There happens to be something called DVCS. People said its gonna solve all our problems. But then, different people had different view on our problems, and they had different ideas how to solve them. Out of the flux came to life generation one, followed by generation two and finally generation three. Evolution, they say. So we have Mercurial, Bazaar, Git as representatives from the latest DVCS evolutionary branch.

Now, there is this project that we’ll call GNOME for needs of this little essay. There are a lot of smart folks there doing some amazing work, pushing boundaries of what can be done with every release at least a little bit further. They’ve been using CVS for quite some time, and altough they didn’t really like its deficiencies, they always somehow managed to chum along with it. The “In good and evil” kind of relationship, you know. And then one day, enough! Decision was made to make a migration to Subversion, since it was supposed to make things better. As it happens to be, like with everything else, it brought good and bad things on the table. But eventually most of the things were polished by the rocking GNOME sysadmins, even if there is only a few of them.

Today, we’ve got some other winds blowing. There is an ongoing campaign to make a switch to a DVCS. Here is where things become a little more trickier. People usually discuss things before doing them, and that is actually good! But it becomes quite useless when all you’re doing is throwing balls at each other, and not actually listening to what others have to say. Personally, I am not a fanboy of any DVCS, am quite involved with both bzr and git communities, and have at least tried other DVCSs like Mercurial and Darcs. So before you even mention it, I’m not promoting any of those. What would *really* rock, and I know that upstream developers would appreciate it, is pointing out all those use-cases where you’ve experienced problems with a certain DVCS. If we can do that, upstream developers could work on improving their tools where needed. Perhaps, just perhaps, we can even give them a hand? At least one developer to the developers pool of some DVCS project would really make a difference, because those projects are quite young and don’t have many hands behind them.

Our administrators are worn out from SVN migration, and they are cautious of yet another migration just for the fun of it. If you really want to migrate to your favorite DVCS, write a plan and develop tools needed for the proper migration, do the required testing, and collaborate with upstream developers to improve their DVCSs in needed areas. Let us talk. For quite some time now I’ve been working with KDE to prepare a GitoriousKDE specification which could be further discussed at Akademy. Basically the specification defines a variation of Gitorious project specifically tailored to the needs of KDE project. Perhaps we can do something similar for the Gnome? Then, a moment before you start thinking about it, forget that it supports only git and let your imagination work. Perhaps we can let the developer of individual module to choose desired VCS from range of {bzr,git,hg} trio? Adding others would be fairly trivial anyway, since VCS stuff would be abstracted through not-yet-existant rvcs.

If you tell me that a common base for one of the key infrastructure pieces for KDE and GNOME is not useful, just stop for a moment, get a glass of water, and read this post again. Lets show FOSS collaboration in all its glory with this effort.

I am far from done with expressing my opinion on this matter, but I challenge you now …

continue reading.....