W3C home > Mailing lists > Public > www-tag@w3.org > September 2003

Re: Arch Doc: 26 September 2003 Editor's Draft (review of some terms)

From: Dan Connolly <connolly@w3.org>
Date: Mon, 29 Sep 2003 14:01:31 -0500
To: "Ian B. Jacobs" <ij@w3.org>
Cc: www-tag@w3.org
Message-Id: <1064862090.2697.399.camel@dirk.dm93.org>

On Fri, 2003-09-26 at 15:05, Ian B. Jacobs wrote:
> Hello,
> The 18 September 2003 Editor's Draft of "Architecture of
> the World Wide Web" is now available at:
>    http://www.w3.org/2001/tag/2003/webarch-20030926/

Reviewing from the index, i.e. looking at the
first-class terms and how they relate:

 * URI ambiguity hmm...
  "URI ambiguity refers to the use of the same URI to refer
   to more than one distinct thing."

seems like it should be tied to the term "message" or something.
Hmm... perhaps it is, somewhat indirectly, thru "The value of
a URI increases with the predictability of interactions using
that URI."

Seems odd to speak of the value of a URI, i.e. a sequence
of characters.

 * URI opacity

Why "People and software making use of URIs ..." rather
than "Web agents ..."? Is that on purpose?

 * URI persistence

This bit about "strong social expectations" and "always a matter of
policy" seems more awkward and arbitrary than it should be. It
seems like there should be a simple, compelling argument
from information theory and economics about URI persistence
and ambiguity.

 * URI reference

"Section 4.1 of [URI] defines a URI reference to be either a URI or a
relative URI reference"

That feels like a category error. Even though "http://www.w3.org/"
is both a URI and a URI reference, it plays a different role
when used as a URI reference.

The way the term is introduced works...

"The browser displays the retrieved information, which includes links to
other information via additional URI references"

It should be covered in the formats section somehow... in
section 4.8, perhaps

 * access method

This term is defined but not used. Is it needed?

 * Backwards compatibility

hmm... this is defined in terms of agents...
  (1) why not refer to the formal term "Web agent"?
  (2) "agents designed to process all instances of M" er..
  does it matter if the design is buggy? This doesn't
  seem like a good definition of compatible.

 * data formats

The term is introduced by example; I don't see a definition
of it... let's see how it's related to other terms...

  "the URI for the secondary resource" in 2.4.1


also... regarding
  * If URI "U" identifies primary resource "R", and
  * a representation of "R" is in the data format "F", and

when I was formalizing this stuff,
I found the only way I could exploit the relationship
between a URI, a resource, and a representation was
via a message:
  for represents(m, i, l) read: m represents i as l

So I'd write that as:

  * if message M represents the resource referred to in that
    message by U by the representation C in format F, and
  * the specification of F says that E is interpreted as R' in C
  * then M uses U#E to refer to R'

ugh... something like that... telcon time.

Hmm... why is "format specification" a formalized term?
That seems overly meta. It's new since 27 June 2003...
It seems to be in the 1 August 2003 batch of changes:

  4. Representations and Formats. Added "Formats" to section title.
  Definition of "Representation" now explicitly refers to sequence
  of octets. Introduce terms "data format" and "format specification".

Stuff like "Good practice: Format specification availability"
seems out of scope, to me... unless it's part of a general
argument about the self-describing web.

That's as far as I got for now.

Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Monday, 29 September 2003 15:03:27 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:32:39 UTC