> I have an LALR(1) grammar and lexer definition derived by machine
>
translation of the middle section of Matz's parse.y that compiles with
>
0 conflicts. However, I suspect that many of these conflicts are
> masked
by the compiler-compiler's overlapping tokens facility.
You mean you extracted
BNF from parser.y? Then how do you deal with the lexer, still use the c code?
> In fact, I have just tried an experiment where I "un-overlapped"
>
tLPAREN and tLPAREN_ARG (As well as '(' which is implicitly treated as
>
a token.) This resulted in 5 distinct shift-reduce conflicts and 27
> reduce-reduce
conflicts.
tLPAREN_ARG and tLPAREN have to be differented to parse the
following code:
puts (1) + (2)
> What progress have the grammarians made?
>
I am using smoke test to measure my progress. When parsing 1500+ .rb
files from ruby's standard libary, I have 48 failures yesterday.
Xue
Yong Zhi
http://seclib.blogspot.com
_______________________________________________
Rubygrammar-grammarians mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/rubygrammar-grammarians
