#4894: Missing improvement for fun. deps.
----------------------------------------+-----------------------------------
Reporter: diatchki | Owner:
Type: bug | Status: new
Priority: normal | Milestone:
Component: Compiler (Type checker) | Version: 7.1
Keywords: | Testcase:
Blockedby: | Difficulty:
Os: Unknown/Multiple | Blocking:
Architecture: Unknown/Multiple | Failure: None/Unknown
----------------------------------------+-----------------------------------
Changes (by simonpj):
* cc: dimitris@… (added)
Comment:
No, GHC has no way to do this. Not because of shortcomings in the
inference engine, but because the FC language can't express it. The
question is this: where is the *evidence* that `(b~c)`. Alas there isn't
any.
This is a place where the "encoding" of fundeps using type functions
actually works better:
{{{
class (b ~ B a) => F a b where
type B a
instance F Int Int where
type B Int = Int
}}}
Now if you have given constraints (ie evidence for) `F a b, F a c)`, then
you also have evidence for `b ~ B a, c ~ B a`, and hence by transitivity
for `b~c`.
Fundeps in GHC are handled just as in Hugs, as "improvement rules" only,
meaning that they narrow the search space by suggesting unifications. But
they don't provide evidence.
Having equalities as superclasses isn't working properly yet, but it's my
current priority.
Does this make sense?
Simon
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4894#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
