[vlc-devel] [PATCH 01/11] ogg: don't call pf_demux() directly

[Steve Lhomme]

On Mon, May 2, 2016 at 6:59 PM, Rémi Denis-Courmont <remi at remlab.net> wrote:
> Le 2016-05-02 18:24, Steve Lhomme a écrit :
>>
>> ---
>>  modules/demux/ogg.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/modules/demux/ogg.c b/modules/demux/ogg.c
>> index 40b26c6..8b47e86 100644
>> --- a/modules/demux/ogg.c
>> +++ b/modules/demux/ogg.c
>> @@ -246,7 +246,7 @@ static int Open( vlc_object_t * p_this )
>>      TAB_INIT( p_sys->i_seekpoints, p_sys->pp_seekpoints );
>>
>>
>> -    while ( !p_sys->b_preparsing_done && p_demux->pf_demux( p_demux ) > 0
>> )
>> +    while ( !p_sys->b_preparsing_done && demux_Demux( p_demux ) > 0 )
>
>
> This patch itself looks OK, but something is wrong if you depend on it.

Yes, the problem is the way a lot of the code is using demuxers. They
often use p_demux->s and in the core there's code that work
differently whether the demuxer as a stream or not. If I go with my
previous way, each filter would be a demux_t, with at least the stream
copied (but it's not the owner). I don't know to what extent
everything else needs to be copied and updated to all these values
need to be monitored or not. That's why I think the demuxer being
untouched might be a better idea, even though it means not accessing
pf_demux and pf_control directly. Maybe they could be made private ?

>>      {}
>>
>>      return VLC_SUCCESS;
>
>
> --
> Rémi Denis-Courmont
> http://www.remlab.net/
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel