[Python-Dev] Move selected documentation repos to PSF BitBucket account?

[Nick Coghlan]

On 23 November 2014 at 16:10, Guido van Rossum <guido at python.org> wrote:
> On Saturday, November 22, 2014, Nick Coghlan <ncoghlan at gmail.com> wrote:
>> The learning curve on git is still awful - it offers no compelling
>> advantages over hg, and GitHub doesn't offer any huge benefits over
>> BitBucket for Sphinx based documentation (ReadTheDocs works just as
>> well with either service).
>
>
> Git may well have a learning curve, but ever since I "got" it I started
> preferring it over hg.

I took the git knowledge I acquired by necessity at Red Hat and
figured out how to apply it to hg. All the same features are there in
hg, they're just switched off by default (mainly because the core
Mercurial devs are adamant that any potentially history destroying
operation in a version control system must be opt-in). In particular,
the evolve extension is an impressive tool that allows history to be
edited in such a way that the edits can be safely shared amongst
repos.

> Too bad for BitBucket, but most people who started contributing to open
> source in the past 5 years already have a GitHub account.

You can log into BitBucket with a GitHub account.

More generally, I'm very, very disappointed to see folks so willing to
abandon fellow community members for the sake of following the crowd.
Perhaps we should all just abandon Python and learn Ruby or JavaScript
because they're better at getting press in Silicon Valley?

>> Note that if folks prefer Git, BitBucket supports both. I would object
>> strongly to unilaterally forcing existing contributors to switch from
>> Mercurial to git.
>
> What about potential new contributors? And the hg-git bridges that git fans
> are always referred to work in the opposite direction too... :-)

We already have lots of potential contributors (if we didn't, review
bandwidth wouldn't be the bottleneck the way it is today), and the
functional differences between GitHub and BitBucket from a barrier to
entry perspective are so low as to be trivial.

For organisations with no vested interest in BitBucket or Mercurial,
sure, they may as well just go with the popular choice.

Unlike most organisations, we have a vested interest in encouraging
the growth of the Python ecosystem and community, and the developers
of both Mercurial and BitBucket are members of that community.

While we have historically done very little to date in reaching out to
the Mercurial developers and seek their assistance in improving the
developer guide, remedying that and providing better Mercurial usage
guidelines is one of my hoped for outcomes in making it easier for
others to contribute to the developer guide.

By contrast, GitHub is a Ruby application, and git is a collection of
C and shell scripts - they may be tools used *by* the Python
community, but they are decidedly not products *of* the Python
community.

Given the choice between two functionally equivalent solutions, one of
which lets us continue using the tools we already use (including all
the associated automation), and is created by members of the same
development community, I think the burden of proof falls on the folks
that want to make the backwards incompatible change that the
additional cost will be worth it.

Moving from self-hosted Mercurial repos to externally hosted Mercurial
repos is a low risk change. It reduces maintenance overhead and lowers
barriers to external contribution, both without alienating existing
contributors by forcing them to change their workflows.

Proposing to *also* switch from Mercurial to git significantly
increases the cost of the change, while providing minimal incremental
benefit.

Regards,
Nick.

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