Re: RFC: Known Failures [Was: RFD: Testsuite cases for inferior function calls]

["Peter.Schauer" <Peter dot Schauer at regent dot e-technik dot tu-muenchen dot de>]

KFAILs would be fine with me, as long as they show up in the testsuite run
output.

I understand the desire of release management to produce a `flawless' release,
but from a user perspective, I'd really like to know about the known failures,
without grepping gdb.sum for KFAILs.

> "Peter.Schauer" wrote:
> > 
> > As the usage of XFAILs throughout the testsuite increased during the last
> > two years, you might as well XFAIL the test on the affected targets.
> > 
> 
> We've been discussing this for ages.  I would like to introduce the KFAIL
> (or known failures) category so we keep those separate.  XFAIL should be used 
> when we expect something to fail on a platform/target because there is a limitation
> on the environment or a bug in the operating system (note the difference: not a bug 
> in the software being tested, but in something we do not have control over).
> If a test does not apply to a target at all because it does not have the capabilities
> it should be marked as "unsupported" (we currently just skip the tests).
> 
> We would like to create a database to keep track of tests, so we would need to 
> improve our output a bit.  For instance, I plan to change things so that tests have
> the same identifier when they pass and for each of the reasons they can fail.
> The idea is to split the text into id and reason.
> 
> Other suggestions?  Comments?

-- 
Peter Schauer			pes@regent.e-technik.tu-muenchen.de