[Neurodebian-devel] Please consider moving common packages to Debian-Science

[Yaroslav Halchenko]

On November 19, 2016 12:03:06 PM EST, Ole Streicher <olebole at debian.org> wrote:
>Hi Yaroslav, Yury and all.
>
>
>On 19.11.2016 17:50, Yaroslav Halchenko wrote:
>> On Sat, 19 Nov 2016, Yury V. Zaytsev wrote:
>>>> * pandas
>>>> * pymc
>>>> * statsmodels
>>>> * scikit-learn
>> 
>>> I'd also add at least those two:
>> 
>>> * mpi4py
>>> * seaborn
>> 
>>>> Would you consider moving these packages to debian-science instead?
>This
>>>> would enable team-maintenance for a larger team and may help to
>keep
>>>> them in a good shape. At least, I don't want to join neurodebian
>just to
>>>> fix the stuff in-place.
>> 
>>> The question, of course, is if moving the packages will really
>magically
>>> attract more maintainers... my understanding is that packages are
>maintained
>>> in NeuroDebian by default to make it easier for NeuroDebian
>maintainers &
>>> leverage the repository infra they have in place. But maybe
>something worth
>>> considering anyways.
>
>I can speak here just for myself: I found the upstream patch that
>solved
>the main issue in statsmodels, and it is more complicated for all to
>upload the patch to the bug, having it back downloaded and applied to
>the git. Having this complicated process in mind, at least I delayed
>working on this quite quite a bit. A repository where I can contribute
>directly is just more inviting. YMMV, however.
>
>> In general I don't mind at all moving (or better -- having moved ;) )
>> any of those packages under debian-science maintenance!    Which one
>> would you like to start with?  e.g. statsmodels is indeed the
>> interesting one.  I have just uploaded fresh pandas (which brought
>new
>> FTBFS :-/) with hope that some might just go away in statsmodels, but
>> that didn't happen.   I am still slowly working with statsmodels and
>> have some non-pushed changes, but if you are keen to start with e.g.
>> pandas and fixing up any of those issues as it being a part of Debian
>> science -- it would be awesome.  Just let me know which package(s)
>you
>> want to start with so we don't overlap.
>
>As I wrote, I looked at statmodels; however the solution suggested by
>Sandro Tosi (tzdata) didn't work. So I am stuck there now; however
>there
>is just this one remaining issue.
>
>Best regards
>
>Ole
>
>_______________________________________________
>Neurodebian-devel mailing list
>Neurodebian-devel at lists.alioth.debian.org
>http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/neurodebian-devel

Fwiw my packaging repository is on github, I would appreciate a PR if you would like to help rapidly with a patch... I am ATM looking into uploading updated maintenance/0.8 iirc upstream branch. Is the fix you are talking about from there?
-- 
Sent from a phone which beats iPhone.