[switching to cafe] On 14 Dec 2010, at 08:59, Sittampalam, Ganesh wrote:
John Smith wrote:
I would like to formally propose that Monad become a subclass of Applicative, with a call for consensus by 1 February.
I would prefer that we have some proposal like class aliases implemented before we start fundamental restructuring of basic type classes. This would help to limit the disruption these changes cause, which will be substantial.
I'm inclined to agree. What's even more galling than not having Functor and Applicative instances for Monads is having to write them! At the very least, can we open this particular vessel of vermicular splendidity and verify that the inmates are still grinning with persistence? If I recall, the class alias proposal was rather more ambitious than necessary to solve this problem (though that is no crime), but still a little wrinkled in the corners. Looking at the relevant webpages, I see the proposal has lots of attractive motivational examples, but less by way of definition. But perhaps, somewhere, it has been defined enough for a prototype? It's surely worth looking at some way to define default instances for superclasses, provided they can be overridden or switched off. Where did this train of though leave the rails, again? Cheers Conor