|Tool support for resolving LR conflict firstname.lastname@example.org (2005-09-07)|
|Re: Tool support for resolving LR conflict email@example.com (2005-09-10)|
|Re: Tool support for resolving LR conflict firstname.lastname@example.org (Karsten Nyblad) (2005-09-17)|
|Re: Tool support for resolving LR conflict email@example.com (2005-10-02)|
|Re: Tool support for resolving LR conflict firstname.lastname@example.org (Sylvain Schmitz) (2005-10-04)|
|From:||Sylvain Schmitz <email@example.com>|
|Date:||4 Oct 2005 01:43:05 -0400|
|References:||05-09-02705-09-038 05-09-077 05-10-016|
|Posted-Date:||04 Oct 2005 01:43:05 EDT|
> Is DeRemer's work being used in popular compiler compilers?
Yes. It is arguably the most popular LALR computation technique around,
used in GNU bison, Berkeley YACC, and many others; the Gardens Point
Parser Generator presented a week ago on this newsgroup is yet another
example. And when the technique is not used as such, a similar one is,
and at least their computation of relations closures using SCCs is being
used; for instance, I would be very surprised if Karsten Nybald's
implementation did not.
The application of their SCC technique to detect non LR grammars is not
used as often; I think I have only seen it once, maybe in jikes. Their
conjecture on this point is proven in
Sager, T. J. 1986. A short proof of a conjecture of DeRemer and
Pennello. ACM Trans. Program. Lang. Syst. 8, 2 (Apr. 1986), 264-271.
> A paper by Bermudez and
> Schimpf [...] disproves the claim of DeRemer, that says NQLALR(1) is
> superset of SLR(1).
The false claim does not harm the validity of their work on LALR(1):
after all, they only introduce NQLALR(1) in order to help others
avoiding the mistake of computing NQLALR(1) lookahead sets instead of
the real LALR(1) lookahead sets!
Return to the
Search the comp.compilers archives again.