Re: Fwd: Biopython 1.86 released

Étienne Mollier <[email protected]> Wed, 29 Oct 2025 23:55:17 +0100
Newsgroups gmane.linux.debian.devel.medical
Message-ID <[email protected]>
Hi Peter,

Thank you for having issued Biopython 1.86 lately, I took some
time to bump the version for Debian sid tonight and it should be
available at some time tomorrow for general use.

Peter Cock, on 2025-10-29:
>  FYI - this adds Python 3.14 support, and removes a few deprecated
> modules.

Thanks, with Python 3.14 around the corner, support should avoid
us some upcoming related bugs.  I see that several legacy
components that were causing difficulties in the past have been
removed, and it has been possible to drop several lines of
workarounds in the packaging code.  Your work has been overall
very helpful, thank you!

> I do not expect any issues, please get in touch if there are
> any surprises.

Will do.  Not sure if that counts as a surprise, but I ran into
some friction with the i386 build[1]: something about the excess
of floating point precision needed to pass test_align_one_char3.
As far as I could witness, the error between 32 and 64-bit is
more than reasonable, so I just skipped the test on 32-bit and
moved on.  For reference, differences look like:

	AssertionError: '<Ali[26 chars] = 0.19999999999999996; substitution score = 1[87 chars]860>' != '<Ali[26 chars] = 0.20000000000000007; substitution score = 1[102 chars]860>'
	- <AlignmentCounts object (score = 0.19999999999999996; substitution score = 1.0; gap score = -0.8; 1 aligned letters; 1 identities; 0 mismatches; 4 gaps) at 0xe29fa860>
	+ <AlignmentCounts object (score = 0.20000000000000007; substitution score = 1.0; gap score = -0.7999999999999999; 1 aligned letters; 1 identities; 0 mismatches; 4 gaps) at 0xe29fa860>

[1]: https://salsa.debian.org/med-team/python-biopython/-/jobs/8522875/raw

Upload to unstable will trigger builds for several processor
architectures and integration tests, we'll see if there is more
friction than just the i386 issue in the upcoming days.

> We are seeing some setuptools deprecation warnings, and are likely to
> move from setup.py to pyproject.toml for the next release.

Yes, it looks like the whole Python ecosystem is moving to
pyproject.toml, so from a package maintenance perspective, this
will give a chance to have the set of modules packaged in a way
more homogene.

> Thanks,

Have a nice day,  :)
-- 
  .''`.  Étienne Mollier <[email protected]>
 : :' :  pgp: 8f91 b227 c7d6 f2b1 948c  8236 793c f67e 8f0d 11da
 `. `'   sent from /dev/pts/2, please excuse my verbosity
   `-    on air: Moon Safari - Lady of the Woodlands
signature.asc (application/pgp-signature, 833 B)
-----BEGIN PGP SIGNATURE-----
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=n6B/
-----END PGP SIGNATURE-----