Nicholas Clark wrote in perl.perl6.language : >> Actually I don't think you can define a grammar where two operators have >> the same precedence but different associativity. Be it a pure BNF >> grammar, or a classical yacc specification (using the %left and %right >> declarations). > > But that would mean only perl6 could pass perl6, which isn't much different > from the perl5 situation, is it?
I meant that if <~ and ~> are going to have the same precedence, you can't parse s ~> t <~ u It's not a well formed phrase of the language (even though this language can't described by a nonambiguous BNF grammar.) In fact, this is different from the Perl 5 situation you're alluding to.