Re: Radical cure for BOS confusion

At 08:57 AM 01/09/97 -0800, Derek Denny-Brown wrote:

>One of the (traditional) difficulties in implementing ilinks is based on the
>fact that any of it's anchors may be in various other documents, some
>parsed, some not.  If link processing can be isolated such that it is either
>guaranteed to be done before or after the parsing of all of it's relavent
>target documents, it becomes much easier.  If it is parsed first than a

Hmmm. I think the problem is mostly that the *intermediate* rungs of the
location ladder can be scatted, not that the end can be far away. But your
point that perhaps the problem can be isolate is a good one.

What if we simplified HyTime location ladders by insisting on the convention
that in XML-LINK documents, all ilinks and all of their subordinate rungs
must be together in a reserved header element (I don't care whether we
reserve a GI or just an architectural form name in this case). Then you
don't have to parse entire documents just to get their ilinks, and you can
simplify/accelerate your implementation a tiny bit by knowing where to look
for some things.