[FFmpeg-devel] [PATCH] lavu: make AV_TIME_BASE_Q work in C++ code

[Tomas Härdin]

fre 2020-06-26 klockan 11:36 +0200 skrev Hendrik Leppkes:
> On Fri, Jun 26, 2020 at 11:31 AM Tomas Härdin <tjoppen at acc.umu.se> wrote:
> > ons 2020-06-24 klockan 21:22 +0200 skrev Carl Eugen Hoyos:
> > > Am Mi., 24. Juni 2020 um 21:10 Uhr schrieb rcombs <rcombs at rcombs.me>:
> > > > ---
> > > >  libavutil/avutil.h | 4 ++++
> > > >  1 file changed, 4 insertions(+)
> > > > 
> > > > diff --git a/libavutil/avutil.h b/libavutil/avutil.h
> > > > index 4d633156d1..c11b33f466 100644
> > > > --- a/libavutil/avutil.h
> > > > +++ b/libavutil/avutil.h
> > > > @@ -257,7 +257,11 @@ const char *av_get_media_type_string(enum AVMediaType media_type);
> > > >   * Internal time base represented as fractional value
> > > >   */
> > > > 
> > > > +#ifdef __cplusplus
> > > > +#define AV_TIME_BASE_Q          AVRational{1, AV_TIME_BASE}
> > > > +#else
> > > >  #define AV_TIME_BASE_Q          (AVRational){1, AV_TIME_BASE}
> > > > +#endif
> > > 
> > > The problem with this patch is that it gives C++ users of the libraries
> > > the feeling that we would care about them and that we would indeed
> > > fix issues.
> > > Given that the contrary is true (at least in the past) and that we do
> > > not do any specific C++ testing, I believe it is better to let users
> > > work around this issue (from their point of you).
> > 
> > Maybe we should put in a #warning to that effect? Inside an #ifdef
> > __cplusplus
> > 
> 
> Now that would just be annoying. 99% of the headers work just fine in
> C++, spewing a warning at us for using them would not serve anything.

Well, if you volunteer to maintain C++ support then sure. Most devs in
this project are C guys. Admittedly this is a bit of a Måns-ism, but
he's no longer involved so.. I don't think anyone is principally
opposed, it's just a question of where to spend developer effort.

/Tomas