[uf-rest] "Casual Web Services" and Well Designed Urls

[Etan Wexler]

Nick Gall wrote to Microformats REST:

> Nothing in REST, nothing in WWW architecture, and certainly nothing 
> in Roy T Fielding's dissertation forbids constructing URIs based on 
> "out of band documentation".

I was discussing REST, which Roy Fielding defined in his dissertation. I 
was not discussing Web architecture as defined in any other document, 
not even other documents by Roy Fielding.

In particular, I was discussing a computer program’s construction of 
URIs, as opposed to a human’s construction of URIs. (If my writing was 
unclear, I apologize.)

I stand by what I wrote. The constraint of hypermedia as the engine (not 
merely an engine) of application state means that using anything other 
than hypermedia to progress from interaction to interaction is not 
within REST. Explanations in natural languages are not hypermedia.

I admit that Roy Fielding’s dissertation 
(<http://www.ics.uci.edu/~fielding/pubs/dissertation/top.htm>) does not 
declare that REST forbids out-of-band explanations as an engine of 
application state. The dissertation leaves many points implicit, and, 
rather than expressing conformance criteria after the mode of RFC 2119, 
simply defines what REST is. Whatever does not fall within the 
definition is implicitly forbidden.

> Here's what [Roy Fielding] has to say on the matter [...]:
> 
> "REST does not require that a URI be opaque. The only place where the 
> word opaque occurs in my dissertation is where I complain about the 
> opaqueness of cookies. In fact, RESTful applications are, at all times, 
> encouraged to use human-meaningful, hierarchical identifiers in order 
> to maximize the serendipitous use of the information beyond what is 
> anticipated by the original application."
> <http://rest.blueoxen.net/cgi-bin/wiki.pl?RestAndUriOpacity#nid1SK>

I fail to find a contradiction of my position. Serendipitous use is 
something in which human beings engage; it is not something in which 
computer programs engage.

As a human being, I frequently practice serendipitous use. For example, 
I used my knowledge of Wikipedia to construct the URI 
"http://en.wikipedia.org/wiki/Representational_State_Transfer". I 
accessed the resource which the URI identifies. In the representation 
returned, there is a link to 
<http://upload.wikimedia.org/wikipedia/en/thumb/8/89/Resttriangle.svg/273px-Resttriangle.svg.png>. 
I went on a hunch, constructed the URI 
"http://upload.wikimedia.org/wikipedia/en/thumb/8/89/Resttriangle.svg/600px-Resttriangle.svg.png", 
and accessed the resource which the latter URI identifies.

As a human being, I am not bound by the rules of the application.

> And here is what the W3C TAG currently has to say about it:
> http://www.w3.org/2001/tag/doc/metaDataInURI-31.

The distinction which I drew between computer programs and humans is a 
distinction which the cited document draws.

> In short, the only time a URI should be considered opaque is when the 
> person  looking at it has no documentation or code (eg a FORM element)
> to support their speculation of what the components of the URI might mean.

Well, RFC 3986 provides documentation for all URIs.

> don't speculate from the text embedded in the URI what the rules are 
> for composing such a URI.

Such speculation is serendipitous use that Roy Fielding encourages 
(albeit not in his dissertation) and that the authors of the draft TAG 
finding acknowledge as “important to Web users”.

--