Robert Ramey writes:

I was considering the situation whereby a particular compiler/configuration fails all or most of the test because of some issue in the test computer's build configurtation.

Which is what we had.

Since the "release package" just points to the test results, they can be updated at anytime after the release.

Of course.

So I don't see that this test problem should hold up the release.

Oh, definitely not.

That's all I meant to say.

OK, thanks for the clarification. -- Aleksey Gurtovoy MetaCommunications Engineering