Re: [C++] named return values and return without value warnings

[Richard Henderson <rth at cygnus dot com>]

On Sun, Oct 08, 2000 at 10:57:16AM -0700, Mark Mitchell wrote:
>     Richard> Frankly, that's my vote as well, but before I the effort
>     Richard> is invested in tweeking that many test cases I'd prefer
>     Richard> to have buy-in from the front-end folk.
> 
> I'm sorry -- I'm a little lost.  We're not seeing warnings on some of
> these test-cases any more because warn_return_type isn't on by
> default, is that the issue?

Yes.  There were 5 test cases that had expected warnings that
we no longer got if warn_return_type was off.  If I turn
warn_return_type on by default, there are about 70 test cases
that fail because of excess warnings.

> One problem here is that I think we wanted to warn about falling off
> the end of functions that never get turned into RTL.  (That happens
> rather much in the presence of inlined functions.)

I agree that that is a nice thing to have, but the bits I
deleted were using can_reach_end, which implies rtl.  So
I'm a bit confused by your statement.

> So, I'm not sure where to go from here.

I dunno.  We could do nothing, but then we'd have no fall off the
end of the function warning by default, which is I think a mistake.
That one's serious (and common) enough that I think we ought not
require -Wall or whatnot for detection.


r~