Sergey Beryozkin <[EMAIL PROTECTED]> writes:

> Another reason a client can benefit from using a wsdl is that it
> might bring extra uptodate information for the client policy engine,
> extra alternative endpoints, etc...

But is the idea here that a client would fetch this WSDL document from
some remote server, or that it will have a WSDL document bundled up,
say, in its JAR?

I ask because fetching the WSDL from a remote server would add one
network round-trip to every client's start-up sequence. That seems
like a strange design: "The first thing the client does is contact
some server to see if the client still knows how to talk to service
and to figure out where and how the service is now offered."

That's a far cry from "contract-first" design that would allow clients
and servers to built independently, having agreed to a fixed set of
assumptions.

-- 
Steven E. Harris

Reply via email to