Ryan Ingram
When I was implementing a toy functional languages compiler I did away with
precedence declarations by number and instead allowed the programmer to specify a partial order on declarations; this seems to be a much cleaner solution and avoids arbitrary precedences between otherwise unrelated operators defined in different modules. I agree. I don't declare operators very often, and when I do I always struggle to remember which way round the precedence numbers go. I usually end up hunting for a Prelude operator that works the way I'm aiming for, then copy its definition. It would be much easier to declare the fixity of myop to be same as someotherop (which would presumably have to be already declared/fixed in an imported module). [It's also slightly counterintuitive that the thing being defined comes last in an infix declaration, and that the stand-alone operator isn't in parens.] infixAs !! .$ -- fixing myop (.$) to be fixed as Preludeop (!!) If you wanted to define precedence relative to some other operator(s), it might be clearer to give some model parsings (grabbing some syntax something like Ryan's): infix .$ (x ** y .$ z .$ w) ==> (x ** ((y .$ z) .$ w)) -- === infixl 9 .$ OTOH, I think Евгений's proposal is getting too exotic. Do we really need such fine shades of binding? Will the reader remember how each operator binds relative to the others? Surely a case where explicit parens would be better. (Anything else we can bikeshed about while we're at it?) AntC