On September 16, 2019 6:21:42 PM UTC, Jonathan Carter <j...@debian.org> wrote:
>On 2019/09/12 18:30, Mo Zhou wrote:
>> Agreed. there are already python2 forks such as:
>> https://github.com/naftaliharris/tauthon
>> It may sound funny but I don't hope any python2 stuff get back with
>> a new name after the python2 removal. But, it could happen if some
>> people is willing to support and maintain...
>> How should Debian react if someone submitted an ITP for python2
>> forks, such as the tauthon above?
>It would probably be handled just like any other piece of new software
>entering Debian. I can't speak for the entire Python team, but the
>Python team would probably not pay much (or any) attention to it
>whatsoever and it would be unfair for anyone to expect that from the
>Python team. Just like with any other software, it would just need an
>active set of contributors behind it.
>Personally I just don't think putting a significant amount of effort
>into a python2 fork is worth while long-term. It's just adding more
>technical debt. Instagram is one of the largest sites in the world and
>switched over to python3 because they say they realised that no
>significant performance work is still going into python2, that's all
>happening in the python3 world now. JP Morgan's Athena[1] project has
>over 35 million lines of python code over 150 000 python modules and
>they regret starting so late but aim to complete migrating the critical
>core pieces by the 2nd quarter of 2020.
>My point is that the world is moving to python3, maybe not as fast as
>everyone would like, but putting in additional work to stick with
>python2 would mean confining yourself to an ever faster shrinking
>universe and earning interest on that technical debt and you'll end up
>with code that no one will want to touch in either terms of working on
>it or using it.
>So, I don't think it would be a problem accepting such fork in Debian,
>think Debian would be indifferent to it, but in terms of a consumer of
>such a fork I believe it's better to pull the plaster fast, do the work
>now to get it updated, or it will just get more expensive in the

It's not that simple.  If such a fork were to be packaged for some purpose that 
only needed the standard library, then maybe, but generally that's not enough.  
The ecosystem of python modules and extensions requires integration.  I don't 
think it's reasonable to ask maintainers of these packages to support such a 
fork, which would limit it's utility in the archive.

I've no doubt there will be many organizations running python2.7 or forks of it 
for many years to come, but I don't think it's something we can usefully 
support in the Debian archive.

Scott K

Reply via email to