[Python-ideas] A python bridge between versions

Nick Coghlan ncoghlan at gmail.com
Sat Mar 1 07:55:37 CET 2014


On 1 March 2014 15:53, ian o <ian.team.python at gmail.com> wrote:
> Steven,
>> *This is not a problem to be solved.* It's okay for people to stick with
>> an old version of Python. There's no rule that says Python has failed if
>> some developers stick to old versions.
>>
> OK. So you are telling us forget moving our developers to python3 until
> every last dependency has moved. And we should also force the people who in
> turn use our code to stay with python2 and it is not a problem.

Ian, please keep in mind that you are asking people to do extra work
for you *for free*, after you have already been benefiting for years
from work that we already made available to you free of charge. The
latter part is OK, and entirely the way non-commercial open source
works, but it *does* mean that you *don't* get to tell the core
development team our priorities are wrong when you haven't paid us a
cent. If that's the kind of relationship you want, pay a commercial
vendor to listen to your concerns - mediating between customers and
the community one of the services commercial open source vendors
provide (and the best ones will then contribute upstream development
effort on your behalf).

By contrast,community oriented open source operates on a time-based
economy - the only ways for people to get things are done to dedicate
their own time, inspire others to dedicate their time, or else to pay
developers (either directly or indirectly through a vendor) to spend
*their* time on projects of interest to those providing the funds.

The CPython core development team has made it clear we don't want to
support running Python 2 and 3 code in the same process because doing
so is *hard* (and perhaps even impossible without completely
redesigning the interpreter) and certainly *not fun*. The inspiration
based approach is unlikely to work in this case (because we've already
considered the possibility and dismissed it as impractical), but that
still leaves open the possibility of finding someone else that is
interested in attempting to prove us wrong, as well as people finding
a way to pay someone to make it happen.

However, in both cases, keep in mind that the people *most* familiar
with the CPython code base think it's a bad idea that probably won't
work. Instead, with the aid of many other members of the community,
we've created a source based migration approach that involves taking
advantage of the large common subset of Python 2 & 3.

If commercial operations higher in the stack would like to migrate to
Python 3, but are relying on dependencies that the community has
provided for free, then it's time to think seriously about *investing*
in the community and helping those projects to migrate (either by
contributing developer time directly or by contributing funds to the
existing developers for those projects). Note that the Python Software
Foundation is able to help mediate requests for assistance with
getting dependencies ported.

Regards,
Nick.

-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia


More information about the Python-ideas mailing list