[FFmpeg-devel] My view of this f***ed up situation

Vitor Sessak vitor1001
Mon Mar 14 20:56:46 CET 2011

First of all, sorry to everybody for joining this discussion so late, 
even though I have already given my opinion on pvt to those who asked. I 
wanted to wait and see the outcome of the "revolution" (I still had some 
hope of a positive outcome).

I have to say that I'm pretty disappointed with both sides for the way 
this issue was handled:

-- to Mans, Ronald et al: --

I found that the way the whole issue was handled was awful: making MN 
know about it through a ML announcement, that such announcement was 
impersonal, corporate-style one of how you were doing the change to 
"facilitate the development of exciting new features" leaving everyone 
to read between the lines the real reasons (a large number of devs not 
being happy with the (former) project leadership). Using this tone is 
plain rude when dealing with people (ie, all the other devs) you have 
known for years.

Secondly, when people asked what would be the way conflicts would be 
resolved when two developers strongly disagree on a patch (for the three 
cases: committer+committer, committer+non-committer and 
non-committer+non-committer), there was no clear answer for this during 
_weeks_! There was also no guarantee that no committer would 
force-commit a patch when _he_ decided there was a "consensus" about it.

Third, knowing that the real reason of the change was having a 
hot-headed, non-compromising project leader, putting Mans in the double 
role of admin and committer is just sketchy. Also, in the committers 
team there was nobody whose views on technical questions was even 
slightly aligned to those of MN, which gives a "turf war" taste to the 
whole change.

Also, changing the project to a new name just a few hours after MN's 
takeover of the DNS looks a bit too fast for a decision that hurts the 
project as a whole (brand dilution, etc).

Finally, I do think the idea of the change would be a good one if you 
could make so that all the devs give in to it, or at least every dev but 
MN. But that was not really the case because of the mistakes above.

-- to Michael --

First of all, one thing I think is true for project leaders in general: 
people get fed up with them sooner or later. Given that you are not the 
most politically able person in the project, it took quite a long time 
and you should have seen this coming.

Also, I do not condone the DNS change. There was no project-wide 
consensus about that (not that was one for the maintainership change, 
but two wrongs do not make a right). I'm pretty sure that letting it 
settle for some time to see how it would end up would be better to the 
project as a whole.

For the rest of the criticism, I think you have already heard enough of it.

Finally, please give up being the project leader. I don't see any 
scenario where it happens and the project does not become a mess.


Finally, I think that no matter which parts "win" the project 
maintainership war, it would be a loss for the project as a whole. The 
"winner" will be seen more as the winner of a pissing contest than a 
legitimate project leader. The leadership question will come back over 
and over again and the project will remain flooded by flames.

What is IMHO a decent solution to this mess borrow one idea from Luca's 
last email:

1- The two trees, (git.videolan.org and git.ffmpeg.org) will keep 
existing, but none of them would be the project "official" tree

2- A third tree, managed by a small group of cool headed people (like 
the "mediators" in Luca's email), will cherry-pick patches from both 
trees. This will be our "official" tree, and nobody will commit to it, 
it will need to pass first through one of the other trees. Every patch 
that gets in both trees have to be cherry-picked per policy. That should 
correspond to the vast majority of the patches, but in some cases they 
will have to step in and end the controversy. People that maintains 
FFmpeg distro packages are also good candidates for this task.

3- Mediators decision is final and non-negotiable for a fixed amount of 
time (say, three months). If they ever had to take a decision, it 
already means that people discussed the patch in the mailing list and no 
consensus has reached (or else it would been committed to both trees).

4- Mediators decide in pvt so nobody can bikeshed about how the decision 
was made.

5- Do a 6-month long ban on controversial patches (ie, those _assured_ 
to cause a flamewar), so everybody can calm down.

Any opinions on this?


More information about the ffmpeg-devel mailing list