Re: The UR Trout: Web Services, REST, SOAP

"Anne Thomas Manes" <anne@manes.net> writes:
>
> Perhaps we'd like to take a step back and define the term "service" before
> defining the term "web service". One of the biig concerns I have about
> saying htttp/xml things are web services is that so many http/xml things
are
> not services at all -- they are designed for GUI consumption (via a
browser)
> not for software consumption (via an api).

I agree that many forms are designed for GUI consumption. However,
what makes them GUI consumable is that the output format is HTML
or some format that's not meant for machine processing.

That's however not a statement about the HTTP/XML technologies at
all. Those are simply form "services" that are not suitable for
interoperable usage from different platforms.

Thus, I'll be happy if you folks say that HTTP/XML/MIME stuff is
a Web service as long as what they consume and produce are machine
processable things.

> There are many ways to implement services -- http/xml, java/rmi, dcom,
.net
> remoting, corba, soap, etc. The wsa describes a particular type of service
> called a web service. Features of this type of services are that it
ensures
> complete language and platform neutrality through a standardized wire
> protocol (soap), and it enables interoperable tooling through a
standardized
> description language (wsdl).

+1!

> I think these two goals should appear in our definition: language and
> platform-neutrality and standardized tooling
>
> I'd be willing to say that an http/xml service described by wsdl qualify
as
> a web service because it supports these two goals, but it doesn't support
> the higher-level infrastructure services defined by the architecture
> (security, reliability, etc). So perhaps we might want to define level 0
> services (http/xml/wsdl) and level 1 web services (soap/wsdl).

+1 to the concept that HTTP/XML supports a different class of QoS.
However I'm not sure whether there's a need to label it with level
numbers. (Now I can see another REST debate saying "of course it
can support any QoS .. you just have to encode that in the URI!" Yes
I agree; if you program everything then it'll do everything.)

> I disagree with Sanjiva that "wsdl = web service" because you can describe
> anything with wsdl, but wsdl doesn't ensure language and platform
> neutrality. wsif is great technology -- it allows a java application to
> access almost anything described by wsdl -- but it doesn't help a vb
> application access an ejb component. Nor does it help a Java application
> access a .net remoting component.

Agreed. I am in no way expecting WSA to say WSDL = Web service. To me
WSDL = service and WSA's definition should be along the lines you said
above.

Sanjiva.

Received on Monday, 7 July 2003 01:42:50 UTC