W3C home > Mailing lists > Public > www-ws-arch@w3.org > May 2003

RE: Proposed text for section 1.6.2 and 1.6.3

From: David Orchard <dorchard@bea.com>
Date: Wed, 7 May 2003 14:33:18 -0700
To: "'Champion, Mike'" <Mike.Champion@SoftwareAG-USA.com>, <www-ws-arch@w3.org>
Message-ID: <001801c314e0$4753f420$6401a8c0@beasys.com>

Indeed, I had rebutted this point earlier.  REST has better visibility only
for single protocol solutions, where visibility is defined to be the ability
to determine the method.  I actually think that this property is redundant,
as it is devolves to either the performance of the intermediary or the
simplicity of the configuration of the intermediary.  Which are covered in
the perf and simplicity properties.

Speaking of the simplicity property...  Given a multi-protocol situation and
complex interactions, it is typically easier to deploy "RESTless" SOAs than
RESTful SOAs.  I can create 1 xpath/xquery to do the match on the soap
representation, rather than figuring out which of the combination of
application methods are required.  As in, check for "getFareQuote as the
SOAP body's first child" rather than check for HTTP GET and SMTP ... and FTP
... and JMS vendor 1 method foo and JMS vendor 2 method bar and .... Which
scales well to doing things like "updateFareQuote" instead of HTTP PUT or
some magic thing inside a SOAP POST.

I can achieve all the samples of visibility by using either dynamic queries
against the content or by standardizing where the method is in the soap
body.

I believe that in most cases, simplicity of development and deployment
outweighs the performance property.

Cheers,
Dave

> >
> > I thought we had agreed that REST had superior visibility than
> > SOAs?  And while I agree that XML improves visibility, I strongly
> > disagree that the visibility of SOAs is "good"; on the
> contrary, it's
> > *very* poor.  I suppose this isn't the right time for that
> discussion,
> > so I'll just raise an issue when it's published.
>
> I've never agreed with you on that issue; I think you thought
> that Roy had
> convinced Dave Orchard of this point at one time, but that's
> for him to
> speak to.  I'm happy to accept "friendly amendments" to the
> way I stated the
> visibility benefits of XML, but I truly believe that XML *is*
> the secret
> sauce that extends visibility to the entire message, not just
> the protocol
> metadata that classic firewalls/proxies use.   I would be happy to say
> something like "as a practical matter, HTTP headers are more
> visible to
> *today's* widely deployed intermediaries than XML is," but I
> think you have
> something more profound in mind. And yes, I would *much*
> prefer for you to
> raise an issue and let the Powers that Be sort it out than
> continue to argue
> the point, because I think neither of us is likely to
> convince the other.
>
>
Received on Wednesday, 7 May 2003 17:31:16 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:18 GMT