[darcs-users] darcs patch: Remove premature optimization.
Jason Dagit
dagit at codersbase.com
Sat May 23 08:18:17 UTC 2009
On Fri, May 22, 2009 at 9:58 PM, Trent W.Buck <trentbuck at gmail.com> wrote:
> Out of interest, I replaced the horribly confusing code with an
> obviously inefficient -- but readable -- version, and timed
> difference. The sum of three runs of "darcs show authors" on the
> Darcs repo and the darcs-wiki repo, on a fast system, is:
>
> real user sys
> 1.090s 0.670s 0.140s -- without this patch
> 3.394s 3.080s 0.160s -- with this patch
>
> So readability will quintuple the CPU time, though the operation's
> total time is negligible. Is it worth applying this patch?
My preference would be to add a comment saying that the optimized code
is just equivalent to your version. That way people reading the
source have a better clue what it is doing, but the people using darcs
don't have to wait any longer than they already wait. Do you think
that would be reasonable?
On a side note, premature optimizations are bad, but I doubt I would
consider this such.
Jason
More information about the darcs-users
mailing list