Re: shift-reduce conflict,

kers@hplb.hpl.hp.com (Chris Dollin) writes:
> Can you make Bison happy by allowing
>
> '(' <expression> ')' <expression>
>
> as legal syntax, and then requiring that the prefix-expression denote
> a type? Off the top of my head, this should be OK, shouldn't make
> error reporting any more confusing, and even allows for future expansion
> of the language in the direction of type-expressions :-)
>
> Consider (a) - b. Is this a cast of -b, or a-b ?


\kosh Yes.


It depends on the bindings. My fragment above makes your example ambiguous;
mea culpa. For Java of course you have to pick for the second <expression>
whatever Java allows after a cast -- alas, my reference is at home -- and
I presume that this is not ambiguous.


Were I choosing, I'd make that right operand "as tight as possible". Well,
that's not true; were *I* choosing, I'd not give casts that syntax at all;
I can think of more useful ways to overload parentheses.
--


Regards,
Kers.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.