Upgrade Givaro to version 3.7.0

[pcpa]

Givaro 3.6.x is a requirement for LinBox 1.3.0.

• Install http://sage.math.washington.edu/home/malb/spkgs/givaro-3.7.0.spkg
• Apply attachment: trac_9511_givaro_3_7_x.2.patch

See https://bitbucket.org/malb/givaro-spkg for SPKG Repository.

Upstream: Reported upstream. Developers acknowledge bug.

CC: @zimmermann6 @jpflori

Component: finite rings

Reviewer: Volker Braun

Issue created by migration from https://trac.sagemath.org/ticket/9511

[ClementPernet]

comment:2

I'm currently working on the upgrade to givaro-3.3.3rc0 together with linbox-1.1.7, since this later version of linbox required very recent changes in givaro (only in 3.3.3rc0).

[malb]

Changed upstream from None of the above - read trac for reasoning. to none

[malb]

Author: Paulo César Pereira de Andrade, Martin Albrecht

[malb]

comment:6

Note this upgrade won't work unless LinBox is upgraded as well.

[kiwifb]

comment:7

Replying to @malb:

Note this upgrade won't work unless LinBox is upgraded as well.

I was going to say that, not to mention we need a givaro-3.3.xx spkg too.

[kiwifb]

comment:8

spkg that I missed in the description somehow. linbox in a new ticket?

[malb]

Dependencies: #11718

[malb]

comment:9

Yep, I opened #11718 for that but I just accidentally killed the last three hours of pretty boring & tedious work trying to compile 1.1.7. So it'll take a while until I upload an SPKG there.

[malb]

comment:10

Status update:

We get errors like this

/sage-4.7.1/local/include/givaro/givzpz16table1.inl:424: error: ‘UINT32_MAX’ was not declared in this scope

on sage.math unless we explicitly pass -D__STDC_LIMIT_MACROS in spkg-install. This works but is weird because Givaro does set __STDC_LIMIT_MACROS before it includes <stdint.h>. Also, I do not have this problem my local machine (running 64-bit Debian/GNU Linux)

[nexttime]

comment:11

Just for the record:

There's a Givaro 3.2.13.rc1.p4 spkg with a couple of fixes at #12761 (currently still needing review).

[malb]

comment:13

I am feeling ambitious today (Brice from the LinBox team is in shouting distance) so I decided to give this update business another try. Givaro 3.6.0 is the easy part so here it is.

[malb]

comment:17

Updated so that the number theory doctest in the "French book" is deterministic.

[jdemeyer]

comment:56

This needs to be rebased to sage-5.3.beta1 (to be released very soon).

[vbraun]

Rebased patch

[vbraun]

comment:57

Attachment: trac_9511_givaro_3_7_x.2.patch.gz

Rebased patch for sage-5.3.beta1. Just a trivial rebase in module_list.py.

[jpflori]

comment:59

Volker, could you tag the last hg commit here as well?

[jhpalmieri]

comment:60

I thought that Jeroen's scripts for Sage releases added those tags automatically, but I'm not sure.

[vbraun]

comment:61

I haven't made any changes to the spkg yet, so I'm not going to partake in the bikeshedding here. The Sage developer guide contains nothing about mandatory hg tagging. In fact, since we frequently make changes in response to reviews before releasing the spkg, really only the release manager can set the tag.

[jdemeyer]

comment:62

Replying to @jhpalmieri:

I thought that Jeroen's scripts for Sage releases added those tags automatically

Indeed.

[vbraun]

comment:63

Superseded by #13164.

[jhpalmieri]

comment:64

Is the Sage library patch still necessary? If so, maybe it should be moved to #13164 and this ticket should be closed as a duplicate.

[jpflori]

comment:65

I did not try without it (in fact I did but without the linbox patches as well, by mistake), but with it there are no errors.

[vbraun]

comment:66

Superseded by #13164, close this ticket as duplicate.

[vbraun]

Changed dependencies from #12840, #12841, to be merged with #12883 to none

[jdemeyer]

Changed author from Paulo César Pereira de Andrade, Martin Albrecht to none