llvm dev - Oct 2015 - RFC: Introducing an LLVM Community Code of Conduct

On 14 October 2015 at 14:32, Krzysztof Parzyszek via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> CoC is not going to stop or prevent any harassment.
True.
> The current administrators of the mailing
> lists, or the organizers of the LLVM conferences already have the means to
> deal with it.
Not really.

We haven't had to ban anyone from the list or kick anyone from a
conference yet, but without some form of written agreement, it'd be
hard to get to any concrete solution if it were to happen.

I'm not saying the CoC is perfect, or is the best tool for the job,
but it's *a* tool, IFF there is agreement in the community about it,
and IFF the whole process is democratic, including future changes.
Like constitutions trump laws, I believe community pressure should
always trump committees and foundations.

cheers,
--renato

On 10/14/2015 8:44 AM, Renato Golin wrote:
> On 14 October 2015 at 14:32, Krzysztof Parzyszek via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
>
>> The current administrators of the mailing
>> lists, or the organizers of the LLVM conferences already have the means
to
>> deal with it.
>
> Not really.
>
> We haven't had to ban anyone from the list or kick anyone from a
> conference yet, but without some form of written agreement, it'd be
> hard to get to any concrete solution if it were to happen.
There is no assumed "right to participate" in the mailing lists, at 
least that's how I see it.  The administrators are not under obligation 
to allow access to everyone.  I agree that having a written CoC would 
make it easier to justify exclusion, but banning someone should not be 
the first remedial action anyway.  It would normally be preceded by some 
sort of communication from the administrators, at which point the nature 
of the offence would be clarified, and only repeated behavior would lead 
to a ban.  This can be done today without any special documents.  This 
relies on the "benevolent authority" of the administrators, which is
not
always granted (although I think this list does not suffer from that). 
On the other hand, a codified CoC could encourage loophole-seeking and 
alternative interpretations of certain passages on the part of the 
intentional offenders.  Ultimately, the administrators would have the 
last word, which is not all that different from the no-CoC approach.

Another problem with a CoC is that if it puts too much stress on "not 
offending" anyone, some people may avoid voicing certain (legitimate) 
thoughts out of fear of being considered offensive.


A conference is a completely different thing, as it is a subject to 
actual law (contract, etc.).  A written CoC for a conference would be a 
good idea, and agreeing to it could be a requirement for participation. 
  The difference there is that this agreement would be legally binding, 
and allow for legal remedies for violations.


-Krzysztof

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, 
hosted by The Linux Foundation

On 14 October 2015 at 14:59, Krzysztof Parzyszek
<kparzysz at codeaurora.org> wrote:
> This can be done today without any special documents.  This relies on the
"benevolent
> authority" of the administrators, which is not always granted
(although I
> think this list does not suffer from that).
For electronic communications, I agree.

> Another problem with a CoC is that if it puts too much stress on "not
> offending" anyone, some people may avoid voicing certain (legitimate)
> thoughts out of fear of being considered offensive.
My strong reaction to the CoC was originally sparked by that feeling.
I'm sure there are a lot of people like me in this list that feel the
same way and won't voice their concerns. Imposing a strong stance for
the offender *will* marginalise the unintentional offenders. If the
idea of the CoC is to not marginalise anyone, we have to be careful.

As I said earlier, if the intent is to "mediate" not
"punish", and the
people mediating it understand this conundrum, we should be fine.

> A conference is a completely different thing, as it is a subject to actual
> law (contract, etc.).  A written CoC for a conference would be a good idea,
> and agreeing to it could be a requirement for participation.  The
difference
> there is that this agreement would be legally binding, and allow for legal
> remedies for violations.
Agreed.

--renato

On Oct 14, 2015, at 6:44 AM, Renato Golin via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
> We haven't had to ban anyone from the list or kick anyone from a
> conference yet, but without some form of written agreement, it'd be
> hard to get to any concrete solution if it were to happen.
This is not true.  We have completely ejected a hostile member from the
community in the past - it just happened a long time ago (sometime around 2007).
This was after trying, very hard, to get them to work in a productive and
professional way.  This was a hard thing for everyone involved, because the
person was an active contributor - but was treating many people in an
unacceptable way.  While it was a hard decision, I am *very* glad that this
happened, because I highly value the LLVM community and how we all work together
as a team.

Back then, the LLVM community was much smaller, and it was relatively easy to
make a unilateral decision to do this.  Today, doing such a thing would be much
harder, because the scale of the project and the number of people involved. 
While I hope we never have a situation like this again, it is clearly possible,
and we as a community need to be prepared for it.

-Chris

On 14 October 2015 at 16:16, Chris Lattner <clattner at apple.com>
wrote:
> This is not true.  We have completely ejected a hostile member from the
community in the past - it just happened a long time ago (sometime around 2007).
This was after trying, very hard, to get them to work in a productive and
professional way.  This was a hard thing for everyone involved, because the
person was an active contributor - but was treating many people in an
unacceptable way.  While it was a hard decision, I am *very* glad that this
happened, because I highly value the LLVM community and how we all work together
as a team.
I stand corrected. This was before my time. :)

> Back then, the LLVM community was much smaller, and it was relatively easy
to make a unilateral decision to do this.
Doesn't sound unilateral to me. With or without CoC, trying hard and
involving the affected people (including the offender's point of view)
is the way to go.

cheers,
--renato