Re: TAG document: SOAP HTTP GET binding available

"Sam Ruby" <rubys@us.ibm.com> writes:
>
> Shouldn't that be every operation which is idempotent, where the request
> has no headers, where the body contains only one child (i.e., no
multiref),
> where there are no structured parameters, and where the ultimate URI is
> relatively short?

Yes, there are a set of constraints on when a GET binding is possible.
Dave's document outlines those constraints.

> I still am wondering what is wrong with the current HTTP GET bindings for
> web services (other than the fact that they are not widely implemented).

I don't think there's anything wrong with them. The concern (if I
understand it correctly) is that by SOAP publishing only an HTTP-POST
binding will encourage avoiding HTTP GET bindings when they are in
fact possible (and maybe even better for the advantages they carry).

The right way to solve this IMO is to publish a separate recommendation
outlining the advantages of a GET binding and strongly recommending
Web service authors to use that. The problem has really little to do
with SOAP, unless I'm missing something basic. The criteria for when
a GET binding can be done is probably easiest explained using WSDL.

Sanjiva.

Received on Friday, 10 May 2002 09:40:01 UTC