[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
bug#2184: 23.0.60; performance issue since CVS 2009-01-30
From: |
poppyer |
Subject: |
bug#2184: 23.0.60; performance issue since CVS 2009-01-30 |
Date: |
Wed, 04 Feb 2009 21:53:25 +0800 |
User-agent: |
Emacs Gnus |
Adrian Robert <adrian.b.robert@gmail.com> writes:
> I have not noticed performance problems here but there have been
> significant changes from 01/28 through today 02/04 and it is possible
> there are problems under some usage conditions. First, could you
> test again using current CVS (including 02/04 changes)? Also, is
> there an easy way to obtain the 100% CPU usage cases in a mode that
> comes with emacs? What OS X version are you on?
Yes, today 02-04 is the same, and I am using OSX 10.5.6.
I am sorry but I am unable to reproduce it in other simple setup.
However, we seem to trace down that the issue is about regular
expression matching, as also mention in:
http://lists.gnu.org/archive/html/emacs-devel/2009-02/msg00012.html,
there are some clue about function "re-search-forward"
I profile in my w3m case, with exactly the same procedure here comes:
Function Name Call Count Elapsed Time
Average Time
====================== with 2009-02-04 Emacs executable =======================
regexp-opt-group 2668 2.5593199999
0.0009592653
re-search-forward 730 1.4294740000
0.0019581835
run-hooks 38 1.3881749999
0.0365309210
regexp-opt 162 0.8015100000
0.0049475925
regexp-quote 1650 0.0171659999
1.040...e-05
remove-text-properties 506 0.0074000000
1.462...e-05
====================== with 2009-01-19 Emacs executable =======================
regexp-opt-group 2668 0.4568619999
0.0001712376
regexp-opt 162 0.1268280000
0.0007828888
re-search-forward 730 0.0308000000
4.219...e-05
run-hooks 37 0.029099
0.0007864594
run-at-time 41 0.0009150000
2.231...e-05
regexp-quote 1650 0.0008870000
5.375...e-07
remove-text-properties 506 0.0007210000
1.424...e-06
Is there any specific changes in regular expression part in 2009-01-30
(not about lisp but the Emacs executable)?
>
> thanks.