2014 Applicative => Monad proposal

[David Luposchainsky]

Hey everyone,

here's a short summary of what's happened during the AMP's first week:

- Adding `join` to the Monad typeclass introduces a cyclic dependency,
  leading to an infinite loop if the function definitions are
  forgotten. Until this is solved, the `join` part of the proposal may
  not be a good idea.
  Ticket #7633 solves this problem by introducing compiler warnings if
  the minimal complete definition is not met. Looking at the Trac page,
  it's probably going to be in HEAD quite soon.

- Apart from the `join` issue, everyone seems to be in favor of the
  AMP as proposed. (+19/-0 according to Ben's writeup)

- A `Pointed` typeclass discussion emerged from the AMP thread with
  mixed opinions. If necessary, this should be made a separate
  proposal.

- I was asked about MonadFail by multiple people via email. Again, the
  reason this is not considered in the AMP is that it is orthogonal to
  the proposed changes; the only similarity is that it changes things
  in the same typeclass we're already editing.


So long,

David