[commit: ghc] master: Clarify comment about data family arities (c6d4219)

[git at git.haskell.org]

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

On branch  : master
Link       : http://ghc.haskell.org/trac/ghc/changeset/c6d4219ae46cddd63aa2b5762efaf99f815009a4/ghc

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

commit c6d4219ae46cddd63aa2b5762efaf99f815009a4
Author: Richard Eisenberg <rae at cs.brynmawr.edu>
Date:   Mon Jul 31 08:29:48 2017 -0400

    Clarify comment about data family arities
    
    as requested in #14045.
    
    [skip ci] comments only


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

c6d4219ae46cddd63aa2b5762efaf99f815009a4
 compiler/types/FamInstEnv.hs | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/compiler/types/FamInstEnv.hs b/compiler/types/FamInstEnv.hs
index f40dabe..b9aa439 100644
--- a/compiler/types/FamInstEnv.hs
+++ b/compiler/types/FamInstEnv.hs
@@ -162,6 +162,11 @@ Over-saturation is also possible:
       see Note [Eta reduction for data families]. Accordingly, the FamInst
       is never over-saturated.
 
+Why can we allow such flexibility for data families but not for type families?
+Because data families can be decomposed -- that is, they are generative and
+injective. A Type family is neither and so always must be applied to all its
+arguments.
+
 Note [Eta reduction for data families]
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Consider this