Petr Rockai <[EMAIL PROTECTED]> writes:
> Hashed
> ------
> total time = 3.14 secs (157 ticks @ 20 ms)
> total alloc = 290,903,144 bytes (excludes profiling overheads)
>
> Oldfashioned
> ------------
> total time = 0.64 secs (32 ticks @ 20 ms)
> total alloc = 149,008,476 bytes (excludes profiling overheads)
Now that I have found a possible culprit for issue973, the new results are
somewhat different.
The original hashed-repository timing was:
12:26:56 | [EMAIL PROTECTED]:~/dev/divine/branch-1.3 -> time
pdarcs-colored-2.0.2 wh
No changes!
pdarcs-colored-2.0.2 wh 3,24s user 0,20s system 92% cpu 3,716 total
And now let's try a version with broken getSymbolicLinkStatus replaced with
getFileStatus, which...
12:27:00 | [EMAIL PROTECTED]:~/dev/divine/branch-1.3 -> time
pdarcs-colored-2.0.2-nolink wh
No changes!
pdarcs-colored-2.0.2-nolink wh 2,88s user 0,16s system 97% cpu 3,121 total
12:27:34 | [EMAIL PROTECTED]:~/dev/divine/branch-1.3 ->
pdarcs-colored-2.0.2-nolink revert
There are no changes to revert!
Finished reverting.
12:27:44 | [EMAIL PROTECTED]:~/dev/divine/branch-1.3 -> time
pdarcs-colored-2.0.2-nolink wh
No changes!
pdarcs-colored-2.0.2-nolink wh 0,36s user 0,07s system 99% cpu 0,429 total
Notice how doing a revert reduces whatsnew runtime from 3.1 to <0.5 seconds on
a mtime-working version of darcs (as it does mtime sync). It might be feasible
to sync mtimes in the working copy on actual whatsnew if they are different --
subsequent whatsnews will be blazing-fast.
For comparison, whatsnew on the old-fashioned repository:
12:33:19 | [EMAIL PROTECTED]:~/dev/divine/branch-1.3_oldfashioned -> time
pdarcs-colored-2.0.2-nolink wh
No changes!
pdarcs-colored-2.0.2-nolink wh 0,11s user 0,03s system 99% cpu 0,145 total
There is still room for improvement, apparently, but this is already much
better. I'm gonna grab GHC repository now to see how goes.
Yours,
Petr.
--
Peter Rockai | me()mornfall!net | prockai()redhat!com
http://blog.mornfall.net | http://web.mornfall.net
"In My Egotistical Opinion, most people's C programs should be
indented six feet downward and covered with dirt."
-- Blair P. Houghton on the subject of C program indentation
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users
