On Thu, Mar 31, 2016 at 8:23 AM, Noel Grandin <noelgran...@gmail.com> wrote: > Of course, it may be simpler and safer to simply surround the relevant tests > with a brute-force-5-min-timeout-and-kill.
yeah.. except the timeout value should probably be longer or smarter, or an optional parameter in gbuild invocation for outlier... we do not want it triggering abusively on slow machines... for ci purpose even 15 minutes would be tolerable... (and really a test that 'normally' run for longer than that should be split anyway) beside this is not necessarily a java only issue... on occasion some other tests not involving java can dead-lock or cpu-loop... Norbert _______________________________________________ LibreOffice mailing list LibreOffice@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/libreoffice
