On Mon, Sep 22, 2008 at 06:04:34PM +0200, pepe wrote:
>
> this patch makes a subtle change to the semantics of tryUser in
> Panic.hs, used in the TcRnMonad.
>
> I wonder if this is intended or just an oversight*.
An oversight.
I've been thinking about exceptions in GHC, though, and I think that
rather than trying to keep GHC stage 1 working with both old and new
exceptions for the next few years we should just put an extensible
exceptions library in the GHC tree (or perhaps even in a Cabal package,
so that things like ghc-pkg can use it).
For tryUser, I think that what we're really interested in is catching
the
ioError (userError "IOEnv failure")
from IOEnv.failM, so what we ought to do is to make a
data IOEnvFailure = IOEnvFailure
datatype and throw that instead.
Anyone got any opinions?
Thanks
Ian
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
