[Debian-science-sagemath] pari-sage or pari ?

[Julien Puydt]

On 21/08/2016 12:44, Jeroen Demeyer wrote:
> Being forced to stick to released versions of packages would be very
> hard for Sage. So we don't plan to do this. PARI is especially bad in
> this regard because they release so slowly. They made a non-bugfix
> release this month, but the one before that was more than 2 years ago!
> We really don't want to wait 2 years to be able to use new PARI features.

1) I don't understand why it would be very hard for sage to stick to 
released versions of packages : each and every other open source 
projects does this. What I understand is that when you depend on hundred 
of packages and you start to package a svn snapshot of that one, a git 
commit of that one, but with this patch added and that one removed, etc 
then *that* is more than very hard : there's no sanity along that path!

2) Releasing development version of some upstream in a stable version of 
sage is more than impolite towards said upstream : it's hostile. If 
there is a bug, the user gets said "Oh, it's a problem with sage and its 
pari 2.8-pre314159~git271828+p42", but understands "Oh, it's a problem 
with pari".

3) One can argue that pari is moving too slowly... that might be true, 
but if the current upstream is too slow... become upstream and make it fast!

Snark on #debian-science