#4159: move Monad and MonadFix instances for Either from mtlto base

[Brandon S Allbery KF8NH]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 7/1/10 04:27 , Claus Reinke wrote:
> The problems start when all of these classes are linked to a single
> Monad instance, and the 'fail' that has been put in it. So far, I had
> thought that there was a way to have all common uses of Either
> compatible with a single 'fail' (representing both non-local returns and
> empty sums), so I've been arguing to keep a sufficiently general Monad

(a) I would guess not.
(b) It strikes me that trying to make one Either do both of these is just
wring.  Either is a perfectly reasonable sum type; "fail" is not a general
attribute of sum types, though.  IMO "fail" should be associated with a
MonadFailure (which might conceivably be a typeclass which has Either as an
implementation; but given the lack of useful information available in its
most common usage as part of the "do" machinery, Maybe is just as reasonable
an implementation).

That said, I'm also not sure if fail maps to the perennially-suggested
MonadZero entirely sensibly, btw.  Okay, obviously it's related to mzero in
some conceptual sense, but it seems to me that it's not related to monoids;
instead, it's an exception.  The fact that fail is usually implemented as a
call to error is symptomatic of this; and treating an exception as a monoid
strikes me as a failed abstraction.

(On the gripping hand, a MonadFailure typeclass would give people who wanted
it the option of doing so.  Of course, if you do this then it simply
replaces the current question with the question of whether the Either monad
should have a MonadFailure instance that is an exception or simply Left, and
what happens if someone disagrees... there are no easy answers.)

- -- 
brandon s. allbery     [linux,solaris,freebsd,perl]      allbery at kf8nh.com
system administrator  [openafs,heimdal,too many hats]  allbery at ece.cmu.edu
electrical and computer engineering, carnegie mellon university      KF8NH
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkwsVkQACgkQIn7hlCsL25VF/QCbBrU+NFx9bEzQva0awdLwZgV5
zyYAoIwemwZUBbLSWEnextXSvjiYR0xG
=egxO
-----END PGP SIGNATURE-----