Dylan writes:

On Tue, Jul 16, 2002 at 04:02:44PM +1000, Bernard James POPE wrote:

...

I would like to use do-notation in the transformed program, but have it refer to Prelude.Monad and not MyPrelude.Monad which is also in scope.

Why do you have a MyPrelude.Monad (different from Prelude.Monad) if you don't want to use it?

Hi, This is part of a debugger for haskell. The approach taken is to transform the original program. Under transformation the types of functions change. If the user's program calls >>=, then in the debugged program that is a call to a transformed version of >>= (ie MyPrelude.Monad - don't worry it isn't really called "MyPrelude"). Btw all do notation from the original program is desugared during transformation. It just turns out that the transformed code is neater if I can use do notation in it. Which is quite helpful because it makes it easier to check that the transformation is doing the right thing! However, for this I have to use Prelude.Monad instead of MyPrelude.Monad, because the latter is transformed and all the types are changed. It really isn't a big deal. I can do without the do notation. I just have to change the transformation. If there were more fine grained control about how desugaring worked then that would help, but that would also add a lot of extra complexity to the compiler and I doubt the benefits would outweigh the costs. Cheers, Bernie.