[FFmpeg-devel] [VOTE] drop support for using libav* compiled with mingw/cygwin in msvc

Måns Rullgård mans
Wed Feb 27 02:54:30 CET 2008


Michael Niedermayer <michaelni at gmx.at> writes:

> On Tue, Feb 26, 2008 at 11:29:15PM +0000, M?ns Rullg?rd wrote:
>> Diego Biurrun <diego at biurrun.de> writes:
>> 
>> > On Mon, Feb 25, 2008 at 01:59:45PM +0100, Michael Niedermayer wrote:
>> >> [...]
>> >
>> > Can we please have a bit more discussion before resorting to votes the
>> > next time around?
>
> Yes, sorry. This is the kind of thing that happens without a formal voting
> procedure ...
>
>> 
>> For the record, my decision to change the code had nothing to do with
>> any "voting" that may have taken place.
>
> Of course not, after all the vote said revert, you did not revert :)
>
>>  I do not believe in democracy.
>> 
>
>> To run a country, democracy is probably the best system devised thus
>> far, although it is by no means perfect.  This is because populations
>> at large have at most average intelligence, probably lower [1]. 
>
> So far i agree with you ...
>
>> In a
>> group like FFmpeg, the average intelligence is high enough that the
>> chances of reaching a consensus around a sensible decision are pretty
>> good. 
>
> We did not really reach a consensus, my oppinion is still that the
> original was best, and that the second best would be as it is
> currently but without the AV_VERSION(_INT) macros requireing the
> reader to search and open a second file.

A macro is necessary to combine the numbers and dots into a single
token.  The necessity of the LIBAV*_VERSION macros (not _INT) is of
course debatable, but that debate is orthogonal to this one.  A quick
google search show that people are using the macros, so simply
removing them would break other projects, which I believe we try to
avoid when possible.

> Its not something i will fight over but its not a consensus.
> If it were one, list the names of the people who do agree with it!

The main complaint I noticed concerned the breakage that was caused
with microsoft compilers.  I don't recall anyone but you opposing the
change as such.

If I understand you correctly, the change does not cause you
personally any difficulties, your objection being based on
the needs of hypothetical readers of the code.

Consensus may not be an accurate description of the situation, but
when you first proposed a vote, the alternatives were keeping the
msvc-breaking code (which I still quite like) and reverting to the
previous version (which you prefer), and the side in favour of
reverting was winning.  Had we simply reverted, the net gain/loss
would have been zero (obviously).  Instead, an alternative was found,
which most seem to see as a gain, and you (unless I am mistaken) do
not see as a major loss.  Thus, the final version would appear to be
overall preferable.

>> In a group of intelligent people it is also much more difficult
>> for a tyrant to gain foothold.
>
> I can not say that i agree or disagree here, just that i have some
> doubt. Also the word "intelligent" is a little vague, what is intelligence?
> I think that the word "educated" fits better.

Educated implies intelligent, while the converse is not true.  An
intelligent person is able to make rational decisions without being
educated.  Education is nothing but a convenience.

>> In other words, by resorting to democratic processes, an
>> intelligent group is likely to make worse decisions than had time
>> been allowed for proper discussion and consensus-forming.
>
> I cannot follow your reasoning here ...

When faced with a decision, it need not be immediately obvious to all
involved, due to factors such as differing background knowledge, which
alternative is the better.  If a vote is held, people are forced to
make a choice, even if they have insufficient information available to
make the best one (by whatever metric).  This essentially introduces a
random element in the decision making process.  A thorough discussion
allows people to consider aspects they may have been previously
unaware of, and entirely new solutions may be discovered.  If memory
serves, such things happen regularly around here.

The less information the voting population has, the more random the
outcome will be.  An extreme example is that of a modern country,
where people on average have limited access to accurate information,
while being bombarded with propaganda (the situation is worst among
members of parliament).  This results in almost completely random
decisions being made, and instead of the steady progress we like to
imagine our society making, we get something best described as
Brownian motion.  The situation is further worsened by the fact that
the population rarely shares a common goal.

At the other extreme is a pure dictatorship, where one person alone
makes all decisions.  In such a system, the dictator can cause rapid
changes in whatever direction he chooses.  In practise, the desires of
the dictator very rarely go beyond personal wealth, and quite often
also involve the misery of the population at large.  In the eyes of
the decision-maker, however, the system is highly efficient.

The FFmpeg project consists of intelligent people, all sharing the
same goal (that of making FFmpeg better), and nobody (I hope) is in it
for reasons of personal greed.  These are conditions in which voting
should only be used as a last resort, when other attempts at
reconciliation have failed.

>> Who wants to join me in founding the Kingdom of FFmpeg?
>
> We all do, but we lack land and the ability to defend it. without these
> 2 the local police would end our rule quite quickly. ;)

I was thinking Oded would be able to supply some weapons, and we have
a terrorist of our own as well. ;-)

-- 
M?ns Rullg?rd
mans at mansr.com




More information about the ffmpeg-devel mailing list