Mailing-List: contact forrest-dev-help@xml.apache.org; run by ezmlm
Precedence: bulk
Reply-To: forrest-dev@xml.apache.org
Message-ID: <3EEDE157.80409@apache.org>
Date: Mon, 16 Jun 2003 17:25:11 +0200
From: Nicola Ken Barozzi <nicolaken@apache.org>
Reply-To: nicolaken@apache.org
Organization: Apache Software Foundation
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
 rv:1.4b) Gecko/20030507
MIME-Version: 1.0
To: forrest-dev@xml.apache.org
Subject: Re: Forrest 0.5 outstanding issues (Re: Cleaning Forrest source
 directory
   madness)
References: <3EE8320E.2010601@apache.org>
 <20030614104743.GE2758@expresso.localdomain> <3EEB35A1.8020202@apache.org>
 <20030615042317.GA1127@expresso.localdomain> <3EEC96D1.60203@apache.org>
 <20030616094852.GB617@expresso.localdomain> <3EED95B7.200@apache.org>
 <20030616112204.GN617@expresso.localdomain> <3EEDD097.2050302@apache.org>
In-Reply-To: <3EEDD097.2050302@apache.org>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit


Nicola Ken Barozzi wrote, On 16/06/2003 16.13:

...
>>  - Speed!  Need to haul out the profiler and find out why normal Cocoon
>>    CLI rendering is 5-10x faster than Forrest.
> 
> 
> +1
> 
> I don't know if/when I'll have time, but I'll surely want to work on this.

Ok, so I ran hprof on it, with the krysalis-site skin ans PDFS, and it 
seems to tell me that most of the time is used in - guess what - in Xalan.

So I though, hey, it's our bl--dy skin that sucks, with wierd templates.
Guess what, this is what I got (all with PDF generation on):

  template:     31 seconds
  krysalis:     35 seconds
  forrest:      34 seconds
  forrest-css:  36 seconds

Wait a sec, it's not only the kind of skin then. It's about the number 
and types of xsl transformations that it has to go through!

So, what I will have to do now is to do a profile on Cocoon, because the 
java one has quite clearly shown that the problem is not in link 
translation or other Forrest-specific component, but in the xsl 
transformations, where they are, and what they do.

Also, IIUC we are still using the old crawling method, is this correct?

-- 
Nicola Ken Barozzi                   nicolaken@apache.org
             - verba volant, scripta manent -
    (discussions get forgotten, just code remains)
---------------------------------------------------------------------