Bug#901807: libmath-gsl-perl: incompatible with GSL >= 2.5

[gregor herrmann]

On Tue, 26 Jun 2018 10:25:06 +0300, Niko Tyni wrote:

> On Sun, Jun 24, 2018 at 04:02:06PM +0200, gregor herrmann wrote:
> > -my $ver2func = do(catfile(qw/inc ver2func/));
> > +my $ver2func = do('./' . catfile(qw/inc ver2func/));
> Yeah, that's better than -I. (hardcoding '/' as the directory separator is
> a bit ugly but works for us, and I see catfile is rather eager to remove
> './' if it sees it.)

Right, I also started by adding '.' to catfile() :)
 
> >  sub is_release {
> > -    return -e '.git' ? 0 : 1;
> > +    return 0;
> >  }
> I think I was testing all the time with inside a git checkout,
> that probably explains it. Happy if that's enough to trigger
> the rebuild. Not sure if it still looks at file mtime stamps
> and would need an explicit clean first?

Hm, good question.
They do get recreated for me in my cowbuilder chroot without any
further intervention; and when I build twice-in-a-row, a clean
happens anyway.

But in theory ... maybe ... Ok, after looking at inc/GSLBuilder.pm a
bit: It uses Module::Build's

    up_to_date($source_file, $derived_file)
    up_to_date(\@source_files, \@derived_files)

to check swig/FOO.i and pod/FOO.pod against the derived
xs/FOO_wrap.VERSION.c which happens to end well for us; but maybe
only because our spelling patch touches about all pod files :)

Maybe it would be safer to make sure that up_to_date() always returns
false ...

When I `touch xs/*' before dh_auto_build, indeed re-swig-ification is
skipped for all files; so on the other hand, touching swig/* should
make sure that it happens. -- Does this make sense? Committed in git
and pushed.
 
> > > The latter one may not turn out to be
> > > necessary if the deprecated functions get reinstated with #902281.
> > Ack.
> It looks like the deprecated symbols will be reinstated for now.
> Not sure if we still want to disable them on our side. Probably not.

I'm a bit confused here; the current upload of gsl activates the
patch which reinstates them but the maintainer sounded like he'd
prefer to disable the patch again? If I understood this correctly we
should probably keep our patch, right?

And I guess at least if we keep "our" patch, we don't need a
versioned build dependency?
 
> In any case, I think we should still do a swig rebuild every
> time as part of the normal package build.

Ack.
  
> > I've pushed your and my patches, but I'd rather have another
> > doublecheck before uploading.
> Looks good to me, thanks!

Thanks for checking!


Cheers,
gregor

-- 
 .''`.  https://info.comodo.priv.at -- Debian Developer https://www.debian.org
 : :' : OpenPGP fingerprint D1E1 316E 93A7 60A8 104D  85FA BB3A 6801 8649 AA06
 `. `'  Member VIBE!AT & SPI Inc. -- Supporter Free Software Foundation Europe
   `-   NP: Beatles: While My Guitar Gently Weeps
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 963 bytes
Desc: Digital Signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-perl-maintainers/attachments/20180627/c9b145b0/attachment.sig>