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

Re: REST, uniformity and semantics

From: Walden Mathews <waldenm@optonline.net>
Date: Thu, 15 May 2003 21:42:55 -0400
To: Francis McCabe <fgm@fla.fujitsu.com>
Cc: www-ws-arch@w3.org
Message-id: <00d101c31b4c$79a49700$1702a8c0@WorkGroup>

> > The context was a little choppy, granted, but my objections
> > remain.  You either mean
> >
> >     (1) that REST imposes meaning on message exchange, or
> >     (2) every REST constraint is about the elevation of meaning
> >     [(3) I don't know what you mean, and my head hurts]
> >
> > My counter to (1) is that constraint lending meaning (at least
> > in someone's eyes) is a universal concept, not specific to REST
> 
> True, but REST imposes particular constraints; not constraints in some 
> vague sense. Sure, constraints lend meaning is a fairly universal human 
> approach to making sense of the world.

I has assumed that every constraint was a particular constraint,
and I admit I don't know what "constraints in some vague sense"
really means.  So I couldn't have been talking about that.
> 
> The particular constraints that REST imposes include enclosing a 
> representation of state, and the use of particular verbs with 
> particular meaning.

I don't know if I should comment on that wording or not, but
I don't think either of those statements is very accurate.

> 
> The stateless server constraint has a lot to do with interpreting the 
> message: the server is not *supposed* to take into account any state 
> information not present in the message. That is squarely a level 2 
> constraint and it has a lot to do with conveying meaning.

I think you're bending things too much to your particular purpose
here.  When you say "has a lot to do with",  you're leading us away
from the true rational (and "meaning") of the 'stateless server'
constraint:  which is the scaleability of the system.  In other words,
that constraint isn't about allowing meaning to coalese in this way
or that way; it's about memory limitations in real computing hardware.
And, of course, the desire to "serve billions".

I have some doubts about the "encodes" relation in your model.
I'm not sure I agree with that term.  Would any of "specifies", 
"describes" or "represents" work?

(Is this conversation still appropriate for this list?)

--Walden
Received on Thursday, 15 May 2003 21:38:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 July 2007 12:25:19 GMT