[commit: ghc] wip/new-flatten-skolems-Aug14: Comments (2c516a3)

[git at git.haskell.org]

Repository : ssh://git@git.haskell.org/ghc

On branch  : wip/new-flatten-skolems-Aug14
Link       : http://ghc.haskell.org/trac/ghc/changeset/2c516a30894bd6be8b9ec882f2e7d2f25feeac40/ghc

>---------------------------------------------------------------

commit 2c516a30894bd6be8b9ec882f2e7d2f25feeac40
Author: Simon Peyton Jones <simonpj at microsoft.com>
Date:   Tue Oct 7 13:37:40 2014 +0100

    Comments


>---------------------------------------------------------------

2c516a30894bd6be8b9ec882f2e7d2f25feeac40
 compiler/typecheck/TcInteract.lhs | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/compiler/typecheck/TcInteract.lhs b/compiler/typecheck/TcInteract.lhs
index 703facf..4948692 100644
--- a/compiler/typecheck/TcInteract.lhs
+++ b/compiler/typecheck/TcInteract.lhs
@@ -902,7 +902,10 @@ equality (b ~ phi) in two cases
     and can subsequently unify.
 
 (2) If the new tyvar appears in the RHS of the inert
-    AND the inert cannot rewrite the work item
+    AND the work item is strong enough to rewrite the inert
+
+    AND not (the inert can rewrite the work item)   <---------------------------------
+
           Work item:  [G] a ~ b
           Inert:      [W] b ~ [a]
     Now at this point the work item cannot be further rewritten by the
@@ -919,6 +922,8 @@ equality (b ~ phi) in two cases
     necessarily idemopotent.  See Note [Non-idempotent inert substitution]
     in TcCanonical.
 
+          Work item:  [G] a ~ Int
+          Inert:      [G] b ~ [a]
 See also Note [Detailed InertCans Invariants]
 
 Note [Avoid double unifications]