[FFmpeg-devel] [PATCH] Fix printf usage in vhook

[Benoit Fouet]

Stefano Sabatini wrote:
> On date Friday 2008-09-05 09:46:20 +0200, Benoit Fouet encoded:
>   
>> Hi,
>>
>> Stefano Sabatini wrote:
>>     
>>> On date Friday 2008-09-05 09:14:41 +0800, JonY encoded:
>>>
>>> [...]
>>>
>>>   
>>>       
>>>> Index: libavutil/internal.h
>>>> ===================================================================
>>>> --- libavutil/internal.h	(revision 15214)
>>>> +++ libavutil/internal.h	(working copy)
>>>> @@ -242,7 +242,7 @@
>>>>  #define strcat strcat_is_forbidden_due_to_security_issues_use_av_strlcat
>>>>  #undef  exit
>>>>  #define exit exit_is_forbidden
>>>> -#if !(defined(LIBAVFORMAT_BUILD) || defined(FFMPEG_FRAMEHOOK_H))
>>>> +#if !(defined(LIBAVFORMAT_BUILD) || defined(AVFORMAT_FRAMEHOOK_H))
>>>>  #undef  printf
>>>>  #define printf please_use_av_log
>>>>  #undef  fprintf
>>>>     
>>>>         
>>> That looks like an hack to quickly fix it, the correct solution seems
>>> to me to *not use* f?printf but av_log as in your previous patch (and
>>> #undef it when it is necessary to write to file), then maybe we can
>>> remove that ifdeffery.
>>>
>>>   
>>>       
>> anyway, if FFMPEG_FRAMEHOOK_H has been renamed to AVFORMAT_FRAMEHOOK_H
>> in the new multiple inclusions guard naming scheme, this should be
>> reflected in internal.h
>>     
>
> Well we have two possible choices:
> 1) fix the header inclusion guard name in internal.h
> 2) #undef fprintf in those files which use it to print to file (rather
> than to log something, which was wrong and has already been fixed) and
> remove that condition from the directive in internal.h
>
> I have a *very* slightly preference for the second solution, but feel
> free to apply the first one as proposed by JonY if you prefer that.
>
>   

if there are no more reason to have a special case for framehook in
internal.h, I'm all for dropping FFMPEG_FRAMEHOOK_H reference from there
totally.

-- 
Benoit Fouet
Purple Labs S.A.
www.purplelabs.com