- From: Sean B. Palmer <sean@mysterylights.com>
- Date: Sat, 20 Oct 2001 18:26:39 +0100
- To: <w3c-wai-er-ig@w3.org>, "Al Gilman" <asgilman@iamdigex.net>
- Cc: <wai-tech-comments@w3.org>
> > In HTML the use of the LINK element seems to be > > appropriate for linking to the report, there are certainly > > no others within the current recommendations. Before I get to Al's rant about <link>, I'll add my own, previously unarticulated, comments. The idea of separating literal valued data from URI-reference valued data is useful, and compares to what was done in RDF, the "de facto" (triple quote it) metadata standard. The head[@profile] attribute adds value in that both meta[@name] values and the link-types datatype are contained within the profile space. HTML does not say how they are contained within the profile space. This is comparable to the fact that the XML Namespace specification does not actually reveal the structure of namespaces. It leads, in some senses, to broken Web architecture, as people start to define - by specification - the workings of the QName (which is also what the profile/value couplets in HTML are, when you think about it), rather than by documentation, or by machine readable assertion. The HTML link type system is a step more broken, because in some situations, link types used in conjunction with others "mean" special things. e.g. "alternate stylesheet". The same goes for an "alternate" link value used in conjunction with a language value (@xml:lang). This is a terrible way of going about things if you want to port to machine readability, because it means you start having to come up with complex predicates almost immediately, and build a case exceptions list for the patterns which don't fit. Indeed, I already ended up having to do this in my latest "RDF-from-XHTML" attempt, and that forms the basis for the rant; this is based on conclusions from a practical experiment. The conclusions is that LINK and META are just poor bits of metadata kit; weak extensiblity, and a mess of custom semantics. [...] > > "Authors may use the following recognised link types, listed > > here with their conventional interpretations. - -" That implies semantics that are dynamic, and based upon the context of use/processing. Ugh. [...] > The META syntax is limiting in having only one name-value pair > at a time. The ideal thing here is something on the order of > > <META KEYWORDS="<list>" HREF="<uri-reference>"> > > This would indicate that for more meta-information in any of the > concern areas mentioned in the list of keywords, it would be > suggested to process the resource indicated in the HREFed > resource. By "<list>", I presume you mean a space separated list of NCNAME tokens, or something along those lines? In that case, you would need to have processor recognition of those values in order to come to the next part of the process, which is that I want to know more about this "topic", dereference the HREF. Perhaps a list of QNames would be better? It gets messy from there on in; I wrote an XSLT parser once to process multiple QNames, so once again, this is based on practical experimentation. > What one perhaps wants is the same effect in a LINK. > > <LINK ROLE="meta" HREF="resource" A-PROPOS-DE="<keywords list>"> > > The availability of processors for languages which are somewhat > self-defining, that is to say which include definitions for the terms > they use, is a hot issue. Relevant here, and controversial in the > community. Very much so. But I won't repeat too much of the discussion here, seeing as how we've iterated through it many many times on RDF IG etc. > [...] X-Link exists as a generic utility in XML, XLink is another story, but now you mention it... * The "role" attribute is great, but it corresponds directly to "rel" in HTML. O.K., so, where's the "rev" equivalent gone to? * The datatype value of "role" is basically xsd:anyURI. That's rather offputting: QNames that resolve to URI-references a la RDF may be controversial, but they are still sensible to investigate in an XML environment. This is another thread that could run off at a tangent. [...] > The proposal that I am working on is that > > a) the element that cites the report indicates why you would be > interested in what it says. So the "accessibility evaluation results" > semantic is somehow in the LINK mode or other syntactic slots > in the citation expression. Heh, "somehow" :-) > b) [...] In particular, schema citations are an area where a > binding type indication may be appropriate. It is appropriate in any case where there is no magic formula for a document type. One example is NQuads: where the extra term could be a statement ID or a context flag! So:- _:a _:b _:c _:d . may be very different to:- _:a _:b _:c _:d . Ah... extensions of RDF. > The idea that the HTTP indicated Content-Type property defines > all the type knowledge about the content forever and aye is rejected. Yes! I'm very glad to read that. It's not as if extensible languages have sprung up overnight, but XML and RDF really do change the way that we think about "x is of format y", because now the semantics, or whatever mechanism the instance is trying to express is not necessarily bound to the serialization itself. That is always true for RDF, and it may be true for any given "dialect(?)" of XML. > So long as the syntactic processing can be completed in > accordance with the [super-] type [...] As contrasted against [syntax-] type? Cheers, -- Kindest Regards, Sean B. Palmer @prefix : <http://webns.net/roughterms/> . :Sean :hasHomepage <http://purl.org/net/sbp/> .
Received on Saturday, 20 October 2001 13:27:50 UTC