DVCSs and DAGs (part 2)

One of the things I thought was interesting about this post was when he said "And what they find there is someone who doesn't seem to get it" (regarding Linus speaking on Subversion in his talk at google: http://www.youtube.com/watch?v=4XpnKHJAok8 ).

When I watched the talk initially, I had the impression that what made Linus' talk tip from "I don't think this is the right way of doing things" to "I think SVN is a pile of crap" wasn't so much on the history model (DAG vs line), but SVN's implementation. Specifically, I was thinking about how the base revisions are stored locally (with the .svn folders in each versioned control folder, which git resolves much more efficiently), the poor rename support etc. It's been a while since I saw that talk though, so it may well be that he was indeed criticising the Line model in general (and I can see how the line model doesn't fit in with his workflow at all.)