W3C home > Mailing lists > Public > www-xml-linking-comments@w3.org > January to March 2000

XLink 1999-12-20 WD: more comments

From: John E. Simpson <simpson@polaris.net>
Date: Mon, 17 Jan 2000 14:05:52 -0500
Message-Id: <>
To: www-xml-linking-comments@w3.org
Cc: xlxp-dev@fsc.fujitsu.com
I'm likely just a victim of jargon ignorance, but I believe some cleanup of 
the Terminology section [1.3] of the WD (especially relating to the word 
"arc") might be in order.

The particular issue here has to do with arcs ("symbolic representation[s] 
of traversal semantics in links, especially the direction and context of 
traversal") and links ("explicit relationship[s] between two or more 
resources or portions of resources"). These definitions imply that the 
terms "arcs" and "links" are always interrelated -- that every link has one 
or more arcs (i.e., one or more symbolic representations of the link's 
traversal semantics).

In a simple link, clearly everything about the link, including the 
characteristics of its sole arc, is defined in the local resource; the 
values of the from/to attributes are implicit. (Presumably "from" the local 
"to" the remote resource.)

Things get conceptually a little mucked up in extended links, however, 
because the local resource (as encapsulated in the resource-type element) 
can *never* be the "from" end of an arc. As described in [3.1.4], the 
direction of an arc is defined using from and to attributes; the value of a 
given one of these attributes is constrained to match the value of a role 
attribute on at least one locator-type element in the same extended link. 
There's nothing at all about matching the role of a resource-type element.

The problem, I think, is that there's the informal term "arc" (as used in 
the Terminology section [1.3]) and the more formal one (as used in e.g. 
"arc-type element"). The former sort of arc can reference local or remote 
resources; the latter, remote only.

Again, I probably am just missing some elemental piece of the puzzle. But 
as it stands now, syntactically the term "arc" is connected only with the 
remote resource(s), which implies to me that the definition of the term 
"arc" in [1.3] should say, explicitly, "...traversal semantics in links 
*among remote resources*."

(P.S. Yes, I'm aware of the special case in which an arc may *implicitly* 
include the local resource as its from/to value, as when values for the 
from and to attributes are not explicitly provided. This just confuses the 
language even further, IMO.)

John E. Simpson            |  I spilled spot remover on my dog.
simpson@polaris.net        |  He's gone now.
http://www.flixml.org      |  (Stephen Wright)
Received on Monday, 17 January 2000 14:05:19 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:32:21 UTC