Re: grammar ambiguity

>A different approach to extending the reach of parsers is the use of
>non-terminal lookahead. I have 3 papers by someone who researched
>that, and called the technique non-canonical LR if I recall correctly.
>The k in this technique is k non-terminals and not just k tokens.
>This technique can guarantee that the grammar is unambiguous and can
>parse grammars that are not LR(k) for any k. However, I'm not aware
>of the author ever releasing his software. We also used that
>technique in the first version of Yacc++, which we never released.
>Foolishly, we replaced the technique with the minimal state LR
>algorithm in the versions that saw the light of day.


It appears that you preferred this implementation. What were the
tradeoffs involved that led you to drop this implementation, and what
has prevented you from reimplementing it in subsequent versions of
YACC++?


William B. Clodius

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