On Thu, Mar 14, 2019 at 10:36 AM Fāng-ruì Sòng wrote:
>
> On Thu, Mar 14, 2019 at 9:59 AM Henrik Gramner wrote:
> >
> > On Wed, Feb 20, 2019 at 8:03 PM Fāng-ruì Sòng
> > wrote:
> > > --- a/libavutil/mem.h
> > > +++ b/libavutil/mem.h
> > >
> > > +#if defined(__GNUC__) && !(defined(_WIN32) ||
On Thu, Mar 14, 2019 at 9:59 AM Henrik Gramner wrote:
>
> On Wed, Feb 20, 2019 at 8:03 PM Fāng-ruì Sòng
> wrote:
> > --- a/libavutil/mem.h
> > +++ b/libavutil/mem.h
> >
> > +#if defined(__GNUC__) && !(defined(_WIN32) || defined(__CYGWIN__))
> > +#define DECLARE_HIDDEN __attribute__
On Wed, Feb 20, 2019 at 8:03 PM Fāng-ruì Sòng
wrote:
> --- a/libavutil/mem.h
> +++ b/libavutil/mem.h
>
> +#if defined(__GNUC__) && !(defined(_WIN32) || defined(__CYGWIN__))
> +#define DECLARE_HIDDEN __attribute__ ((visibility ("hidden")))
> +#else
> +#define DECLARE_HIDDEN
> +#endif
Ping :)
In short, when people build .so from ffmpeg .o/.a files, they no longer
need -Bsymbolic
-Bsymbolic is bad because it breaks C++ semantics, e.g. address uniqueness
of inline functions and their static local variables, uniqueness of
template specializations, address uniqueness of type_info
Sorry, wrong signed-off-by line..
(isn't using my usual mail client mutt and i get very sick of the
webmail...)
On Thu, Mar 7, 2019 at 11:36 AM Fāng-ruì Sòng wrote:
> On Wed, Mar 6, 2019 at 4:14 PM Carl Eugen Hoyos
> wrote:
>
>> 2019-02-21 2:37 GMT+01:00, Fāng-ruì Sòng <
>>
On Wed, Mar 6, 2019 at 4:14 PM Carl Eugen Hoyos wrote:
> 2019-02-21 2:37 GMT+01:00, Fāng-ruì Sòng >:
> > Sorry if this doesn't attach to the correct thread as I didn't
> > subscribe to this list and don't know the Message-ID of the thread.
>
> The thread works fine here with gmail but your
2019-02-21 2:37 GMT+01:00, Fāng-ruì Sòng :
> Sorry if this doesn't attach to the correct thread as I didn't
> subscribe to this list and don't know the Message-ID of the thread.
The thread works fine here with gmail but your patch was corrupted,
you have to attach it.
Carl Eugen
On Thu, Feb 21, 2019 at 10:06 AM Fāng-ruì Sòng wrote:
>
> > Why is it a good idea to remove them from the linker command line?
>
> In short, it improves portability. I'm not suggesting removing
> -Bsymbolic or --version-script from the ffmpeg build system. I mean
> users will no longer have to
> Why is it a good idea to remove them from the linker command line?
In short, it improves portability. I'm not suggesting removing
-Bsymbolic or --version-script from the ffmpeg build system. I mean
users will no longer have to specify the two options to link ffmpeg
object files into their own
Sorry if this doesn't attach to the correct thread as I didn't
subscribe to this list and don't know the Message-ID of the thread.
> The word "also" indicates here that this should be an independent patch.
I added `#if defined(__GNUC__) && !(defined(_WIN32) ||
defined(__CYGWIN__))`, not `#if
2019-02-20 10:13 GMT+01:00, Fāng-ruì Sòng :
> Inline asm code assumes these DECLARE_ASM_ALIGNED declared global
> constants are non-preemptive, e.g.
>
> libavcodec/x86/cabac.h
> "lea"MANGLE(ff_h264_cabac_tables)", %0 \n\t"
>
> On ELF platforms, if -Wl,-Bsymbolic
>
11 matches
Mail list logo
