On Sun, 2010-01-10 at 14:13 -0500, Steinar Bang wrote: > And then later, when I'm online, I do a an update against the > repository, with virgin versions of the changed files, and then I copy > in the copied-aside files, and do my checkins in stages.
FWIW, a lot of people seem to use git as a Subversion client (using git-svn) for pretty much exactly this workflow. bzr also claims to work well against a Subversion upstream branch but I don't know anyone who uses it that way. It's sort of interesting that, despite fairly large discrepancies in data models, "atomic whole-tree versions of a single branch" is enough of a lingua franca for modern version control tools to interoperate to some degree.
