Re: 1.4 f: terms for links colocated with their ends

> | Is there a reason why we can't just call these "direct" and "indirect"
> | links?
> I agree with Liora.  These work very well intuitively and are about as
> apropos as anything else that's been suggested.

They are good terms, but not the right ones (sorry).

An indirect link ought to be the case where I have a link (a thingie)
that points to another link, and that link is the real link that is used.


Link called Priscilla:
    I am a link betweeen
	word 3 of paragraph 1 of chapter 2 in document 301
	all of footnote 27 in document 906

Link Called Ignatius:
    I am an indirect link.
    When you use me, you are really supposed to use Priscilla instead.

    That means that if you are going uphill, you get to
	word 3 of paragraph 1 of chapter 2 in document 301
    and if you are coming back downhill, you get to
	all of footnote 27 in document 906

The TEI Pointers support such indirection, for example, although without
explicit use of the names Priscilla and Ignatius :-)

As I understand the proposal, there are three things conflabulated:
(1) Internal or External: whether the link is contained within the
    document that uses it;
	a HyTime "ilink" can be internal or external, but
	a "clink" (as far as I know) must be internal.
	An HTML <A HREF=...> must be internal -- it must be contained in
	the document that's at one end of the link.

(2) With an implicit end document: whether the link is between
    this document and others, or between two arbitrary documents not
    necessarily including this document;
	<LINK> in HTML establishes a link between the current
	document and another one; one end of the link is implied by
	putting the <LINK> in a particular document, and HTML software
	(such as Mosaic or Lynx) inferrs this.

(3) participating or detached: whether one end of the link is where the
    link occurs or whether the link's markup itself is not part of the link.
	The HTML construct <A HREF=...>xxx</A> surrounds a region of
	a document (xxx) that's one end of a link, so the <A> markup
	participates in the link.  Of course, <A> has to be in the document
	that's at one end of the link, and hence is also Internal.

	A HyTime "clink" is contained in the document that's one end
	of the link, so it's Internal, but the clink markup can go
	anywhere in the document, and is detached; you establish the
	position within the this-end-document using an ID/IDREF link to
	point to the clink, a subject of much confusion amongst HyTime
	learners.  (HyTime doesn't describe it like this -- I am
	concerned here with giving an example of a detatched link,
	not with describing HyTime correctly)

An annotation held in a separate file might be an exmple of a link
that's External.

If the actual annotation is placed inside the markup of that External link,
the markup is Participating in the link -- i.e. the link is itself one of its
own end points.

If all the annotations in that separate file refer to the same document,
perhaps as determined by a header in the annotation file, then the
annotation links in tht file needn't name the other document each time;
in that case, our annotation has an Implicit End Document.

Not all combinations are useful: a unary external participating link with
an implicit endpoint is a binary link waiting to happen :-)

Having sorted out the concepts (I hope), perhaps we can find terms for
them that we like, and preferably that don't use analogies like harpoons
(pointy bits on the ends of ropes) or railway stations (terminuses --
although checking the dictionary I see that in English a terminus can
also be a milestone as well as a railway station or a town or city
containing one, in English it can't be the end of a link) or airports.

I've got lots of analogies.  Full of them.   I can tell you why an
External Unary Link is like an odd sock in an otherwise empty sock
drawer if you like :-)

For now, I have
    Internal if it's contained in a document it refers to;
    Participating if it's one of its own end points.
    With an Implicit End if one end point (or more) is deduced from
    the context in which the link occurs.