversion-1.14.5/subversion/bindings/swig/python/tests/repository.py",
>  line 438, in test_replay_batons_refcounts
>     self.assertEqual(sys.getrefcount(e_ptr), 2,
>     ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^
>                      "leak on editor baton after replay without errors")
>                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> AssertionError: 1 != 2 : leak on editor baton after replay without errors
>
> ----------------------------------------------------------------------
> Ran 161 tests in 2.398s
>
> FAILED (failures=2, skipped=1)
> *** Error code 1
>
>
> --
>   Dan Langille
>   [email protected]

Hi Dan,

Thanks for your report.

If I'm reading this correctly, this is a problem reported to us in
July 2025, see [1]. There is a patch in the thread, tested on FreeBSD.
It was committed to our trunk in r1926575 and r1927715. It was also
backported to the 1.14.x branch in r1929147 however we have not made a
new release of 1.14 since that time, however you could probably bring
these patches to the FreeBSD port. These changes will also be included
in the upcoming 1.15.0 release, you can get RC tarballs from [2].

Can you check if this solves your problem?

Kind regards,
Daniel Sahlberg

[1] https://lists.apache.org/thread/tfo366jztyvj4vsnj4pl0lwcdhtvr116
[2] https://dist.apache.org/repos/dist/dev/subversion

Reply via email to