An interesting post from Dave. A few comments…
I’ve been saying for a while now that I think it’s a shame that SOAP 1.2 didn’t define a general SOAP to HTTP binding that used HTTP as a transfer protocol, for the previous 2 reasons.
It does, Dave. The default binding is a transfer binding; I made sure of that. I think you’re confusing how people use it with how it’s defined. Web services proponents generally think that a SOAP envelope is a SOAP message, yet that interpretation is not licensed anywhere in the spec, and is even explicitly rejected in the HTTP binding where the state transition table clearly shows HTTP response codes affecting SOAP message semantics. It’s also alluded to in the glossary where the definition of the two terms differ (you think this was accidental? Hah! 8-).
I would love it if there was a reasonable way to bridge the SOAP/WS-Addressing world and the HTTP Transfer protocol world, but I just don’t see that each side really want the features of the other side. The SOAP/WSA folks want the SOAP processing model for Asynch, and don’t care about the underlying protocol. The Web folks want their constrained verbs and URIs and don’t care about SOAP processing model.
Avert ye eyes! False dichotomy alert!! You can get the SOAP processing model, and HTTP as transfer protocol (including asynch, which HTTP handles just fine despite insistance from many that it doesn’t) simply by using SOAP in the manner prescribed in the SOAP 1.2 spec and default HTTP binding. In order to do so though, you need to give up on the idea of a new (non-URI) identifier syntax. This is really not a big deal!. We are, after all, primarily talking about syntactical differences here. What EPRs are trying to do is comparable to inventing a new alphabet for the english language; perhaps there are benefits, but I think the phoenician alphabet has a, ahem, rather large and insurmountable head start in deployment, making those benefits – if they exist at all – completely inconsequential.
Dave then makes a really interesting statement of the “protocol independent” variety;
Here’s a test case: Would the Atom protocol switch to using WS-Addressing and then use the HTTP as Transport binding(s) and HTTP as Transfer binding? Seems to me not likely. The Atom folks that want to use HTTP as Transfer have baked the verbs into their protocol, and they won’t want to switch away from being HTTP-centric. And same as I don’t see the SOAP centric folks wanting to “pollute” their operations and bindings with HTTP-isms.
Emphasis on “baked the verbs into their protocol”. Seriously – no matter how you slice it you’re always baking verbs into a “protocol”, because an application developer has to know what verbs they’re using. The problem as I see it, again, is one of nomenclature; that Web services proponents have a very narrow RPC-inspired definition of “protocol” (transport), and their mental models built around this definition simply can’t fully absorb the implications of the broader definition used in the IETF and W3C (transfer). They simply can’t conceive of something called a “protocol” playing such an enormously significant role in a distributed system, yet this is precisely how all existing Internet scale systems are built, and precisely why Web services proponents haven’t yet realized that the Web is what they’ve been trying to build, at least since the quest for “document oriented” services began in 2001/2002.
One might also look at Dave’s statements and ask themselves, well, if they’re going to be dependent on a protocol, then it might as well be the most successful one ever developed rather than one which has struggled for deployment anywhere except behind the firewall. And somebody please remind me; why is it desirable to be independent of a transfer protocol, but dependent on SOAP the protocol?