clarifying the href attribute in the language reference

["Grosso, Paul" <pgrosso@ptc.com>]

The description of the href attribute in the language 
reference says:

  Non-DITA targets use standard URL syntax. 
  DITA content is targetted as follows:

  Target elsewhere in the same file:
  href="#topicID"
  href="#topicID/elemID
  Target in a different file:
  href="filename.dita#topicID"
  href="fname.dita#topicID/elemID"

That makes it sound like references to DITA content
do not necessarily use URI syntax.  Specifically,
many operating systems allow the # character in a
file name, and the above description does not seem
to disallow that (though it would sure make a mess
of trying to parse such a value).

In the DITA 1.1 Arch. Spec near the start of Chapter 4 
in the section on IDs and references it says:

 The complete identifier for a topic consists of the 
 combination of the URI for the document instance, a
 separating hash character, and the topic id (as in
 http://some.org/some/directory/topicfile.xml#topicid). 
 URIs are described in RFC 2396. 

[Actually, 2396 is obsolete and superceded by 3986 at
http://www.ietf.org/rfc/rfc3986.txt
in 2005 which we should reference instead.]

And a little later in the same section:

 The complete identifier for a map element consists of
 the combination of the absolute URI for the map document
 instance and the element id (as in
 http://some.org/some/directory/mapfile.xml#topicrefid). 

So it sounds like it is already the case that the href values
are always URIs and we just need to clean up the language a
bit in the DITA 1.1 Language Reference Manual to make sure
this is clear.

It might be helpful to add a note somewhere suggesting
percent-encoding (per RFC 3986) if there is a need to
represent a certain character in an href value that is
not allowed in a URI.

paul