OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] REST as RPC done right

[ Lists Home | Date Index | Thread Index ]

On Saturday 02 March 2002 09:10 am, Paul Prescod wrote:
> There's no accounting for taste. ;) Of all of the service
> description languages I've ever seen, WRDL is the simplest. The
> normative part of the spec is about five pages.

The interface specification language for the DOM was probably simpler, 
but anyway, the point is that once you start layering wrdl etc. on top 
of REST, it's not substantially easier than anything else.

> > ... you're basically
> > writing a vtable in XML....
> That's an extremely poor analogy. A vtable is an implementation
> technology, not a specification one. It's like comparing reference
> counting and monads.

Data is just a dumb form of programming (forget who said that). A 
vtable is a dispatch specification, as is (essentially) WRDL. They can 
both be interpreted declaratively or at runtime, perhaps I should have 
thrown IDL into the mix for completeness. One way or another, you're 
going to end up defining interfaces...

> No I am not. I use the methods exactly as they are defined in the
> HTTP specification, with exactly those semantics.

Sure... and then depending on the query parameters etc. you dispatch 
differently. Sounds like overloading to me.

> I'm waiting for a technological argument.

We've already said that at the turing completeness level, they're 
equivalent, but that's not the point. Business objects tend to have 
much more sophisticated behaviour, and more complex interfaces at the 
level of abstraction that people think of "business objects".  So what 
you end up with, like in untyped scripting languages, is a quick win, 
but less complete overall specification. 

> Those systems are built in a very top-down fashion. Web services was
> supposed to *fix* EDI so that it could grow more organically...like
> the Web.

Depends on your POV. A lot of people *using* EDI don't really see that 
much benefit in web services. Anyway, my point is that there are 
*huge* systems that scale well, perform well at scale, and which don't 
need web technology or REST. Some of these systems are more tightly 
coupled/strictly typed than current web services, which generally 
means that the evolve more slowly. That is necessarily a bad thing.... 
and is part of the design decisions one makes.

Your argument that REST is needed for scalability is negated by 
existing implementations.

> > > Plus, the W3C folks have always used a definition of hypermedia
> > > that meant: "all information, everywhere, hyperlinked."
> >
> > In *your* history books anyway.... but I remember a WWW before
> > images and before CGI scripts.
> "1989, while working at the European Particle Physics Laboratory, I
> proposed that a global hypertext space be created in which any
> network-accessible information could be refered to by a single
> "Universal Document Identifier"." ... "Its universality is
> essential: the fact that a hypertext link can point to anything, be
> it personal, local or global, be it draft or highly polished."

I think the term "information" has evolved in the above... as has 
"distributed hypermedia".


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS