[
Lists Home |
Date Index |
Thread Index
]
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
/ "Simon St.Laurent" <simonstl@simonstl.com> was heard to say:
| It sounds a lot like what "SGML for the Web" sounded like to me in 1997,
| but never was. This gives me a bit more hope for those kinds of
| visions.
I see two central arguments in this piece:
1. Globally unique names aren't necessary. The context is, in
practice, always well known. It follows that it's not a problem that
your <forsale> for a distributed ebay application and my <forsale> for
a distributed real estate application mean entirely different things.
You're not really going to point your bidder at my files and I'm not
going to point my real-estate searcher at your files.
If I want to build a document that inventories the things I've
purchased, I won't be able to mix the <forsale> elements for things I
bought by distributed ebay with the <forsale> elements for investment
properties I've purchased. But how often do I need to do that in
practice? Where the rubber really meets the road, embedding SVG and MathML in
DocBook for example, there really is enough context (<svg> only occurs in
<imageobject>s, <mml> only occurs in <equations>, etc.)
I think this would represent a serious challenge for vocabularies that
want you to be able to mix "your vocabulary here" (I'm thinking of
schema annotations and XSLT top-level elements.) But again, you don't
need to do that globally.
You may be right.
Similarly, the semantics of a vocabulary are always known to the users
of the vocabulary and there's no benefit in associating semantics in
an independent, programmatic way. And really, if I wanted to
communicate the semantics of <forsale> and <bid>, I could do it by
transformation.
You may be right.
But what's preventing you from implementing this vision today? No one
says you must use namespaces. No one says you must use any technology
you don't want to.
Now, if you're developing a technology that you want to interoperate
with other technologies that have adopted namespaces for global name
disambiguation and RDF for describing semantic relationships, I might
encourage you to use those technologies as well. I might even express
the opinion that I think you should.
I think there are significant benefits (engineering, training, etc.)
in reusing technologies. Even technologies that aren't exactly the way
I'd like them to be.
Be seeing you,
norm
- --
Norman.Walsh@Sun.COM | In every work of genius we recognize our own
XML Standards Architect | rejected thoughts; they come back to us with
Sun Microsystems, Inc. | a certain alienated majesty.--Emerson
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.7 <http://mailcrypt.sourceforge.net/>
iD8DBQE9m0sDOyltUcwYWjsRAg4aAJ47Ss/TFpJckKq2BWMW6OtQCNxCBQCgjI8R
d20ET69CElYBfBMtDIiaJqQ=
=FWFw
-----END PGP SIGNATURE-----
|
