[FFmpeg-devel] [VOTE] Equality and leader team

[Felipe Contreras]

On Sat, Feb 5, 2011 at 6:31 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> On Sat, Feb 05, 2011 at 05:47:18PM +0200, Felipe Contreras wrote:
>> On Sat, Feb 5, 2011 at 1:31 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
>> > It was said the new system is similar to the linux kernel, but really it has
>> > nothing to do with it. Linus and Andrew have not been locked out of the
>> > process by 7 of their friends and asked to send patches.
>>
>> This shows you have absolutely no idea how linux development works.
>
>> Andrew Morton sends all his patches:
>> http://article.gmane.org/gmane.linux.kernel.mm/57767
>>
>> In fact, he has to, because nor he, nor anybody else has commit access
>> to Linus's tree, but Linus. That's the whole point of a *distributed*
>> DSCM; every person has his own repo. But according to your definition,
>> Linus has evilly locked out all kernel developers and maintainers.
>
> nope, linus is the most qualified and it makes alot of sense that he is the
> gatekeeper, what we have here is different, the 7 barely know maybe 1/3 of
> the code. Iam alone the author of 1/3 and know alot more.

Linus most certainly has not written 1/3 of the kernel, most likely
not even 1/100. Pretty much all he does is merge branches. Moreover,
he is not looking into each single commit, he has a few people he
trusts, and as long as the patches come with the ack of those pople,
he merges. Those lieutenants in turn don't look into every patch, they
have people they trust, and as long as the patches have their ack,
they accept them. And so on.

FFmpeg doesn't need a Linus that would be merging 130 subtrees and
making sure nothing explodes, a bot can do that.

>> In fact Linus sends his own patches too:
>> http://article.gmane.org/gmane.linux.network/176588
>>
>> Not only does he sends his patches, but he himself doesn't trust his
>> own patches on subsystem X, it's the duty of the maintainer of
>> subsystem X to pick and acknowledge the patch, and then send the pull
>> request to Linus with the patches of many people including Linus
>> himself.
>
> Absolutely and 100% agree, I also send patches and want to send patches where
> iam not the maintainer, the guy who knows the code must look at my patch and
> check its ok. The terms ans structuring is different in linux and ffmpeg but
> the idea WAS the same.

You are confused, Linus is not the maintainer of _anything_. He sends
_every_ patch he intends to commit.

Again, everybody has to send patches.

> But the new maintainers are not maintainers in the sense of knowing the code.
> The actual people who know the code have been completely locked out, they dont
> even have private repositories left where they can commit or where they could
> work together. For example 2 people wanting to work on the mpeg audio decoder
> have to send patches to be approved by the 7 (none of them knows mpeg audio)
> The maintainer of mpeg audio (yeah me too, i did alot) is locked out of the
> process. That simply is not the linux development model

The only change in MAINTAINERS I see is that you are no longer the
project leader, but actually you remain as maintainer in many pieces
of code. Presumably, if somebody sends a patch that touches your code
and you ack it, or nack it, it would count the same as any other
maintainer.

Cheers.

-- 
Felipe Contreras