[Debian-science-sagemath] Getting sagemath back into testing

[Julien Puydt]

Hi,

Le vendredi 30 octobre 2020 à 21:53 +0000, Tobias Hansen a écrit :

> - giac needs to transition to testing. The problem are failing tests
> on ppc64el and s390x.

First there was a bug I couldn't track on S390X. I tried to get
upstream to sign his key so he could get access to a porterbox, but:
"My key is on the university server and you know it's me, so why
bother?"... I tried to argue, to no avail. So I was thinking I would
disable this architecture, unsupported upstream.

Then upstream included a (forked) micropython in the source code, with
hand-written compilation for the library. That broke PPC64EL. Again,
upstream didn't really want to work on the matter. I wanted to try
writing a better compilation of that library (libtool is complex, but
there's a couple reasons why!), but that didn't happen for personal
reasons. So I think I'll disable that architecture too (same reason).

Now that raises a couple of questions :

- is it possible to write something like: "Achitecture: all !s390x
!ppc64el" or do I really have to list all architectures? (if the answer
is: list, then that's a good wishlist bug...)

- do I need to add Architecture: to all packages or will doing this for
the libgiac package be enough -- all others depend on it after all?


Cheers,

JP