Re: 1.5: Discuss link characteristics?

Certainly it's termini and not links that in the first instance appear
somewhere, and therefor may need to be displayed, if you have that
kind of application (e.g. a browser).

But neither traversal or action is at the right level of generality.
[See next message about traversal]
Both you and Martin are too focussed, in my view, on browsers and
mouse clicks.  As a possibly useful example, consider the case of
external, non-participating 2-ended links to annotate the translation
relationship between the sentences of french and english versions of a
bilingual document.  A terminology extraction tool will exploit the
semantics of such a link (i.e. behave in a certain way) completely
differently to a translation tutorial program.

I realise this is dragging us back towards the whole link
{semantics/type/behaviour} debate, which I have done my best to
follow, but it leaves me feeling that although we need to clearly
distinguish between

1) intrinsic XML semantics for links, i.e. termini, participation, etc.;
2) XML Application (in the strict SGML sense of 'application')
   semantics, e.g. the bilingual alignment example above;
3) Implemented application behaviour

we are only in the business of specifying (1).