Bernd Nowak, on Mar 17 2009, 08:56 PM, said:
having nearly a decade of experience with CVS and a fair share of time with SVN and GIT i would like to give some comments. CVS is incredible old school and runs simply everywhere and with every IDE there is. providing a repository is as simple as installing CVS on the server and giving the users a ssh-account. providing public repos is more tricky. SVN is rather newish and fix some of the very few problems of CVS (or at least stuff, that one might see as inconvinience). yet stuff like public repos with WebDAV access come with the package - but also need knowhow in other areas like webservers to config properly. GIT is quite insane to start with - but also very mighty (focused on patch based development, easy to branch, easy to just work as a local repo (directly where your source is)). migrating from one to the other (which nowadays usually means CVS -> SVN, CSV -> GIT, SVN -> GIT) is rather easy.
yet what all of these systems share - and i even dare to say lots of the commercial systems even if just know few - is the focus on source-code. handling .xml, .cc and .txt is their big strength. having lots of binary files around in the world of GPL editing turns them just into "logs". not that this is a bad idea also to have all the versions of a binary file and know what or why something changed, but beeing able to see the changes on some 100k SLOC project is quite a treat.
one other Good Thing of such a repo would be the "global" backup in one place for all the good stuff, that is around.
well i am all for version controlling - i even hold my home directory in one to make new servers/computers comfy in no time. yet you will have some good share of user education and evangelism ahead in an artistic environment like here.