Dan Kegel wrote:
No fair, Mike, you should have given the link to the PR:
http://gcc.gnu.org/PR19430
:) If you know a gcc person that has time, nagging them would be great.
Many similar errors will get caught, as long as you turn on the
optimizer with -O (otherwise the compiler can't detect them).
It does warn on the use of x as long as you avoid the specific case of
PR19430.
In the case where you use &var, ignoring compiler bugs, it always seems
like a good idea to initialize var to a known value to avoid
indeterminate behaviour.
If the compiler is clever enough to give an uninitialized variable
warning, then it will likely be clever enough to optimize out the extra
assignment too.
Mike
