Re: Tests pass (was: Re: [Gzz] Why the mirror test fails)

[Tuomas Lukka]

On Wed, Oct 09, 2002 at 10:54:20AM +0200, Benja Fallenstein wrote:
> Benja Fallenstein wrote:
> 
> >The mirror test currently fails because MasterImpl, after getting the 
> >changes from its space, applies them back to the space; this assumes 
> >that checkpointDelta() reverses the changes it returns. 
> 
> 
> I've commented out that line for now, so that the test passes until this 
> issue is decided.

Yes, thanks. This is the right choice: the semantics changed but I didn't
notice when trying to port the mirror code and didn't have time to look
deep. Excellent find.

> Now, *all tests pass*. PLEASE keep them passing! DO NOT commit code that 
> breaks the tests from now on!

*except* if you find a bug. Committing a test that exercises a clear bug
is explicitly allowed and even encouraged. If not sure whether something is a 
bug,
ask someone.

        Tuomas