Link Details

First, a minor point: conflating Clojure and REST as "shiny new things" gives this piece a distinctively "Get off the grass!" feel to it. They are very different technologies with very different histories and domains. They're both things that have the attention of the cool kids right now, but that's no more an indictment of either than the cool kids' opprobrium towards SOAP is a mark against that technology. The article's real beef seems to be with REST, so I couldn't quite understand why Clojure was lumped in there too. What did it do to deserve this? Surely its Java interop is equal to the task of calling CXF!

That said, I'm the last person to deny that we sometimes get a bit utopian about our favorite technology, and REST certainly lends itself to this kind of thinking. Shame on any technology evangelist who sells snake oil in this way. I also need to acknowledge that I haven't spent my career in large enterprise systems, so my opinion is admittedly easy to dismiss.

That said, the crux of the article seems to be the statement, "Define your interfaces, nail them down, get them right." I find this troubling. Does this happen over the long term, in anyone's experience? It seems to be the majority report that massive enterprise systems are painful precisely because they are unresponsive to change—and they are located, like everything else, in non-static business contexts.

Maybe if we could cordon off parts of the world and set them in glue, unchangeable top-down contracts are a good idea. Given that such a state of affairs rarely obtains, do the arguments against REST hold?

REST is bad? Because WS-* was so freaking awesome? Are you kidding? You probably like CORBA!!
,

Dynamic programming in enterprise is new? It is actually very very old. It is smalltalk-80 old. Just ask those banks using Gemstone/S http://www.gemstone.com/products/gemstone

It also shows that NOSQL has been around for a while. But, as Alan Kay said, "this is not secret knowledge. It’s just secret to this pop culture." (from http://queue.acm.org/detail.cfm?id=1039523 )

Maybe I could take this article seriously If the whole WS-(Death)* stack didn't give me so many grey hairs as a developer.

I don't know what Clojure did to deserve this rant, other than being a really interesting piece of technology.

I logged in after 6 months just to downvote this.

The article's author seems to live in some vacuum. Putting the author's rant of REST and Clojure aside, here's some realities of WS-* SOA that the author himself laments.
1) WS-* can not be understood by the "90% of average developers." Smart people realize that it's overly complex. Smart people yearn for simplicity and avoid things that force them into complex implementations *even before* they start developing anything.
2) XML is just about the worst language to model data. Even with a fancy tool.
3) Model the contract in XML? Another joke.
4) Performant WS-* & XML based SOA - sure, if you buy 3X the hardware
5) Dynamic languages actually make it *easier* to deal with the nightmare of complex WS-* implementations

I could go on with another dozen or so points, but I have to get back to my shiny new toy (Clojure).