On 3 Aug 2007, at 12:51, Hans (Req man) skrev:
Hans Aberg schreef:
Don't expect to get much out of error recovery. It is a research
topic, and particularly bad under LALR(1), as the parser states
are merged in a way that when an error token appears. some
reductions may still take place.
Instead, uses GLR (the Bison %glr option) to deal with grammar
ambiguities. But it does not help with ambiguities on the lexical
level, and Paul Hilfinger, who wrote the GLR parser, says it is
not adapted to use with a lexer that splits the lexemes.
Hans, I see, and thanks for the suggestions. For now I will
continue testing this release of diction (1.11). I will look into
your hugs suggestion as well as the bison %glr option afterwards.
Regards, hans
I am cc'ing Paul, in case he might have something to say about the
topic.
Hans Aberg
_______________________________________________
[email protected] http://lists.gnu.org/mailman/listinfo/help-bison
