Tests failing kdb5_util create apparently broken on the trunk

[Ken Raeburn]

On Oct 4, 2006, at 13:33, Will Fiveash wrote:
> and plain.length is ending up with a huge value (4294967294) so  
> malloc()
> blows up.  Ken tells me that this is very likely a ASN.1 bug he is
> working on.

The ASN.1 problem is in the LDAP-specific code, which I assume was  
not what Sam was testing.

I wasn't able to reproduce Sam's problem in my current source tree  
(which has my LDAP ASN.1 changes and a number of debugging printfs).   
It also didn't show up in the nightly testing reports I looked at,  
though we do seem to have some other "Heisenbugs" there.  I didn't  
try valgrind though.

>> Running valgrind shows several problems in the plugin code.
>
> I also intend to do test using libumem on Solaris to catch memory
> issues.

I think it's probably worthwhile to spend a little time making these  
tools easier to use in conjunction with our test suite.  We have a  
few hooks for Purify, specifically, and of course some tools like  
Electric Fence can easily be enabled for the whole "make check" run  
and ought to have no ill effect, but I don't think I care to see what  
valgrind reports for "make" or "sh" or ....