On Sat, Mar 08, 2008 at 03:39:40AM +0100, Vincent Lefevre wrote: > You missed the point: gcc emits the warning even in the cases where such a > test is necessary.
Agreed. However, in this case the test is *not necessary*. It serves no purpose. If someone later coming along changes the code so that mbox isn't an array, they should add appropriate tests for NULLity. And frankly, I think the changes of someone changing the code in that way are almost zero, so you're arguing to keep something which causes a warning right now, in case we get an incompetent programmer at some random point in the far future. -- Paul
