On Wed, 14 Mar 2007, Grant Goodyear wrote:
> Underlying the draft code of conduct is an assumption that aggressive
> and less-than-nice behavior on gentoo-dev is seriously harming Gentoo.
> On the other hand, LKML is famous for its flamewars, and nobody claims
> that Linux is in serious trouble. Does anybody have a good feeling for
> where the difference lies? Are we sure that we're solving the right
> problem? (That's not a rhetorical question; I really don't know the
> answer.)
One thing that matters a lot is that development mostly goes through the
list, or is at least reported on it. Even if there's a huge flamewar going
on, the infiniband patch dump is still more messages. Sure, most people
don't actually read the infiniband patch sets, but there's this constant
sense that work is getting done.
Tied into that is the sense that you can win flamewars by posting code
that's just so good that people can't help but accept it. Or benchmarks
showing that your version is significantly faster on important things. Or
that you can eliminate a common class of bugs. Or that you can make a lot
of code simpler. But, in any case, that flamewars are ultimately decided
on the technical merit of the code, as judged sort of objectively by
people widely accepted to have good taste.
Conversely, if somebody's getting work done and posting the results, and
lots of people are flaming that person, it ultimately doesn't matter, if
nobody else produces an alternative. If nobody can say how the code should
be instead of how it's been written, it eventually wins. And people are
therefore forced to be productive instead of just flaming.
With all the development, it's so high traffic that someone being really
annoying can be ignored with no more difficulty than the piles of messages
that just aren't interesting to a particular subscriber. People fight
about the issues as long as the conversation stays on the issues, but
personal attacks are generally mostly ignored, because they're low-value
in how flamewars are scored.
People are willing to be at odds on one issue while working together on
another issue. Adrian Bunk is maintaining 2.6.16.x, and there have been
ongoing flamewars over his patch acceptance policy for that series; at the
same time, he maintains the list of regressions in the latest development
kernel, and the same people are really grateful for his work in following
up on bug reports and getting the right issues to the right people's
attention.
It comes down to this: everybody sees what you post. If you post mostly
useful stuff, you're taken seriously. If you post mostly flames, you're
not. If you're taken seriously, your critiques of other people's code
matter more. If you want to be taken really seriously, you have to make
constructive comments on other people's code. If your comments are
sufficiently technically correct, you can flame people and not be ignored.
-Daniel
*This .sig left intentionally blank*
--
[email protected] mailing list
