[GHC] #10675: GHC does not check the functional dependency consistency condition correctly

[GHC]

#10675: GHC does not check the functional dependency consistency condition
correctly
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  7.10.1
      Resolution:                    |             Keywords:  FunDeps
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by AntC):

 Some afterthoughts:

 * What if the instances from the O.P. were declared in separate modules?
 Then GHC couldn't apply the mutual improvement. Would it reject the
 instances as inconsistent?

 * I find the statement of '''Definition 6''' in the CHR paper a bit
 imprecise. Mark Jones 2000 paper is clearer [Section 6.1]. It says "if
 `tX` and `sX` have a most general unifier `U`, then `UtY = UsY`." where
 `tX, sX` are the determinant types from the instance decl, `tY, sY` are
 the dependents.

 * I wonder if what GHC is doing to apply that test is rather than `UtY =
 UsY`, it's checking `UtY ~ UsY` -- that is, checking unifiability rather
 than equality(?) That would explain the behaviour with the `TTypeEq`
 examples.

 * Re the "dysfunctional" Functional Dependencies ;-), we could do with
 some better way of writing instances to give maximum help for type
 improvement. Currently you keep running up against instance-FunDep
 conflicts. And I don't think Injective type functions help much. For
 example type-level `Plus` over `Nat`s, with three-way FunDeps. [Yes I know
 Oleg worked it out years ago, but it's a ''tour de force''.]

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10675#comment:11>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler