- From: Jeni Tennison <jeni@jenitennison.com>
- Date: Wed, 12 Apr 2006 11:38:15 +0100
- To: public-xml-processing-model-wg@w3.org
Rui Lopes wrote: > Norman Walsh wrote: >> An XSLT 1.0 processor can only have a single principle input document >> (I don't know how, or if, we deal with the fact that the stylesheet >> can load auxilliary documents), but it can produce multiple result >> documents. (Ok, technically, it can't, but every XSLT 1.0 processor >> except the MS one has an extension that does so.) > > Should we dismiss stylesheet's auxiliar documents? The same question may > be asked regarding an XInclude processor, should we dismiss the > documents specified in xi:include elements? I raise the question, > although I haven't got an answer to that :) My take on this is that we *must* support XInclude, and however we support XInclude's access to auxiliary documents, the same method should apply to XSLT. An XInclude processor needs to somehow get hold of a document given a URI. Either the pipeline processor controls the retrieval (acts as an entity resolver) or the component does. The current design implies the latter. We also need to decide whether/how to support the situation where the auxiliary document is itself the result of a pipeline, or whether we explicitly exclude this. We need to say what happens when a "Saved" document is "Loaded" or requested as an auxiliary document, particularly if we want to allow parallel processing strategies. Cheers, Jeni -- Jeni Tennison http://www.jenitennison.com
Received on Wednesday, 12 April 2006 11:15:00 UTC