[
Lists Home |
Date Index |
Thread Index
]
On Fri, 28 Oct 2005 13:29:09 -0700 (PDT)
Tech Rams <techmailing@yahoo.com> wrote:
>Whether or not we like it, XSD/WSDL are based on
>programmatic and object-oriented data structures. If
Err, what?
I s'pose it depends on which version of WSDL you're talking about; 1.1
was designed to support SOAP when SOAP was still an acronym. For that
matter, WSDL 2.0 can be viewed through the prism of object-oriented
programming or even (ugh) RPC. It can also be viewed, however, as a
contract for the exchange of messages. And it isn't tied exclusively to
W3C XML Schema, either.
As for W3C XML Schema, describing it as based on programmatic and
object-oriented data structures is, at best,... kind. In fact, the
requirements given to the working group that ultimately produced it were
as psychotic as the ultimate result (hardly a surprise), requiring that
the language be able to specify traditional "markup" types, database
types (possibly one could refer to this as "programmatic", but I
wouldn't), certain over-hyped internal pseudo-pointer types, as well as
the noted 'programmatic' types (which I think might be better labelled as
'register types', conforming to bit patterns in CPU registers). W3C XML
Schema has absolutely nothing object-oriented about it, though. Its
techniques for deriving data structures from previously-defined types
(extension and restriction) does *not* model inheritance in
object-oriented languages. Worst of all, many of its types differ in
significant detail from the types (in various and sundry languages) to
which they are mapped, with consequences that are amusing or disastrous
(depending upon your sense of humor and your job security when it occurs).
Possibly we're stuck living with this. And probably we're stuck with
procedural and object-oriented languages ruling the roost and defining
the consensus set of datatypes. That doesn't, however, make W3C XML
Schema's set of types match the consensus set, nor does it make it
object-oriented. Likewise, it's hard to regard WSDL 1.1 as
object-oriented when it lacks the facility to base one portType on the
definition of another, though there are OO dialects that use aggregation
to the exclusion of inheritance. Nor is it likely (in my opinion, which
is undoubtedly biased given that I've spent many hours teleconferencing
on the subject) that we'll be "stuck with" WSDL 1.1 once 2.0 is completed
(soon, I hope), nor is it necessary to view WSDL 2.0 as exclusively
representing procedural or OO programming models.
Amy!
--
Amelia A. Lewis amyzing {at} talsever.com
Confidence: a feeling peculiar to the stage just before full
comprehension of the problem.
|
