On 3/24/2025 10:29 AM, Gyan Doshi wrote:
> 
> 
> On 2025-03-24 06:42 pm, James Almer wrote:
>> On 3/24/2025 9:31 AM, Martin Storsjö wrote:
>>> On Mon, 24 Mar 2025, Gyan Doshi wrote:
>>>
>>>> ffmpeg | branch: master | Gyan Doshi <ffmpeg@gyani.pro> | Sat Mar 15 
>>>> 16:51:05 2025 +0530| [cbbc927a67f142a02bfc5d047c03587599ee5b55] | 
>>>> committer: Gyan Doshi
>>>>
>>>> ffmpeg: add per-stream input option drop_changed
>>>>
>>>> This is a replacement in ffmpeg for the deprecated avcodec flag 
>>>> AV_CODEC_FLAG_DROPCHANGED.
>>>>
>>>> This option is meant to be used when the filtergraph should not be
>>>> reinited upon input parameter changes as that leads to loss of state
>>>> in the filtergraph potentially leading to broken or aborted output,
>>>> e.g. inserting of silence with first_pts specified in aresample.
>>>>
>>>> Generally useful to avoid corrupted yet decodable packets in live
>>>> streaming inputs.
>>>>
>>>> This option when enabled takes precedence over reinit_filters
>>>>
>>>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/? 
>>>>> a=commit;h=cbbc927a67f142a02bfc5d047c03587599ee5b55
>>>> ---
>>>
>>> This change is causing failed fate tests in many different 
>>> configurations; I'm seeing null pointer dereferences and similar, on 
>>> both msvc/arm, msvc/x86_32, macos/aarch64. Please investigate.
>>
>> I pushed a quick fix, but nonetheless, the log message in question is 
>> being printed for one stream (the last one handled by the scheduler) 
>> rather than every stream where the option was the for, so it needs to 
>> be properly done still.
> 
> Thanks. FATE passed here and on Patchwork as well.
> 
> If ifp can be NULL, how come this pre-existing line above,
> 
>            if (ifp->type_src == AVMEDIA_TYPE_SUBTITLE) {
> 
> didn't lead to FATE failures?

The goto read_frames before ifp is ever set.

> 
> Re: log, the final log is printed for each stream:
> 
> e.g., from the log of a command with option applied to aud and vid,
> 
> [af#0:1 @ 000001b4af668380] Total changed input frames dropped : 234
> [vf#0:0 @ 000001b4af3e0540] Total changed input frames dropped : 149

Oh, i see.