Re: FW: draft findings on Unsafe Methods (whenToUseGet-7)

Mark,

On Wed, Apr 24, 2002 at 10:34:06AM -0700, Mark Nottingham wrote:
> I think you might be taking this quote out of context;
> 
>    "And on the seventh day, He rested."
> 
> Seriously, isn't it a bit hubristic to say that REST is the only 
> possibly successful architectural style for any kind of Web-scale 
> application,

There's "Web-scale" and then there's "Web-like".  I'm referring to
the latter.

I'm not saying that other architectural styles can't succeed on the
Internet, because I know that to be false; Usenet, FTP, email are
all Internet scale, but did not use the REST style.

What I am saying is that the Web is defined by a single architectural
style, and that REST is the best attempt around to define it.  If you're
doing something different, then it's not the Web.  But there's a caveat
there; because REST is so general, it can sometimes absorb other less
general systems.  For example, it's absorbed FTP, POP, Gopher, etc...
So those systems are now part of the Web, despite not being designed to
be so.

> and that the W3C should be prohibited from approaching 
> problems from any other direction?

The W3C's mandate is to lead the Web to its full potential.  Given
what I said above, that means REST-specific.

> I (yet again still) haven't 
> completely read Roy's dissertation, but I wasn't aware that the W3C had 
> accepted it as the One True Word of Web Architecture. I know that you 
> and others feel that way, and I agree that REST is a great thing, but 
> I'm not yet ready to assert that it's appropriate for every possible 
> application to the exclusion of other solutions.

Constraints suck! 8-)  But they're also there for a reason; that it is
impossible to accomplish anything without them.

> This isn't to say that REST shouldn't be evangelised or even preferred 
> in most WGs' work; just that it shouldn't preclude other styles.
> 
> I despair to see the REST fundamentalist view; it indicates that efforts 
> to come to a accommodation (like defining a HTTP-GET "binding") are 
> useless, because they don't RESTify all Web Services.

Well, there is a time and a place for pragmatism.

MB
-- 
Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA.      mbaker@planetfred.com
http://www.markbaker.ca   http://www.planetfred.com

Received on Wednesday, 24 April 2002 14:09:42 UTC