1.6.6 popen4 regression: returns incorrect exit status

Thomas E Enebo added a comment - 09/Feb/12 11:16 AM Trying the same thing on my machine: Darwin mrpopo 11.3.0 Darwin Kernel Version 11.3.0: Thu Jan 12 18:47:41 PST 2012; root:xnu-1699.24.23~1/RELEASE_X86_64 x86_64 I don't see your issue: system ~/work/jruby jruby-1_6 963% jirb irb(main):001:0> IO.popen4("sh -c \"echo\"; echo $?") {}; p $? IO.popen4("sh -c \"echo\"; echo $?") {}; p $? #<Process::Status: pid=34886,exited(0)> => nil I am unclear how pre-Lion could do this. Which version of Java are you using?

Xavier Shay added a comment - 09/Feb/12 8:28 PM Uhm, I can't reproduce now either :S NFI what is going on, closing for now.

Xavier Shay added a comment - 28/Feb/12 1:57 PM Argh this is non-deterministic :S jruby-1.6.6 :001 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94720,exited(0)> jruby-1.6.6 :002 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94725,exited(0)> jruby-1.6.6 :003 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94730,exited(143)> jruby-1.6.6 :004 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94735,exited(143)> jruby-1.6.6 :005 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94742,exited(143)> jruby-1.6.6 :006 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94747,exited(0)> jruby-1.6.6 :007 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94752,exited(143)> jruby-1.6.6 :008 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94757,exited(0)> jruby-1.6.6 :009 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94762,exited(143)> jruby-1.6.6 :010 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94767,exited(143)> jruby-1.6.6 :011 > IO.popen4("sh -c \"echo\"; echo $?") {}; $? => #<Process::Status: pid=94774,exited(0)> I also see non-deterministic results on 1.6.5 and 1.6.7 also (though with an exit code of 0 or 1, not 143), so my initial labeling of this as a regression was incorrect. I was not able to reproduce a changing exit code just using a shell. Wish I could provide more insight into this problem. Darwin koopa.local 11.3.0 Darwin Kernel Version 11.3.0: Thu Jan 12 18:47:41 PST 2012; root:xnu-1699.24.23~1/RELEASE_X86_64 x86_64 java version "1.6.0_29" Java(TM) SE Runtime Environment (build 1.6.0_29-b11-402-11D50) Java HotSpot(TM) 64-Bit Server VM (build 20.4-b02-402, mixed mode)

Xavier Shay added a comment - 28/Feb/12 1:57 PM See most recent comment.

Xavier Shay added a comment - 28/Feb/12 6:52 PM Can also replicate occasionally on Linux somehost 2.6.18-274.3.1.el5 #1 SMP Tue Sep 6 20:13:52 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux

Charles Oliver Nutter added a comment - 17/May/12 12:49 PM FWIW, this seems consistently correct on OS X. I let a script run for many hundreds of process launches and it was 0 result every time. Could be specific to Linux, since Tom could not reproduce either.