[FFmpeg-user] I found the bugs

Mark Filipak markfilipak.imdb at gmail.com
Sun Jun 16 11:19:31 EEST 2024


On 16/06/2024 03.51, Jim DeLaHunt wrote:
> On 2024-06-15 23:04, Mark Filipak wrote:
> 
>> On 15/06/2024 23.39, Jim DeLaHunt wrote:
>>> On 2024-06-15 19:27, Mark Filipak wrote:
>>>
>>>> It would be nice if folks from here went here:
>>>> https://trac.ffmpeg.org/ticket/11055
>>>> and saw what is going on. It's up to 76 comments now, so what I ask will take you a while.
>>>> What's going on is a crime.
>>>
>>> What's the crime, detective?  All I see is two people talking past each other and not being clear 
>>> about their evidence. I posted a comment with my feedback for you two.
>>
>> You posted an irrelevant comment supporting inadmissible info from one of the accused: FFprobe 
>> show_frames.
> 
> _Who_ posted the criminal comment (the comment you describe as "What's going on is a crime")?  _I_ 
> posted it?  Check again.

No, Jim. I posted "it's a crime" here way before your comment on ticket 11055. "It's a crime" that 
MasterQuestionable is ignoring what would be legal proof in a court of law and is instead kicking up 
a load of dust. How anyone could argue that actual timestamps from actual packets plus actual 
timestamps from framecrc that match them perfectly are somehow disqualified or insufficient is 
beyond my comprehension. showinfo and show_frames are wrong and there's no two ways about it. The 
actual timestamps and what showinfo and show_frames report are miles apart. The bug in some internal 
routine that they use is wrong in this case, of a video with these particular properties, and I 
think that's true of '-ss' and '-to' and of transcoders in which AVC/H.264 is the source.

> Note that the XML file which MasterQuestionable attached is from ffprobe's -show_entries, not 
> -show_frames.

Same thing.

> If you want to assert that those are the same thing, and that -show_entries data is 
> untrustworthy for the same reason that -show_frames data is, then you should have some evidence for 
> that assertion.

It doesn't matter, Jim. Do I have to find EVERY problem? I found two. Isn't that enough? Those two 
will lead to the others but it looks like it's going to be closed with no action, no code look, no 
testing. This all stinks and I'm sick of it.

I know how to make perfect cuts and splices that play perfectly, open GOPs or closed GOPs, DTS-order 
or PTS-order, and I don't give a damn if FFmpeg wants to bury the news and leave everything 
unchanged. I'm finished with it. I got no support here and I see the picture clearly.



More information about the ffmpeg-user mailing list