I’ll try to play this straight. It seems that a posse of industry titans (IBM, Oracle, CA, and EMC) want a W3C working group to standardize WS-Transfer, WS-ResourceTransfer, WS-Enumeration and WS-MetadataExchange. Because, as they say, “There is still some work to be done”, and “Accessing data about a resource through Web services is an area of the Web services architecture that has yet to be fully realized.” I guess that if you really do want to implement HTTP on top of the SOAP stack on top of HTTP, these are clearly the Right Vendors For The Job. There is, however, real danger in this move, as outlined by Mark Nottingham in The WS-Empire Strikes Back... feebly.
Contributions
Comment feed for ongoing:
From: Mark Nottingham (Jul 03 2008, at 22:55)
I found it VERY difficult to keep even a semi-straight face writing my entry... you're a better man than I.
[link]
From: Jason R Briggs (Jul 04 2008, at 00:22)
...And the Award for the longest death scene in a public forum goes to... WS-*.
Whoo... pee...
I'm abruptly reminded of the death scene in Snatch with Boris. Who just won't die no matter how many times he's shot.
[link]
From: Peter J. (Jul 04 2008, at 06:47)
Please forgive a digression (although I suppose it could be considered "data about a resource"): this entry appears to have been published about 19 hours (2008-07-03T02:00:00-07:00, which is 2008-07-03T19:00:00+10:00) before Mark wrote his piece (2008-07-04T14:16:27+10:00). I wouldn't mention it except that your new posts often appear well into the "past"; perhaps there's a glitch somewhere in the publishing process?
[link]
From: David Smith (Jul 04 2008, at 09:15)
And, after ridiculing the whole WS-* world, when you have a non-trivial service that has to interact with a wide variety of non-trivial applications, your alternative is...?
We're three years into our project, and only weeks away from release, so your answer won't be immediately acted upon. But we will be rewriting one day and inquiring minds want to know?
It's fun to kvetch (I've worked hard to be an outsider lo these 60 years), and I'm certainly no fan of any of the Big Standards projects, but when you cut to the chase and have to talk with something big, like a hospital's electronic patient record system, what's the option?
[link]
From: John Cowan (Jul 04 2008, at 13:25)
A year or two ago in an XML Core WG telcon, there was a little bit more of the desultory talk we have from time to time about assembling One Big Standard that covers XML 1.0, XML 1.1, XML Namespaces, XML Base, XML ID, the Infoset, and maybe XLink and all the other low-level odds and ends that the WG is charged with maintaining.
I capped the discussion thus:
"And then we can call it WS-Core."
General laughter, and the topic hasn't arisen since.
[link]
From: Mark Nottingham (Jul 04 2008, at 15:29)
David, Tim was right: if you want something with with slick developer tools, shiny packaging and support contracts from huge vendors, WS-* is probably what you want at the moment.
Using HTTP for services doesn't have massive frameworks and development environments behind it, and it requires serious thought and more than a little interpretation to get right.
That doesn't mean that it isn't possible; lots of places are doing so today with great success.
We kvetch because it's painfully obvious to all that WS-* is a evolutionary dead end; it's bad, over-engineered technology that has serious interoperability problems, and is now trying to badly ape a technology that got it right in the first place.
[link]
From: Jon Ellis (Jul 04 2008, at 16:12)
<i>It's fun to kvetch (I've worked hard to be an outsider lo these 60 years), and I'm certainly no fan of any of the Big Standards projects, but when you cut to the chase and have to talk with something big, like a hospital's electronic patient record system, what's the option?</i>
HL7. How many other wheels are you trying to re-invent?
Personally i find the the "so, mr. critic, what's your alternative" line of rebuttal one of the worst. It might be better to face up to the fact that you can't please everyone, and move on.
[link]
From: David Waite (Jul 04 2008, at 16:39)
Aren't there people in the REST camp trying to replace SOAP with HTTP? I wonder if there will be a WS-REST to run SOAP over HTTP over SOAP over.. oh wait, now I'm confused.
[link]
From: Don Park (Jul 06 2008, at 15:41)
You guys are overlooking one teeny yet essential fact: there are people out willing to pay good money to have WS-* extended to cover their needs whether the need is real or better met by REST.
If you got a heap of trash and someone comes up to you waving money and saying "I got lots of trash already and what you have here will complement my collection perfectly", are you going to say no?
Another thing is that, for now, SOAP sells better than REST because potential buyers understand it better, probably because that's how their legacy systems communicate already.
[link]
From: Jeffrey A. Young (Jul 09 2008, at 11:39)
The thing I find interesting and essential from WS-ARCH is that it appears to have started out with a strong SOAP/WSDL bias, but the domain model itself is very abstract and can easily profit a REST perspective.
The computing world needs a common vocabulary and WS-ARCH (stripped of a lot of gory detail and assumptions) appears to be a good start. For example, the word "service" is casually thrown around here to mean everything from WorldCat in general, down to SRW (SOAP), down to a single RESTful CRUD operation on a resource and everything in between. We need similar authoritative definitions (with reliable URIs) for terms like resource representation.
Believe it or not, I find a RESTful interpretation of these and other terms as defined inside of WS-ARCH quite satisfying. Why can't we focus on identifying basics that everyone agrees on (whether we realize it or not)?
[link]
From: Stephen Masters (Jul 12 2008, at 02:33)
I suspect it's because these are the vendors who are trying to push web services via SOAP over MQ and other middleware. They have just realised that by doing so they are losing all the meta-data that HTTP was giving us in the first place.
[link]
From: Alex (Jul 18 2008, at 11:25)
"Accessing information about a resource" != "Accessing a resource".
[link]