3.2 performance measure

Bradley Baetz bbaetz at acm.org
Sat Sep 20 01:09:25 UTC 2008

On Sat, Sep 20, 2008 at 3:12 AM, Atsushi Shimono <shimono at mozilla.gr.jp> wrote:
> Hash: SHA512
>  hi,
> Frédéric Buclin wrote:
>> Atsushi Shimono a écrit :
>>>   From this graph, I might be able to say that 3.2 shows less performance
>>> than 3.0 by 1.2 to 1.3 times, but this might be explained by new features
>>> in 3.2.
>> It would be interesting to know which parts of the code slow down the
>> process. Maybe using DProf would help.
>  Yeah, i think to try with DProf. I think i've heard that it has some
> problem related on perl TT. So, i'll report something including that.

Last time I tried dprof on bugzilla, it didn't break down stuff inside
TT that usefully. One thing that comes to mind would be if for some
reason something was doing a bad job with multi-byte characters, eg
some of our templates use regexs, .contains, and so on that aren't as
easy to optimise when each character takes a different number of
bytes. Has anyone done a similar look using the english locale? What
perl version is this - 5.8 or 5.10?


More information about the developers mailing list