Re: lack of consensus on httpRange-14

Roy Fielding writes:
> Ahem... it's not just a joke -- I've been using that example since the
> Matrix came out to force people to think about the generic interface
> provided by HTTP.  The point is that the resource does not exist -- a
> resource is, essentially, an expectation that future representations
> obtained via that interface will have a sameness in relation to past
> representations.

This Platonic Form of a resource can, however, leave us without a spoon
to bend, even given a representation of spoons, as soon as we cross into
the (perhaps) more concrete territory of URI references.

Take, for example, this URI:
http://www.example.com/spoons1234

The "resource" described by this URI is a particular set of spoons.  The
server at our always-fictional example.com is running some environment,
perhaps AxKit or Cocoon, that can produce different representations in
response to different browsers or explicit content negotiation.  Working
from a base XML document that we'll pretend is a raw representation of
the resource, it can produce plain text, HTML, SVG, PNG, XSL-FO, WML,
and PDF renditions of such spoon.

The easiest of those formats to "bend" is likely SVG.  I feed that URI
into my SVG viewer, get a graphic of all the spoons, and pick one to
bend in XYZ graphic tool.  For convenient reference, I note that I've
selected
http://www.example.com/spoons1234#svgView(viewBox(0,200,200,200))

As I have friends who are fond of bending spoons, and I find this
particular spoon especially exciting, I post that URI reference some
place shared.

Unfortunately, my friends don't all have SVG, so while they might see
all the spoons, they don't reach the spoon to which I was referring.

A more robust approach, assuming that my transformations preserve ID
information (and convert it into appropriate forms for particular media
types), is a bare name identifier:
http://www.example.com/spoons1234#spoonusMaximus

but even that approach is unreliable in a variety of circumstances, and
it's very hard to argue that implementers must always provide consistent
handling of fragment identifers applied to a given URI regardless of the
content-type returned as a representation.  (It would be nice, perhaps,
but full XPointers into GIF files sounds unlikely.)

It is perhaps possible to create a new URI which represents the
particular spoon (subresource?) I've chosen, but approaches for doing
that are non-obvious at present.

Given references to The Matrix, none of this may seem insoluble, but in
practice I think it suggests how diverse in practice these forms of
reference are, even within the relatively limited scope of HTTP.

I'd also recommend that the TAG read Dilbert today:
http://www.dilbert.com

The reason for my pointing to that resource in this message will be far
clearer if you visit it on 4 October 2002 than it is likely to be on any
other date.


-------------
Simon St.Laurent - SSL is my TLA
http://simonstl.com may be my URI
http://monasticxml.org may be my ascetic URI
urn:oid:1.3.6.1.4.1.6320 is another possibility altogether

Received on Friday, 4 October 2002 16:49:16 UTC