|
From: Bart V. A. <bar...@gm...> - 2009-01-15 07:23:31
|
On Wed, Jan 14, 2009 at 10:17 PM, Nicholas Nethercote <n.n...@gm...> wrote: > On Wed, Jan 14, 2009 at 7:08 PM, Bart Van Assche > <bar...@gm...> wrote: >> IMHO a more elegant solution would be to have automake variables available >> that allow to conditionally compile the regression tests that depend on >> barriers and spinlocks. > > Could you give an example of what you mean? I'm not familiar with Darwin, so I don't know whether it's the compile time, link time or the run time support for e.g. POSIX barriers that is not present in Darwin. In the first two cases support for e.g. POSIX barriers can be detected via an extra test in configure.in, just like the test for Qt or Boost. I'm also curious to know which issues were triggered by the trylock and sigalrm tests on Darwin ? Bart. |