[
Lists Home |
Date Index |
Thread Index
]
Adam Turoff wrote:
>If I'm doing my math right, that leaves roughly 1% of XML users
>actually *trying* to do something with XML Schema. From there:
>
We're using MSXML 4, .NET and XML Spy. I think the only really
irritating problem I've had is that XML Spy doesn't detect ambiguous
content models - this should be at least an option.
In general I'd say that anyone who wants to do typed Web Services is
likely to be doing XSD. That's going to be a lot more than 1% of users
in the near to medium future, in my prediction.
How are they going to do it? I've just experimentally dragged a database
table onto a schema in VS.NET - the result's not perfect (got primary
keys but can't see referential integrity) but it's a pretty good start.
So that gives at least three options to those designing XML Schemas -
[1] start by designing the schema, either textually or graphically
[2] start by designing a programming language class, then
auto-magically convert the class to a schema
[3] start by auto-magically converting one or more tables to schemas,
plus of course
[4] start by getting a third-party schema and simply putting it
through your toolkit
An interesting point is that when the .NET utility xsd.exe (I imagine
that there is or will shortly be a java equivalent) generates a
schema-specific deserialiser, the generated code not only parses XML (or
an XML event stream) blazingly fast, but also provides implicitly quite
a lot of structure and data-type validation.
Strangely my tests show the generated code de-serialising XML into
objects faster than it serialises objects into XML - is this just me?
Francis.
|