[
Lists Home |
Date Index |
Thread Index
]
On Thu, Oct 13, 2005 at 09:45:42PM +0800, Henry Luo wrote:
>When we define the types for XML, of course we don't have to make it a
>superset of every possilbe data type in the world.
Obviously. Or maybe not. Whose baby are you throwing out with that
bathwater? The W3C XML Schema folks had this problem.
>The predefined data types in XML Schema is a good basis.
<snicker /> You're kidding, right? <chuckle /> W3C XML Schema doesn't
have a type system, it has a bag. <guffaw /> Is the point of this to raise
*every* possible permathread, at some point?
Nineteen "primitive" types. Another twenty-five types derived from those
nineteen. Nine of the primitives are date/time related. The numeric types
aren't all related to one another. NOTATION exists, with no relation to
either of the encoded primitive types, which are not related to each other.
URIs are not related to strings, and neither are QNames. Booleans are
unrelated to anything else in the world.
As type systems go, this one is mostly useful as a bad example.
>And we just need to define the syntax for the atomic types. The complex
>types can be mapped into element-attribute structure.
You've completely ignored the point that others have made, that a label for
a type is potentially much more dynamic and interesting than a
bondage-and-discipline type specification. Removing the ability to apply
different schemata to the same document is *reduction* in capability, and
one that can hurt.
>If the syntax in XML Schema is more or less sufficient to express all
>the data values we want to exchange in XML, then we are not far off alreay.
<giggle type="hysterical" />
Amy!
--
Amelia A. Lewis amyzing {at} talsever.com
The flesh is strong. The spirit stronger. So shed your skin, baby.
Let it through. Come on over.
-- Amy Ray
|
