Re: About pynac libtool versioning

[Burcin Erocal]

Hi Julien,

CC: pynac-devel

On Tue, 19 Jun 2012 15:04:02 +0200
Julien Puydt <julien...@laposte.net> wrote:

> I played with pynac today, and was a little surprised by the result :
> libpynac-0.2.so.4 -> libpynac-0.2.so.4.0.0
> libpynac-0.2.so.4.0.0
> libpynac.so -> libpynac-0.2.so.4.0.0
>
> I think it's incoherent :
> (1) either I'm supposed to link to some unversioned thing like -lpynac
> and then I should have:
> libpynac.so.4 -> libpynac.so.4.0.0
> libpynac.so.4.0.0
> libpynac.so -> libpynac.so.4.0.0
>
> (2) or I'm supposed to link to a versioned thing like -lpynac-0.2 and
> then I should have:
> libpynac-0.2.so.4 -> libpynac-0.2.so.4.0.0
> libpynac-0.2.so.4.0.0
> libpynac-0.2.so -> libpynac-0.2.so.4.0.0
>
> After some poking around, it seems the problem stems from the fact
> that both -release and -version-info are used in ginac/Makefile.am
> (after some things where defined in configure.ac) ; I suggest to
> apply the attached patch, which puts pynac in situation (1) above.

You're right. AFAICT, the patch you proposed is equivalent to these two
commits from ginac:

http://www.ginac.de/ginac.git?p=ginac.git;a=commitdiff;h=f9a14723ef80c4f38b814556fb16084e2b64d426

http://www.ginac.de/ginac.git?p=ginac.git;a=commitdiff;h=f320e27e9cfe287168c879af5991babffaa1e9c8


I will adapt these, probably by adding support for ginac_lt_age after
your patch.

Thanks for your help.

> PS: http://pynac.sagemath.org is 503 today, so I played using what I
> found in sage-5.1.beta3... it might not be the best.

pynac.sagemath.org wasn't updated since the sagemath virtual machine
was taken down. I will need to arrange access to the current server
with the server admins at UW.


Cheers,
Burcin

[Burcin Erocal]

On Tue, 19 Jun 2012 15:46:46 +0200
Julien Puydt <julien...@laposte.net> wrote:

> Le mardi 19 juin, Burcin Erocal a écrit:

> > You're right. AFAICT, the patch you proposed is equivalent to these
> > two commits from ginac:
> > http://www.ginac.de/ginac.git?p=ginac.git;a=commitdiff;h=f9a14723ef80c4f38b814556fb16084e2b64d426
> > http://www.ginac.de/ginac.git?p=ginac.git;a=commitdiff;h=f320e27e9cfe287168c879af5991babffaa1e9c8
> >
> > I will adapt these, probably by adding support for ginac_lt_age
> > after your patch.

In the end I just applied your patch:

https://bitbucket.org/burcin/pynac/changeset/c2abbebf9042

The instructions we have to change the library version are much simpler
than those in current GiNaC. I don't think there is any need to
complicate the release process.

> Ok. When is a new version out? :-)

I can tag pynac-0.2.5 now and create a tarball. Any volunteers to make
an spkg?

With Titus, we recently fixed a segfault when working with Python longs:

http://trac.sagemath.org/sage_trac/ticket/13107

This justifies a new release.


Of course, we should get to the order improvements (#9880 on Sage trac)
at some point, but that will need some time. It might not even make it
to Sage 5.2 at this rate.


Cheers,
Burcin

[Jean-Pierre Flori]

I think I'll be able to make a new spkg next week.
I'll be at a conference and should finally have some spare time.

Best,

[Burcin Erocal]

Hi Jean-Pierre,

On Fri, 29 Jun 2012 01:18:33 -0700 (PDT)
Jean-Pierre Flori <jpf...@gmail.com> wrote:

> I think I'll be able to make a new spkg next week.
> I'll be at a conference and should finally have some spare time.

Perhaps it's too late, but I managed to find time to roll a release
only now. Here is the tarball:

http://sage.math.washington.edu/home/burcin/pynac/pynac-0.2.5.tar.bz2


Cheers,
Burcin