Simon Marlow <[EMAIL PROTECTED]> wrote:
> the other meaning, that you were expecting, is
>
> fail = "exits with a non-zero exit code"
> expected = "exiting with non-zero exit code is the correct behaviour"
>
> "desired" wouldn't be right - all failures are undesired.
Not so. It is perfectly desired for a program to reject incorrect
input, and this is certainly something for which one might wish to have
a regression test. For example, imagine a bug in the compiler's
optimisation transformations, that causes an exception not to happen
when/where it was expected. Or a bug in the runtime system, where
it invokes the wrong handler for an exception.
But I admit these are far rarer than the other kind of tests.
Regards,
Malcolm
_______________________________________________
Cvs-ghc mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/cvs-ghc
