git, duplicity and dupinanny

We use duplicity at work to manage backups for a few Linux systems. I maintain a script to automate the process (error handling, email report, rolling backups), which I've dropped into a git repository at github.

The script may be useful to other duplicity users. I mostly set this up to get a chance to play with git, and so far I'm quite pleased by the experience. Maybe I'll find some other code to release which I'll drop in a mercurial repository.

Those new SCMs beg for some stress testing against large asset trees, to see how they would hold up for say, game development. My experiments with yam already hint that SVN 1.4 works nicely for large data trees, but suffers in checkout/update performance and distribution. If I ever get a few weeks of free time I'll get some more serious testing done with git.