Re: Patch for various warnings in gcc

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Zack Weinberg <zack@codesourcery.com> writes:

| On Thu, Aug 08, 2002 at 01:57:29PM -0400, Kaveh R. Ghazi wrote:
| >  > Personally I agree with Dale that 'int a = a;' is too obscure a way of
| >  > silencing this warning.  If we must have a way to disable it without
| >  > generating extra code, let us create __attribute__ ((initialized))
| >  > which states explicitly what is going on.
| > 
| > The attribute would be more readable but it would still suffer from
| > the non-determinism, right?
| 
| Yes.  This would not be for use in GCC itself.  Paul has a good point
| that in some contexts the extra initialization is too costly, but the
| 'a=a' idiom is not documented, nor is it obvious to a reader that its
| sole effect is to silence this warning.  So I think we should have a
| more readable way to get that effect.

I agree in principle.  While your __attribute__(()) proposal is
certainly a far better alternative of what has been proposed, I think that
the compiler should be instructred about value/condition propogation
and not warn in cases the variable is "manifestly" initialized -- but
GCC is insisting on warning just because it didn't do a good job at
propagating conditions.

-- Gaby