On 18 Feb 2015, at 12:44 am, Daniel Macks <dma...@netspace.org> wrote: > >> as the entire test suite fails with an import error with 1.8 (apparently >> crashes the >> interpreter, thus the failure was not caught be the original Test field). > > I think the usual pattern for TestScript is: > > some command sequence || exit 2 > > rather than an intermediate TESTFAIL variable. If "some command sequence" > doesn't succeed, exit 2 is what fink wants to see to be fatal. If it does > succeed, then a normal success exit-code (e.g., not explicitly 2) is default. > But your package already looks like the test suite requires manual tricks, so > it's possible even the exit-handler needs special tricks too.
The idea of first catching the return value of ‘some command sequence’ was a) to allow to return with 1 for a limited no. of (possibly expected) errors, thus only raising a warning b) to let the remainder of the test script run after an exit 1, in case some cleanup is necessary (not required in this package, but many python test suites leave byte-compiled files in the source tree behind). However in this case python seems to actually segfault rather than exiting with an exception, but with an exit value of only 1, so it was also only triggering a warning. The current setting takes care of failing the tests due to two failures when running bottleneck 0.8 + numpy 1.9 failing after the python crash showing up in the combination bottleneck 1.0 + numpy 1.8 It's supposed to give an exit 1 warning, should the test suite produce a single failure. Cheers, Derek ------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk _______________________________________________ Fink-devel mailing list Fink-devel@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.apple.fink.devel Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-devel