Re: [PATCH] Documentation/CommunityGuidelines

From: Ramkumar Ramachandra <artagnon@gmail.com>
Date: Jun 11, 2013, 08:56 UTC
Message-ID: <CALkWK0mWA_aWuWYYV8O5QKUMB_GWXRKB1X_TGuoG-ESgN5fjOg@mail.gmail.com>
In-Reply-To: <7vzjuxj21b.fsf@alter.siamese.dyndns.org>

Junio C Hamano wrote:

> The intent behind the document might be a noble one, but I am afraid
> that the text is too broad and vague and does not address the real
> issue to be of practical use.

Drafting something like this is shit work, which explains why nobody has attempted it yet. I have no intent of collecting feedback and doing iterations: it's going to be an extraordinarily hard and boring task; _much_ worse than any technical documentation.

Let me be clear that I have no hopes of landing this "patch": I just wanted to create a calm and rational atmosphere for people to discuss the problem, in the hopes of minimizing the chances of large frequent fires. If you think we should put _something_ in our tree, I suggest dumping a few raw emails from this thread into contrib/CommunityGuidelines/ (or something).

Show 10 quoted lines

> Taking one bullet point from the top for example:
>
>     0. You do not take offense, no matter what.  If someone attacks
>     you irrationally, you do not respond.  This is a public mailing
>     list, and we are all rational people: the attacker has already
>     humiliated herself in public, and everyone can see that.
>
> What does saying "we are all rational people" help when "the
> attacker" poses a risk to destroy the community?  What does "we are
> all rational people" even mean in this sentence?

I intended it as a way to reassure everyone that we will make unbiased, rational judgements to the extent possible.

Show 8 quoted lines

> It does not address the real cause of flamewars---why do rational
> people feel the need to respond when an irrational comment is made,
> e.g. when a reasonable review comments were responded not with
> either "Yeah, you are right, thanks." or "Not really, because you
> missed this case, I think..."  but with nitpicks with immaterial
> details or repetition without justification that takes account that
> the reviewer is in disagreement and there must be some reason behind
> it, i.e. a poisonous behaviour?

There is no great truth about some hidden "real cause" to be found. For instance, in the one we just had, I would argue that it "started" with your non-patch "administriva" email with a huge number of people marked in the initial CC. Disaster waiting to happen, if you ask me. I'm not "blaming" you, but the lesson to be learnt is: avoid non-patch emails, and CC conservatively; if you want to discuss some changes, send a patch. That would explain why this very email is disguised as a "[PATCH]", with exactly one person in the initial CC.

In short, the "reason" is a complex mix of various people's interactions under the current circumstances. Fires happen, and that is a fact; we can only look for common patterns and attempt to avoid fires by documenting these patterns as violations. Which is exactly what I have done (or attempted to do).

Show 7 quoted lines

> I suspect it mostly has to do with the desire to make sure that
> bystanders do not get an impression that the one who speaks last
> gives the conclusion to the discussion, so stating "The attacker
> being the one who speaks last in the discussion does not mean the
> conclusion is his." explicitly might be one way to make it more
> practically useful by alleviating the urge to respond, instead of
> saying "no matter what".

That is one pattern, but by no means the only one or even the "most important" one. I thought 0 was a nice generalization.