[cfe-dev] PR15105: __attribute__((noreturn)) and noreturn function types

[Richard Smith]

On Tue, Jan 29, 2013 at 4:13 PM, Sean Silva <silvas at purdue.edu> wrote:
> How often does our "improved" behavior cause us to reject valid code?

That sounds very hard to quantify. But at least I don't believe anyone
has ever encountered the name mangling issue in the wild and reported
it as a bug.

> If this is causing problems for someone in the wild (the PR suggests
> that it is), IMO we should just be compatible with GCC.

Well, there are a number of ways we could proceed. I'm currently
swaying towards being largely GCC-compatible, but keeping our ban on
implicitly converting pointer-to-function to
pointer-to-noreturn-function (except in the cases where we can
trivially see that the source is a noreturn function).

> Since as you pointed out it has ABI and language semantics effects,
> incompatibility could have potentially serious consequences and the
> safest bet is to just be compatible.
>
> -- Sean Silva