Re: Beginner help with LALR(1) closure

Daniel J. Salomon wrote:
> Their method is more efficient than the ones given in the Dragon book,
> but quite complex to understand from their paper.


Gianni Mariani <gianni@engr.sgi.com> wrote:
|> Do we care any more ? Are LR(1) tables so big that todays processors
|> are unable to deal with them effectivly ? (seem X lately ?).


The method of DeRemer and Penello produces exactly the same LALR(1)
parse tables, it just does it faster than the algorithms in the Dragon
book. So only compiler implementors would care about this and only if
they were reprocessing their grammars over and over.


Grammars for modern programming languages seem to be getting bigger
than olden day grammars and parser generation time is worse than
linear in the length of the grammar (perhaps as bad as n**2). So
maybe we still need to worry about parser generation speed. However,
since processors are getting faster too, maybe it is all balancing
out.


--
Daniel J. Salomon -- salomon@cs.UManitoba.CA
              Dept. of Computer Science / University of Manitoba
              Winnipeg, Manitoba, Canada R3T 2N2 / (204) 474-8687
--

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