[Incubator] Pros and cons of a DVCS (was: New Application: GeoToolkit)

[Daniel Morissette]

This DVCS exploration discussion is off-topic for the original thread 
which was about "GeoToolkit Incubation"... but I still think a DVCS 
discussion can be of interest for the Incubation Committee, that's why I 
started a new thread with a new subject.

Vincent Heurteaux wrote:
> Hello ,
> 
>> Perhaps some concrete examples of well-known large scale projects based on DVCS could help us see the light.
> 
> The first projects that are comming into my mind are :
> 
>     - Linux
>     - Firefox
>     - Java7 


Thanks. And to start answering my own questions I found this email from 
Linus Torvalds explaining DVCS to KDE developers (SVN users):

http://lwn.net/Articles/246381/

An interesting read, but I'm not sold yet.  ;)

The main benefit to me seems to be the possibility to create local 
experimental branches without asking for permission, much like the 
personal sandboxes that we have in some projects (OpenLayers and 
MapServer for instance), except that SVN sandboxes visible/public things 
and personal repositories are private. So that helps newcomers who want 
to play with the code in private which is good since that could lead to 
new contributors more easily, but we kind of do that with our SVN 
sandboxes already so we're not that far.

The need for a central/master repository for the official release is 
still there so from a social point of view I don't see how DVCS helps 
the release management bit. Actually the release management may be even 
more painful since you need to pull/merge from several repositories for 
every release... yuck!

Daniel
-- 
Daniel Morissette
http://www.mapgears.com/