[
Lists Home |
Date Index |
Thread Index
]
REPOST
(Sorry, but I'm not seeing them. Added some minor edits.)
---------------------------------------------------------
-----Original Message-----
From: Sean McGrath
> The trick is knowing where to draw the line.
> Try and implement everything in declarative grammers is impossible.
In this particular case, I'm thinking that NOTHING can be specified in a
declarative grammar for this data model. I have to look again at RELAX NG,
though, as it might handle the case where ordered elements are interrupted
by unordered insertions of other elements (some unspecified). I'm thinking
not.
> There is a sweet spot, its
> application and case by case specific. You develop a nose for it but
> it takes a long time.
I understand that. What worries me is not having any widely-used XML
processor or architecture available to handle this particular XML.
> Pragmatists err on the side of little gray boxes because the boxes
> are "little" (do something small and do it well), "gray" (plug
> replaceable
> with something else, commodity) and "boxes" (plural - use numerous
> little boxes to make an intractable problem tractable via
> divide and conquer).
Actually, I mispoke (paraphrasing Hercule Poirot, I suppose): this is a BIG
gray box. Does all validation, except perhaps well-formedness checking.
> <Plug>
> I'll be talking about this and other aspects of the industrialisation
> of XML processing at XML 2001 next week
> http://www.xmlconference.org/xmlusa/2001/thursday.htm#7
> </Plug>
Wish I could be there...
|
