Re: now://example.org/car (was lack of consensus on httpRange-14)

Roy T. Fielding writes:
>>> It isn't necessary for the section on how to use the http scheme as
>>> a locator to also define all of the different ways that any locator
>>> can be used as an identifier.
>>
>> It would be useful for the registration of the http scheme to define
>> its use, which I believe it does, and see no useful cause to extend
>> that.
>
>It is suicidal for a specification to define how the standard will be
>used.

It is, however, quite useful for specifications to define what they
define.  MIME Content-type registrations do have to specify what is
found in content identified by that type.  I would expect URI schemes to
specify what they identify - and RFC 2616 plainly does that.

>People will use them in ways that they find useful.  Consider how many
>ways that HTTP is being used today that would have been "disallowed"
>by such requirements in the specification.  I would have been burned
>at the stake.

We could just stop bothering with specifications.  That would avoid
creative reinterpretations both commercial and philosophical.

>In any case, if you think that the section "defines its use", then
>clearly all the current uses of http URI that are not locators (such
>as HTTP cache keys, xmlns, DOI, etc.) must be a figment of my
>imagination as well.

They're not figments, but to they extent that they divorce http URIs
from HTTP listeners (not a problem with cache keys or xmlns when RDDL is
used), they're at best unfortunate abuse of a URI scheme the nature of
whose resources is plainly specified in its registration. 

>Of course not, since RFC 2616 defines HTTP listeners and not what
>people can identify with http URIs.  The http scheme definition was
>supposed to move into a separate document (with https) at the last
>revision, but did not for reasons of editorial convenience.

While I'm aware that it's your story, the statement "RFC 2616 defines
HTTP listeners and not what people can identify with http URIs" is a
direct and unnecessary contradiction to "The semantics are that the
identified resource is located at the server listening for TCP
connections on that port of that host".

>> I'm a lot happier with:
>> http://lists.xml.org/archives/xml-dev/200210/msg00610.html
>
>Go ahead and try to be happy with a model in which people identify
>listeners instead of resources.  You will find that your model is
>inadequate for describing the impact of time (change) and variant
>content within the user-perceived model of hypertext.  

No, listeners are perfectly capable of responding differently over time
and with different content.  Simply because you speak to me in Chinese
doesn't mean I won't reply in English, French, or Dutch.  The URI
philosophy has abstracted HTTP URIs well beyond what is actually
necessary to address those issues.

>Likewise,
>you will be unable to explain why http URI can be used to identify
>XML namespaces, for which there are abundant examples and proven use.

Yes, and wise people who use HTTP for namespaces use RDDL and an http
listener instead of pretending an http URI is a namespace.

>Why you would be happy with an inadequate model is beyond me.  In any
>case, I certainly won't be using it to guide any of my design
>decisions, since I already have a model that suffers from none of
>those limitations.

It suffers from a distinct lack of connection to understood reality, and
encounters stiff resistance every time it proceeds beyond the small
community of the URI illuminati.

>I've pandered to this broken record too many times already, so I won't
>be responding to this thread further.

I'm happy to report that there are in fact coherent and simpler
alternatives to additional layers of abstraction piled on top of the
overly used http scheme, and think I've made my point.

>It is time for folks to shut up so that we can get some real work done.  

Nice.

>If you don't like it, then
>go ahead and deploy a better architecture

Funny thing is, Web architecture works just fine with http URI schemes
identifying listeners.  Code works and people can figure it out - until
you try to pretend that http URIs are abstract forms that can be used to
identify anything.  Maybe the world needs an "anything" scheme...

>and elect a different TAG
>to define it the way it makes sense to you.

I think that last option may be necessary if the Web Architecture is to
have a solid foundation.  Not because the basic architecture doesn't
work, but because some folks want to justify the reuse of girders as
clouds.


-------------
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, 11 October 2002 09:29:08 UTC