Python Goes Mercurial

David Cournapeau cournape at
Sun Apr 5 10:45:25 CEST 2009

On Sun, Apr 5, 2009 at 3:48 PM, Jeroen Ruigrok van der Werven
<asmodai at> wrote:
> -On [20090405 06:05], Michele Simionato (michele.simionato at wrote:
>>P.S. the thing I do not understand if why we are moving
>>away from Subversion. Will all the new features entered
>>in 1.5 and 1.6 Subversion is now not so bad as it used to
>>be and it has the advantage of being already there and
>>familiar to most people.
> I don't fully get the whole 'we MUST go DVCS!' meme going around the
> Internet either.

I think it is often poorly 'marketed', and not well explained. But as
someone whose first source control system which made sense was a DVCS,
I can assure you that svn feels clumsy, slow, inadequate and extremely
primitive. Most people justification for svn is based on the fact they
are used to the workflow. Using DVCS to mimic the svn workflow is an
error IMHO. What would have happened if people tried to mimic the
visual sourcesafe workflow, with locks and co, with svn ? To me,
defending svn against DVCS is like defending the lock-per-file
approach of visual sourcesafe and the likes. Believe or not, there are
some people who think this lock-based workflow is better.

A simplistic analogy: svn it to CVS what C++ is to C, but try selling
C++ to python developers :)


More information about the Python-list mailing list