This is in the language spec, right?
There's a consolidated topic on referencing
in the architectural spec, first topic under "Behaviors".
Michael Priestley
IBM DITA Architect and Classification Schema PDT Lead
http://dita.xml.org/blog/25
"W. Eliot Kimber"
<>
01/24/2007 12:00 PM
To
<>
cc
Subject
[dita] Discussion of conref href= "value"
syntax
In the section "Using conref to refer to an element
within a topic" is
this text:
"The conref value follows the same conventions as HTML for what HTML
calls a ″fragment identifier″(B—a required ″#″ separator separates an
optional filename from the fully qualified id (in the form
topicid/elementid). Note that the ID of the topic must be included in
the reference before the ID of the element. To refer to target content
in a different file, put the full URL of that topic before the # character."
This is not technically correct in that the syntax of URLs is defined by
the HTTP specification and not the HTML specification. That is, what
HTML does is not a "convention" but what the URL specification
*requires*.
Using the terminology from RFC3986, I think this paragraph should say
something like:
The value of conref is a URI that includes (or consists entirely of) a
fragment identifier consisting of the ID of the topic that contains the
element that is the target of the content reference, a slash ("/"),
and
the ID of the target element. If the URI consists of only a fragement
identifier, the target element must be in the same XML document as the
reference.
Likewise, the following section on conref within maps should read
something like:
Within a map, the conref attribute references an equivalent element in
the same map or another map. [NOTE: I purposefully omitted the "will
be
copied" text.] The value of conref is a URI that includes (or consists
entirely of) a fragment identifier consisting of the ID of the target
element. If the URI consists of only a fragement identifier, the target
element must be in the same XML document as the reference. If the URI
addresses a different resource that resource must be a DITA map document.
Note that as far as I can tell RFC3986 does not define a term that means
"the part of the URI that is not the fragement identifier or query".
If
there is such a term the above might be clear by saying "a URI that
consists of an (optional) thingy plus a fragment identifier...".
Also, I think that the following section "Using conref to refer to
a
map" could be combined with the preceding section to avoid having
to
repeat what was just said for conrefs within map.
I haven't taken the time to find all the places that specs talk about
addressing syntax, but anywhere that the value is a URI the same sort of
language should be used.
It would probably be useful to have a separate general statement about
addressing and what forms of URI processors are expected to support. In
particular, I would think that it's a requirement that all DITA
processors support the HTTP schemes but are not required to support any
other schemes for DITA-to-DITA references.
Cheers,
E.
--
W. Eliot Kimber
Professional Services
Innodata Isogen
8500 N. Mopac, Suite 402
Austin, TX 78759
(214) 954-5198
www.innodata-isogen.com
