Re: [boost] type_traits: fixed is_POD, testing methodology questions

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-02-22 07:49:42

----- Original Message -----
From: "John Maddock" <John_Maddock_at_[hidden]>

> >I also made the following change. It causes tests to fail if they don't
> >match their expected number of failures exactly. I needed to do that in
> >order to see that the change fixed anything:
>
> I know this is fragile stuff, I don't know what the solution is, we need
> some kind of "partial success" result.  Really the trouble is that you
have
> to inspect the results by hand to see if there is anything that fails that
> can be fixed, as opposed to something that fails but can't be fixed due to
> a compiler bug.

That seems orthogonal to what I was complaining about. If something gets
fixed, don't you want to be forced to update the number of expected
failures?

The number of expected MSVC6 failures in object_type_traits_test has now
dropped to zero (from something like 31)! I would never have known that had
I not made the change.