On Mon, Jun 04, 2007 at 08:54:52PM -0700, eliott wrote:

Just some random information, that I thought I would share. I was taking a look at how much traffic different protocols take to accomplish 2 things.

1) Sync the abs tree with a fresh pull 2) Perform a 'sync-check' (test to see if there are any new changes) with no updates available.

http://stdout.cactuswax.net/cactus/tests.txt

I was using fresh checkouts, so I don't have any other data other than what appears above. I would like to test some of the protocols against trees that have updates, git repos with more history than fresh imports, and for potential degradation as size increases. I don't know if I will get around to it though.

If anyone else is interested in testing such things, then let me know. I would be very interested in the results.

That's an interesting set of tests. You could keep the git repo in sync with git-cvsimport (I believe it will create a patch for ever cvs commit). That could be automated pretty easily. Abs will just work w.r.t. staying up to date. And rsync will just have to be updated any time you want to test. The hardest part is remembering to do the tests over and reproducably (keeping copies of the previous test directories). Jason