Cabalize ext-core tools

[ghc-hetmet.git] / utils / ext-core / Parser.y

diff --git a/utils/ext-core/Parser.y b/utils/ext-core/Parser.y
index 4ff3d1d..add3ef0 100644 (file)
--- a/utils/ext-core/Parser.y
+++ b/utils/ext-core/Parser.y
@@ -175,7 +175,9 @@ exp :: { Exp }
                { Let $2 $4 }
        | '%case' '(' ty ')' aexp '%of' vbind '{' alts1 '}'
                { Case $5 $7 $3 $9 }
-       | '%cast' exp aty 
+-- Note: ty, not aty! You can cast something to a forall type
+-- Though now we have shift/reduce conflicts like woah
+       | '%cast' exp ty
                { Cast $2 $3 }
        | '%note' STRING exp 
                { Note $2 $3 }
@@ -232,6 +234,12 @@ mnames :: { [Id] }
 -- it sucks to have to repeat the Maybe-checking twice,
 -- but otherwise we get reduce/reduce conflicts
 
+-- TODO: this is the ambiguity here. mname '.' name --
+-- but by maximal-munch, in GHC.Base.Bool the entire 
+-- thing gets counted as the module name. What to do,
+-- besides z-encoding the dots in the hierarchy again?
+-- (Or using syntax other than a dot to separate the
+-- module name from the variable name...)
 qname  :: { (Mname,Id) }
         : name { (Nothing, $1) }
        | mname '.' name