Re: lasagne and keras

2020-06-25 Thread Christian Kastner
On 2020-06-24 19:13, Stephen Sinclair wrote:
> I have submitted an issue upstream but there has been no response, and
> the last real non-trivial commit was in 2018.  I believe this is
> because it depends on Theano, which we know well is already
> end-of-life, so it's possible the authors have lost interest.
> 
> So, although I'd like to fix the strange bug with python mock, I feel
> this may not be worth worrying about if there is no cooperation from
> upstream.

In my own personal experience, I have found that the maintenance burden
of abandoned software is simply not worth the (increasing) effort if the
package has a very low popcon, as it does in this case.

Sure, if I had unlimited time, I would try to salvage it. But my time is
limited, and it is better invested in new projects, or projects that
actually reach users. Even more so when the project in question has a
known EOL, and has reached it.

> If so, this would be the first package I abandon, what is the correct
> procedure? Do I just file as orphaned?
Normally, yes, you would just file as orphaned.

However, in this particular case, with the software (or one of its
dependencies) being EOL'ed, I would consider outright asking for its
removal from the archive.

You might orphan it now, and ask for removal before the next release in
case nobody steps up.



lasagne and keras

2020-06-24 Thread Stephen Sinclair
Hello Debian Science members,

These two packages are having trouble, lasagne more than keras.

lasagne has an autopkgtest bug that I have been so far unable to fix:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=962938

As a result the package has been marked for autoremoval from testing.

I have submitted an issue upstream but there has been no response, and
the last real non-trivial commit was in 2018.  I believe this is
because it depends on Theano, which we know well is already
end-of-life, so it's possible the authors have lost interest.

So, although I'd like to fix the strange bug with python mock, I feel
this may not be worth worrying about if there is no cooperation from
upstream.  If so, this would be the first package I abandon, what is
the correct procedure? Do I just file as orphaned?

As for keras, I am happy to keep it going as long as it is possible.
In my view it would be worth keeping at least until Tensorflow is
available.  I have currently filed an RFS for recent breakage if
anyone is interested to sponsor:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=963621

kind regards,
Steve