[
Lists Home |
Date Index |
Thread Index
]
From: "Mike Champion" <mc@xegesis.org>
> Sooner or later I have to think that it will ... either
> the spec will be revised by the W3C or (more likely?) some sort of informal
> "schema interoperability organization" will figure out the profile of the W3C
> spec that meets the real needs of the Big Guys.
Perhaps XSD alread meets the real needs of the Big Guys.
Perhaps the rest of us might have different needs, though. I wonder if there
is some set of guidelines people can use which would future-proof their
schemas, if we think there will be some simplification or market split in the
future.
Mani probably will jump in here, but I guess
1) avoid abstract types and equivalence
2) avoid deriving data types except from built-in types
3) avoid derived complex types
4) avoid local declarations: use namespaces
5) if you need PSVI type information for elements, just use a fixed attribute instead
6) avoid infoset augmentations
7) model any constraints that are not straightforward for grammars and facets
with Schematron assertions
8) try to model so that you expect a program to only need, when processing the document:
i) the element name, ii) the element's attributes, iii) the element's parent's name,
iv) the element's parent's attributes, and v) the previous element's name.
Cheers
Rick Jelliffe
|
