Re: IBM CICS rejects REST :-) :-)

Don,

At the risk of being a pain in the butt, there is absolutely nothing in the
article that invalidates a REST architecture, nor indicates that IBM has
categorically rejected REST. There are domains where REST works well
(personally, I believe there are a lot of them) and domains where REST works
poorly. I think it remains to be seen as to whether CICS is one of the
former or the latter. SOAP works fairly well in building against legacy
applications, because such applications are usually strongly API based to
begin with. This is analogous to upgrading a computer; you can in theory
continue to add layers upon layers of intermediation code, which is in
essence what RPC/SOAP does, much like adding more memory, a second hard
drive, and so forth, but eventually, the core system will become obsolete,
and the intermediation code will only decrease the performance of the
system.

REST works upon the assumption that if you keep your core architecture
simple and place more of the burden of processing on the semantic layers of
the content, your system will remain more flexible and viable in the face of
changes to the outside world. It's not an easy sell to make, because most
programmers and IT people have been conditioned to think about building big,
complex hierarchical class structures rather than take advantage of the
semantic richness of XML messages, but even RPC systems that have a strong
latency factor to them tend to become more restful over time, as the latency
punishes complex APIs.

My two cents worth, though I look forward to seeing more examples of both
types of architectures; we're all charting new ground here, and I'll be the
first to admit that I may be wrong about this.

-- Kurt Cagle

----- Original Message -----
From: "Don Box" <dbox@microsoft.com>
To: <xml-dist-app@w3.org>
Sent: Wednesday, March 05, 2003 11:34 AM
Subject: IBM CICS rejects REST :-) :-)


>
> http://www.infoworld.com/article/03/03/05/HNcicssoap_1.html
>
> Seriously, this is actualy very cool.
>
> DB
>
>
>

Received on Wednesday, 5 March 2003 15:24:16 UTC