#4166: unsafeCoerce#'s kind is not as liberal enough to inspect tag bits.
---------------------------------+------------------------------------------
Reporter: ekmett | Owner:
Type: feature request | Status: new
Priority: normal | Component: Compiler
Version: 6.12.3 | Keywords:
Os: Unknown/Multiple | Testcase:
Architecture: Unknown/Multiple | Failure: None/Unknown
---------------------------------+------------------------------------------
Inspecting dynamic pointer tagging bits cannot be done with unsafeCoerce#
alone. You have to introduce an extra box:
{{{
import Foreign
import Unsafe.Coerce
import Data.Bits
data Box a = Box a
unsafeIsEvaluated :: a -> Bool
unsafeIsEvaluated a = unsafeCoerce (Box a) .&. (sizeOf (undefined :: Int)
- 1) /= 0
}}}
To see this operation in action, see the 'speculation' package.
There is of course the unsafeCoerce# primop, but it can't change kinds:
{{{
unsafeCoerce# :: a -> b
}}}
However, if unsafeCoerce# was able to switch to an unboxed kind:
{{{
unsafeCoerce# :: forall (b :: ??). a -> b -- 1
}}}
this could be done in one fell swoop without the extra box.
{{{
unsafeIsEvaluated a = unsafeCoerce# a `and#` (sIZEOF_INT# -# 1#)
}}}
Of course, this is far from the most liberal type, but
{{{
unsafeCoerce# :: forall (a :: ??). (b :: ??) -> b -- 2
}}}
would allow conversion to and from addres-sized unboxed types, although it
would likely be risky from a GC perspective.
and worse
{{{
unsafeCoerce# :: forall (a :: ??). (b :: ?) -> b -- 3
}}}
might also be possible, but the interpretation with coercing into an
unboxed tuple doesn't make much sense.
But overall the type given by (1) above is the most permissive possible
without potentially disastrous interpretation consequences.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4166>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
