Upgrade PARI to 2.5.1

[jdemeyer]

PARI-2.5.1 has been released: http://pari.math.u-bordeaux.fr/pub/pari/unix/pari-2.5.1.tar.gz

We should upgrade PARI in Sage to this latest version. It also happens that this is needed to compile PARI with gcc-4.6.2 (see #12369).

spkg: http://boxen.math.washington.edu/home/jdemeyer/spkg/pari-2.5.1.p0.spkg (changes for reviewing: attachment: pari-2.5.1.p0.diff)

apply: attachment: 12363_doctest.patch

In an earlier version of this spkg, an upstream bug was discovered which is now fixed: http://pari.math.u-bordeaux.fr/cgi-bin/bugreport.cgi?bug=1276

Upstream: Completely fixed; Fix reported upstream

Component: packages: standard

Author: Jeroen Demeyer

Reviewer: John Palmieri

Merged: sage-5.0.beta4

Issue created by migration from https://trac.sagemath.org/ticket/12363

[jdemeyer]

Upstream: Reported upstream. Little or no feedback.

[jhpalmieri]

comment:2

Two quick comments:

• the linked gzipped tar file is missing the 'data' directory, which causes some doctests to fail. (This caused me a lot of confusion on OS X Lion: pari fails self tests #12315.)
• your spkg file is huge!

[jdemeyer]

comment:3

Well, there is a reason this ticket isn't "needs review", sorry for the confusion.

[jdemeyer]

Changed upstream from Reported upstream. Little or no feedback. to Reported upstream. Developers acknowledge bug.

[jdemeyer]

Attachment: 12363_doctest.patch.gz

[jdemeyer]

Author: Jeroen Demeyer

[jdemeyer]

Changed upstream from Reported upstream. Developers acknowledge bug. to Completely fixed; Fix reported upstream

[jdemeyer]

Diff for the pari spkg, for review only

[jdemeyer]

comment:8

Attachment: pari-2.5.1.p0.diff.gz

[jhpalmieri]

comment:9

In principle this looks okay and it works on a few different machines. Skynet seems to be down right now, so I can't test it on a very wide range of platforms, though.

[jdemeyer]

comment:10

Replying to @jhpalmieri:

Skynet seems to be down right now.

It seems to work now. Anyway, I already tested this on most of the Skynet machines and once it's merged it will be tested again before the release.

[jhpalmieri]

comment:11

Regarding the doctest patch, can you explain mathematically why this is the right thing to do?

[jdemeyer]

comment:12

Let K be the number field Q[y]/(y² + y + 1).

Let L be the relative number field K[x]/(x⁴ + y*x + 2).

We can also write L = {Q[y]/(y² + y + 1)}[x]/(x⁴ + y*x + 2) (braces added for clarity).

Now consider the element x*y in L, we ask for its relative representation in PARI. Given the above, the correct answer is

Mod(Mod(y, y^2 + y + 1)*x, x^4 + y*x + 2)

So, the new answer to the doctest is more correct than the old.

[jhpalmieri]

Reviewer: John Palmieri

[jhpalmieri]

comment:13

Okay, that sounds good to me.

[jdemeyer]

Merged: sage-5.0.beta4