[
Lists Home |
Date Index |
Thread Index
]
Dare Obasanjo wrote:
>>-----Original Message-----
>>From: Bill de hÓra [mailto:bill@dehora.net]
>>In your world maybe, and good luck to ya. In my world XML
>>Namespaces are being discarded as a bad idea.
>
>
> Obviously we live in different worlds. Mine is described here
>
> http://msdn.microsoft.com/library/en-us/dnxml/html/understxml.asp
Well, you lost me, or maybe I lost you. The mention at all of XML
Namespaces has to do with extensibility:
'XML is Extensible'
Surely you understand XML. It's not extensible, it's a fixed
grammar. I though first you were trying to say:
'XML is Extensible with XML Namespaces'
but 1) that's not XML anymore, 2) whatever it is, it is still not
extensible since it's still a fixed grammar. I claimed this
yesterday, feel free to refute, or handwave. I suspect you're trying
to say:
'XML vocabularies are Extensible with XML Namespaces'
but, 3) that's not about XML anymore, that's about vocabualaries (or
content models), 4) XML Namespaces don't make vocabularies
extensible they just allow the them to be partitioned.
The example around attributes. It doesn't require XML Namespaces,
which is to say, it didn't require XML to be changed the way to was
by the namespaces productions and processing rules. No doubt the
problem is solved, but it's an expensive way to solve what is
someone's data modelling problem/kluge, not really a coordination
one between parties. Nothing is being extended, but it surely would
let one tunnel an attribute annotation through someone else's vocab.
Ultimately I'm not sure it's just a way of justifying a bad
architecture by dressing it up as a pattern (something like a J2EE
best practice) - 'at least three people are doing it, we wrote it
down, so it's ok'. Is this the justification for warping XML and all
its tools? Golly. It's like watching Java folk going about about
how cool attribute oriented programming and dynamic proxies are,
perhaps because they've never seen C# metadata, or Lisp and don't
see the feature for what it is - a patch.
No flame on you, I liked the article. I just think the justification
and claims made for namespaces are pointless and somewhat wrongheaded.
Separating two vocabularies via two namespaces instead of one; what
could be further from an extension of either? The whole point of a
splitting all the names into spaces of names is to stop the
extension of names in any one space (commonly called namespace
pollution), Did you mean to conflate extensibility with modularity?
About the only valid conclusion regarding exensibility we could come
is that XML Namespaces are an extension (a somewhat backwards
incompatible set of productions), of XML. Or perhaps that XML
Namespaces allows us to extend the set of XML Namespace names
through the URIs.
Bill de hÓra
|
