[microformats-discuss] REX! Re: REST-discuss list, naming

[Dr. Ernie Prabhakar]

Hi Dimitri,

On Oct 21, 2005, at 10:51 AM, Dimitri Glazkov wrote:
> On 10/21/05, Dr. Ernie Prabhakar <drernie at opendarwin.org> wrote:
>>
>> I think there are actually two different issues at stake.   To
>> Stephen's point, one aspect of this really is simply how to encode
>> REST information inside XHTML, which is a valid topic for  
>> microformats.
>
> But what _is_ this REST information? I guess I don't have a firm grasp
> on this just yet to understand what you mean. IMHO, there is no such
> thing as a REST information -- there's just a proper way to use the
> Web.

I personally prefer to avoid terms like "proper" -- that's a level of  
dogmatism even I (objectivist as I am!) don't aspire to. :-)   
Instead, I prefer to label these as specific styles or, well,  
"preferences."  That way, people can feel free to opt-in to sharing  
this perspective if they agree it is useful, rather than being told  
they're "improper" if they opt-out.  A subtle linguistic point,  
perhaps, but an important difference in psychological attitude.

Thus, just as REST is a particular "style" of doing web services, REX  
is a particular *style* for doing REST.   As sunch, my goal is to  
introduce the minimal number of additional constraints necessary on  
top of "conventional" REST.  Of course, even *that* is a somewhat  
nebulous concept, so it isn't always clear what is "additional" vs.  
"clarification" vs. "redundant."

 From "REX" perspective, we accept the usual definition of REST:
	a) nouns are URIs
	b) verbs are HTTP methods
and refine it further:
	a') nouns are structured URLs which dereference to XHTML Basic  
documents
	b') verbs are simply GET & POST (no PUT or DELETE)
	c) there are no input documents, only url-encoded key-value pairs
	d) links are always encoded as either hyperlinks or forms
	e) all data is encoded using XOXO and/or other relevant microformats

I think those principles are both necessary and sufficient to ensure  
both interoperability and browser-compatibility. You could certainly  
argue that, if REST is "web services done the way the web actually  
works", then this style is the logical and inevitable embodiment of  
that sentiment.

But, while I don't mind *you* making that argument, *I* don't want  
to. :-) In other words, I'm trying to do the absolutely least  
possible amount of work with as little controversy as possible.

Of course, the hard part is figured out how little is possible...

Does that clear things up at all?

-- Ernie P.