Re: [MPlayer-dev-eng] [MPlayer-users] Build regression on FreeBSD
- Date: Sun, 20 Dec 2015 22:05:01 +0100
- From: Alexander Strasser <eclipse7@xxxxxxx>
- Subject: Re: [MPlayer-dev-eng] [MPlayer-users] Build regression on FreeBSD
AFAICT there are multiple problems at work, that make a
quick solution rather involved or hackish. That's why I
have Cc'ed MPlayer development mailing list.
On 2015-12-19 20:21 +0100, Thomas Zander wrote:
> porting an up-to-date snapshot to FreeBSD I noticed two build
> regressions and want to kindly ask for resolving those upstream:
> 1) Makefile (at root dir)
> codec-cfg.c now includes <vdpau/vdpau.h>
> On FreeBSD this file ends up in LOCALBASE (by default
> /usr/local/include) which is not included in the default compiler
> search path, so in its build target
> codec-cfg$(EXESUF) .... the compile command
> $(HOST_CC) $(HOSTCFLAGS) -o $@ $<
> is not sufficient. The include path for localbase needs to be added.
I guess r37454 introduced this problem, by including vdpau.h in
our img_format.h header. It cannot be reverted because that would
probably break the build with future FFmpeg versions.
The problem is if we want to use external stuff, which cannot be
forward declared, in img_format.h we need to add additional flags
to HOSTCFLAGS as well.
Furthermore it is interesing that vdpau gets detected at all on
your system. I suspect this is because of manual addition to CFLAGS
or because of modification by previous configure tests.
I might be missing a few things, but all in all I do not see why
we need the structure vdpau_frame_data in codec-cfg.c source. So
maybe the structure should be defined in a different header?
All in all I see no really easy way to set HOSTCFLAGS correctly.
Additionally I am not sure if the work is justified, because if
I am not mistaken above it is not needed in codec-cfg.
MPlayer-dev-eng mailing list