cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stuart Roebuck <>
Subject Re: Profiling Cocoon for cheapskates
Date Fri, 06 Dec 2002 00:11:13 GMT

On Thursday, December 5, 2002, at 09:39  pm, Stefano Mazzocchi wrote:

> Vadim Gritsenko wrote:
>> Stuart Roebuck wrote:
>>> I need to do some code optimization / profiling on Cocoon and some   
>>> additional Cocoon libraries we've built.  This raises an issue which  
>>> I  thought might be of general interest, and hoped might open up  
>>> some new  avenues of profiling discovery!
>>> If there was an open source based approach to profiling Cocoon code,  
>>> I  would be happy to write up the process to aid others in the  
>>> future.
>>> x--------------------x
>>> x  Unfortunately...  x
>>> x--------------------x
>>> * We're on a budget and can't afford to pay $400+ per seat for a   
>>> commercial profiling tool.
>> ...
>> OptimizeIt costs money, but I've got a license... I *will* profile  
>> Cocoon sometime again, and would do it sooner, but my unfortunate  
>> schedule moves this event to the 1Q or 2Q of the next year.
> I got a 30-days evaluation license of OptimizeIt and run it against  
> Cocoon a while ago and I found out that cocoon (talking about CVS  
> HEAD) is pretty well optimized already, expecially for memory  
> consumption over requests.
> Look into the mail archives for that message since I sent a report  
> here to the list.

Thanks for that.  I did read through those mailings both at the time  
when they were posted, and yesterday when I was searching the archive  
for hints and tips.  My selfish interest was in finding ways of  
profiling some of our own code with works on top of Cocoon, but the  
principles of profiling Cocoon and related code should be the same, so  
I hoped it was appropriate to ask for any hints and tips.

 From my point of view, whenever I have been writing code for Cocoon I  
have done a lot of unit testing on components utilizing 'Mock' classes,  
and I have done more broad brush system performance testing, but I have  
always shirked away from full blown profiling because attempts at it  
have met with failure or masses of incomprehensible or useless data.

If I could find a simple way of profiling code 'in context' it might  
help to improve the performance of our code.

So far, jProfiler (which offers an 8 day trial) has proven very easy  
and effective and seems to run fine on Java 1.4.1 on Mac OS X.  Now I  
just have to find some way of paying for it!


            Public Key - 1024D/88DD65AF 2001-11-23 Stuart Roebuck  
      Key fingerprint = 89D9 E405 F8B1 9B22 0FA2  F2C1 9E57 5AB1 88DD  
Stuart Roebuck                          
Systems Architect                             Java, XML, MacOS X, XP,  

To unsubscribe, e-mail:
For additional commands, email:

View raw message