|
Hi Richard,
It's a new day, we can try for a fresh crop of errors. I'll lead off :-).
> If this was not a rhetorical question, then to me it indicates that you have a fair amount of learning still to do.
That is an indisputable fact. For example, I know little about the inner workings of parsers, although
I have benefited from using them. I have a feeling that I may know more about that topic by the time you guys are done with me. I relish the journey.
>One could rephrase the "XML namespace or external namespace" discussion in terms of real and perceived affordances. Your thoughts?
Perception is tricky, as it is in the eye of the beholder (vocabulary designer). A bucket, turned upside down, is a stool.
>The semantics of HTML's A/@href are pretty well known, albeit kinda mushy -- if a human clicks, they'll see something related. What are the semantics in XML, where there is often no user,
no display, or nothing to click?
Crawlers use
a/@href to access the referenced content. No human is involved there either. The semantics of anything in XML are defined by the vocabulary designer, but I would tie the xml:href to the most current relevant RFC, which may be
the IRI RFC today. Same for its siblings - refer them all to the best of breed specification which has been vetted by the community. No specification? Doesn't exist.
Picking analogies can be tricky, because they sometimes give an exactly wrong idea. Especially when you lack knowledge of the domain of the analogy, which is a dicey business when you're talking
to this crowd. So, I'll try to be careful here, and I hope that someone with goodwill will not give me too much of a jolt to get me on the right track if I'm off it. I though about atoms and molecules, but there is a really large number of atom types, so
I don't want to leave the impression that the xml: namespace would balloon if we decided to incorporate these concepts into it. So atoms and molecules are out, as an analogy, for now.
xml:href, xml:src, xml:rel, xml:type, xml:hreflang, xml:method, and xml:tref seem to me not unlike the "vowels", not the words, in a hypermedia affordance
vocabulary (language)(are vowels specific to a language?). IOW, they can be recombined to make your own affordance, but they are shared by languages. In Atom, the app:collection element is a powerful such affordance. In html, a similar, but possibly more
powerful affordance is "form". "a" and "link" and "img" are other types of affordance. The semantics of those affordances are defined by the vocabulary designer, who if he says a bucket is a stool, who's to say it's not? The role of the "vowels" is to allow
him to use words that can be understood by programmers, but whose vowels have meaning at the protocol level of the (web) infrastructure, user agents, crawlers, caches, intermediaries etc.
I saw a mathematically-oriented (possible analogy) from Andrew Wahbe regarding hypermedia affordances. I can't find the comment just now, but unless I
misinterpret his ideas, which is indeed possible, there are two main types of hypermedia controls, where the type is a function of the hypermedia format and the client: Adaptive and Referential. Adaptive controls map a domain (defined by the hypermedia format
and client) to the uniform interface. eg VoiceXML, Atom. I take this to mean OTW XML applications can be included, so long as they use the uniform interface, ie web servers. The other type, Referential hypermedia controls, such as "a" in html provide a different
affordance to different users, such as humans and crawlers.
I think I have a 'gift' of (sometimes) being approximately correct, and (sometimes) precisely wrong. In fact, if someone had said
'Sure, that's a great idea!', when I first arrived [1], I think the conclusion would have been _both_ at the same time, as I pointed out yesterday to Pete Cordell [2]. In light of that,
and in light of the fact that people's patience must be wearing rather thin, I have a new proposal :-).
How about I take the day off tomorrow :-), and write a request for comments about xml:href and friends. I mean, I have never written one of those things, so how hard can it be ? ;-). Seriously, if
any member of this community has actually written one, I would appreciate their advice and I would welcome collaborators. Volunteers, please take one step forward :-). But anyway, once this community agrees on such a device, perhaps we could post it in
some neutral territory and ask the hypermedia crowd to present their advice on the issue, to this list? Once received and discussed, the XML Committee would be in a better position to make a wise decision on the matter.
Sincerely,
Peter Rushforth
|