On Nov 3, 2010, at 7:44 PM, Jed Brown wrote: > On Wed, Nov 3, 2010 at 12:50, Barry Smith <bsmith at mcs.anl.gov> wrote: > Because of the || *r2 may never get set, but it gets used later in the > routine. Maybe compilers shouldn't give the warning but some did. > > But the uninitialized value can't be used if PetscMallocX failed because the > error is checked and the error handler is invoked. Was the problem only in > some case where the error code was not being checked, or was it more than > that? Spurious uninitialized warnings are obnoxious, are they worth > pandering to?
I would say it was spurious, but users building PETSc won't know it is spurious and I like clean compiles with no warnings (even if the warnings are wrong). Barry > I can't reproduce them with any of my compilers. > > Jed
