Initial draft of XML-Link spec now available

I have placed the HTML version of the initial draft of the linkage spec
at a URL that appears at the bottom of this message - please read the rest
of this before you pull down the spec and begin fulminating.

1. Don't Panic

2. This doesn't represent the opinion of the ERB; not one item in here
has been voted on and I can honestly say that I just don't know what 
people's leanings are.  To make this point even more forcefully, there are 
a couple of sections in this draft, written by Steve, that I substantially 
disagree with (e.g 5.2 LINKS and LINKSETS); similarly I assume Steve is 
not happy with some of my prose.  I *think* that Steve and I mostly agree 
as to the outline and explicatory strategy.  [One key meta-decision: there
are no productions, we assume that we can say these things are SGML elements
and attributes, and no further syntax elucidation is required.  Seem OK?]

3. Note that one of the areas remaining undecided is: what do we call
this puppy?  In the draft, it veers amusingly between being called
XHL (Extensible Hyper Linkage) and XHA (X. Hypertext Architecture), 
depending on who wrote the section; Jon Bosak always says "XML-Link", but I 
don't know if that represents an opinion.  I *think* [Jon, shout if you 
disagree] that this would be an area to which the WG could usefully turn 
their attention, during the very short time before we start to emit large 
numbers of votable items in Michael's A., B., C., style to provide a 
structure for the meat & potatoes debate.

4. I hereby propose to the ERB that this draft spec be used as the raw
material to generate a large number of votable items.  I believe that
by proceeding through the outline in a fairly linear fashion, we can
generate on the order of 50 concrete questions, and then drop back into
a rhythm like that of last fall.  I guess I'm volunteering to do the
question generation, unless Steve wants to [please, Steve?].

A corollary of this proposal is that if anyone on the WG feels that
the draft spec is *not* suitable for use as a votable-item generator,
i.e. the outline is just missing a huge required area of discussion, or
is organized deeply wrong, or whatever, this is the time to speak up.

And just to make it clear, *everything* is on the table.  Just because
the spec happens to use subelements for link-ends a la TEI as opposed to
token lists in attributes a la HyTime is not evidence that it has to
be this way... so votables will range in flavor from "should there 
be links of type X" to "should links carry BEHAVIOR information" to
"should link-ends be tokens in attribute values, or subelements" to
"should multilinks be corralled, or allowed to wander throughout docs" to
"should link-ends have content" to "should other documents to be processed
at the same time be identified" to "should they be ID'd by entity name or 
URL or what" to "what should the name of the ROLE attribute be".

5. The spec is painfully short on examples, has completely missing
subsections, and doubtless contains any number of inconsistencies and 
typos and thinkos... while editorial input from anyone is always gratefully 
accepted, it might be a good idea to hold off until we've had a chance to 
agree on what the spec says before we start whittling away at how it says it.

The draft spec is at:


Cheers, Tim Bray
tbray@textuality.com http://www.textuality.com/ +1-604-488-1167