Comments on arch document (15 November draft)

* 1.1 Audience of this document - Suggest adding another audience, "Those
that implement W3C specifications, and those who use the resulting
products (e.g., agent vendors, owners of Resources)." Many of the
statements in the draft are directed at them.

* 1.1 Audience of this document - This lists the statements made as
"properties", "constraints", "principles", "rationale" and "examples". In
1.3 and elsewhere, another kind of statement is made; "practice." These
should be rationalized and ideally explained and related to one of the
audiences of the document (e.g., a "practice" is a guideline aimed at
implementers, as described above).

* 2. Identification and resources - The list here may give the impression
that the URI scheme implies the type of the resource; e.g., "MAILTO URIs
identify mailboxes." Does this give the scheme more importance in
characterising the resource than is due?

* 2.2.2 Dereferencing a URI - "Resource" is defined in the first paragraph
of section two, then redefined here as "an abstraction for which there is
a conceptual mapping to a (possibly empty) set of representations."
Probably just need to substitute "A resource is" with "A resource can be
though of as."

* 2.2.2 Dereferencing a URI - In the last paragraph, it's asserted that
metadata accompanies the representation; I thought that a good deal of
metadata (e.g., Content-* headers, AKA entity headers) were part of the
representation, not external to it.

* 2.3.3 Retrieving a representation - The wording of this section does not
make the (critical) distinction between dereferencing a resource and
retrieving it clear; I *think* that it's saying that retrieving is the
"default" operation when dereferencing, and this must be safe. This and
the previous section need to be reworded to clarify.

* 2.2.4 Consistent representations and persistence - The first paragraph
asserts that a representation establishes the authoritative meaning of a
resource. Is this representation of meaning complete (some readers may
come away thinking so), or is it a snapshot?

* 2.3. URI Schemes - The phrase "URI processing" is introduced; I think
'dereferencing' is meant (although this is a good explanation of what
dereferencing is, and probably should be incorporated into the appropriate
section).

* 2.5 Some generalities about URIs - "trust derives from social
behaviour"; suggest adding "and from previous interactions with it related
resources (due to the 'Consistent representations' principle)."

* Status of Authority - I think it would be good to highlight (somewhere
in section two) the special status of the authority URI component, when
present; it serves to identify the publisher (in a Web sense) of a
resource, and therefore a large number of assertions stem from it
(especially surrounding trust and policy, e.g., P3P). Problems are
introduced when truly unrelated Web resources are grouped under the same
authority (we saw a lot of this in P3P). So,
http://www.publicwebhosting.com/~mary/ should be
http://mary.publicwebhosting.com/ if it doesn't have *any* relationship to
other content on the same server.

--
Mark Nottingham

Received on Thursday, 19 December 2002 18:33:28 UTC