Re: The Zen of Shift/Reduce and ambiguous grammars

[Bruce Frederiksen]

David Beazley wrote: I still stand by my earlier statement that reduce/reduce conflicts are almost always bad. Yes, good advice. But on a different topic, I was wondering whether you went over the material I had included in the P.S. of my message.  Basically, I'm not sure whether you are thinking that PLY should report unreachable rules or not.  If not, no problem.  If so, then either the second example in the P.S. doesn't seem to produce any such notification; or the first example produces an erroneous notification (depending on what the notice looks like).  I only bring this up because you say that PLY is used in a teaching setting and is designed to catch lots of errors... -bruce On Mar 23, 2009, at 4:06 PM, Bruce Frederiksen wrote: David Beazley wrote: Reduce/reduce conflicts are almost always bad---they mean that certain rules in your grammar will never be triggered. This isn't always true. Consider the following grammar: # foo.py from ply import yacc tokens = ('A', 'B', 'C') def p_grammar(p): ''' rule1 : rule2 B | rule2 C rule2 : rule3 B | rule4 rule3 : A rule4 : A ''' yacc.yacc() This has a reduce/reduce conflict, but all rules are used: A B B => rule3 B B => rule2 B => rule1 A C => rule4 C => rule2 C => rule1 -bruce P.S. Should PLY 3.1 report unused rules? On this input, it outputs: WARNING: no p_error() function is defined Generating LALR tables WARNING: 1 reduce/reduce conflict WARNING: reduce/reduce conflict in state 1 resolved using rule (rule3 -> A) WARNING: rejected rule (rule4 -> A) The final line says that (rule4 -> A) is rejected, but (rule4 -> A) is still used in other situations (as noted in the second example input, above). Maybe the WARNING just means that the rule was rejected in this one state. But if that's the case, the following input never uses (rule5 -> A) under any circumstance: # foo.py from ply import yacc tokens = ('A', 'B', 'C') def p_grammar(p): ''' rule1 : rule2 B | rule2 C rule2 : rule3 B | rule4 | rule5 rule3 : A rule4 : A rule5 : A ''' yacc.yacc() And yet the WARNING message is the same: WARNING: no p_error() function is defined Generating LALR tables WARNING: 3 reduce/reduce conflicts WARNING: reduce/reduce conflict in state 1 resolved using rule (rule3 -> A) WARNING: rejected rule (rule4 -> A) WARNING: reduce/reduce conflict in state 1 resolved using rule (rule3 -> A) WARNING: rejected rule (rule5 -> A) WARNING: reduce/reduce conflict in state 1 resolved using rule (rule4 -> A) WARNING: rejected rule (rule5 -> A) And there is nothing is parser.out indicating the (rule5 -> A) is never used. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "ply-hack" group. To post to this group, send email to ply-hack@googlegroups.com To unsubscribe from this group, send email to ply-hack+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/ply-hack?hl=en -~----------~----~----~----~------~----~------~--~---