Mansour Moufid <mansourmoufid%gmail.com@localhost> writes:
> I've also upgraded to -current now so the check for 10.6 that was in
> -stable is gone.
>
>
> Index: Makefile
> ===================================================================
> RCS file: /cvsroot/pkgsrc/audio/pulseaudio/Makefile,v
> retrieving revision 1.94
> diff -u -p -u -r1.94 Makefile
> --- Makefile 6 Feb 2015 07:20:14 -0000 1.94
> +++ Makefile 16 Feb 2015 23:34:03 -0000
> @@ -59,6 +59,11 @@ CONFIGURE_ARGS+= --sysconfdir=${PKG_SYSC
> PLIST.evdev= yes
> .endif
>
> +.if !empty(MACHINE_PLATFORM:MDarwin-1[2-9].*) && !empty(SDKROOT)
> +CONFIGURE_ARGS+= --with-mac-version-min=`xcrun --show-sdk-version`
> +CONFIGURE_ARGS+= --with-mac-sysroot=`xcrun --show-sdk-path`
> +.endif
> +
> .include "options.mk"
>
> PKGCONFIG_OVERRIDE+= libpulse-mainloop-glib.pc.in
I can appreciate that this makes it work, but I'm trying to understand
how much of this really belongs in the pulseaudio makefile vs being
generic mac support. In theory buildlink arranges includes and libs,
including the sdpath.
So, I wonder if you know where pulseaudio is going astray - is it
reaching into the fs explicitly rather than using compile tests?
Do you know if the --with-mac-sysroot switch appears in a lot of other
programs?
Possibly related, I find that on a 10.9 system, xcrun --show-sdk-version
shows 10.10, unless I force SDKROOT=macosx10.9 in the environment. That
just feels like a mac bug.
Attachment:
pgpeMjJx9xsRM.pgp
Description: PGP signature