build failures when hiding non-visible imports

[Bertram Felgenhauer]

Simon Peyton-Jones wrote:
> | Would it be reasonable to change ghc's behavior to treat this 
> | (ie an 'import' statement that hides something that isn't exported) as a
> | warning instead of an error?
> 
> Yes, that would be easy if it's what everyone wants. Any other opinions?

+1 for a warning from me

Bertram