Re: keyword tag, REL, REV, TYPE, INDEX....

Ian Graham (
Tue, 2 May 95 10:40:10 EDT

This discussion seems to have several HEAD issues running in parallel, and
disagreements seem to be arising because we are looking for different things
in the same place (namely the LINK REL attribute). To clarify things
(for me, anyway) I have summarized below what I *think* are the models that
have been proposed: Help me out here folks!

1. The desire for simple (binary) LINK element relationships, that can be
used by a browser to (a) construct customized button bars linked to
certain resources, (b) preload special document, (c) other, as
yet undefined, actions.

Typical relationships of this type might be: Previous, NExt, ToC,
LocalHome ..... (choose names you like, folks!) This is the simplest
REL/REV scheme, and probably the only one that can easily be
implemented. Browsers don't even need to know what the REL names
mean -- they could simply put up a menu button, with a label equal to
the REL attributes value, and link it to the associated URL.

2. Complex LINK relationships. Several examples have appeared, such
as REL = "Join(Glossary,Author)" (Bert Bos's example, which might refer
to the author of a document's glossary), and so on. These are nonbinary
relationships, and can be heirarchical or relational. However, they are
not bound to a defined object model.

3. Object Model -- The idea being that the collection of HTML
(or other) objects can be related as objects, with inheritance,
subtyping, etc. This seems related to the TYPE element suggestion
posted by Steven Fought. The problem of course is -- which object
model (foundation class....) do you use? Clearly many different ones
would be appropriate. Does this imply an "object" style sheet?

4. INDEXING features. I've always liked the idea of having document
keywords, and the META tag seems appropriate here. This does not
preclude more sophisticated full-text indexing, but allows for fast,
simple indexing of large collections.

5. Relationships between thes things. These all seem related. For example,
LINK could link to different indexes, one generated from META data, and
another from a full-text search, and another to an object model for
the collection.

Any other thoughts? Am I way off line here?


Ian Graham .....................................