Re: is lex useful? Or how about these other tools...

Regarding PCCTS:


-- non-commercial (and seems to be staying that way)
-- pretty-well supported
-- some future development on-going, but somewhat on hold
(see newsgroup)
-- new book on it coming out (soon?) written by author of tool
(pretty much a reference manual)
-- well integrated, or you can use flex if you'd like
-- C and C++ output code (soon (?) Java as well)
(I strongly recommend C++ output)
-- LL(k) using syntactic predicates for infinite lookahead
(if syntactic predicate present, parser will take a
guess and try the alt) Note that the predicates can
eliminate the need for left-factoring, although it's
cleaner to left factor where it's easy to do so
first
-- generates recursive-descent parser -- easy to debug and
set up trace functions to see how things are reducing
-- semantic predicates allow selection of an alt based on
evaluation of an expression (can determine "type" vs.
"function" in parser instead of in scanner.)
-- automatic generation of "lisp-ish" ASTs
(a separate tool, Sorcerer, can be used to walk
these trees using a second grammar)


-- you may want to follow comp.compilers.tools.pccts
-- you may want to look at
http://www.igs.net/~mtr/software-development/pccts.htm


-- as for future maintenance people, I don't see it as a big
problem. Yes, it's different, but it's easily as
readable as yacc (take that as you like). Just make
sure that you leave a copy of the appropriate PCCTS
docs and some web pointers (such as the newsgroup and
caravan of deams page (the one above) ). Also make
sure you have a copy of Tom Moog's "Notes for New
Users" with those docs as well.




Regarding Visual Parse++:
-- don't know much about it, but isn't it basically yacc with
a few graphical tools to analyze conflicts and show
parsing states as a file is parsed?




Go with PCCTS! (Am I biased? I'm not related to Ter, but I love the
tool! What else can I say?)


OK, so it is a bit biased, but I hope it's some help anyway...


-- Scott




> I've been looking at compiler tools for a few weeks now and
> I'd rather like an answer to your question as well. I would
> also like to add a bit to it...


--
Scott Stanchfield McCabe & Associates -- Columbia, Maryland
--

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