These patches remove the --valgrind-parallel=N option that was broken from the outset (shame on me). Peff's judgement at the time that its usefulness would approximately be "meh" turns out to be correct. What's not in the commit message, but drives part of my reasoning in doing a revert instead of a fix: I did fix it up locally only to notice that it was too slow in this case for what I actually wanted to use it for. The only valgrind-test workflow that I find bearable is to run all the tests in the background under prove (takes hours), and then use the prove output (which says exactly which subtests fail) in --valgrind-only=<subtest>. So the latter is -- again Peff was right -- the really useful thing. The only consolation is that I apparently didn't break any other use of the test suite -- otherwise it would presumably have been fixed very quickly. Thomas Rast (2): Revert "test-lib: support running tests under valgrind in parallel" Revert "test-lib: allow prefixing a custom string before "ok N" etc." t/test-lib.sh | 133 +++++++++++++++------------------------------------------- 1 file changed, 34 insertions(+), 99 deletions(-) -- 1.8.4.1.810.g312044e -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html