>>>There will be great pressure from the publishing industry, where I come
>>>from, to focus primarily on the richness of the markup of individual
>>documents. For the most part, those people should be ignored! They don't
>>
>> Define markup, and I'll decide whether to agree or not.
>
>I guess the difference is the explicit purpose of the markup. For example,
>tagging the words, "Bill Gates," with the information that he is chairman
>of Microsoft seems to me to be explicitly for navigational purposes. It
>doesn't define the structure of the document in which is contained; it can
>be used to define the relationship of the document to other documents.
OK. If, as it seems, you are saying that document structure markup
is unimportant, I'll have to disagree strongly. The value of
structured data cannot be underestimated. I *do* agree that
navigational markup should be an integral part of any networked
documents.
>I've never really made the distinction between net navigation and
>sub-document navigation. That's a matter of ownership and physical
>location, primarily, I think.
I agree. So why build the one without the other? The best thing I
could imagine would be a single method for naming a collection of
documents, a document, or a part of a document.
>I'm not sure I understand about the tree structure. Surely you aren't
>implying that *all* SGML documents should have a hierarchical structure,
>which is how I interpret "tree." Are you?
At the risk of upsetting everyone here, I'll point to HTML as an
example of a "flat" DTD, so certainly, *all* SGML is not structured.
I will say that I think almost all documents have some inherent
structure that *could* be made explicit. Such structure can be used
tp qualify searches, and to aid navigation...
>On the other hand, I have created documents like the one you describe,
>which clearly are a great benefit of hypertext. I love documents and
>sub-documents, as you say, that expand and collapse according to complex
>rules (complex to the programmer, that is).
Given structured documents, and stylesheets, such things are easy to
accomplish. One could generate TOC's in the same way...
It sounds to me that for the large part, we agree, but for some
reason, I feel that what has been said has not answered any questions,
nor brought any difference of opinion. Did I miss something?
nick