W3C home > Mailing lists > Public > public-wai-ert@w3.org > September 2008

some comments on http://www.w3.org/TR/2008/WD-HTTP-in-RDF-20080908/

From: Pat Hayes <phayes@ihmc.us>
Date: Tue, 09 Sep 2008 21:34:48 +0000
Message-Id: <p06240804c4ec946314e5@[]>
To: public-wai-ert@w3.org
This document appears to be somewhat confused about the basics of 
RDF, or may simply be poorly worded.

"2.1. Connection Class

A resource of type Connection represents a connection that is used 
for the HTTP transfer. The following properties may appear in nodes 
of type Connection:"

This is wholly unclear. It seems to conflate the notions of resource 
and RDF description. First, if Connection is an RDFS class (as seems 
to clearly be intended, though it is not explicitly asserted) then a 
resource of that type does not ordinarily "represent" anything: 
rather, it is itself represented (that is, described) by being an 
element of the class. I believe that what this is intending to say is 
that the elements of the class Connection are (not represent) 
connections used for HTTP transfers.

However, that issue aside, the next sentence is wholly inappropriate, 
as the use of the 'may' construction seems to imply that the 
properties are 'permitted' to be applied to the elements of the 
class. But in RDF there are no permissions, as there are no 
prohibitions. Thus, the 'may' construction is meaningless.

     Property representing the connected authority (server host and 
port) as a Literal."

Again, the wording is inappropriate. RDF properties do not 
"represent": they have values when applied to resources.  I presume 
that the final phrase "as a Literal" is intended to mean that the 
property value is a literal value: that is, in OWL-DL terms, that the 
property is a literal-valued property. However, the example given 
immediately below (example 2.1) does not have a literal value in the 
relevant position, which casts doubt on this interpretation (and 
makes the intended meaning quite opaque.)

"2.2. Message Class

A resource of type Message represents an HTTP message. The following 
properties may appear in resources of type Message:"

This repeats the confusions noted earlier in a more egregious form, 
as to speak of properties "appearing in" resources is meaningless.

"2.2.1. Body Property

The body property represents an HTTP entity body as defined in [RFC 
2616]. It can appear in resources of both type Request or Response. 
The object for this property must be a resource of the type 
cnt:Content or a subclass thereof."

(This also repeats the confusions noted earlier, which permeate the 
document and will not be commented on further.)
The phrase "must be" is inappropriate. RDF does not support 
constraints or restrictions such as this. It would be more 
appropriate to use the RDFS vocabulary to assert domain and range 
properties of the relevant properties, though this would not carry 
the same force as that which seems to be intended here.

"2.3.1. Request URI Property

The requestURI property represents the request URI as specified in 
section 5.1.2 of [RFC 2616]. The value of the property is either the 
constant value http:asterisk or a Literal value of the absolute URI, 
the absolute path, or an authority. "

The phrase "Literal value of ...absolute URI, .." is meaningless. A 
literal value is the value of an RDF literal. Absolute URIs, absolute 
paths and authorities are not RDF literals.


Similar errors to the above are found throughout the document, but 
these examples will suffice for now.

Pat Hayes
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
http://www.ihmc.us/users/phayes      phayesAT-SIGNihmc.us
Received on Tuesday, 9 September 2008 23:54:24 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:55:57 UTC