Re: Full LR(1) parser generator Hyacc 0.9 release

Thanks for letting me know.


I actually was in contact with the Bison maintainer's team last April,
unfortunately one member (Joel) was not interested in combining this
work into it, because he thought it couldn't handle a problem caused
by using precedence rules to handle ambiguity (Bison couldn't do it
either at least by then). Personally I don't think it matters, since a
LR(1) algorithm is designed for LR(1) grammars, and all LR(1) grammars
are unambiguous. Any solution to ambiguity are side patches, and
should be orthogonal to LR(1) or LALR(1) algorithms. So incorporating
Pager's LR(1) algorithm, and solve the above issue independently
should not be a problem in my opinion.


The discussion back then can be found at the end of page
http://www.mail-archive.com/bison-patches@gnu.org/
in the thread "Re: LR(1) paser generator based on Pager's algorithm"
and http://www.mail-archive.com/bison-patches@gnu.org/msg02006.html


Bison has been the combined efforts of many people. I would
appreciate the idea of incorporating this into Bison. But since one
of their members was reluctant, I just keep up with my parallel
work. So far the performance of Hyacc is comparable to Bison except
for really complex LR(1) grammars (then LR(1) could be much slower
than LALR(1) by nature), but of course still lacks some interface
features. Bison also supports GLALR besides LALR. Hyacc will
concentrate on LR(1).

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