[FFmpeg-devel] Fwd: [PATCH] Support alternative spellings of the inline keyword.

[Michael Niedermayer]

On Sat, Jun 16, 2012 at 07:30:43PM -0400, Ronald S. Bultje wrote:
> Hi,
> 
> On Sat, Jun 16, 2012 at 7:08 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > Hi
> >
> > On Sat, Jun 16, 2012 at 06:05:03PM -0400, Ronald S. Bultje wrote:
> >> Hi,
> >>
> >> same here, part of a MSVC port.
> >>
> >> Ronald
> >>
> >> ---------- Forwarded message ----------
> >> From: Ronald S. Bultje <rsbultje at gmail.com>
> >> Date: Fri, Jun 15, 2012 at 7:04 PM
> >> Subject: [PATCH] Support alternative spellings of the inline keyword.
> >> To: libav-devel at libav.org
> >> Cc: "Ronald S. Bultje" <rsbultje at gmail.com>
> >>
> >>
> >> From: "Ronald S. Bultje" <rsbultje at gmail.com>
> >>
> >> This is similar to how we support e.g. __restrict as an alternative for
> >> restrict, and allows us to support compilers that use __inline instead
> >> of inline. The define is placed in avconfig.h instead of config.h since
> >> we use inline in public header files.
> >
> > [...]
> >>  #ifndef AVUTIL_AVCONFIG_H
> >>  #define AVUTIL_AVCONFIG_H
> >> +#define inline $_inline
> >>  EOF
> >
> > Is this safe if ffmpeg is compiled and installed with one compiler
> > and an application with a different compiler ?
> >
> > I mean configure is run with one compiler but a application that may
> > include the header and thus #define inline (__)inline could use a
> > different compiler.
> 
> No, that would break. A solution for that would be to statically
> define inline in one of the main headers (similar to how attributes.h
> is set up) depending on the compiler, i.e.:
> 
> #ifdef _MSC_VER
> #define inline __inline
> #endif
> 
> I'm not sure why I didn't put this in attributes, but I can move it
> there, I don't particularly care where it goes, as long as it's
> available for use in installed headers, since we use inline all over
> the place in installed headers.
> 
> Also, if we're scared of symbol collision, we could even rename to
> av_inline, but I'd prefer not to since that's a lot of sed left to do.

The inlines in the installed headers could be changed to
av_always_inline or av_inline, this would
avoid the need to mess with #define inline in public headers and
the private code could then use either the #fdef _MSC_VER or a
configure based check whichever you prefer.

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20120617/9a81133b/attachment.asc>