Commit 41d098e3 authored by simonpj's avatar simonpj
Browse files

[project @ 1999-03-17 08:26:30 by simonpj]

Complete the hack that lets us give 'error' the type

	error :: forall a:?. String -> a

The kind '?' (also known as 'Type AnyBox') means that
error can be applied to unboxed types too.  unsafeCoerce is
similar.  But the 'real' type of error is:

	error :: forall bv. forall a:Type bv. String -> a

That is, it's really polymorphic in the boxity of a; that's
different to saying that the type bound to 'a' must have
kind '?'.  

The fix was in TcType.inst_tyvar, where we look for the kind '?'
and generate a boxity variable instead.  Sigh.
parent e1a83341
......@@ -181,7 +181,6 @@ tcInstTyVars tyvars
inst_tyvar tyvar -- Could use the name from the tyvar?
= tcGetUnique `thenNF_Tc` \ uniq ->
kind = tyVarKind tyvar
name = setNameUnique (tyVarName tyvar) uniq
-- Note that we don't change the print-name
-- This won't confuse the type checker but there's a chance
......@@ -189,8 +188,26 @@ inst_tyvar tyvar -- Could use the name from the tyvar?
-- in an error message. -dppr-debug will show up the difference
-- Better watch out for this. If worst comes to worst, just
-- use mkSysLocalName.
kind = tyVarKind tyvar
tcNewMutTyVar name kind
-- Hack alert! Certain system functions (like error) are quantified
-- over type variables with an 'open' kind (a :: ?). When we instantiate
-- these tyvars we want to make a type variable whose kind is (Type bv)
-- where bv is a boxity variable. This makes sure it's a type, but
-- is open about its boxity. We *don't* want to give the thing the
-- kind '?' (= Type AnyBox).
-- This is all a hack to avoid giving error it's "proper" type:
-- error :: forall bv. forall a::Type bv. String -> a
(if kind == openTypeKind then
returnNF_Tc kind) `thenNF_Tc` \ kind' ->
tcNewMutTyVar name kind'
@tcInstTcType@ instantiates the outer-level for-alls of a TcType with
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment