#5978: Type error in one function causes wrong type error report in another
function in the presence of functionally dependent types
----------------------------------------+-----------------------------------
Reporter: Lemming | Owner:
Type: bug | Status: new
Priority: normal | Milestone:
Component: Compiler (Type checker) | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: GHC rejects valid
program
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
----------------------------------------+-----------------------------------
Changes (by simonpj):
* cc: dimitris@… (added)
Comment:
OK I see what is happening. From the two definitions we get these
"wanted" constraints:
{{{
monoFoo: mf1: C Float alpha
monoBar: mb1: C Float beta,
mb2: C Double beta
}}}
where `alpha` and `beta` are otherwise-unconstrained unification
variables.
Now if we first do fundeps wrt the `instance` decls, we get `alpha = Char`
from `mf1`, `beta = Char` from `mb1` and `beta = Bool` from `mb2`, the
latter two yield an error.
But if we first do fundeps ''between'' the wanted constraints, we get
`alpha = beta`, so that leaves us with `(mf1: C Float alpha, mb2: C Double
alpha)`. Now we can solve `mb2` via the fundep with the `instance` to
give `alpha=Bool`, and that leaves an error for the innocent `mf1`!
Even this isn't quite what happens. Functional dependencies are a pain.
Simon
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5978#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
