Re: Radical cure for BOS confusion
Jon Bosak wrote:
> Intuition tells me that this must be too simple, but I would very much
> like to know why. What's wrong with this picture?
You are at the heart of it: two levels of specification.
One captures what the user of XML must know to use XML; the
other, captures what the implementor must know. We want ease for
both with extensible power of expression.
No matter what we do to define it, the XML user is the
SGML or HTML user and understands elements, attributes, entities (some)
and files. Some of them, the SGML designers, know what DTDs are and how
to write one. Some of them, SGML systems engineers, know how to write
parsers for that. So it goes. We need an XML hyperlinking design
that takes this into account.
For the user, use the minimal tricks of SGML to identify
a hyperlink. If they insist on editing by hand, make it
I say, adopt TEI linktypes for XML 1.0. Express them with a HyTime
specification. That leaves the door open for more should XML need
more HyTime tricks. It gives the HyTime implementors a
lucrative target for inexpensive versions of their technology,
and a guaranteed set of network-ready protocol engines.
Does this have a downside?