Investigating this issue since autoremoval of pychopper from sid
kicked in: when reading the source code of python-parasail, I've
hit a comment in the decode implementation saying:

        # The C interface allocates unaligned memory but does not provide a
        # means of deallocating it.  The parasail_free() is for aligned
        # memory only.  On Windows, this is an error.  On OSX/Linux, free()
        # is the same for aligned and unaligned.

Since memory alignment is also important on arm64 and a few
other architectures, my guess would be that something is going
wrong here and needs adjustment to handle more platforms like
the Windows one.  Why did I look at decode?  Because this is the
method invoked during the crash:

        […]
          File 
"/build/reproducible-path/pychopper-2.7.10/.pybuild/cpython3_3.13_pychopper/build/pychopper/parasail_backend.py",
 line 30, in process_alignment
            fo = first_cigar(aln.cigar.decode.decode())
                             ^^^^^^^^^
          File "/usr/lib/python3/dist-packages/parasail/bindings_v2.py", line 
403, in cigar
            return self.get_cigar()
                   ~~~~~~~~~~~~~~^^
          File "/usr/lib/python3/dist-packages/parasail/bindings_v2.py", line 
390, in get_cigar
            raise AttributeError("'Result' object has no traceback")
        AttributeError: 'Result' object has no traceback. Did you mean: 
'_cigar'?

I may continue tomorrow, in the context of the Advent calendar
bug squashing party.  Hopefully this is really just a matter of
adjusting the decode implementation of python-parasail, but I'm
not sure of the status yet.

In hope this helps,
-- 
  .''`.  Étienne Mollier <[email protected]>
 : :' :  pgp: 8f91 b227 c7d6 f2b1 948c  8236 793c f67e 8f0d 11da
 `. `'   sent from /dev/pts/6, please excuse my verbosity
   `-    on air: 039 - Blood

Attachment: signature.asc
Description: PGP signature

Reply via email to