Friday, May 19, 2006


Savas has a few things on this paper from ThoughtWorks. I wonder if Jim wants to step in ;-) I'm on my way back from JavaOne so haven't had a chance to read it yet. However, my first question would be: What's wrong with the interoperability we've been achieving successfully with Web Services? Seems like we're doing a good enough job so far!


Jim Webber said...


The paper shows a pattern which we in the SOAP space have been using for a long time to avoid the limitations and brittleness associated with WebMethod and its kin. The fact that the authors used messages to explicitly decouple business and integration domains is a good thing, even for POX-based solutions.

The hyperbole in the paper about being POST-REST made me "de-prest" because I believe it muddies rather than clarifies an already confusing situation.

If you can pare this hyperbole and rampant meme-isms from the paper, then you see the pattern is not so different from those we'd use in Indigo (or indeed in certain WS transactions code :-)). Even the notion of using code + XStream metadata as a kind of contract is mirrored in WCF.

So to the crux of the issue: I believe the authors did their work an injustice by coupling it to an ill-understood and completely pointless SOA versus REST-like argument. But the pattern, even if obfuscated, is sensible.


Mark Little said...

Jim, although I agree with the sentiments (that there's something interesting within the paper), the majority of the text makes me cringe and leaves me frustrated. I'm currently reading 20+ papers for 4 difference conferences/workshops and if this paper came across my desk, I wouldn't hesitate to reject it.