| Subject: | 03_run-forked.t randomly failing in blead |
| Date: | Mon, 18 Nov 2013 13:01:35 +0000 |
| To: | bug-IPC-Cmd [...] rt.cpan.org |
| From: | Dave Mitchell <davem [...] iabyn.com> |
Since IPC-Cmd 0.86 was pulled into blead, the run_forked() loop tests
have been randomly failing on various bleadperl smoke platforms, including
linux. I'm unable to reproduce the failure myself, but from what I've seen
of smoke reports:
1) its failing after a random number of times round the
for (my $i = 0; $i < 100; $i++) {
loop, and on a particular smoking platform/configuration, it randomly
sometimes succeeds and sometimes fails.
From a couple of the smoke logs, e.g.
http://m-l.org/~perl/smoke/perl/linux/blead_clang/loge2aa288a0f563ddb8e6f5bb400a8e1f82e623b4e.log.gz
http://m-l.org/~perl/smoke/perl/linux/blead_clang/log659c4b9601a634cd50919970684ee46143183238.log.gz
a consistent error is reported, namely:
# Failed test 'reading /tmp/03_run_forked.t.9942: run_forked output length [282624], backticks output length [286720]'
# at t/03_run-forked.t line 88.
Died at t/03_run-forked.t line 90.
# Looks like you failed 1 test of 52.
# Looks like your test exited with 9 just after 52.
The two sizes reported are consistent across 9 failures in those two smoke
logs; namely that $f_backticks has the right length for the file, but
run_forked()->{'stdout'} is returning 4096 bytes too few.
--
In my day, we used to edit the inodes by hand. With magnets.