So our choice is either check every trace call for null args or put a check in trace to protect against null args going to the printf's. I prefer the latter; trace is for debugging, and debugging means sometimes you get null values.
I'll work on looping over va_args in trace. Aaron On Sat, 2006-05-20 at 20:48 +0200, Kneschke.Lars wrote: > Aaron Stone <[EMAIL PROTECTED]> schrieb: > >I might also add that free(NULL) is required to be legal. Solaris > >apparently provides a compatibility library for null values: > >http://technopark02.blogspot.com/2006/04/solaris-null-pointer-bugs-usrlib00so1.html > > > >I'm very amused at Sun's lib/[EMAIL PROTECTED] hackaround. > > Ok, at least i can confirm that this solved my problems regarding sieve. > > I first added the line you suggested earlier, but this solved the problems > only when using dbmail-smtp. > > When feeding messages trough dbmail-lmtp it was still crashing. > > Now added this lines to the init script: > LD_PRELOAD_32=/usr/lib/[EMAIL PROTECTED] > export LD_PRELOAD_32 > > And now everything seem to be working. > > At least it indicates what the source of the problems is. >
