cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Grzegorz Kossakowski <gkossakow...@apache.org>
Subject Re: pipelineComponent scope troubles
Date Fri, 14 Sep 2007 21:27:38 GMT
Grzegorz Kossakowski pisze:
> Giacomo Pati pisze:
>> You'll find the log for request scoped at http://people.apache.org/~giacomo/request-log4j.log
>> and singleton scoped at http://people.apache.org/~giacomo/singleton-console.log.
Singleton scope
>> doesn't even start the app.
> 
> Ok, that is all weird and definitively worth looking at. I'm just about creating similar
setup to
> yours and debug a little.

Giacomo, I have some news. I managed to reproduce your problem by creating two blocks and
setup them
as we discussed earlier. It turned out that I was right suspecting it's a problem with a scope
but
with pipelineComponent only call scope. Your problem is as simple as the fact that flowscript
put on
different instance of OM than jx was pulling from.

It is caused by the fact that call scope (and pipelineComponent scope is meant to be "inside"
call
scope) is empty whenever new internal call to another servlet is made. That means, when Cocoon
makes
servlet:bfabric-core:/screen/{1} request OM containing status variable is not available in
a
bfabric-core. It means that call scope and whole SSF is designed to NOT share any environmental
data, at least at this stage. I discussed it in detail in this e-mail[1] but I got only one
response, sadly enough.

I'm almost sure that if you change scope of OM to request it's going to work. Just make sure
that
old junk is not sitting somewhere in the classpath. If it's not the case, I must get working
example
of problematic behaviour in order to help anyhow.

[1] http://article.gmane.org/gmane.text.xml.cocoon.devel/74571

-- 
Grzegorz Kossakowski
Committer and PMC Member of Apache Cocoon
http://reflectingonthevicissitudes.wordpress.com/

Mime
View raw message