Performance regression on case-insensitive filesystems (Windows)

[Martin Geisler]

Matt Mackall <mpm at selenic.com> writes:

> On Fri, 2012-07-20 at 10:24 +0200, Adrian Buehlmann wrote:
>> On 2012-07-20 09:56, FUJIWARA Katsunori wrote:
>> > Hummm, it's mysterious.
>> 
>> FWIW, I don't trust 'hg --time' that much on Windows.
>
> I don't know why you wouldn't. I would instead say that the results of
> any single run on a modern operating system aren't terribly meaningful.
>
> The right performance test is the perf extension in contrib/,
> specifically the perfstat command.

I've tested again with 'hg perfstatus' and I get numbers that are
inconsistent with 'hg status --time':

  version  perfstatus  status  overhead
  2.0      2.31        3.0     30%
  2.3-rc   3.48        4.2     21%
  patches  2.65        5.0     89%

So the patches make perfstatus much fastr, but the overall time is not
improved. I had expected a (more or less) constant ratio between
perfstatus and status.

-- 
Martin Geisler

aragost Trifork
Commercial Mercurial support
http://aragost.com/mercurial/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20120720/e72bc1f8/attachment.pgp>