Re: Syntax diagram driven parser

On 24 Sep 2000 13:43:43 -0400, "Brian Webb" <brian_d_webb@hotmail.com>
wrote:


>Does anyone out there have any experiences or opinions on this
>approach to parsing? It seems like a more natural way to specify a
>language than using BNF, lex, and yacc (which I can claim ignorance
>of).
>
>Before I spend too much time trying to turn my code into something
>other people might want to use, I'm trying to answer the following:
>
>Would anyone use it, even if it worked well?


Maybe. I'd sure like to see it. I'm curious to know how you'd attach
actions to productions.


>Are the current methods (BNF, LL?, LALR?) good enough?


I'd say BNF is good enough, but I wouldn't mind having a visual tool
that could take syntax diagrams and produce the corresponding BNF,
just as a convenience. (Going the other way is a trivial exercise, but
I can't think of any tools off the top of my head that do it.)


My understanding of machine-parsable grammars -- and I'm sure there
are plenty of people here with broader understandings -- is that the
limits of grammars which can be parsed in linear time are pretty well
known and firmly circumscribed by solid proofs. That being said, I
don't think anyone would deny the possibility of revolutionary new
methods of building compilers, though opinions on the likelihood of
such methods are likely to vary widely. I'm of the mind that trying
certainly doesn't hurt, and you might find something new.


>What advantages does this approach have?
>Are there useful grammars that it excels at?
>What disadvantages are there? Is it too slow?


My first question would be what sort of grammars your tool produces.
It sounds like you may be confusing the representation of the grammar
with its underlying properties. Whatever the graphical representation
of the grammar, the generated parser is probably either top-down or
bottom-up, for example, and the grammars that are possible with your
tool could probably be represented by an LL or LALR grammar.


Anyway, it sounds like an interesting tool, and I'd certainly like to
see it.


--Eric

Return to the comp.compilers page.
Search the comp.compilers archives again.