Re: advice needed re: parsing C decl syntax

LR parsing implies bottom up parsing.


The standard trick in LL (recursive descent) parsing is to delay doing
something about a token if it could be one of several things.
Basically you are rewriting your grammar so that it is LL(1) instead of
LL(k), k > 1, by left factoring. For example:


statement -> identifier := expression
-> identifier ( paramlist )


is not LL(1). But when you factor it out thus:


statement -> identifier stmttail
stmttail -> := expression
-> ( paramlist )


Now one token of lookahead suffices.


Ken
[This is true, but in real recursive descent compilers that I have seen, it's
as likely that you parse these things by kludgery as by adding productions.
Most recursive descent compilers are written by hand, so you'd parse the first
syntax by eating the identifier, remembering it, eating the next token, and then
going ahead with the appropriate construction. But I'd rather let yacc do the
work for me. Yacc can parse anything with enough help -- I have written a yacc
parser with a moderate number of kludges that correctly parses Fortran 77, a
language in which a statement can't be tokenized until you know what kind of
statement it is. -John]

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