Am 26.11.15 um 07:12 schrieb Luca Barbato:
> On 25/11/15 09:15, Thilo Borgmann wrote:
>> Hi folks,
>>
>> I wonder why you have done a reimplementation based on FFmpeg's device with
>> reduced features and OS support. Can you enlighten me?
> 
> The current implementation from Alexandre and the one from you are quite
> different:

> - This one references the devices using their stable names, your seems
> to use some index.
> - This one uses ARC, yours uses autorelease pools.
> - This one is just for video capture so far, yours is
> audio+video+screen_capture.

a) True. Although I don't share your opinion that this makes the implementations
'quite different'.

> Probably during his initial tries he used more code from yours, but the
> final version[1] I'm looking at keeps from your implementation just the
> pixel format -> CoreVideo formats table...

b) False. The very core is equal: Run-time approach for format detection, frame
receiver delegate, frame buffering in context, thread synchronization, PTS/DTS
generation... all the same, all copied.

> Surely you should be credited in the patch, but yours and Alexandre's
> are quite different implementations and from what I can read, you
> wouldn't move to ARC so I guess their purposes are even different.

c) For proper credit you should have started to add the device from FFmpeg into
the Libav tree; then apply your patches to remove what you didn't like; then
apply your patches to add what you want in addition or what you want to change
(for example ARC); and then add your copyrights next to the existing one (mine).

d) I wrote because I wanted to know the reason for a reimplementation and that
question has not been addressed. I can't see a reason. Also, I wonder why you
are talking about Alexandre's implementation while you are listed yourself as
the first author.


However, I'm not writing because of a nor b nor c nor d. I don't really care and
please let us not discuss a,b,c or d any further. Stop. Please.


Instead, I want to offer both of you to collaborate on this device. Let's setup
a github-based FFmpeg branch, patch the FFmpeg device towards ARC, discuss and
apply further changes you did in the Libav implementation, setup another Libav
branch, pull over the device and if you still don't want audio&screen drop it -
finally we can pull that device into FFmpeg|Libav HEAD.

I would very much like to see this happen in the aftermath of VDD 2015 and have
finally avoided yet another useless divergence between FFmpeg and Libav.

This is my honest offer. Now, it is up to you two to collaborate or not.

-Thilo

p.s. That you pulled the device into Libav an hour or so ago although review on
libav-devel is still in progress does not change anything regarding my offer nor
does it prevent us from doing it.

p.p.s. I keep Alexandre in CC just because I don't know if he follows on
libav-devel and can read this. Sorry for spam if you do, Alexandre.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to