Lists Home |
Date Index |
firstname.lastname@example.org (Shelley Powers) writes:
>If the questions arise that RDF has foisted the concepts of URI and
>namespaces on to the XML world, what do we see as an alternative to
>both? In what way is URI a taint, and what would be an alternative to
>prevent collision if we didn't have namespaces?
Architectural forms was one option back in the day, but even a simpler
context-aware approach could be a useful option. X in one context might
be an HTML X element, in another context might be an SVG X element, etc.
Somehow the notion that we had to be able to precisely identify every
element and attribute in case the tags got put in a blender seems to
have dominated the discussion for reasons I can't say I appreciate.
As for URIs, I feel quite strongly that they're now carrying baggage
they aren't capable of supporting. A lot of RDF usages feel abusive in
particular, but XML namespaces do many of the same things. I've written
on this at:
http://lists.xml.org/archives/xml-dev/200210/msg00610.html (http URIs)
http://simonstl.com/articles/XMLfrag.txt (URI refs)
>I realize that these are tired and old questions that have been hacked
>in the past, again and again. However, perhaps with the new RDF spec
>releases, now is good time to revisit them. Or not, and just accept
>that the damage has been done and move on best we can.
I think we're only now getting to triage. I detest namespaces but I
implement them, manipulate them, and use them in my tools and parsers,
as tools which don't are at a disadvantage when they do encounter
namespace-using documents, especially sets of namespace-using documents.
Joe English has written some nice pieces on "sane namespaces" - see:
My personal efforts at this point are mostly directed at halting the
spread of QNames into new contexts. If I could put a tourniquet on the
problem I'd halt the bleeding at the point where QNames are used to
identify anything other than element and attribute names. Everything
else would be left to rot.
>As for myself, I don't see XML as messy, but I am a data person, and I
>do look for a meta-data structure when I want to work with data, even
>data that's organized with markup. This interest of mine in meta-data
>existed long before XML was a twinkle in Tim Bray's eye, and long
>before someone said, "Hey! What the world needs now is RDF, sweet RDF."
And I was a hypertext guy before the Web. Not as many cheerful Burt
Bacharach references, but XML (and, at one point, XLink and XPointer)
looked like a great answer to me when I found it. RDF has never
interested me except as a supplement to HTML META, and that barely.
>Is there really that much of a divide? I've used straight XML
>processing, RSS, and RDF processing with a RSS 1.0 document and found
>all three tool sets to be quite easy. My preference was the RDF tool
>set because I was able to use the least amount of code. I am that
>ultimate of a good programmer -- a lazy one.
>From my perspective, marked up hierarchies and graphs are completely
different approaches to information. Some people seem better at keeping
them all in their head simultaneously, but I find the switching overhead
>Personally, I don't think XML should be impacted by RDF (or RDF/XML)
>in anyway. I think this is a huge mistake. XML is a meta-language, and
>as such, shouldn't be impacted by structures built on top of it.
>Telling people "Well, if you just do this and this and that and that
>and this and don't forget to do _that_" to coerce them into ensuring
>their XML is RDF-compliant, when they don't see a need or a purpose
>for it, is a mistake. That's not the way to promote the use of RDF,
>and is nothing more than a way of winning enemies and disinfluencing
>friends. The resulting effort will reflect this, becoming nothing more
>than a bastardized, incomplete, piss poor instantiation of RDF/XML.
I agree with you, but feel that the damage has already been done. Liam
Quin's presentation last August (from my perspective, on how to finish
the RDF hatchet job on XML and make XML a disaster area) basically
pushed every angry RDF button I have. Echoes of that still taint my
perspective on John Cowan's and Bob DuCharme's more moderate (and yes,
useful) recent piece.
>Years ago, I saw this same attitude about the relational data model in
>my work converting network-based or hierarchical-based databases over
>to RDBMS. I would see tables that were nothing more than an exact copy
>of the previous implementations, guaranteeing the absolute worst
>performing databases in the entire world. A case of using the
>technology because one is forced to, rather than using the technology
>because its use is right and proper and one takes the time to use it
>carefully and correctly.
XML has all the same issues. In fact, the RDF serialization in XML
feels a lot to me like a similar problem to dropping an XML document
into an RDBMS and figuring out which pieces connect to which other
>Well, that's cool. I promise that when RDF and I head out for a night
>of swinging, and hit the dance floor to cut a rug, I promise not to
>step on XML toes in my exuberance. And no slammin' either.
Sounds like a good party.
Simon St.Laurent - SSL is my TLA
http://simonstl.com may be my URI
http://monasticxml.org may be my ascetic URI
urn:oid:18.104.22.168.4.1.6320 is another possibility altogether