[1661] in Kerberos_V5_Development
Re: testing
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Wed Aug 28 21:48:22 1996
To: "Theodore Y. Ts'o" <tytso@MIT.EDU>
Cc: "Barry Jaspan" <bjaspan@MIT.EDU>, krbdev@MIT.EDU
From: Ken Raeburn <raeburn@cygnus.com>
Date: 28 Aug 1996 21:47:55 -0400
In-Reply-To: "Theodore Y. Ts'o"'s message of Wed, 28 Aug 1996 19:22:23 -0400
"Theodore Y. Ts'o" <tytso@MIT.EDU> writes:
> Agreed; I'm not sure I understand what is the advantage of requiring
> dejagnu for those tests which are currently being done using make
> check....
Right now, if a db test fails, the "make check" run stops. No
applications get tested. No other unit tests that would be completely
independent of db get tested. This sort of thing has been wasting my
time recently.
If it were all managed by a single dejagnu process, you'd get
something like:
Running util/db/db.exp...
UNTESTED: test3 (no dictionary)
UNTESTED: test5 (no dictionary)
...
Running lib/krb5/os/aname.exp
FAIL: aname check
Running lib/kadm5/unit-test/blah/blah...
...
Running krb-standalone/gssftp.exp...
=== krb Summary ===
# of expected passes 2159
# of unexpected failures 1
# of untested testcases 2
and all the details would be in the dejagnu log file.
So, now I know that the other stuff works. If I get lots of errors,
then I may still need to do some analysis to see if they're cascade
errors or independent problems.
Right now, if there are certain failures I don't care about (like the
ones above), I have to do testing piecemeal, or change the sources to
remove those tests. Maybe with "make check -i" I could get it to run
through all the tests, but I'd have to scan the log file to check that
everything else passed.
