[PATCH v2][BZ 14561] Testcase
Joseph S. Myers
joseph@codesourcery.com
Tue Jun 4 14:45:00 GMT 2013
On Tue, 4 Jun 2013, Ondrej Bilka wrote:
> On Tue, Jun 04, 2013 at 09:08:17AM -0400, Carlos O'Donell wrote:
> > This is a good first pass, but we need to make this easier to debug
> > when it fails and when we switch to PASS/FAIL test we should be able
> > to explain which ordering failed.
> >
> I plan to add a breakpoint() function to test which counts times it was
> called before failure(or signal). Then we can output something like:
>
> FAIL, Type following for debugging:
I think the goal with PASS / FAIL information is for each test to have a
well-defined, fixed (i.e. not depending on the results of the tests) set
of test assertion names, such that for each assertion it outputs either
PASS: <name>
or
FAIL: <name>
(or other variants such as UNRESOLVED, UNSUPPORTED etc.), resulting in
both such information for overall tests (the units run from the makefiles)
and for individual assertions within those tests. Ideally the <name>
would be prefixed in some way with the name of the overall test, so
PASS: stdlib/bug-srand-srandom-dependency some name
(for example) in the log for the individual test.
Additionaly information might be printed, but should be separate from the
basic PASS / FAIL for individual assertions, so that results from multiple
test runs can be properly compared.
--
Joseph S. Myers
joseph@codesourcery.com
More information about the Libc-alpha
mailing list