[
Lists Home |
Date Index |
Thread Index
]
Mike Champion wrote:
> On Thu, 30 Jan 2003 06:11:12 -0800, Paul Prescod <paul@prescod.net> wrote:
>
>
>>
>> If we can't agree that pervasive use of URIs is a defining
>> characteristic of the Web as we know it, then I really can't imagine
>> how we have the building blocks for any meaningful conversation at all!
>
>
> Sure, no dispute there.
You said that "resource oriented programming" is unproven as a basis for
the Web-as-we-know-it. I admit I only used the context of the one
message and not the whole thread (my name floated by...). In my personal
opinion, whether one thinks of resources as abstract or concrete,
resource oriented programming is fundamental to the Web-as-we-know it.
If by "resource oriented programming" you mean something more
metaphysical about abstract resources then I won't spend much effort
arguing about it...but woudl prefer you use a different term!
I don't think that there is that much PRACTICAL difference between the
two views because best practices will lead people to give concrete
representations to otherwise abstract resources, and to NOT use
pre-existing concrete URIs to mean something abstract.
> ... The open question for me is whether "the Web as
> we know it" proves the concept of URL's that *locate* something or other
> to be determined by all sorts of context, MIME types, ad hoc conventions
> and out of band agreements ... or whether it proves the concept of URIs
> that *identify* abstract resources with representations. The former is
> much less general and abstract than the latter, and I'm skeptical of the
> argument that the success of the less general form proves the validity
> of the more general form.
Google uses URIs as identifiers. When you ask it for a cache of some
page it returns you appropriate data _whether or not_ there is currently
an HTTP listener available at that location. If it uses the URI without
"locating" the data, how is it using it as a locator?
>...
> "everything has a link" doesn't equate to "has an abstract identity
> specified by a URI". Or more to the point, Google as we know and love
> it would work just as well if links were just plain ordinary URLs rather
> than nice abstract URIs. This get's back to Miles Sabin's "what would
> break if we stopped thinking about abstract resources?" question.
Of course Google doesn't care whether I think that the strings are URLs
or URIs, but if Google treats them only as locators and not as
identifiers, then it cannot provide its service. A locator is a thing
with exactly one operation: "dereference." If URIs had only this
operation, then we couldn't cut and paste them and compare them. We
couldn't cache based upon them. An identifier is a thing with exactly
one operation: "identify for 'client-side' comparison". If URLs had only
this operation then we couldn't use them to retrieve web pages. URIs do
both things.
More to the point, I suspect, I'm unwilling to extrapolate from Google
> and RSS to applications where machine processing is much more important,
> e.g. automated order submission, bill payment, etc. In such cases I
> suspect that the "contract" between the producing and consuming programs
> about the syntax and semantics of the data being exchanged is much more
> important than the architectural style of the communication (POST vs
> GET/PUT/DELETE) between them.
In the realm of services hard-coded to talk to each other, you are
entirely right. In the realm of services that are more loosely connected
(for instance through pipes and filters), I do not think you are.
Paul Prescod
|
