This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [C++ Patch/RFC] Final part of c++/26099, __is_convertible_to
On 6/24/07, Richard Guenther <richard.guenther@gmail.com> wrote:
;)
I suppose you tried that - from a quick look that argument to
perform_implicit_conversion would need to be passed down to
a _lot_ of stuff. I agree that a this global flag in diagnostics is
not the prettiest thing though.
I just looked into perform_implicit_conversion and it does not look
that bad since you can use LOOKUP_NORMAL & ~LOOKUP_COMPLAIN for
implicit_conversion and before calling convert_like, you can check if
bad_p is true on the conversion and make sure convert_like_real's
issue_conversion_warnings is false. And maybe a little more is needed
in convert_like_real but it should not be that painful when I looked
into the code.
What you did is something which I would considered to be a prototype
patch. It is not like we don't control all of the C++ front-end. Yes
the patch would become bigger but most of the infrastructure for not
warning/erroring out is already there.
I was not trying to be funny but I was trying to say you added a
feature to the generic code that is just ugly and makes the front-end
comunicating with a global variable which is just the wrong approach.
Also note I think it is wrong to put the variable in flags.h/toplev.h
anyways. A more obvious place is to put it inside
diagnostic.c/diagnostic.h.
-- Pinski