On Fri, Jul 13, 2012 at 10:59:44PM +0100, Måns Rullgård wrote:
> Diego Biurrun <[email protected]> writes:
> > On Tue, Jul 10, 2012 at 08:22:01PM +0300, Martin Storsjö wrote:
> >> On Tue, 10 Jul 2012, Diego Biurrun wrote:
> >> >--- a/libavcodec/dct-test.c
> >> >+++ b/libavcodec/dct-test.c
> >> >@@ -78,9 +78,11 @@ static int cpu_flags;
> >> >
> >> >static const struct algo fdct_tab[] = {
> >> >    { "REF-DBL",        ff_ref_fdct,           NO_PERM    },
> >> >+#if CONFIG_ENCODERS
> >> >    { "FAAN",           ff_faandct,            NO_PERM    },
> >> >    { "IJG-AAN-INT",    ff_fdct_ifast,         SCALE_PERM },
> >> >    { "IJG-LLM-INT",    ff_jpeg_fdct_islow_8,  NO_PERM    },
> >> >+#endif
> >> 
> >> Looks straightforward enough even though I don't know the different
> >> dcts well enough to know whether it is right or not.
> >
> > In libavcodec/Makefile we have
> >
> >   OBJS-$(CONFIG_ENCODERS) += faandct.o jfdctfst.o jfdctint.o
> >
> > so the files containing those DCTs are compiled under the same
> > condition as the #ifdef I use to conditionally add the entries
> > into the table of DCTs to test.  Looks correct to me :)
> 
> The dependency on CONFIG_ENCODERS is quite dubious.

It's the dependency that we currently have in libavcodec/Makefile and
libavcodec/dsputil.c.  So in libavcodec/dct-test.c we need the same
dependency to avoid linker failures.

>  I'd really like to decouple that rather than cement it further.

I worked on decoupling the DCT selection/dependency in March this year.
It's not all that easy to separate, but this #ifdef will not notably
complicate my work when I pick it up again.  It will, however, let some
configurations compile that are currently broken.

Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to