🚀 go-pugleaf

In article <23266A@erik.naggum.no>, erik@naggum.no (Erik Naggum) writes:
>The problem, as posed, was one of overlaps.  It wasn't really about
>overlaps at all, but about set relations.  I think, consequently, I have
>solved the problem, by redefining the solution space from what Darrell
>wanted it to be, because he had intentionally made it empty by the form
>of his question.  

  I think it's the problem space you're redefining, and not the solution
space.  I said "overlaps", and I meant overlaps of sequences.  X is not 
the set {"now", "is", "the", "for"}, it is the sequence "now is the for",
a sequence which happens to be embedded in another sequence (the total
sentence). X is a discontiguous sequence, not a set of unique words. 

>My solution to this problem uses the fact that SGML does, indeed, not
>specify any semantics for the individual elements or for the element
>structure.  Relatively intuitive, however, are some relations between
>elements, such as "precedes", "succeeds", "contains", and "is contained
>by".  Each distinct element is also seen as a separate instance of the
>element type.  Further, the content of one element is not also the
>content of parent elements, although the child element itself is.

  "Relatively intuitive" is a new one on me.  What does it mean? Does 
it mean SGML can communicate this information, but only informally?  
Or that you can infer the relationships if you want, given a tagged 
document, but you have no idea if they are actually meant to hold?  

-Darrell.