Thanks Carsten!
I got my source working now, but it's awfully slow. The source gets a specially modified zip-file from any cocoon source (context used) and after working on it, extracts a file from this zipped package. The slow section seems to be the unzipping part. Also, no caching seems to happen by Cocoon. A web page that called this source 10 times took 20 seconds to load on a fairly fast machine. The zip file is only 84Kb!
Any optimization tips? Didn't think Java's zip-utilities are so slow...
The Cocoon caching needs a valid SourceValidiy object from your Source implementation in order to detect if the cached content is still valid. Perhaps you're not returning this?
Carsten
