On Mon, Jan 03, 2005 at 11:09:09AM +0100, Falk Hueffner wrote: > Bastian Blank <[EMAIL PROTECTED]> schrieb am 03.01.05 10:57:34: > > > I just read a buildlog for gcc-3.4 and saw large amount of test failures > > but the build themself is marked as successfull. I don't think this is > > the proper use of a testsuite and have to asume that nothing in the > > package may work. > > So what do you propose to do? Fail the build if there are test failures? That
Well, there's a reason that test suites exist, you know. If your tests are failing spuriously, then it's time to fix the tests, not ignore them. > would pretty much ensure that the package never, ever builds. And the Well, if it's always broken, we don't really want it, do we? > bootstrapping already makes sure that the compiler isn't totally broken. "Not totally broken" might be a suitable marketing strategy for Microsoft, but I'm not convinced it's a level we want to be particularly aiming for. - Matt
signature.asc
Description: Digital signature
