Thanks, I will be off for 2 days and try that afterwards.

Many thanks again!


> -----Ursprüngliche Nachricht-----
> Von: Bruno Dumon [mailto:[EMAIL PROTECTED]
> Gesendet: Mittwoch, 3. September 2003 22:10
> An: [EMAIL PROTECTED]
> Betreff: Re: Incremental Processing in Cocoon 2.1
> 
> On Wed, 2003-09-03 at 20:40, Hauke Ernst wrote:
> > Very good idea, but not yet the solution. When I switch to "caching",
> > ThreadIdentifyingTransformer.class prints both thread-id's directly at
> > the beginning instead of printing the second one at the end, which was
> > the behaviour with profile-caching. This seems to be a good sign.
> > However, the output on the screen starts to appear if the
> > transformation is finished completely.
> > Even if I remove all transformations having only a generator and a
> > html serializer, it does not stream. Maybe it is the serializer?
> > Its configuration is
> >
> >       <map:serializer logger="sitemap.serializer.html"
> > mime-type="text/html" name="html" pool-grow="4" pool-max="32"
> > pool-min="4" src="org.apache.cocoon.serialization.HTMLSerializer">
> >         <buffer-size>1024</buffer-size>
> >
> > The buffer size seems appropriate. The testfile itself is HUGE.
> 
> I almost couldn't believe it, but indeed you're quite right. Cocoon will
> indeed buffer the entire output before sending it to the client. I guess
> this was done to make it always possible to reset the response in case
> an error occurs during execution of the SAX-pipeline.
> 
> The buffer-size parameter you show above doesn't really exist, I removed
> it from the default sitemap in CVS (it used to exist but that's changed
> now).
> 
> I was a little wrong in the syntax of the buffersize parameter before,
> it should be specified like this:
> 
> <map:pipe name="noncaching"
> src="org.apache.cocoon.components.pipeline.impl.NonCachingProcessingPipeli
> ne">
>   <parameter name="outputBufferSize" value="8192"/>
> </map:pipe>
> 
> If -1 is specified, the whole output is buffered before sending it to
> the client (= the default). Better don't use 0 or a very low value
> because that will have a negative impact on performance too.
> 
> I'll add this parameter commented out to the default sitemap together
> with a short explanation.
> 
> --
> Bruno Dumon                             http://outerthought.org/
> Outerthought - Open Source, Java & XML Competence Support Center
> [EMAIL PROTECTED]                          [EMAIL PROTECTED]
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to