[Debian-science-sagemath] Upgrade Cython in Debian / sage 8.2 status

[Tobias Hansen]

On 06/05/2018 03:07 PM, Tobias Hansen wrote:
> Hi Julian,
>
> On 06/05/2018 02:02 PM, Julian Rüth wrote:
>> Hi Tobias,
>>
>>> If none of the maintainers of Cython will update it, I'll look at it
>>> when finishing up the sage 8.2 package. The official way to ask them
>>> for an update is to file a bug with severity "wishlist" against the
>>> package. Maybe it helps to do that in addition to the pull request.
>> Thanks for the pointer. Will do so next time around.
>>
>>> You said in the pull request you don't use Debian. I wanted to comment
>>> that in case you use Ubuntu you can use the same development tools as
>>> Debian developers, sbuild [1] to build packages on Debian unstable and
>>> schroot which we use in particular to work on sagemath on Debian
>>> unstable [2].
>> Thanks for the hint, but I actually use Arch Linux (the Debian tools in
>> the AUR seem to be broken.) Anyway, I just use Debian in docker
>> now and that works fine.
>>
>>> I uploaded your updated cython package to our development apt repo
>>> (see [2]) and started to do test builds of sagemath.
>> That page mentions
>> https://debian-science.alioth.debian.org/apt/sid-sage/ but this seems to
>> be gone now. Did this move somewhere else?
> This was on the alioth server, which was shut down. I don't know if there's another Debian server we can use for this. Ximin, do you have an idea?
>
> Meanwhile we can try to use Debian experimental to upload everything we need. In the moment the only package that is needed to build sagemath 8.2 and not on experimental yet is cython 0.28.2. I uploaded it on Sunday but it failed to build, I think because the buildd chroots were not updated yet after a gcc update. I'll upload it again soon and then it should build.
>
> You can of course just build cython yourself and then build sagemath 8.2 following the Wiki page.
>
>>> Last time I tried (April 24) sagemath 8.2.rc4 built and ran the test
>>> suite, but with too many failing tests. One problem could be the
>>> mismatch in ipywidgets versions.
>> I would be happy to fix some of the failing doctests. I am also on
>> #debian-science (as saraedum) if you want to coordinate what needs to be
>> done still.
>>
> After cython 0.28.2 is available in experimental, sagemath 8.2 is ready to be uploaded to experimental as well. I solved the problem related to ipywidgets on Sunday, there are still some other doctests to be looked at though.
>
> I had to undo sagemath's changes for ipywidgets 7, because we don't have that version yet and updating the package does not seem to be trivial.
>
> Also a note for Jerome: sagemath 8.2 needs singular 4.1.0-p3, so please don't upload singular 4.1.1 to unstable for now.
>
> Best,
> Tobias

I did a first upload of sagemath 8.2 to experimental today. There are 65 failing doctests and I could really need some help to get this number down. If you have any questions on how to get set up for building sage let me know. Apart from the Wiki page there is also the README.Debian file with some tips for debugging doctests.

Another thing that should be resolved before uploading everything to unstable is that flint-arb 2.12 FTBFS on half of the release architectures...

Best,
Tobias