Re: Ann: let’s create better norms for ensuring all important libraries have several active maintainers

Sat Jun 6 00:26:46 UTC 2020

I've sometimes found it difficult to select people to help with
maintenance. Specifically, I really like to choose folks who've already
made significant contributions to the project, demonstrating familiarity
with at least some of its technical aspects. Ideally, they will also have
participated substantively in discussions of issues and pull requests,
pointing in helpful directions and offering constructive criticism.

All this means that we need to help more folks get more *involved* in
important libraries. What are the important barriers? One major challenge
is that as libraries mature, they can often get *more difficult* to dig
into and understand.

In the case of vector (which I'm only occasionally tangentially involved
in), I blame the fusion bundle mechanism and the framework supporting it.

In containers, most things are still reasonably approachable, but there are
some exceptions. IntMap and IntSet have never had quite enough comments to
explain their machinery. Major optimizations we've made in Data.Sequence
have made that always-tricky module quite intimidating. And the alterF
mechanism for Data.Map is not for the faint of heart.

In unordered-containers, Johan improved internal documentation
substantially a couple years ago, but it could still use some work.

In lens, well, everything. Ed targeted maximum flexibility, at the cost of
a bunch of barely-documented and sometimes ad hoc type classes, and types
barely anyone understands.

What is my point? I'm not sure. There's clearly some room to combine
software and human-factors engineering in library internals to make them
friendlier places for people to get to work.

On Fri, Jun 5, 2020, 11:55 AM Carter Schonwald <carter.schonwald at gmail.com>
wrote:

> I’m very happy that we now agree to have several active comaintainers for
> all core/systematically .  (though the specifics of how it was done this
> week i view as anthetical to what our community represents and values.)
>
> There’s ways of doing that can be done in a good faith way.  I’m
> definitely not perfect.  I also have spent much of the past five years
> trying to learn to manage my somewhat aggressive and personal flavor of
> generalized anxiety disorder (certain flavors of anxiety or depression in
> men present as aggressive behavior ).  Some of my approaches on tickets
> that people find frustrating are my attempt to avoid getting fixated on
> getting into Interpersonal conflicts rather than trying to shut those
> voices down. Also I find it hard to have those dialogues in that
> communication format.
>
> For asynchronous conversations in different time zones I fully welcome
> discord or what’s app or signal or freenode irc.  Conversations with
> emotional dimensions are challenging in email.  Let alone with issue
> trackers!
>
> In the past month I was working with Simon Jakobi and Andrew lelechenko as
> a sort of pilot for a new Haskell action team (HAT) with them as the
> initial leadership for 2020, to have some folks we all recognize to have
> excellent taste and engineering to help support and triage all maintainers
> and efforts. We spent a bit of time helping out on bytestring and I think
> that was a success. and I hoping they and other members of  HAT can make a
> big difference across all the important libraries we have. I look forward
> to support HAT via my role as CLC.
>
> More broadly, I was also privately in discussion with some folks before
> this week's tornado about how to move to make it the norm for all core /
> important libraries have 3-5 variously active and diversely different
> comaintainers. What happened this week is not how id have liked it to be
> rolled out, and I’m in the best of head spaces at the moment, but I think
> it is VERY good that we collectively agree there should be a norm of
> actively making sure there are several diverse co maintainers for every
> such library.
>
> how it was handled this week is not how it should have been handled, i
> pray that no one ever views this as grounds for similar treatment of people
> inour community who have been trying to be in their own way, good
> caretakers of important community resources
>
> In terms of what i like and appreciate about this community, I would like
> to challenge us all to think about how we can turn frustration into
> concrete and positively actionable feedback that supports each other in a
> good faith positive way. I feel that we all failed colectively, and thats
> fine, but i hope we can learn to engage in this differently.
>
> perhaps most importantly, i think its GENUINELY important to strive to
> support and enable every project to have a genuinely diverse maintainer set
> for all of our important libraries, both in terms of age, creed and
> whatever. Haskell has failed to do a good job of that in the past, and i
> challenge us all to work to address that starting with our core going
> forward
>
> On Fri, Jun 5, 2020 at 8:32 AM Mathieu Boespflug <m at tweag.io> wrote:
>
>> Zemyla,
>>
>> I would expect more charitable reception of anything said by anyone, but
>> especially when the original author stated clearly in the very same message
>> that they are not a native English speaker. And even if the words are to be
>> taken on face value, do note that no one has called anyone a tyrant.
>> Calling out specific behaviour (that predates the current pandemic) as
>> "tyrannous" (or "tyrannical") and naming someone a tyrant is simply not the
>> same thing.
>>
>>
>> On Fri, Jun 05, 2020 at 13:06:20, Zemyla <zemyla at gmail.com> wrote:
>>
>>> So you're saying that the fact that there's basically a civil war and
>>> pandemic going on is no excuse to respond to e-mails. And you're calling
>>> his behavior "tyrannous" when there are literal tyrants shooting tear gas
>>> at civilians.
>>>
>>> I gotta admit, this makes me disinclined to believe you're acting in
>>> good faith. Sometimes shit happens to good people, and someone who can't
>>> show a little compassion in these (pardon the cliche) unprecedented times
>>> is not someone I want anywhere involved in package management.
>>>
>>> _______________________________________________
> Libraries mailing list
> Libraries at haskell.org
> http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/libraries/attachments/20200605/a08889eb/attachment.html>