home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ebt-inc!uhura!sjd
- From: sjd@uhura.uucp (Steve DeRose)
- Newsgroups: comp.text.sgml
- Subject: Re: Straw man proposal for overlapping markup
- Message-ID: <5444@ebt-inc.UUCP>
- Date: 28 Jul 92 13:17:26 GMT
- Sender: news@ebt-inc.UUCP
- Organization: EBT
- Lines: 22
- Originator: sjd@uhura
-
- Darrell wrote in response to my overlapping-markup proposal:
-
- > Can the parser validate balancing tags for overlapping elements without
- > ceasing to be context free? I don't think so. (A sketch of a proof:
- > imagine a pushdown automaton stacking the opening tags, and then having
- > to match up an end tag with some opening tag that's inside the stack).
-
- True, this isn't context free (except that given a concrete syntax and
- quantity set, there is only a finite number of possible ID values, and
- so the problem is actually finite-state; but only *barely* finite).
-
- However, ID/IDREF validation is also not context-free. I don't see this
- as a problem, but as being like type-checking in programming languages;
- checking that the types of two variables in an assignment are compatible
- is not context free, but is left to the semantic part rather than
- the syntactic part of the language's grammer. Readers unfamiliar with
- either linguistics or compiler construction would do well to read a bit
- of Chomsky to follow the current thread (not suggesting Darrell has such need!).
-
- By the same token, it is fair to say that ID/IDREF validation is a place
- where SGML defines semantics, not syntax; when various posting have claimed
- SGML deals "purely" with syntax, they were incorrect.
-