On Sat, 11 Apr 2020 at 12:54:57 +0100, Jessica Clarke wrote: > Indeed, I just forced parallel=1 on the buildd (kamp) for dbus-python > and the builds succeeded. If it helps anyone debugging this, kamp has 2 > i7-4790K cores allocated to it, which is a desktop processor not a > server processor so has a higher clock speed (and despite being a 2014 > model performs very well; this buildd alone can easily keep up with both > kfreebsd-amd64 and kfreebsd-i386). The buildd is also configured with > parallel=2, and has a habit of exposing race conditions in test > scripts...
If there's an incompatibility for two of the tests running in parallel, I'm surprised not to have seen it on any of the Linux machines where I've been building dbus-python in parallel mode for several years. Presumably this is kFreeBSD-specific, but I can't think of a reason why it would be... smcv