</snip>
> 
> A bit wild guessing, isn't it? ;)

Kind of, yes :-)))

> 
> AFAIK there is something like URLHandlers which have to be 
> registered on the
> XSLT processor - but I actually never worked with that API 
> directly, so I do not
> really know. What I wonder about is that Saxon behaves 
> differently than Xalan.
> But I'd put my shirt off that it is not because Xalan is also 
> from Apache :) 
> 
> > Seeing its implementation, I am curious about what it means 
> for the caching
> > impl. I can not imagine that the validity of the cocoon:// 
> protocol in
> > document() function is added to the stylesheet validity 
> (TraxTransformer
> > validity), so I would never chose for this solution anyhow 
> (unless by some
> > magic it is added, but I don't think changes in the 
> pipeline called by
> > cocoon:// are picked up in the cached result, unless of 
> course you chose to
> > have no caching at all)
> 
> Exactly the caching has always been the reason to not 
> recommend the usage of
> document(). It should only be used for REALLY static 
> documents, as it is not
> included in any cache validity (AFAIK).

And you end up with the really annoying "having to clear your cache whenever 
the include cocoon://foo" might change. Just take it with an aggregate or an 
include I would suggest...

Ard

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

Reply via email to