- From: Jim Ley <jim@jibbering.com>
- Date: Tue, 11 Dec 2001 12:17:04 -0000
- To: <w3c-wai-er-ig@w3.org>
"Danny Ayers" : > Hi Sean & all, > I've been playing with XPointer a bit (trying to hack annotations a la > Annotea) and though some of the points you raise are exactly the things I've > had trouble with, I'm not sure about your reading of the spec. The way I > read it is that an XPointer expression simply adds a bit of fine detail to > what you'd get with XPath or DOM2 - presumably going back to the Infoset > (but those are murky waters that I find a bit scary). But if document a has > XPointer expression B, then it's tied to that document whether we like it or > not. Yep, which means it can't be a fragment id on a http:// resource as that isn't a document, it could be anything the document is only one representation, and XPointer are only valid for that representation, current fragment IDs are valid for all. http://LordofTheRings/#FrodoGetsTheRing makes sense whether the document you get back is the film, the book, the film script or whatever the representation of the resource the request returns (based on accept headers or whatever.) whereas http://LordofTheRings/#xpointer(page=57) only makes sense for the book. This means I can't send you that url and have it mean anything unless I also tell you what accept-headers to send (and what if you get the large print version?) XPointers are very useful for specifying things in documents, but they shouldn't be considered a fragment of the url, they are pointers within documents not pointers within urls. Jim.
Received on Tuesday, 11 December 2001 07:18:13 UTC