Bug#1072012: libxml-libxml-perl: new upstream release 2.0210, addressing test failures with libxml2 >= 2.11
Paul Gevers
elbrus at debian.org
Thu Oct 3 09:35:54 BST 2024
Hi Aron,
On Wed, 29 May 2024 14:16:47 +0800 Aron Xu <aron at debian.org> wrote:
> > They all seem to fail with:
> >
> > Warning: program compiled against libxml 212 using older 209
> >
> > and this comes from libxml:
> >
> > https://sources.debian.org/src/libxml2/2.12.7+dfsg-2/parserInternals.c/?hl=79#L79
> >
> > if ((myversion / 100) < (version / 100)) {
> > xmlGenericError(xmlGenericErrorContext,
> > "Warning: program compiled against libxml %d using older %d\n",
> > (version / 100), (myversion / 100));
> > }
> >
> >
> > Not sure if this is should be fixed in libxml2 or if we should add an
> > artifical dependency on a newer libxml2 (to avoid testing against the
> > version in testing). The former sounds more logic to me.
> >
>
> Although it looks trivial to remove the warning from libxml2, I'm
> reluctant since this piece of code existed for a very long time, a
> random check shows that version 2.2.3 (in 2000) has the logic:
> https://gitlab.gnome.org/GNOME/libxml2/-/blob/04698d9e1c56467007fcbb9472e5db67cf5938f5/parserInternals.c#L66
It seems wrong to me for a library in a distribution like Debian to warn
on version mismatch. I know we're having the discussion about SONAME
bumps in libxml2 for a reason, but if a library breaks its ABI, it must
bump its SONAME. If it doesn't break its ABI, the warning is useless (in
a distribution like Debian). To me the presence of the warning in the
upstream code is in line with the impression that libxml2 upstream
doesn't understand how distributions handle libraries.
Paul
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-perl-maintainers/attachments/20241003/0a159765/attachment.sig>
More information about the pkg-perl-maintainers
mailing list