Re: [PATCH] Use __gnu_inline__ for __extern_always_inline in g++-4.2

[Richard Smith <richard at metafoo dot co dot uk>]

Ping.

On Tue, Jan 8, 2013 at 1:29 PM, Richard Smith <richard@metafoo.co.uk> wrote:
> Updated patch attached.
>
> On Mon, Jan 7, 2013 at 11:02 PM, Paul Eggert <eggert@cs.ucla.edu> wrote:
> > On 01/07/2013 09:04 PM, Richard Smith wrote:
> >
> >> Index: include/sys/cdefs.h
> >
> > Shouldn't that be misc/sys/cdefs.h in the glibc source?
>
> Thanks, attached patch is produced by git diff against the glibc sources.
>
> >> +# if defined __GNUC_STDC_INLINE__ || \
> >> +    (defined __cplusplus && defined __GNUC_GNU_INLINE__)
> >
> > Shouldn't that be "#ifdef __GNUC_GNU_INLINE__"?
>
> No -- in C++, __GNUC_GNU_INLINE__ is defined, even though __inline
> does not use the GNU C inline semantics. I've added an explanatory
> comment in the revised patch. I also simplified this test slightly by
> removing the `defined __cplusplus' check -- it will now use the
> __gnu_inline__ attribute whenever it exists, rather than just when it
> exists and is not the default, which I hope will avoid future
> confusion here.
>
> On Tue, Jan 8, 2013 at 1:10 AM, Andreas Schwab <schwab@suse.de> wrote:
> > Richard Smith <richard@metafoo.co.uk> writes:
> >
> >> +# if defined __GNUC_STDC_INLINE__ || \
> >> +    (defined __cplusplus && defined __GNUC_GNU_INLINE__)
> >
> > Line break before operator, not after.
>
> Thanks, new revision of the patch needs no line break.

Attachment: extern-always-inline-gnu-inline-2.diff
Description: Binary data