[Jean Delvare] > I don't understand the test for (ofmt != NULL) in pr_finish(). I can't > see how ofmt could not be set, if that was really possible then I > believe things would break much earlier. > > I also don't understand why you explicitly initialized ofmt to NULL, > this shouldn't be needed.
I can not really speak for Jiri, but to me such approaches are good programming practice and what is commonly known as defensive programming, <URL: https://en.wikipedia.org/wiki/Defensive_programming >. I just wanted to air a small defence against the practice, as it in my experience defense against future surprises when many people update a code base. :) -- Happy hacking Petter Reinholdtsen
