Re: [RTF] AC019 proposal to WSA WG

Joe,

On Tue, Jul 09, 2002 at 08:10:56PM -0700, Joseph Hui wrote:
> Mark,
> 
> My take is that if RM is a requirement, then it's in --
> in the doc, in someone's face (no derogative intended) --
> regardless of what architecture style it'll turn out to be.
> The architects will just have to make the style fit (RM).

Am I writing in a foreign language? 8-) Do you not believe me when I say
that reliable messaging is not appropriate in all styles?  If you do,
why would you want to put in a requirement for something that wouldn't
fit?  We're not talking about anything trivial here, we're talking about
a fundamental feature that would have huge impacts on performance, the
ability to deal with partial failure (read; brittleness), simplicity of
implementation and deployment, etc..  You don't just make something like
that fit by wishing it did. 8-)

> Letting style-yet-to-be influence a requirement is like
> letting the tail wag the dog.  Sometimes it happens;
> but it shouldn't be the norm, as in this case.  

I'd say the opposite; that driving a solution before there's an
architecture is the tail wagging the dog.

If we want to pick a style now, I'd be happy with that.  As I've said
before, there appears to be an implicit assumption that we're rebuilding
OMA/CORBA.  So let's just say that.  If we do, then I'd accept reliable
messaging as a requirement, because that style pretty much requires it.

MB
-- 
Mark Baker, CTO, Idokorro Mobile (formerly Planetfred)
Ottawa, Ontario, CANADA.               distobj@acm.org
http://www.markbaker.ca        http://www.idokorro.com