RE: REST; good for humans and machines

> -----Original Message-----
> From: Mark Baker [mailto:distobj@acm.org]
> Sent: Sunday, January 05, 2003 1:58 PM
> To: Champion, Mike
> Cc: www-ws-arch@w3.org
> Subject: Re: REST; good for humans and machines
> 

> Can you not see how much easier it is to coordinate the communication
> of information between untrusted parties with URIs and GET?  Or do you
> see it, but don't feel that the difference matters?  An answer to that
> would help me focus my responses.

How about if you explain it, in nice simple terms that even the boss can
understand :-)  

Here's a scenario:

Example.com traditionally used proprietary mainframe systems to handle their
invoice processing.  The invoices came into Bronto, and that software had to
communicate and coordinate with the accounting system on T-Rex, and did so
with proprietary middleware that offered a secure and reliable
infrastructure.  They web-enabled the front end at some point so that
invoices could be POSTed to an HTTP gateway into Bronto.

Now, Example.com wants to outsource the invoice processing but not the
accounting, so they want a more loosely coupled relationship between Bronto
and T-Rex.  They brought in two consultants, one a SOAP/WSDL drone, and the
other a REST geek.  The SOAP/WSDL drone says to decouple the data from the
processing on T-Rex and Bronto with XML, SOAP, WSDL, the usual drill ... but
continue using the proprietary middleware to ship the XML messages between
systems.  He also recommends that the data should be POSTed over the HTTP
gateway irrespective of whether it is being stored or requested, and whether
the operation is idempotent or not.... 

OK, you're the REST consultant ... at this level of abstraction/fantasy,
what do you recommend?  How do you coordinate the communication between
Bronto and T-Rex using URIs and GET (now that their trust relationship is
being strained)?  Remember that there is lots of money on the table here,
and security and reliability are paramount, so whatever you don't get from
the proprietary MOM system has to be made up somewhere.  Feel free to
elaborate on or modify the scenario, so long as you start from the
assumption that the current system works, the actual code runs on mainframes
that nobody wants to mess with, and that reliability and security are sine
qua non.

Received on Sunday, 5 January 2003 15:49:22 UTC