Brief article on benchmarks of Python repository with leading DVCSen

[Paul Moore]

2009/2/12 Barry Warsaw <barry at python.org>:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Feb 11, 2009, at 12:58 PM, Paul Moore wrote:
>
>> OK, here's a question. Given the existence of the current "unofficial"
>> DVCS branches, what benefits are there to non-core committers of the
>> core moving to a DVCS?
>
> Lots, many of which are outlined in PEP 374.

You may have misunderstood me - at the moment, I hack on Python using
Mercurial, based on the available hg mirror of Python trunk. The fact
that the "official" repo is in Subversion doesn't make any difference
to me - my interaction is by means of hg clone, local work in hg, and
submission of patches.

If Python core development switches to a DVCS, what difference would
that make to me? (Assuming that (a) the hg mirror doesn't disappear,
and (b) I'm happy with Mercurial and have no wish to change my local
practices). It's hard to see any, given that I currently ignore the
fact that the core is in Subversion :-)

>> If someone confirmed that regardless of which DVCS was chosen, a
>> maintained, up to date Mercurial mirror of Python would be provided
>> (and that patch submission processes would remain VCS-neutral) then
>> I'd stop bothering.
>
> To the extent that such mirrors can be run by people outside the small and
> overworked python.org admin team, I'm all for it.  Maybe this will motivate
> people to make Tailor rock and then provide that as a service to the Python
> community.  I seriously doubt the py.org admins will take any of that work
> on.  Dedicated, long-term volunteers are always welcome and might influence
> the decision to pull that service in-house.

To an extent, that's what I'm trying to establish - whether
svn-as-master makes it easier for some volunteer to maintain a (hg,
for my particular preference) mirror, as opposed to
bzr/git/hg-as-master. Clearly, my ideal would be for Antione to
continue managing the hg mirror, but given that it's based on his
hgsvn, he may be less inclined to mirror a bzr/git master similarly...

I guess I've made more than enough of this point now (and it's
severely off-topic for the Bazaar list by this stage!) so I'll leave
it at that. I've suggested to the PEP authors that adding a section
evaluating "ease of maintaining foreign mirrors" would be useful -
we'll see what comes of that.

Paul.