[Debian-science-sagemath] Required tasks for sage in stretch

[Julien Puydt]

Hi,

On 09/11/2016 12:36, Tobias Hansen wrote:
> I'm a bit worried that we might not achieve the goal to get sage into
> stretch, if we don't focus on the essential tasks.[1] If we do not get
> all of the following things done in the next 3 or 4 weeks, there will
> simply be no sage in stretch.

I'm a bit worried by this Zeitnot : rushing can give us more headaches 
down the line actually maintaining things.

> * gap/libgap - Jerome, do you have a plan to resolve this?

No clue.

> * maxima - even updating sage to maxima 5.38.1 is hard (see [2]) and
> then there is the gcl vs ecl issue. I begin to think the only way to
> solve this in time is to have our own package of maxima 5.35.1 with ecl
> and fasl library.

That is a difficult point. I got the technical part settled years ago ; 
we're left with the human part.

> * ecl - the one fix I applied does not migrate to testing due to a
> deadlocked situation with libatomic-ops (unfixable (?) RC bug). Maybe
> this can be resolved by asking for removals of libatomic-ops packages
> for the failing architectures from testing. Since the RC bug is also in
> testing, the package should then migrate, right?

No clue.

> * planarity - We need have to figure out how to use the new version
> packaged in Debian.

I might try to have a look. It's an upstream sage work.

> * cython - We need to fix RC bug #833288 for testing migration and make
> sure cython 0.25 gets packaged.

No clue.

> * pyzmq - RC bugs need to be fixed for testing migration.

No clue.

> * linbox - I might need to close RC bugs by disabling tests.

Those tests might be failing for a good reason.

I disable tests in my javascript packages (and I'm not the only one), 
but the reason isn't that I don't like their results : it's that the 
testing framework isn't packaged in Debian and is terrible to get in 
(hundreds of packages with cyclic deps everywhere) !

> * sagenb - Is someone taking care of this? What's the status?

No clue.

> * jupyter-sphinx-theme - Jerome, can this be uploaded?
>
> * jupyter-notebook - Gordon, can this be uploaded?
>
> * ipywidgets, widgetsnbextension, thebe - ? (Are these essential?)

For ipywidgets, jpuyter-sphinx-theme, jupyter-notebook and 
widgetsnbextension : if we just want sage to work as a terminal 
application, we can probably get away with patching their use off our 
sagemath package.

For thebe, I think it's essential to build the documentation. :-/

> * sagemath - fix doctests, disable doctests we can't fix, revise
> patches, polish packaging, etc.

Perhaps add to the list to disable the notebook, see above?

> [1] I really hope noone gets offended by this, but singular 4.0.3-p5 and
> pynac 0.7 are not on this list. We should probably also stop worrying
> about jmol. It would be a pity if we can't get sage into stretch because
> we get carried away doing other things.

I packaged the latest pynac because I saw it was out and I was wondering 
if weren't going to need it. I open the door to the possibility, not 
forcing to get it uploaded : no problem if it's not needed yet.

For jmol, it's used for the notebook part and for the documentation if I 
understand well. Or perhaps it's possible to use it in a terminal to 
produce image files too?

Cheers,

Snark on #debian-science