[
Lists Home |
Date Index |
Thread Index
]
Alaric Snell wrote:
>
>...
>
> That's a higher level issue. If some object in your slice of the server's
> state is meaningfully shareable, then it should support a suitable standard
> interface, with a URL you can give to the other web service.
Agree 100%.
> WSDL is just an interface description language. That's all. It's not a *bad*
> IDL because it doesn't address certain issues (although I'd be interested in
> adding type information to URIs for this kind of work; not just 'this value
> is a URI' but 'this value is a web service URL supporting the WSDL at this
> other URL')
That's what I'm talking about. WSDL needs to support that.
> ... if those issues can be handled in another system that's
> orthogonal to WSDL.
What system? If WSDL is for defining web services and a web service is
composed of multiple dynamically created and destroyed data objects with
methods then WSDL needs to describe their interface.
> > And when you send information into this service, how do you get it it
> > out again? How do you communicate to some other service how THAT SERVICE
> > would get the information out. You own the information.
>
> Not necessarily - this depends on the terms of your contract with the service
> provider!
Fine. The standards should not be biased in favour of terms of service
that make it such that it is difficult to get your information out. This
is the sort of thing that is going to drive Stephen Newcomb to an early
grave.
>...
> Now, you have an XML-RPC service URL for the phone book and another for each
> person. The PhoneBook interface might be implemented as simply as:
>
> return "http://www.phonebook.com/......foo.....?person="; + name;
>
> ...mapping from people's names into the URLs of the Person objects.
Fine, so what value is XML-RPC providing? Raw HTTP is doing all of the
work. You've got another layer in there to no avail.
And how do I specify it in WSDL? What's the point of a language designed
to allow static type annotations if I'm switching back and forth between
the statically typed, predictable world and the world of strings and
dynamic binding?
Paul Prescod
|
