This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Fix implicit-fallthrough warnings in tst-setjmp.c
On Thu, 14 Feb 2019, Paul Eggert wrote:
> On 2/14/19 2:23 PM, Joseph Myers wrote:
> > > Shouldn't we use the more-standard _Noreturn for this sort of thing?
> > > cdefs.h
> > > defines _Noreturn for pre-C11 GCCs.
> > The attribute seems much more widely used in glibc .c files at present.
>
> Sure, but that habit predates C11. Now that we have a standard way to say
> "this function does not return" it's surely better to use the standard way, at
> least for new code and arguably even for old code.
The consistent, universal style in glibc, for all files not coming from
external sources, other than a single use of _Noreturn in
nptl/thrd_exit.c, is to use the attribute. You can of course propose
(with corresponding patch) a change to the style used (recalling that for
installed headers _Noreturn would reduce portability because of prefix
attributes not working with GCC 2.7), but it seems clear my patch is
consistent with the current style and _Noreturn would be inconsistent with
it.
--
Joseph S. Myers
joseph@codesourcery.com