But do we really need to accumulate all available tests under 1 super-target? 'build test' is supposed for pre-integration testing, which must balance between better coverage and lesser time-to-run. Current regression suite hardly fit to this trade-off, so exercising it just under CC makes more sense IMO. BTW, 'build test' either does not include ehwa.test (automated EHWA) which is far more effective/important.
2007/4/12, Sergey Dmitriev <[EMAIL PROTECTED]>:
Ok I see now. Thanks. On 4/12/07, Vladimir Ivanov <[EMAIL PROTECTED]> wrote: > On 4/12/07, Sergey Dmitriev <[EMAIL PROTECTED]> wrote: > > Vladimir, thanks for your answer. Why the "reg.test" target isn't > > linked to "test"? Any reason? > > Regression tests infrastructure was added not long ago... And we > needed some time to verify it. Seems, now it may be added to the > common 'test' target. > > Thanks, Vladimir > > > > > > > On 4/12/07, Vladimir Ivanov <[EMAIL PROTECTED]> wrote: > > > Results of tests run uploaded to the harmonytest.org manually (or by > > > user's scripts). Actually, I upload results for classlib tests only. > > > And yes, 'test' target does not depend on 'reg.test' target for DRLVM. > > > > > > thanks, Vladimir > > > > > > > > > On 4/12/07, Sergey Dmitriev <[EMAIL PROTECTED]> wrote: > > > > Hello > > > > > > > > recently I've checked the harmonytest.org and didn't find the DRLVM > > > > regression tests. > > > > I.e. test results contain nothing from > > > > org.apache.harmony.drlvm.tests.regression.*. > > > > > > > > I'd suppose "build.sh test" doesn't invoke the "build.sh reg.test". > > > > > > > > Was it designed so knowingly? > > > > > > > > > > > > Thanks > > > > Sergey > > > > > > > > > >
