| Subject: | exec.t - 'execute fail mode 3' failing |
In exec.t, the 'execute fail mode 3' test is failing. I have debugged into it
and I can't see how it could actually be passing for anyone.
I wonder if the expected value of the test shouldn't be *inverted*, i.e. we
should expect the execute to fail rather than to succeed.
Here is what I see happening:
0001 $sth->{syb_do_proc_status} = 1;
0002 $dbh->{syb_flush_finish} = 0;
0003 # ...
0004 $rc = $sth->execute("raise", 1, 3.2234, "jan 3 2001", 5.4);
0005 # ...
0006 get_all_results($sth);
0007 $rc = $sth->execute(undef, 0, 3.2234, "jan 3 2001", 5.4);
In line 4 we call the dbitest procedure. It returns the value of @two, which in
this case is 1. This means an error state for the procedure. Of course,
execute() itself succeeds, as described in the DBD-Sybase docs for
syb_do_proc_status, but we then proceed to fetch all the results (line 6), and
the last fetch returns the non-zero return status of dbitest().
This non-zero return status is fetched in st_next_result. We then check that
restype == CS_PARAM_RESULT (== 4042)
and set the failFlag. Then, based on the failFlag, we call syb_st_finish(). Yes,
this actually finishes the $sth. As a result, the subsequent call to execute
(line 7) refuses to ct_send.
My question is: should we
A) rebase the test, expecting the execute on line 7 to fail?
B) try to fix this in DBD-Sybase so that we don't finish the statement handle?
Any comments welcome.
Thanks,
Sam