>> At a guess, sizeof(XmGadget) < sizeof(Widget) so casting a gadget to Widget
>> will trigger a bounds check?
Alexander> What's the purpose on checking bounds on read not write access?
You mean why check reads at all? I can imagine all kind of damage to an
algorithm that can result from reading beyond the end of an object. Also,
having detected a bounds violation somewhere, you need to know where it
originated, so the cast is an obvious source of such problems.
Alexander> And finally: do we need to keep this code??
Alexander> I don't consider it good style to have code depending on some "strange"
Alexander> compiler features in public headers.
Since it predates 1997/02/20, there is a good chance that the compiler has
changed since it was written anyway!
The way that Widget and XmGadget conveniently overlap reminds me of FORTRAN
common blocks :-)
__Martin
