My apologies for taking so long to reply to this (I was onto something totally different).
> There are a number of algorithms that extends PEGs to support left > recursion, but do they agree on how ordered choice should be interpreted if > one of the alternatives is left recursive? The situation I am thinking of: > > A <- A a / a A / empty > Vic Nightfal finished implementing left recursion in TatSu (with the help of Nic Laureant and others). While testing for associativity, I remembered this note, and came back to the above example. I'm pondering if a left-recursive rule that is nullable, as in the above case, is PEG, or something to be rejected, as done in TatSu, because `A a` matches `empty*` with that rule. This version of the same language seems OK for PEG. S <- A / empty A <- A a / a A / a Regards, -- Juancarlo *Añez*
_______________________________________________ PEG mailing list PEG@lists.csail.mit.edu https://lists.csail.mit.edu/mailman/listinfo/peg
