- From: Eric van der Vlist <vdv@dyomedea.com>
- Date: 28 Sep 2002 23:12:42 +0200
- To: Didier PH Martin <martind@netfolder.com>
- Cc: "'Jeni Tennison'" <jeni@jenitennison.com>, www-tag@w3.org
Didier, On Sat, 2002-09-28 at 21:48, Didier PH Martin wrote: > > Hi Eric, > > Eric said: > No, I don't think that XSLT would need to be updated: XSLT processes > trees but do not define which operations may or may not be done before a > tree reaches it. > > Didier replies: > Eric why do say that? Its obvious that XSLT will have to be updated > since if the mapping is defined in a css document it will need to get it > in order to know the mapping. Can you explain in detail how, if the > mapping is contained in a CSS document, an XSLT engine will be able to > process the mapped object. Concretely speaking, let's say that the src > attribute is inherited from the xlink:href attribute. How will you do to > process a document containing only elements with the scr attribute and > having them matched with a template like the following <template > match="@xlink:href" >. How would you do that? If I feed the XSLT processor with the output of the SAX filter which I have proposed this morning, the XSLT processor will "see" xlink:href instead of src and I don't need to change anything in XSLT (nor even to the processor) for this. Why would it be different if the filter was reading a HLink declaration using a CSS format than it is when it reads HLink/xml? > > Eric said: > Jeni is right (IMO) when she says that a processing model could be > defined and this processing model could add CSS properties to the tree > if needed (pretty much like a SVG DOM consolidates information from XML > and information from CSS). > > Didier replies: > Yes I requires a modification of the existing XSLT specs and processing > model. It needs to import a CSS document to know the mapping. No, all it requires is a modification of the layer which reads a document and send it as SAX events, DOM or whatever to the XSLT processor. > Eric said: > This kind of operation which can be done by a SAX filter as I have shown > for HLink would be slightly more complex but could be done if CSS was > used. I think that these are implementation details and several orders > of magnitude below the level of complexity of current UAs! > > Didier replies: > Yes I know, we can about everything with code. But this is not the > point. The point is that to support the mapping in CSS document you'll > need to modify the existing XSLT implementations and you'll need to > modify the existing XSLT spec. Instead of modifying the XSLT processor, you can modify what the XSLT processor will see, that's much simpler (and common practice). > Eric said: > What is more difficult IMO is to define the *right* architecture and > model for the hyperlinks. > > Didier replies: > Yes I agree and this is precisely what are all trying to do, this time > by measuring the ecological impacts to the other W3 domain languages and > in general to the actual XML architecture (also to the XML legacy out > there). Great! Eric > Cheers > Didier PH Martin > > > -- Rendez-vous à Paris. http://www.technoforum.fr/integ2002/index.html ------------------------------------------------------------------------ Eric van der Vlist http://xmlfr.org http://dyomedea.com (W3C) XML Schema ISBN:0-596-00252-1 http://oreilly.com/catalog/xmlschema ------------------------------------------------------------------------
Received on Saturday, 28 September 2002 17:12:46 UTC