[antlr-interest] Multiplication and pointers in C-style language

[Gavin Lambert]

At 01:06 11/04/2008, David Olsson wrote:
 >Unfortunately, my language is not like C in all cases.
 >Specifically, my language is a two pass language. In the
 >first pass, all the parsing is done. In the second, type checks
 >and such is done. This means that I don't need a preprocessor
 >nor includes. This, however, also means that there is no easy 
way
 >of telling if an identifier is a type or a variable during
 >parsing; that information isn't available until the second pass.
 >
 >I've looked at Terence Parr's example Java grammar but since 
there
 >are no pointers in Java it is of little help.

You could possibly have a look at a C# grammar; I think there's a 
C# ANTLR grammar floating around somewhere, but failing that you 
could look at the official spec.

C# has pointers (though they're rarely used) and doesn't require 
pre-declaration of types (no header files), so that's probably the 
closest example to what you're trying to do.

I don't know offhand, but I think C# gets around the problem by 
declaring that expressions cannot be statements in their own 
right; combining this with the fixed signatures for operator 
overloading means that a * on the left of an assignment or on its 
own can only possibly be a pointer declaration, not a 
multiplication, since a multiplication cannot produce an 
lvalue.  (Unlike in C++, where you can overload it to return 
whatever you want.)

 >Maybe the only solution is to change the pointer syntax so that
 >there are no conflicts with other operators (the benefit of
 >inventing your own language ;-))...

That might be the easiest option, but it's not the only one :)