On Fri, 23 May 2008, Michael Wieher wrote:
Ok, so I've been struggling with a strange issue and a broken destructor.
btw, anyone know a single-word for the idea "situations so complex
they cannot be reproduced"
http://en.wikipedia.org/wiki/Heisenbug
Here is situation:
---------------------------
class {
datax ** y;
data * b;
int numbofY;
}
destructor:
free all in y (via for-loop & numbofY)
causes seg-fault unless i syslog something in there
however
destructor:
free all in y(same)
freee b
works OK
any idea why syslog makes this failure to free(b) acceptable during
execution/exit of program?
Valgrind might tell you. With memory bugs, the link between cause and
effect is often difficult or impossible to determine without the right
tools.
