act totally broken, but only for programs run from cron. the waitpid
loop in t_api.c times out eventually, and the testcases are killed
and reaped, but the net effect of the problem is to cause
each testcase to take 60 seconds, which is the default timeout. As
a result, the tests take forever to run and often don't finish before
the report generator is run, causing the report generator to incorrectly
reports an overall test status based on an incomplete journal ...
so ... I've added some more simplistic journal correctness checking
for now so that the FreeBSD test status is correctly reported - if
a journal doesn't end with the correct end stanza, a test result of
inspect(incomplete journal) is reported.
obviously, at some point in the future, a test case count needs added
so that the report generator can do more complete correctness checking
of the journal.
in test2 of bin/tests/dst/t_dst looked like it either messed up the
stack or there is a compiler bug of some sort because the results
didn't get reported, ie code got skipped. this changed causes the
report generator to flag tests that don't report results.
David Lawrence