Le 22/03/2013 19:10, Bill Allombert a écrit :
On Fri, Mar 22, 2013 at 03:09:26PM +0100, Julien Puydt wrote:
Le 20/03/2013 14:25, Bill Allombert a écrit :
On Wed, Mar 20, 2013 at 11:53:56AM +0100, Julien Puydt wrote:
Hi,
in fact, looking at the pari 2.5.3 in sage, there are quite a few
patches in it, some of which might be interesting for upstream...
and which are needed to build sage.
I know, this is why I think it is premature to package SAGE for Debian.
Time would be better spent improving the packaging of SAGE components inside
SAGE so there is no random patching.
They have a particular patch (polred.patch) which does more than
packaging modifications ; its description in patches/README.txt is:
* polred.patch: Fix polred(), add polredbest() based on upstream commits
- 2d00a24fbb1ffe8eba35b9a04763e36eef8a5f7b
- a3596c56f9439144a0dbed4c47bd6ff9485e3fc8
- 1a00ca416de4daebccaab2be1a4b8a061a9f2fde
- ad550d9bbfee8113087407c3262bffc27a020c98
This changes the library ABI. Instead, Sage could write its own PARI
extensions without patching libpari.
Some of the sage developers have a tendency to consider that as sage
ships all dependencies, it is ok to implement new features by
half-patching deps and half-patching sage itself... add to that that
they also have a tendency not to forward their patches upstream and
you'll end up with this type of situation. See for example
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=703263 for another example.
Anyway, if we want to package sage in debian, we'll somehow need the
pari package to be compatible with what sage needs.
What do you suggest?
Snark on #debian-science
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
