Changes since v1: - Handle forking processes properly (there is a test in the last patch for this which probably deserves review, I've learned a lot about how POSIX processes work for this series and may or may not have gotten it right) - Split the timeout code into a Mixin. This simplifies handling support for timeouts on systems that are either windows or don't have subprocess 32. - Better unit-testing - Actually terminate (and kill) processes that run too long.
This leaves much of the groundwork for timeouts on windows, but doesn't implement them. I don't have a windows dev system, nor does my team really care about windows. However, windows should continue to work timeout free until such a time as someone cares to implement support. This is available at my github: https://github.com/dcbaker/piglit submit/subprocess32 Thomas, Hopefully I've addressed all of the issues you raised with the first iteration of this series, please let me know if there are any cases I'm not handling correctly. _______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
