[GHC] #4259: Relax restrictions on type family instance overlap
GHC
cvs-ghc at haskell.org
Fri Jan 11 03:57:14 CET 2013
#4259: Relax restrictions on type family instance overlap
----------------------------------------+-----------------------------------
Reporter: lilac | Owner:
Type: feature request | Status: new
Priority: normal | Milestone: 7.8.1
Component: Compiler (Type checker) | Version: 6.12.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
----------------------------------------+-----------------------------------
Comment(by goldfire):
Yes. The desire for injective type families (#6018) and this report seem
something like treating symptoms instead of the disease. (Don't take that
simile too far -- I'm not suggesting GHC's current treatment is somehow
diseased, just not as powerful as some want.) What we all seem to want
(even if we don't know it) is the ability to define an arbitrary rewrite
system over types and for GHC to somehow, magically check that our rewrite
system is consistent and then to apply it. I'm pretty sure that goal is
impossible. But, we can certainly make inroads in that direction. Relaxing
restrictions on instance overlap as described here is one way, and
injective type families is another.
My current thought on all of this is that we should consider some unified
approach with a unified goal, instead of attacking this problem piecemeal.
This unified approach may end up devolving into solving these two problems
separately, but there should be an articulated reason why these two
problems (this ticket and injectivity) are the two problems to solve.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4259#comment:27>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
More information about the ghc-tickets
mailing list