On Thu, Jul 07, 2011 at 07:36:15PM +0100, Matthias Scheler wrote: > Please undo that change. The point of two different tests was in > particular that one binary is linked with "libpthread" and the other one > isn't.
Additionally: we have been discussing this in PR toolchain/44848. We have nearly five hundred test files. Already because of sustainability, I think it is not plausible to replicate "test.c" and "test_pthread.c" or "test_random_compiler_flag.c". In my opinion a hypothesis about errors due linking should go somewhere else, whereas the libc-tests should be strictly about functional verification. - Jukka.
