[patch] syntax:plain for .hgignore
em at nerd.ocracy.org
Tue Sep 11 13:07:09 CDT 2007
Jonathan S. Shapiro suggerì:
> The numbers reported by "time" are notoriously inaccurate.
> Unfortunately, this is a case where running the test multiple times
> gives a very skewed view, because it will obscure the disk overheads
> that (in the real world) probably dominate this situation. Still, it is
> worth a try if our goal is to understand the pure RE overhead. Try
> running each test 5 times and send the numbers from the last run.
I'd also recommend repeating the tests dropping the fs caches before
each run, to test the cold cache performances:
# echo 3 > /proc/sys/vm/drop_caches
Complimenti per l'ottima scelta.
More information about the Mercurial-devel