Aristid noticed these broken implementations of (>>) in the two
copies of the Perhaps monad - the 'Unknown' case is not
consistent with (>>=).
It seems best just to remove the specialisations as the compiler
ought to be able to optimise these anyway.
Aristid tested with replacing the faulty case with 'undefined'
and couldn't find any extra failing tests, so it also seems like
a very safe change.
2 patches for repository darcs-unstable@darcs.net:screened:
Fri Sep 26 17:45:01 BST 2014 aristidb@gmail.com
* Perhaps was not a valid Monad
Sat Sep 27 08:37:53 BST 2014 aristidb@gmail.com
* Remove faulty (>>) in second Perhaps