[Catalyst] Speed issue ?

[Sebastian Riedel]

Am 02.09.2005 um 18:07 schrieb Bernard FRIT:

> 2005/9/2, Andrew Ford <A.Ford at ford-mason.co.uk>:
>
>> What is the speed on the second request for the SpeedyCGI engine?
>>
>
> SpeedyCGI (same requests)
> 1.
> [Fri Sep  2 17:57:56 2005] [catalyst] [info] Request took 1.578377s  
> (0.634/s)
> 2.
> [Fri Sep  2 18:00:34 2005] [catalyst] [info] Request took 1.430226s  
> (0.699/s)
> 3.
> [Fri Sep  2 18:01:40 2005] [catalyst] [info] Request took 1.166644s  
> (0.857/s)
>
> FastCGI  (same requests)
> 1.
> [Fri Sep  2 17:52:25 2005] [catalyst] [info] Request took 1.656971s  
> (0.604/s)
> 2.
> [Fri Sep  2 17:53:16 2005] [catalyst] [info] Request took 1.211765s  
> (0.825/s)

I must admit that i personally never tested the SpeedyCGI engine, but  
this definately doesn't look like a persistent interpreter! (compared  
to the pure CGI results)

>
>
>> Speedy should be about the same speed as CGI for the first  
>> request, but
>> on the the second and subsequent requests (for each Speedy  
>> process) it
>> should be faster as it will not need to re-load any modules used or
>> re-initialize the application.
>>
>
> It sounds we could not expect to gain a lot. The "subjective" time in
> front of the browser window is much more important as I guess catalyst
> doesn't log tt2 template generation time.

You should really try mod_perl or FastCGI, even the built in server  
should perform better than the results above!!!
Every action call (including forwards) get profiled, just look at the  
table in debug mode.
You can even get profile data for each template (as comments in HTML  
output) by activating the Template::Timer support.



--
sebastian