[FFmpeg-devel] [PATCH][0.5] Fix building on the hurd

[Reinhard Tartler]

On Mo, Feb 01, 2010 at 22:38:30 (CET), Reimar D?ffinger wrote:

> On Mon, Feb 01, 2010 at 09:38:41PM +0100, Reinhard Tartler wrote:
>> On Mo, Feb 01, 2010 at 21:22:10 (CET), Reimar D?ffinger wrote:
>> 
>> > On Mon, Feb 01, 2010 at 09:04:52PM +0100, Reinhard Tartler wrote:
>> >> 
>> >> Author: Andres Mejia <mcitadel at gmail.com>
>> >> Fix build failures for the GNU Hurd OS
>> >> 
>> >> please approve for branches/0.5
>> >> 
>> >> --- a/configure
>> >> +++ b/configure
>> >> @@ -1651,6 +1651,9 @@ case $target_os in
>> >>          ;;
>> >>      gnu/kfreebsd)
>> >>          ;;
>> >> +    gnu)
>> >> +        disable dv1394
>> >> +        ;;
>> >
>> > How does it fail?
>> 
>> It fails to build.
>> 
>> > There certainly should be a proper autodetection for that feature?
>> > Actually, considering it only seems to require ioctl and mmap is there
>> > much of a point to disable this feature?
>> 
>> The hurd does not seem to have firewire support at all. Moreover, the
>> proposed patch is actually a backport of trunk, and therefore I think it
>> is valid on its own.
>
> If it is then I must be blind again today, because I can't see anything
> like that in trunk.

------------------------------------------------------------------------
r18938 | diego | 2009-05-25 07:39:58 +0200 (Mo, 25. Mai 2009) | 3 Zeilen

Add Hurd to OS list and disable dv1394 in the Hurd case.
patch by Andres Mejia, mcitadel gmail com

Index: configure
===================================================================
--- configure	(Revision 18893)
+++ configure	(Revision 18938)
@@ -1697,6 +1697,9 @@
         ;;
     gnu/kfreebsd)
         ;;
+    gnu)
+        disable dv1394
+        ;;
 
     *)
         die "Unknown OS '$target_os'."


However, upon further inspection the next relevant patch reads:

------------------------------------------------------------------------
r18946 | diego | 2009-05-25 22:18:33 +0200 (Mo, 25. Mai 2009) | 2 Zeilen

dv1394 is disabled by default, do not redundantly disable it on Hurd.

Index: configure
===================================================================
--- configure	(Revision 18938)
+++ configure	(Revision 18946)
@@ -1698,7 +1698,6 @@
     gnu/kfreebsd)
         ;;
     gnu)
-        disable dv1394
         ;;
 
     *)

>> For completeness, I think this is the relevant debian bug report:
>> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=372290. It was reported
>> by a hurd porter, and I generally tend to trust porters.
>
> If that's all you have you can't know that enabling dv1394 causes the build to fail,
> it may just have been the --enable-dc1394 alone that caused the issue, or the v4l
> or the bktr code.

Hm, it seems that you're right, and I should rather fix that build
failure by special casing the hurd in debian/rules instead of configure.

thanks for the review!

-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4