Re: GitHub organization discussed at FOSDEM

[Hisham <h@...>]

On 20 February 2015 at 16:38, Thijs Schreijer <thijs@thijsschreijer.nl> wrote:
>
>> -----Original Message-----
>> From: lua-l-bounces@lists.lua.org [mailto:lua-l-bounces@lists.lua.org] On
>> Behalf Of Pierre Chapuis
>> Sent: vrijdag 20 februari 2015 16:03
>> To: lua-l@lists.lua.org
>> Subject: GitHub organization discussed at FOSDEM
>>
>> Hello list,
>>
>> at FOSDEM we discussed the possibility of creating a GitHub organization
>> to maintain some Lua modules for which there should be several maintainers
>> (for instance LuaSocket, and probably some libraries from the Kepler
>> project).
>>
>> I wanted to float that idea here again, and maybe start moving on it.

Thanks for bringing it up!!

>> So I have a few questions:
>>
>> 1) How do you want to call this organization?
>> 2) Who wants to be part of it?
>> 3) Who wants to create it?
>
> Can't the existing Kepler organization [1] be used?
> Wouldn't that be easiest?

Yeah, I remember that suggestion from the fruitless
new-LuaForge/new-Lua-for-Windows discussions (which were long and
didn't produce results, so I'm willing to go to something way more
focused this time).

I brought up "Kepler" as a possibility at FOSDEM, but IIRC more than
one person mentioned that this name carried too much historical
baggage, and/or would be confusing, and/or brings the issue of
unmaintained Kepler modules getting in the mix.

> If one of the current owners [2] (Fabio, Andre or Hisham)
> could create a new team for LuaSocket and Diego would
> hand it over...
>
> Now who should be on that team? Diego obviously, but
> probably also someone else... looking at the contributors [3]
> there are some Lua commoners there. What is mostly needed
> I think is someone that can intelligently discuss PR's and
> merge them. Let the community do the development, if
> nobody codes a fix for an issue, the need isn't big enough.

What you described is quite in line with what we discussed at FOSDEM.
(I hope someone took pictures of the blackboard at the end of the
meeting :) )

When I mean going with something focused, I mean to avoid the old
Grand Plans we tried in the past that produce lots and lots of
discussions/opinions/bikeshedding and little results. So no trying to
fix the perennial issues of "batteries", "all-in-one packages",
"one-stop-shop for high quality modules", "set of modules made to work
for each other" or "blessed modules". (All of these are nice wishes,
but beyond the scope.)

What we came up with there then was, if my memory serves me right, the
following assessment of the situation:

* There are modules which have people who are interested in them, but
which don't have a clear maintenance status (are they maintained? by
who? is the 3-year old code in github lying there because it's
abandoned or is it because it "just works"? etc.)

* Diego says he wants to hand over maintenance of LuaSocket. A few
people offered to be part of a maintenance team, but no one stepped
forward "I'll be the one maintainer". Understandably, no one wants to
risk themselves to be in Diego's position (out of free time and with
the responsibility of Lua's most important module). The obvious
solution is to switch to a "maintenance team" model.

* The status of some Kepler projects is unclear. Since I have write
access to them all, I've been merging trivial/obviously-right pull
request that people send to them lately, but I'm not really a
maintainer to these projects. We need to figure out what to do with
them (if anything).

* Sometimes there are different people working on the same project but
in a uncoordinated/alternating manner. LuaExpat is an example of that.
I had the opportunity to meet Matthew at FOSDEM (and then have lunch
with Tomás after I returned to Rio) so it was nice to get both
perspectives. It's clear to me that collaboration can easily happen,
it's just a matter of having the mechanisms. (Also, we need to drag
Tomás to one of the Lua Workshops — he's one of the unsung heroes of
the Lua module world :) ).

So, the concrete proposal boiled down to this:

* We create an organization in GitHub (essentially a group account,
like http://github.com/keplerproject ) which will host some projects

* The criteria for hosting them in this org will be simple: it will
host modules being maintained by two or more maintainers. This way,
people can join, leave and the project keeps its continuity, without
people having to figure out which fork in GitHub is the maintained
one.

* If maintainers leave the project to the point that there's a single
remaining maintainer, the remaining one can ask for volunteers to join
in lua-l, and if no one comes up, the project is moved out of the org
and into the sole maintainer's account.

* This way, instead of a subjective selection of modules (some
maintained and some abandoned) like in Kepler, looking at the projects
in the org you'll have at least the guarantee that (a) they are not
abandoned, (b) there are at least two people who use/like/maintain
this project. (So it is to an extent a metric of curation, but a
concrete one.)

If I'm misrepresenting anything that was discussed in the BoF, please
do correct me! It's been weeks!

There are probably other practical details to decide (For example, how
long does it take until a project is abandoned? My suggestion is to do
a yearly review to check if maintainers are still involved/reachable.)

And then we get to Pierre's questions:

1) How do you want to call this organization?

As I told Justin over lunch after the BoF meeting, this is probably
the hardest question. :) luateam is obviously taken by the Lua Team,
luausers was suggested in the BoF might be confusing with the
lua-users.org wiki... luacommunity? I like the sound of it.

2) Who wants to be part of it?

I think it's a matter of seeing which projects we start moving, and
who volunteers to their maintenance teams. LuaSocket and LuaExpat
would be two obvious initial candidates.

I'm pretty swamped by the bunch of stuff that I already maintain so I
won't joining maintenance of any new projects, but since I have admin
access to the keplerproject org and we'll probably do some migrating,
I volunteer to perform general account administrivia (again, in the
spirit of the idea, we need at least one more admin, IMO preferrably
someone outside of LabLua — Pierre?).

3) Who wants to create it?

I can kickstart it and I can also move LuaRocks to this new org, as
its development is already done collectively (and in particular I'm
grateful to Thijs and Ignacio for allowing me to not have to take care
of the Windows parts!). Also, http://github.com/luacommunity/luarocks
is a cute URL. :)

-- Hisham