On Wed, Aug 27, 2008 at 4:06 PM, Eric Y. Kow <[EMAIL PROTECTED]> wrote: > On Tue, Aug 12, 2008 at 10:35:34 -0700, [EMAIL PROTECTED] wrote: >> I'd just point out that while low-level optimization is easy, it's >> much more valuable to do high-level optimization. However, an >> improved hunk or binary patch format would be a worthwhile low-level >> optimization. > > Are there any high-level optimizations you think we should attempt?
There's always the long-standing, well-understood darcs annotate problem. We know what to do, it's just that no one wants to do it. In general, figuring out where the high-level performance problems are. The low-level code is generally fine, all significant darcs-2 regressions come from doing more work than darcs-1 did. In some cases, this is a result of a bug-fix, and can't be easily reverted, but in other cases who knows? But in general, low-level optimization is absolutely the wrong thing to do (and that's not just true of darcs). David _______________________________________________ darcs-users mailing list darcs-users@darcs.net http://lists.osuosl.org/mailman/listinfo/darcs-users
