Brief article on benchmarks of Python repository with leading DVCSen

[Paul Moore]

2009/2/13 Barry Warsaw <barry at canonical.com>:
>> No, it doesn't make a difference to me because my only interface to
>> the core repository is via submitting patches. I can merge or use
>> branches as much as I like for myself, but I *have* to submit a patch
>> at the end of the day because I am not a core committer.
>
> In my ideal world, no one would ever submit a patch.  Patches are dead
> things.  Much better would be to request a review and/or merge of a branch.
>  A branch is a living thing.  You can update a branch much more easily when
> you notice a typo or your reviewer gives you comments,  You can update a
> branch when trunk changes render moot some of your branch's changes.

That would be nice, but it does imply that everyone who contributes to
a project have relatively stable, long-term public hosting for patches
(after all, plenty of tracker items are round for months, if not
years). That's a pretty high barrier to entry.

Of course, if there were a bug tracker which allowed you to host a
branch *in* a bug report (exactly the same as uploading a patch) that
might be pretty neat. Although the storage for all those full
branches, each of which held a 2-line doc patch, might be an
interesting challenge to manage :-)

Paul.