Hi
If mzero, there's less gained over Maybe, but still no guarantee of safety.
If a program has a safety problem with mzero, that anyway needs to be detected. So I don't think that will have an effect on overall program safety, it's an orthogonal safety concern.
I don't think this is the case. mzero vs. Nothing One doesn't crash, the other may crash depending on what is floating around. Perhaps I've used type inference extensively, and the actual decision of what mzero means is 10 functions away. Now its really hard to see the program is safe, versus not. Of course, if you use an automated thing, this is irrelevant (traditional plug of Catch: http://www-users.cs.york.ac.uk/~ndm/catch/) - but I think its still useful for humans to be able to decide this easily.
Other than minor convenience adavantages one way or the other depending on what program I happen to be writing, there is only one thing I can think of that might make a difference. Current compilers might produce programs that are slightly less efficient with MonadPlus, and there might be no easy way to work around that if MonadPlus is hard-wired into the library. Efficiency experts?
Discussions about the ideal and perfect API should not come down to such minor considerations as efficiency :-) In reality, I don't think it makes any difference, and in fact I think the Monad class might make some things more efficient, as the class is likely to be specialised away but returns of Maybe with a liftMaybe aren't likely to be optimised out, as its essentially deforestation on a non-list type, which GHC can't do (but something like Supero could http://www-users.cs.york.ac.uk/~ndm/supero/) Thanks Neil