> > I suspect that the offside rule is acting up.
>
> Another offside rule difference I've found is that Hugs will accept
> this code. (The important thing is that the case alternatives are at
> the same indentation as the case expression.)
>
> main = do
> x <- return Nothing
> case x of
> Nothing -> print x
> Just y -> print (y :: Char)
>
> Whereas ghc rejects it:
>
> Foo.hs:4: parse error on input `->'
Again, GHC is correct on this example. However, I suspect this is one
subtle aspect of layout parsing that will change in Haskell 2.
A slightly more common example:
main = do
x <- something
seq x $ do
y <- something else
this is actually illegal, but turned out to be so prevalent in existing code
that I put a special hack into GHC to handle it. While we cope with this
kind of non-indentation of nested contexts for 'do', we don't handle it for
'case', which is why your example doesn't work with GHC.
Cheers,
Simon
