[GHC] #13593: Unexpected behavior from Data.List.groupBy

[GHC]

#13593: Unexpected behavior from Data.List.groupBy
-------------------------------------+-------------------------------------
        Reporter:  dsf               |                Owner:  (none)
            Type:  task              |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Core Libraries    |              Version:  8.0.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Documentation     |  Unknown/Multiple
  bug                                |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by svenpanne):

 I still claim that `groupBy` is by no means special: We have literally
 tons of functions which expect that some argument, be it implicit or
 explicit, obeys some laws, and those laws are not enforced by the type (so
 you can actually write down the type without having a Ph.D. in theoretical
 CS ;-). Take e.g. `Functor`: We expect that the Functor laws are obeyed by
 all instances. If they're broken, all odds are off and probably most of
 the function expecting such an argument do strange things. Non-transitive
 comparison for `sortBy`? Have fun... etc. etc. Nevertheless, we don't
 repeat such law-breaking examples all over the library documentation.

 Do we really know which of the 10 `fooBy` functions do something remotely
 sensible if they don't get an equivalence/total ordering? I doubt that.
 Without looking at the implementation I would have a hard time figuring
 that out.

 Putting the `notBoth1` example below the 'User-supplied equality
 (replacing an Eq context)' header and just repeating the equivalence
 requirement in each function documentation is probably the right thing.
 Similarly for `Ord`.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13593#comment:15>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler