Bug#938494: skimage (build-)depends on python-cloudpickle which has already been dropped

[Andreas Tille]

On Fri, Sep 13, 2019 at 11:09:58AM +0200, Ole Streicher wrote:
> > Sure.  But do you want to do team uploads of python-numpy and its > 300
> > rdepends as well?  I was considering the amount of packages that are
> > affected by RC bugs which is way lower if we leave skimage RC buggy
> > rather than making python-numpy RC buggy.  I mean, in the end we need to
> > port everything but we can decide about a noisy mailing list flooding us
> > every now and than with 300 mails about packages getting removed from
> > testing or keeping that number at about 20.
> 
> I am not sure I understand you here.
> 
> skimage is now scheduled for autoremoval; I really think we should go
> forward here. And as far as I remember, numpy is secured from
> autoremoval, so even an RC buggy numpy will not lead to autoremoval.

Ahhh, I was not aware that this safety line exists.  In this case it
might make sense to proceed from here as you said since the harm is done
anyway and the other rdepends from skimage might be not as important.
I have not checked further.
 
> I now pushed skimage_0.14.2-3 to salsa, with Python 2 removed. Just
> awaiting the salsa CI tests: shall I upload or shall I wait? We should
> go forward, shouldn't we?

I'm definitely not against progress even a forced one.  I was just
concerned about finding 300 useless mails in my inbox in a regular
frequency.  BTW, I even found some false positives in the calculation
where rdepends are fixed in unstable.  However I also found rdepends
without any open bug (arden) which I simply fixed.
 
> > BTW, you can start with droping rdepends by droping astro-python
> > metapackage.
> 
> Oops, I thought that I already uploaded it. Will do ASAP.

:-)

Kind regards

      Andreas.

-- 
http://fam-tille.de