On Wed, Oct 04, 2006 at 08:11:00PM -0700, Justin Patrin wrote: > Hmmm, no, this still seems to take a long time to me.
Damn. :-) > The update now > quickly figures out the branch (well, outputs), and selects (outputs) > the update target Excellent; certainly this is the part I expected to be much faster in 0.30 >, but the update still takes a while to seem to do > anything else. It definately takes longer on our .dev branch than on > our .oz354x branch (.dev has many more revisions). Perhaps it's just > me being impatient (I don't think so, I don't particularly mind the > wait, I'm just making observations) and not realizing the amount of > work that goes into figuring out how and what to update. Could you see if inodeprints helps here please, just out of curiosity? It seems likely that at this point it is simply seek bound examining the repository for local changes, and this could be very dependent on hardware, filesystem memory/cache size, etc. I know these things can make an *enormous* difference to "cvs update" speed, which can be very heavily disk bound if the workspace won't all fit in cache. -- Dan.
pgpx4dxKS33e3.pgp
Description: PGP signature
_______________________________________________ Monotone-devel mailing list Monotone-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/monotone-devel
