[
Lists Home |
Date Index |
Thread Index
]
2/13/2002 4:30:50 PM, Gavin Thomas Nicol <gtn@rbii.com> wrote:
>
>If you're talking in the context *of* the web as a single
>application... it's very different from talking about network
>computing in the general sense. What I've been trying to point out is
>that the choice of network architecture is dictated by the
>application. I would argue that web services/rpc etc. are really a bit
>outside the scope of "the web" per se. and rightfully should be.
I think I agree ... you can either put your application at the center
or the web at the center, and neither is "right" or "wrong."
The message I take from Paul Prescod's and Mark Baker's numerous
attempts to educate us about this is that you CAN put XML and
the Web at the center of very many distributed applications rather than
simple treating XML as just a generic object serialization
format and HTTP as just a convenient transport protocol. Which
approach you choose is a matter of perspective, design philosophy,
other constraints, etc.
Nevertheless, if you treat your application as outside the scope of the
web, you probably can't leverage a lot of the web's goodies to build
it. If your web service is defined in some detail by a URL, then Google
may help people find it; if people use GET to access your service, they
benefit from the cacheing support, and so on. But if it is defined by
a complex WSDL document in a UDDI repository somewhere, Google won't
point people at it, and if you use POST to sent a message to access
your service, caches won't help your users.
Likewise, if you really want to liberate your data from the application
that created it, store it natively in XML exposed at some URI (which
could be physically inside an application or DBMS) and document it
with whatever HTML pages, Schemas, DTDs, RDDL documents, RDF or Topic
Map metadata you think appropriate. If you want to keep your data tied
to an application, just use XML/HTTP as an RPC format/transport. But
if you do so (unless you are a Marketing weasel looking only for
buzzword compliance) don't expect to enjoy the full benefits of XML and
the Web. If you don't need those "benefits", fine ... but understand
the choices you make and don't let your IDE make them for you.
|
