Re: Are we losing out because of grammars?

[Rick Jelliffe <ricko@allette.com.au>]

From: K.Kawaguchi <k-kawa@bigfoot.com>

>the sequence of "a c a b b b b" satisifies all of your rules, but is
>invalid compared to your original grammar.

Yes. That is my point (in answer to Len's question),
   * that grammars and path-based rule systems are not equally expressive;
   * that grammars and path-based rule systems are not a subset, either of
the other; and
   * that the structures that grammars can handle which  path-based rule
systems cannot are
     less useful and of dubious merit (because there are structures present
which are not
    labelled and therefore inaccessible to standard API-based systems that
just use the
    infoset; to say these structures are significant puts us on the slippery
slope to PSVI.)

So I am not trying to say that Schematron is strictly more powerful than
RELAX or TREX, because they can model different things.  I am saying,
however, that I think the structures that path-based rule systems can model
are better than those that grammar-based-systems can model.  I don't buy the
ancestor-only rule for context-processing here: there is no reason why any
arbtrary set of data should form a tree rather than a graph, and
consequently there is no reason why some data or structure in one part of a
document may not constrain the data in another part in some important way.

>I still can't understand what you were tring to show by your example...

The example was trying to show that it is very easy to infer by various
heuristics
simple rules to model a grammar (and to give simple natural language
explainations).

But that to model everything in any arbitrary grammar was either
  - impossible, because of unbounded repetitions of sequences that can occur
in
     other parts of the grammar (I don't have proof this is so, but I cannot
see otherwise
     how it could be done), or
  - nasty, because the resulting XPaths could have to trace through long
strings simulating the paths possible in the FSM, or
  - not useful for the purposes of Schematron, because the natural language
description is
valued just as highly as the artificial language expression and long complex
chains may be too difficult have nice descriptions.

Cheers
Rick Jelliffe