[Piglit] Use subprocess32 for test timeouts, or fake it

Daniel Vetter daniel at ffwll.ch
Mon Oct 12 00:17:45 PDT 2015

On Fri, Oct 09, 2015 at 12:53:36PM -0700, baker.dylan.c at gmail.com wrote:
> This changes the way that test timeouts work to be more robust than the
> current implementation. This uses the external module subprocess32 which
> provides the same interface as python 3.2+, a timeout parameter for
> Popen.communicate. This is much simpler than our implementation, and has
> the advantage of working correctly with threads, unlike our
> implementation.
> Since the interface is like python 3.x, this provides us a path forward
> for building piglit has a hybrid python 2.x/3.x application, as we look
> toward a python 3 only future.
> Thomas, Daniel:
> Will this work for you? I would really like to move forward with
> building a hybrid python 2/3 codebase for piglit, and this would be
> really nice, since it would make timeouts work very nicely in python 3.

fwiw (and that's not much since I completely lack python fu) here's my
ACK. I guess Thomas will have the more useful opinion ;-)
Daniel Vetter
Software Engineer, Intel Corporation

More information about the Piglit mailing list