Re: [patch/rfc] KFAIL gdb.c++/annota2.exp watch triggered on a.x

[Daniel Jacobowitz <drow@mvista.com>]

On Fri, Jan 03, 2003 at 01:48:25PM -0800, David Carlton wrote:
> On Fri, 3 Jan 2003 16:39:20 -0500, Daniel Jacobowitz <drow@mvista.com> said:
> 
> > May I recommend at the least "i?86"?
> 
> That makes sense.
> 
> > Also, I really don't see the point of the kpass's; before doing
> > this, you need to establish if those patterns are acceptable
> > results; if so, they are passes, period.
> 
> Sorry, I should have explained my reasoning there.  My theory behind
> that is that they're a reminder to people who fix bugs that they
> should update the test suite.  If somebody fixes this bug a year from
> now, doesn't know that there's a test case for the bug, and doesn't
> pay attention to gdb.sum (just to the naked 'make check'), then that
> person might easily forget to update the test suite.  (Especially
> since the test case in question is in gdb.c++/annota2.exp, whereas the
> bug doesn't involve either C++ or annotations!)
> 
> So it seems to me that, if the failure isn't reliable, then we should
> leave the success case as a PASS, but if the failure is reliable, then
> KPASS is slightly better.

How do you envision them updating the testsuite?  Certainly not by
removing the KFAIL's pattern; that defeats the point of having a
regression test.  That's why I like Michael's approach of having a pass
pattern and a kfail pattern.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer