On 15.12.2007 19:35 Uhr, Boisvert, Eric wrote:
I agree in principle, this is how cocoon works, in theory. But, for example, as soon as one puts an XSLT transformer in the pipeline, the document is indeed read completly in memory (in a DOM structure) simply because XSLT (XPath) allows to reference any part of the document - which force the whole document to buffered in memory prior to execute XSLT.
That's not quite true. Yes, you can write XSLT in a bad way (using XPaths starting with double slash for example), which indeed forces the whole document to be loaded. But if you write it in a good way the XSLT processors have some optimizations to only load the most necessary parts of the document.
Now I'm redesigning the architecture and exploring the idea of using Cinclude to dispatch WFS queries and I was afraid Cinclude was 'buffering' the document in memory
From the implementation of the CIncludeTransformer: It uses SAX to stream through the included documents - except when you use the select attribute. Then it creates a DOM and executes the select on it.
Joerg --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
