"Wade Hampton" <wade(dot)hampton(at)nsc1(dot)net> writes:
> I have a multithreaded application using Postgresql and am testing for
> memory leaks. I compiled and linked with dmallocth. My PGconnect() is now
> causing a dmalloc abort. The error appears to be a calloc of 0 elements of
> 40 bytes in libkrb5, called from libpq.
(1) Why is that a bug? calloc of 0 elements is explicitly allowed by
the Single Unix Spec:
http://www.opengroup.org/onlinepubs/007908799/xsh/calloc.html
(2) If you do think it's a bug, why are you complaining to us rather
than to the Kerberos people?
regards, tom lane