(Not quite sure what components this should ultimately be filed under, Fedora 7 at least has the perl-IPC-Shareable component) Description of problem: I'm trying to rebuild the FC6/F7 perl-IPC-Shareable Perl module for CentOS 5 x86_64 (like the EPEL RPMs), and it stalls as part of the "make test" phase: ... + make test PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t t/00base......ok t/05sv........ok t/10av........ok t/15hv........ok t/20ref.......ok t/25ipc.......ok t/30lock......ok t/35clean.....ok t/36ipcav.....ok t/38ipchv.....ok 1/5 Version-Release number of selected component (if applicable): perl-IPC-Shareable-0.60-3.fc6.src.rpm How reproducible: Always. Steps to Reproduce: 1. rpmbuild --rebuild perl-IPC-Shareable-0.60-3.fc6.src.rpm Actual results: Stalls at 1/5 in the 38ipchv tests Expected results: Should complete cleanly with no errors Additional info: Build host is fully updated CentOS 5 x86_64 installation. Build completes correctly on a Fedora 7 installation.
Actually, I lie, it will also hang on a Fedora 7 installation. Running that bit of the test suite in a loop will get it to lock in the same place eventually: while true ; do PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/38ipchv.t ; done t/38ipchv....ok All tests successful. Files=1, Tests=5, 0 wallclock secs ( 0.05 cusr + 0.02 csys = 0.07 CPU) ---8<------8<--- t/38ipchv....ok All tests successful. Files=1, Tests=5, 0 wallclock secs ( 0.04 cusr + 0.02 csys = 0.06 CPU) t/38ipchv....ok 1/5 Suggests some sort of subtle race condition somewhere, given this test is also forking child processes.
Seems someone reported this on CPAN already: http://rt.cpan.org/Ticket/Display.html?id=19169
I've sent a ping to Ben S. to see if he is planning to do a new release anytime soon.
I'm pretty sure I fixed this in rawhide as part of the perl 5.10.0 work...
I tried build on RHEL-5 successfully from F-7 branch.