[FFmpeg-devel] [PATCH 1/2] fftools/ffmpeg: exit application when decoding returns AVERROR_EXIT

[Soft Works]

> -----Original Message-----
> From: ffmpeg-devel <ffmpeg-devel-bounces at ffmpeg.org> On Behalf Of
> Gyan Doshi
> Sent: Tuesday, October 19, 2021 6:21 AM
> To: ffmpeg-devel at ffmpeg.org
> Subject: Re: [FFmpeg-devel] [PATCH 1/2] fftools/ffmpeg: exit
> application when decoding returns AVERROR_EXIT
> 
> 
> 
> On 2021-10-19 09:38 am, Soft Works wrote:
> >
> >> -----Original Message-----
> >> From: ffmpeg-devel <ffmpeg-devel-bounces at ffmpeg.org> On Behalf Of
> >> Gyan Doshi
> >> Sent: Tuesday, October 19, 2021 6:01 AM
> >> To: ffmpeg-devel at ffmpeg.org
> >> Subject: Re: [FFmpeg-devel] [PATCH 1/2] fftools/ffmpeg: exit
> >> application when decoding returns AVERROR_EXIT
> >>
> >>
> >>
> >> On 2021-10-19 04:54 am, Soft Works wrote:
> >>> Introduce a way for decoders to request application exit via
> error
> >> return
> >>
> >> Why? The ffmpeg app may be processing multiple inputs and outputs.
> At
> >> most, you can close the codec and end the stream.
> > When a hardware device fails unexpectedly, why should the app
> continue
> > running? In this case, it's clear that things do not work as
> intended.
> 
> Because the application may be doing other things and the hardware
> function may not be central to its task.

Is that really a realistic scenario that a user wants ffmpeg to
continue even though his setup doesn't work?
All over the code there are so many less critical conditions that
lead ffmpeg to exit.


> See the condition just before the one you added. It is
> `exit_on_error`
> which implies that the user can *choose* to exit, or not, upon error.
> Your patch lets the decoder unilaterally decide to quit without user
> input.

Yes, exactly.

> In fact, I don't think your patch is necessary. You added the
> condition
> in process_input_packet() . During decoding, check_decode_result() is
> called, which has
> 
>      if (ret < 0 && exit_on_error)
>          exit_program(1);
> 
> So, the user can choose to set exit_on_error and get this behaviour.

Well, but usually nobody sets this.

So when it isn't set, we should keep ffmpeg 
alive to let it write a 10 GB logfile?

Maybe there's a better way than my patch, but what happens is 
a bug IMO, and I wouldn't say that it would be a fix to say that 
a certain option needs to be set by the user to "fix it".

Kind regards,
softworkz