RE: Alternative dereference behavior

A URI (or URL or IRI or whatever) is a communication from a
URI producer to a URI consumer, which identifies a resource
the URI producer wants the URI consumer to identifier.

If the URI consumer wants to understand the communication,
the URI consumer needs to understand what the URI producer 
intended.

" does it leave open the possibility of conforming agents
  using mechanisms that give answers at variance with what
  the Web would give?"

Consumers of URIs that use mechanisms other than the
one indicated by the URI itself must somehow be using
additional information that isn't contained within
the URI itself.  I don't think it should be part of
the "Web canon" to forbid the use of additional 
out-of-band information, but certainly such use is
"non-conforming" within the domain of applicability.

One additional piece of information which is often
necessary but not part of the URI is the timeframe 
in which access is expected. So 100 years from now,
if an interpreter of a 100-year-old document comes
across "http://www.netscape.com", the interpreter
(a URI consumer) may well use some out of band
information to understand and interpret what was
likely meant by the producer of that URI.

Other examples of out-of-band information come from
local configuration information (intranet users 
allowing WINS resolution of domain names) and 
"transparent proxies", where organizations intercept
HTTP traffic and try to redirect requests to local
caches as a way of reducing net bandwidth use
and improving latency.

There is no simple way to normatively ALLOW or
DISALLOW a URI consumer from using out-of-band
information to pick a different resolution method,
but this doesn't change the fact that the URI
producer's *meaning* is the one indicated by the
URI scheme itself, even if the producer is aware
of and takes into consideration the consumer's
likely additional behavior.

I think the discussion of "meaning" without
being clear about "to whom" and "when" is
very difficult and leads to incorrect
conclusions and contradictions; be
careful to use "meaning" as verb (someone
means something at some point in time)
rather than an attribute (X has meaning
Y).

Sometimes within a given document you don't
need to qualify all those things, if the
document itself is clear (alas, not the
case for most of the documents we're discussing.)

Larry
--
http://larry.masinter.net


-----Original Message-----
From: www-tag-request@w3.org [mailto:www-tag-request@w3.org] On Behalf Of Jonathan Rees
Sent: Monday, September 28, 2009 2:29 PM
To: Dan Connolly
Cc: www-tag@w3.org
Subject: Re: Alternative dereference behavior

On Mon, Sep 28, 2009 at 4:04 PM, Dan Connolly <connolly@w3.org> wrote:
> On Mon, 2009-09-28 at 13:00 -0400, Jonathan Rees wrote:
>> This message is pursuant to ACTION-312 which I took on at the F2F.
>>
>> Roughly speaking, the question is: Does the canon say that the Web is
>> the authority for http: URI "dereference" (GET), or does it leave open the
>> possibility of conforming agents using mechanisms that give answers
>> at variance with what the Web would give?
>
> [...]
>> Summary:
> [...]
>>   . General advice (AWWW, IAB TC) is that if you "split the web" by making URIs
>>     non-global you are doing something really tragic.  A change
>>     in the rules for dereference would theoretically be OK, as long as everyone
>>     made the change in step (ha!).
>
> Exactly. That seems like a "yes" answer to the question above,
> inasmuch as an authority is something that helps you avoid something
> really tragic.
>
> I don't see any contradiction with my reading of webarch;
> maybe the description of the action should change or something?
>
> "Find a path thru the specs that I think contradicts Dan's reading of
> webarch"
>  -- http://www.w3.org/2001/tag/group/track/users/38732 

I will not be able to complete this action, because you were right and
I was wrong.
So I think the action does need to be changed or dismissed.

Jonathan

Received on Monday, 28 September 2009 22:05:01 UTC