| Subject: | Time-HiRes-1.92 tests 34-37 fail unpredictably on Cygwin |
After noticing a test failure for Time::HiRes in bleadperl, I attempted
to recreate the failure from a command line. Unfortunately, the re-run
showed a different test failed. A second re-run had two tests failed.
A third, no test failed.
The failures all occur with tests 34 through 37, although I usually saw
failures from 35 and 37. They also are not limited to bleadperl, as the
results below came from my Perl 5.8.8 installation.
# t0 = 1160762425.6875
# ualarm(100000)
# t1 = 1160762425.78765
# dt = 0.100152969360352
ok 34
# t0 = 1160762425.78841
# ualarm(1100000)
# t1 = 1160762426.89062
# dt = 1.102215051651
ok 35
# t0 = 1160762426.89136
# ualarm(2200000)
# t1 = 1160762429.09375
# dt = 2.20239210128784
ok 36
# t0 = 1160762429.09486
# ualarm(4300000)
# t1 = 1160762433.39062
# dt = 4.29576706886292
not ok 37
# ualarm(4300000) close enough
# 1160762433.5625 1160762433.6875 1160762433.8125 1160762433.92188
1160762434.10938 1160762434.3125 1160762434.48438 1160762434.57812
1160762434.76562 1160762434.95312
ok 38
P.S. Also note the HiRes stat results. Cygwin is a type 4. jhi++