Re: Bug#942106: python3.8 / pandas py2removal

2019-11-13 Thread Andreas Tille
Hi,

On Sun, Nov 10, 2019 at 03:18:31PM +0100, Matthias Klose wrote:
> tnseq-transit is a leaf package, raising the severity now, could be removed
> and re-enter later.

Besides I fully agree to this in general python-pypubsub was accepted
yesterday and thus new upstream of tnseq-transit with Python3 support
uploaded.  I also re-uploaded python-pypubsub as source-only upload so
testing migration should be soon.
 
> IMO, we should care about the recommends later ...

+1
 
> > (patsy isn't a leaf package for py2removal, it's in a cycle with pandas
> > and statsmodels, i.e. for a non-breaking removal they all have to go
> > together.)
> 
> ok I'm undoing the NMU from the delayed queue, you'll find it at
> https://people.debian.org/~doko/tmp/

Regarding those packages where I'm listed as Uploader feel free to do
non-delayed NMUs.

Kind regards

 Andreas. 

-- 
http://fam-tille.de



Re: Bug#942106: python3.8 / pandas py2removal

2019-11-10 Thread Rebecca N. Palmer

I have uploaded pandas and statsmodels.

On 10/11/2019 14:18, Matthias Klose wrote:

https://people.debian.org/~doko/tmp/


The patsy one has a bug: as debian/tests/nosetests3 was a symlink to 
nosetests2, it should have deleted this link and renamed nosetests2 to 
nosetests3, not deleted nosetests2.




Re: Bug#942106: python3.8 / pandas py2removal

2019-11-10 Thread Matthias Klose

On 10.11.19 14:46, Rebecca N. Palmer wrote:
mdp isn't in testing either, but if you're using a policy of "no py2removals 
that break packages in testing", tnseq-transit (Depends: statsmodels) and 
possibly stimfit (Recommends: pandas) need to be done as well.  Those are both 
thought to need new upstream versions.


tnseq-transit is a leaf package, raising the severity now, could be removed and 
re-enter later.


IMO, we should care about the recommends later ...

(patsy isn't a leaf package for py2removal, it's in a cycle with pandas and 
statsmodels, i.e. for a non-breaking removal they all have to go together.)


ok I'm undoing the NMU from the delayed queue, you'll find it at 
https://people.debian.org/~doko/tmp/




Re: Bug#942106: python3.8 / pandas py2removal

2019-11-10 Thread Rebecca N. Palmer
mdp isn't in testing either, but if you're using a policy of "no 
py2removals that break packages in testing", tnseq-transit (Depends: 
statsmodels) and possibly stimfit (Recommends: pandas) need to be done 
as well.  Those are both thought to need new upstream versions.


(patsy isn't a leaf package for py2removal, it's in a cycle with pandas 
and statsmodels, i.e. for a non-breaking removal they all have to go 
together.)




Re: Bug#942106: python3.8 / pandas py2removal

2019-11-10 Thread Matthias Klose

On 10.11.19 14:22, Matthias Klose wrote:

patsy is a leaf package, no problem.

scikit-learn, needs mdp, pymvpa2, I think that's manageable, so I'm volunteering 
to do the NMUs, preferably with a 0 or 1 day delay.


PyMVPA has other RC issues, is removed from testing, so ignore it for now. and 
I'm raising the severity of the py2removal bug.




Re: Bug#942106: python3.8 / pandas py2removal

2019-11-10 Thread Matthias Klose

[CCing debian-science]

On 10.11.19 13:18, Rebecca N. Palmer wrote:

Matthias Klose wrote:
yes, please do [raise pandas 0.25 blocking bugs to "important"] 


Done, but only 2 of them have been fixed since.

This leaves 13:
has patch or Ubuntu fix: matplotlib2 patsy python-apptools scikit-learn
may need more extensive work: cnvkit python-feather-format python-skbio stimfit 
tnseq-transit

already not in testing: mdp psychopy pymvpa q2-types


If you can get that done with [pandas] 0.25, fine,


It took longer than I was expecting, but pandas 0.25 / statsmodels 0.10 now 
appear to be working, including with python3.8.  (Though we won't actually know 
if #943732 is gone until mipsel tries to build it.)


\o/

and I wouldn't worry too much about the other four breaking packages at this 
point.


Was this intended as...

... a request to upload pandas 0.25 / statsmodels 0.10 to unstable (and apply 
the Ubuntu py2removal patches to patsy and scikit-learn?), overriding normal 
py2removal rules?


patsy is a leaf package, no problem.

scikit-learn, needs mdp, pymvpa2, I think that's manageable, so I'm volunteering 
to do the NMUs, preferably with a 0 or 1 day delay.


So yes, please go ahead.

... a request to split pandas into a pandas2 0.23 and a pandas 0.25? (since 4 is 
only the number of non-py2removal breakages, and the wiki page 
https://wiki.debian.org/Python/Python3.8 says to do this if 2.7 and 3.8 need 
different upstream versions)  Should be technically easy, but means going 
through NEW.


... a statement that once pandas 0.25 works, this is no longer my problem, i.e. 
that I don't have to fix 0.23?


I don't think that's necessary at this point.


matplotlib and pandas don't have Python2 packages in Ubuntu
   anymore, so I can't tell much what is needed here


matplotlib has already been split into Python 2 and Python 3 source packages. 
(matplotlib2 is in Ubuntu, and unbuildable there due to #943924.)


Python2 matplotlib are already removed in Ubuntu, I shouldn't have synced 
matplotlib2 from experimental.



According to its Ubuntu build log:
https://launchpad.net/ubuntu/+source/matplotlib/3.0.2-2ubuntu1/+build/17942804

matplotlib has one python3.8-specific test failure, 
test_axes.py::test_pathological_hexbin.  This is currently being ignored 
(#942766) along with (many) errors that also happen on 3.7.