[ I'm reproducing here a comment I left on Groklaw in response to someone claiming that BK was bad because you couldn't extract the metadata. ]
I’ve been using the free (as in beer) BK client for a while, about since Linus started using it I think. Did you? Where did you (or anyone else, for that matter) get the impression that you couldn’t extract all the metadata information with it? You can get diffs between any two revisions. You can export changesets as diffs, with all the comments and other data: try “bk export -tpatch -r1.2291,1.2292″ on a recent linux-2.5 tree for example.
What is undocumented is the actual format of the BK files and the protocols between client and server, but there’s no reason to get at them since you can export everything you need in other ways. Or, if you don’t want to use BK yourself, you can ask someone who does to export what you need. In this respect, and I’m quoting Linus here, it’s exactly the same as someone not wanting to use CVS: (s)he could ask someone else to export what (s)he needs. And I bet Linus would have been glad to export just about everything if asked politely.
When I first saw news of this matter, I thought “Larry is a bad guy”, but after I read the points of view of those involved I completely agree with Linus. Those who did not want to use BK had absolutely no reason to feel “left out in the cold”: as Linus himself said when he started using BK (and repeated ad nauseam), nothing would change for the worse for them. To the contrary, it was so easy to get snapshots out of BK that the -bk tarballs were made available daily.
BK is head and shoulders above other SCMs. If you dont’want to use it for ethical reasons, fine; if you want someone else to stop using it, either write a better tool (or at least a tool “good enough”, as Linus often said) or shut up. Tridge was not writing a better tool, only another tool to extract data from a repository, and that was useless. When asked to stop, he refused. When asked to stop Or Else, he refused. If what he wanted was really acess to the metadata from BK, I’m sure Linus and Larry would have worked out something, like exporting all changesets in a reasonable format. By doing what he did after knowing that he would disrupt a well-working and provably useful system, he did only a disservice to the community that he has otherwise served so well with his work on Samba.