The question I wish to ask is succintly given in the title. Let me give an example of the grammar in question:
identifier_list
: identifier
| identifier_list identifier;
lambda_arguments
: '(' identifier_list ')'
| identifier;
lambda
: lambda_arguments '=>' expression
Then we add in the normal C expression grammar- particularly,
primary_expression
: '(' expression ')'
| identifier
| lambda;
The real question is, is this grammar LALR(1) parsable, i.e., capable of being parsed by automated parser generators? Or does it require a hand-rolled or GLR parser? Note that I wish to know specifically about this subsection, not the context-sensitive keywords stuff or any other section.
What I'm thinking right now is that if the parser sees '(' identifier ')', this has two valid parses, so if the parser sees identifier, looks ahead to ')', it won't be able to decide which parse tree to go down. This could just be a shift/reduce conflict, though, that I could eliminate by assigning some arbitrary precedence (probably favouriing '(' identifier ')' ).
Edit: Actually, I was considering stealing using this subsection of the grammar for a similar feature in a new language. I already have anonymous functions similar to JavaScript in grammatical form but my guinea pigs squeaks feedback complains they are too verbose for many uses, and pointed out the C# lambda expressions as a more ideal solution. I was concerned about potential ambiguity resulting from this solution. So, really, I was only interested in that subsection. The other stuff like generics and casts are non-issues for me.
Previous editions of my grammar are mechanically parsable and I wouldn't want to lose this property, and my previous experience with a mechanical generator tells me that it's best to check he