[EMAIL PROTECTED] wrote:
> > serious problem, and is almost alway bogus) it would
> > make finding the legitamate warnings much easier.
>
> How about just taking out the declarations
> of variables not used? Instead of shutting
> up the compiler, why can't clean code be
> written?
Many of these warnings are from code like:
void SomeClass::genericHandler(SomeWidget * window, SomeEvent * e)
{
UT_ASSERT(window && e);
doSomethingGlobalWith(e);
}
You'll notice that "window" isn't used, except that it != NULL is checked.
Often this helps me find problems when genericHandler is invoked with
a bogus "window" argument. This can happen when the genericHandler is
bound to events through a windowing system's event loop (like GTK's).
Most of GTK's handlers are like that; they take widget and object pointers
that often aren't used, but are nice to check.
> I know several of you would be irritated
> by what I said above, but it will be nice to
> have a cleanly compiling code in the long run.
I think someone should probably take a look at toning down GCC's
warning level. It really is set to throw every warning it can find,
except for some really, really obscure ones.
Once we tune the options, I think we should attack the ones that look
suspicious.
--
Shaw Terwilliger
