[876] in Kerberos-V5-bugs
Re: Make check in tests/asn.1 always 'succeed'
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Wed Oct 19 12:02:08 1994
Date: Wed, 19 Oct 1994 12:01:58 +0500
From: Theodore Ts'o <tytso@MIT.EDU>
To: epeisach@MIT.EDU
Cc: krb5-bugs@MIT.EDU
In-Reply-To: [868]
From: epeisach@MIT.EDU
Date: Tue, 18 Oct 1994 09:49:25 -0400
make check will never bomb out as the decode test never returns with a
non-zero exit status. It may be me, but I prefer a make check to run
until something is wrong and then stop there.
Yes, that's the way it's supposed to be. It's just that this set of
test programs wasn't adhereing to that convention. Well, they are now!
(BTW: On osf 1.3, using the cc compiler, ccopts=-O will break asn.1.
ccopts-O1 will not.).
Great. More compiler optimizer bugs.... apparently AIX's 1.3.0.19 cc
will also cause asn.1 to break. We should track down exactly what's
buggy with these compilers, to make sure it's not some subtle,
non-portable assumption that we're making in the asn.1 code.
The following patch will make the tests die after noticing an error.
(actually after running all tests). I do not know what the other test
programs do.... I suspect the results are mixed...
Most of the other test programs should be exiting with a non-zero exit
status after they detect a problem. At least, that's the plan.
- Ted
