Bug#829570: FTBFS on hurd-i386

[Felipe Sateler]

Hi Svante,

I noticed I did not CC you in my previous reply, so doing this now.
Full quote below.

On 4 July 2016 at 15:19, Felipe Sateler <fsateler at debian.org> wrote:
> Hi,
>
> On Mon, 04 Jul 2016 12:50:13 +0200 Svante Signell
> <svante.signell at gmail.com> wrote:
>> Hi,
>>
>> webrtc-audio-processing fails to build on GNU/Hurd and GNU/kFreeBSD due
>> to missing $host cases for Hurd and kFreeBSD, with the corresponding
>> changes in the file webrtc/base/platform_thread.cc. The attached patch,
>> non-linux.patch fixes the build problems for both Hurd and kFreeBSD.
>> Build tested on hurd-i386 and kfreebsd-i386.
>
> Thanks for the patch. I have one suggestion though:
>
> Index: webrtc-audio-processing-0.3/configure.ac
> ===================================================================
> --- webrtc-audio-processing-0.3.orig/configure.ac
> +++ webrtc-audio-processing-0.3/configure.ac
> @@ -44,6 +44,11 @@ AS_CASE(["${host}"],
>           OS_CFLAGS="-DWEBRTC_LINUX"
>           PLATFORM_CFLAGS="-DWEBRTC_POSIX"
>          ],
> +    [*-k*bsd-gnu* | *-*gnu*],
> +        [
> +         OS_CFLAGS="-DWEBRTC_NONLINUX"
> +         PLATFORM_CFLAGS="-DWEBRTC_POSIX"
> +        ],
>
>
> NONLINUX appears to me to be a too generic word. Maybe change to
> WEBRTC_GNU? After all, this module is supposed to be buildable on
> windows and osx platforms.
>
> Also, it would be great if you could forward this upstream as well[1].
> I can do this as well, but it will probably faster if you forward it.


Also, I'm not sure I understand why the reinterpret_cast is required
(AFAICT it is a noop, because it is a reinterpret_cast from pthread_t
to pthread_t), that would probably benefit from upstream comments.

>
> Saludos
>
> [1] https://bugs.freedesktop.org/enter_bug.cgi?product=PulseAudio



-- 

Saludos,
Felipe Sateler