Stream: t-compiler/wg-meta

Topic: meeting 2019.02.21


nikomatsakis (Feb 21 2019 at 17:29, on Zulip):

so we have the meeting later today

nikomatsakis (Feb 21 2019 at 17:29, on Zulip):

let me review what @davidtwco added, first

nikomatsakis (Feb 21 2019 at 18:24, on Zulip):

OK I was looking things over and wondering about how to manage the prioritization bit

nikomatsakis (Feb 21 2019 at 18:25, on Zulip):

maybe making a spreadsheet or something would let people kind of leav notes

nikomatsakis (Feb 21 2019 at 18:34, on Zulip):

OK, so I created a google spreadsheet

nikomatsakis (Feb 21 2019 at 18:35, on Zulip):

I tried to create an entry for the major areas and to kind of break them down a bit:

nikomatsakis (Feb 21 2019 at 18:35, on Zulip):
nikomatsakis (Feb 21 2019 at 18:37, on Zulip):
nikomatsakis (Feb 21 2019 at 18:37, on Zulip):
nikomatsakis (Feb 21 2019 at 18:38, on Zulip):

Right now, that spreadsheet should be world editable.( I might try to tighten the later.)

nikomatsakis (Feb 21 2019 at 18:38, on Zulip):

Please feel free to add things that aren't covered

nikomatsakis (Feb 21 2019 at 18:38, on Zulip):

See y'all at 14:00!

davidtwco (Feb 21 2019 at 18:49, on Zulip):

Looks good to me.

davidtwco (Feb 21 2019 at 18:53, on Zulip):

Here's a link to the agenda so there's one in this topic.

nikomatsakis (Feb 21 2019 at 19:00, on Zulip):

Ping @davidtwco @centril @Santiago Pastorino @Cem Karan -- these are just names I culled from recent conversations -- Compiler Team Meta meeting starting soon.

nikomatsakis (Feb 21 2019 at 19:00, on Zulip):

Clearly, first order of business is creating a Zulip user group

Cem Karan (Feb 21 2019 at 19:01, on Zulip):

I'm here

nagisa (Feb 21 2019 at 19:01, on Zulip):

/me will read passively

nikomatsakis (Feb 21 2019 at 19:01, on Zulip):

I just created @WG-compiler-meta -- I added @davidtwco and myself -- if you want on, write me a privmsg and I'll get to it

nikomatsakis (Feb 21 2019 at 19:01, on Zulip):

I wonder if Zulip can make those "self-maintainable" somehow

nikomatsakis (Feb 21 2019 at 19:02, on Zulip):

So, welcome all! =)

nikomatsakis (Feb 21 2019 at 19:02, on Zulip):

As @davidtwco already posted, we have a DropBox paper where we collected some initial thoughts, which I also tried to collate a bit into a Google spreadsheet.

simulacrum (Feb 21 2019 at 19:02, on Zulip):

/me notes that a PR to https://github.com/rust-lang/team/ adding this WG is probably in order :)

nikomatsakis (Feb 21 2019 at 19:03, on Zulip):

Heh, that's very germane to one of the topics

nikomatsakis (Feb 21 2019 at 19:03, on Zulip):

Which is the "how to deal with all the sync overhead"

nikomatsakis (Feb 21 2019 at 19:03, on Zulip):

But yes, thanks :)

nikomatsakis (Feb 21 2019 at 19:03, on Zulip):
nikomatsakis (Feb 21 2019 at 19:03, on Zulip):

That was the rough agenda we had in the paper

nikomatsakis (Feb 21 2019 at 19:03, on Zulip):

I highlighted the "bootstrap process" because it seemed like the most urgent thing for us to discuss, since I think that is kind of what we are aiming to do right now -- get the compiler working groups into a functioning steady state

nikomatsakis (Feb 21 2019 at 19:04, on Zulip):

But maybe we can start a bit by talking about the "meta working group" (maybe not the best name...) and what exactly its role should be.

centril (Feb 21 2019 at 19:04, on Zulip):

(this meta wg is just for t-compiler, right? we have a different for lang?)

nikomatsakis (Feb 21 2019 at 19:05, on Zulip):

I had personally imagined it as a place for us to work out the details of what being a working group means, hence the name meta, but I've been wondering if it makes sense to generalize mildly to "governance" -- i.e., to generally take a look at how the compiler team runs itself and try to document and (where necessary) create procedure (but always taking that to the full team for consensus and discussion)

nikomatsakis (Feb 21 2019 at 19:05, on Zulip):

@centril yes I think this is specific to the compiler team, though I sort of want to make a similar effort for lang team, which I would expect to overlap in many ways

nikomatsakis (Feb 21 2019 at 19:06, on Zulip):

but obviously there are separate concerns in each case

Cem Karan (Feb 21 2019 at 19:06, on Zulip):

I personally think that the bootstraping process is what this group should concentrate on.

nikomatsakis (Feb 21 2019 at 19:07, on Zulip):

It seems like it's the highest priority regardless

Cem Karan (Feb 21 2019 at 19:07, on Zulip):

So, what do working groups always need?

nikomatsakis (Feb 21 2019 at 19:07, on Zulip):

Once that seems to be "working" we can certainly decide what to do next

nikomatsakis (Feb 21 2019 at 19:07, on Zulip):

Right now we have a kind of tentative list of working groups we came up with at the all hands

nikomatsakis (Feb 21 2019 at 19:07, on Zulip):

that i've been meaning to write a blog post about ;)

Cem Karan (Feb 21 2019 at 19:07, on Zulip):

And can we automate the creation of that (e.g., github, zulip, zoom?, etc.)

nikomatsakis (Feb 21 2019 at 19:08, on Zulip):

but which are documented in a dropbox paper ... somewhere

davidtwco (Feb 21 2019 at 19:08, on Zulip):

I've been working on the assumption that the latter was the purpose of the group - to handle the administrative side of the compiler team where that exists. Of course, number one on that list is the former - working out the details of a working group.

nikomatsakis (Feb 21 2019 at 19:08, on Zulip):

Dropbox paper from the Organizational meeting at all hands

Cem Karan (Feb 21 2019 at 19:08, on Zulip):

Actually, just keeping track of what working groups are active should be something that can be automated...

nikomatsakis (Feb 21 2019 at 19:08, on Zulip):

I feel like we shouldn't get too worked up on automation yet

nikomatsakis (Feb 21 2019 at 19:09, on Zulip):

though eventually it would definitely be nice

nikomatsakis (Feb 21 2019 at 19:09, on Zulip):

but to start having a list of the things is good

nikomatsakis (Feb 21 2019 at 19:09, on Zulip):

(Once we know what we want, we can also automate)

nikomatsakis (Feb 21 2019 at 19:09, on Zulip):

https://github.com/rust-lang/compiler-team/

nikomatsakis (Feb 21 2019 at 19:09, on Zulip):

So I've created this repository, and it seems like an initial thing we can work on is the "template" for a working group in there

nikomatsakis (Feb 21 2019 at 19:10, on Zulip):

Well, maybe we should start by just kind of brainstorming about what the .. questions are to be answered?

nikomatsakis (Feb 21 2019 at 19:10, on Zulip):

We had this in the dropbox paper:

davidtwco (Feb 21 2019 at 19:11, on Zulip):

I had the following questions trying to come from the perspective of a prospective contributor looking to find out more:

What working groups are available? What is working group X doing? Where can I find a working group? Who is in a working group? Are there meetings? Do I need to attend? Do I need any prior experience? Are there resources that would help me get up to speed? Are there labels on the issues for this working group? Who is leading the working group? What working groups might start in future? How can I register my interest in them?

nikomatsakis (Feb 21 2019 at 19:11, on Zulip):

I think another good question would be one of membership -- do we want to maintain lists of members? On what basis? etc. It may be something a WG can answer for itself, but it seems like we should have a default. Personally, I would like that to be someting that people can sign themselves up for

nikomatsakis (Feb 21 2019 at 19:12, on Zulip):

Yes, ok, that's a helpful way to look at it. Another thing I"ve been thinking about, which gets at @Cem Karan's question about automatiion, too:

What do WG members need to "do their job"?

For example, I think it is useful if we can assign issues to you, so you have to a member of the github org. I think you would want to be part of the Zulip User Group. Maybe some other things

Cem Karan (Feb 21 2019 at 19:13, on Zulip):

Membership will likely depend on the purpose of the particular WG; I'm not sure if a one-size-fits-all policy would work there.

nikomatsakis (Feb 21 2019 at 19:13, on Zulip):

I'm creating a section at the end of the paper to take a few notes and try to cull down into .. a set of deliverables, things to decide?

nikomatsakis (Feb 21 2019 at 19:13, on Zulip):

Membership will likely depend on the purpose of the particular WG; I'm not sure if a one-size-fits-all policy would work there.

Maybe. Regardless, I think we should have a default policy.

davidtwco (Feb 21 2019 at 19:13, on Zulip):

NLL had a stream here, GitHub group, Zulip user group and some GitHub labels but not really much else that I can remember?

nikomatsakis (Feb 21 2019 at 19:13, on Zulip):

But I suspect it will work better than you think

nikomatsakis (Feb 21 2019 at 19:13, on Zulip):

If membership should be limited, I think it will limit itself naturally

nikomatsakis (Feb 21 2019 at 19:14, on Zulip):

Regardless, I don't care to make rules, just helpful defaults

nikomatsakis (Feb 21 2019 at 19:14, on Zulip):

NLL had a stream here, GitHub group, Zulip user group and some GitHub labels but not really much else that I can remember?

Seems right.

Cem Karan (Feb 21 2019 at 19:14, on Zulip):

Got it, helpful defaults are good!

nikomatsakis (Feb 21 2019 at 19:14, on Zulip):

Issue labels are important

centril (Feb 21 2019 at 19:14, on Zulip):

Regardless, I don't care to make rules, just helpful defaults

That framing is good, useful also for T-Lang!

nikomatsakis (Feb 21 2019 at 19:15, on Zulip):

So, on the topic of GitHub Labels

nikomatsakis (Feb 21 2019 at 19:15, on Zulip):

NLL had WG-nll-foo stuff

davidtwco (Feb 21 2019 at 19:15, on Zulip):

I don't know there is much meaning to being a member of a working group. Being in the GitHub group was only to be able to get assigned to issues and, like the Zulip group, get pinged about new issues and the like.

nikomatsakis (Feb 21 2019 at 19:15, on Zulip):

we also had A-nll

nikomatsakis (Feb 21 2019 at 19:15, on Zulip):

it was confusing

nikomatsakis (Feb 21 2019 at 19:16, on Zulip):

I don't know there is much meaning to membership of a working group.

yeah so this is why I don't want to harp on it too much. I've found it is often very "fluid" -- people are involved for a bit, then have other things to do. Some folks stick around a lot, and that's fine.

nikomatsakis (Feb 21 2019 at 19:16, on Zulip):

Should we generally do A-wg-foo or something? What is the role of this label :)

nikomatsakis (Feb 21 2019 at 19:16, on Zulip):

I guess it will depend a bit on the WG

nikomatsakis (Feb 21 2019 at 19:16, on Zulip):

maybe just saying "WG can define labels" is fine for now

nikomatsakis (Feb 21 2019 at 19:17, on Zulip):

but it'd be useful for WGs to document the labels they are using

centril (Feb 21 2019 at 19:18, on Zulip):

@nikomatsakis feels like there's a distinction in purpose... A- labels are for areas, e.g. the release team will assign these as a matter of triage... whereas e.g. T-compiler and I-nominated say that X are working on things and X is important to the group

centril (Feb 21 2019 at 19:18, on Zulip):

A-nll is sorta like A-inference or A-closure

nikomatsakis (Feb 21 2019 at 19:18, on Zulip):

I had the following questions trying to come from the perspective of a prospective contributor looking to find out more:

Coming back to these questions, some of them seem to be things a working group can do, but others are basically part of our own infrastructure and how well we advertise it, right?

nikomatsakis (Feb 21 2019 at 19:19, on Zulip):

A-nll is sorta like A-inference or A-closure

Yeah, so A-foo is for "part of the code", and sometimes that aligns with the WG, but sometimes not

nikomatsakis (Feb 21 2019 at 19:19, on Zulip):

Is that what you mean?

davidtwco (Feb 21 2019 at 19:19, on Zulip):

A working group could probably just specify labels that represent the areas they focus on - eg. A-traits, A-NLL, A-const-eval - and they can create other labels around that to help them categorize or prioritize. As long as they are listed, I don't think we should be too prescriptive but just provide some guidelines.

nikomatsakis (Feb 21 2019 at 19:20, on Zulip):

So take a look at e.g. the Traits README

Cem Karan (Feb 21 2019 at 19:20, on Zulip):

How do they create new labels that make it easy to match to known working groups? A-trait-foo?

centril (Feb 21 2019 at 19:20, on Zulip):

@davidtwco ok, but that needs to be communicated clearly to the release-team then

nikomatsakis (Feb 21 2019 at 19:20, on Zulip):

(er, wait, let's finish up w/ labels maybe)

centril (Feb 21 2019 at 19:20, on Zulip):

things oftentimes have many A- labels too

nikomatsakis (Feb 21 2019 at 19:21, on Zulip):

I'm not sure what needs to be communicated to the release team exactly

nikomatsakis (Feb 21 2019 at 19:21, on Zulip):

Though I do think that some things probably do

centril (Feb 21 2019 at 19:21, on Zulip):

@nikomatsakis the release team does issue triage and labels things :slight_smile:

nikomatsakis (Feb 21 2019 at 19:21, on Zulip):

Especially if there are triage concerns, it would be helpful to have a standard way to bring something to the attention of a WG

nikomatsakis (Feb 21 2019 at 19:21, on Zulip):

Yes, right

davidtwco (Feb 21 2019 at 19:21, on Zulip):

@davidtwco ok, but that needs to be communicated clearly to the release-team then

I'm not suggesting that a working group create any new A- labels just that they can list the existing A- labels that best describe the parts of the compiler they are focusing on. Any new labels are things like NLL-* that only have meaning to the working group and are outside of the existing A- labels that have a meaning.

Cem Karan (Feb 21 2019 at 19:22, on Zulip):

Got it

nikomatsakis (Feb 21 2019 at 19:22, on Zulip):

So e.g. suppose there is some regression that relates to a new traits feature or whatever

centril (Feb 21 2019 at 19:22, on Zulip):

@davidtwco oh... I got the impression that each wg would get their own label

nikomatsakis (Feb 21 2019 at 19:22, on Zulip):

I imagine that release team would tag this with suitable A-labels (A-traits) as well as T-compiler and the regression labels

nikomatsakis (Feb 21 2019 at 19:23, on Zulip):

We could certainly "route" it to the working group as part of the normal compiler team triage meeting, if they don't notice it themselves earlier

nikomatsakis (Feb 21 2019 at 19:23, on Zulip):

You could also imagine some other technique

centril (Feb 21 2019 at 19:23, on Zulip):

@nikomatsakis something like that yeah; but other labels could easily apply too

davidtwco (Feb 21 2019 at 19:23, on Zulip):

@davidtwco oh... I got the impression that each wg would get their own label

You could, but I don't think it would be strictly necessarily - I imagine that most working groups could pick a existing A- label that would just be what they care about.

centril (Feb 21 2019 at 19:23, on Zulip):

like A-typesystem

nikomatsakis (Feb 21 2019 at 19:24, on Zulip):

My concern with each WG having a label is that .. well, it seemed annoying for NLL somehow, but maybe it was ok. There was always uncertainty about A-nll vs WG-nll

nikomatsakis (Feb 21 2019 at 19:24, on Zulip):

we had to have specific searches to kind of catch "union of things tagged A-nll and WG-nll"

nikomatsakis (Feb 21 2019 at 19:24, on Zulip):

Anyway, maybe this is in the weeds.

davidtwco (Feb 21 2019 at 19:24, on Zulip):

I think the only new labels a working group would need would be the equivalent of the NLL-* labels that help with their own triage.

nikomatsakis (Feb 21 2019 at 19:24, on Zulip):

It feels important but also like something I don't want to spend entire meeting discussing :)

nikomatsakis (Feb 21 2019 at 19:25, on Zulip):

Yeah, I'm sort of inclined to say that WGs should pick and list some A labels that they are monitoring for now. If there are GH teams too, then release can also cc said teams (though this just goes into a GH notification blackhole for me, but that's a separate problem)

nikomatsakis (Feb 21 2019 at 19:25, on Zulip):

It seemed to work ok for NLL that way tho

centril (Feb 21 2019 at 19:26, on Zulip):

(though this just goes into a GH notification blackhole for me, but that's a separate problem)

I think eddyb told me you watch the repo for all the issues and PRs... don't... do that :P

nikomatsakis (Feb 21 2019 at 19:26, on Zulip):

I don't do that

davidtwco (Feb 21 2019 at 19:26, on Zulip):

It seems like unresolved questions surrounding working groups come from one of two places - processes and practices involved in running a working group; and information that is desirable by prospective contributors.

nikomatsakis (Feb 21 2019 at 19:26, on Zulip):

I try to at least look at notifications with @nikomatsakis in them, but even that is really hard for me to do.

nikomatsakis (Feb 21 2019 at 19:27, on Zulip):

Yeah. So one thing I specifically want todo this week

nikomatsakis (Feb 21 2019 at 19:27, on Zulip):

is to go to each of the prospective WGs we have planned

nikomatsakis (Feb 21 2019 at 19:27, on Zulip):

find and identify leads

davidtwco (Feb 21 2019 at 19:27, on Zulip):

I don't think I've seen the first category of questions enumerated quite as much as the second category.

nikomatsakis (Feb 21 2019 at 19:27, on Zulip):

and have them fill out some template

nikomatsakis (Feb 21 2019 at 19:27, on Zulip):

so one thing I would sort of find useful is agreeing on what said templaet should be :)

nikomatsakis (Feb 21 2019 at 19:28, on Zulip):

(I sort of imagine that the "what WGs find useful" is also something we can kind of work on a bit as we go, i.e., learning from questions and problems that arise -- though we have some experience already)

nikomatsakis (Feb 21 2019 at 19:28, on Zulip):

I guess the question is whether there is info that would be useful which is not on e.g. this README

davidtwco (Feb 21 2019 at 19:28, on Zulip):

I think the questions I listed previously would form the basis for most of the template's questions?

Santiago Pastorino (Feb 21 2019 at 19:29, on Zulip):

sorry, something very important come up and I'm just arriving will read the log quickly

centril (Feb 21 2019 at 19:30, on Zulip):

(added question about labeling to release team's agenda for next wednesday)

nikomatsakis (Feb 21 2019 at 19:30, on Zulip):

The list @davidtwco added earlier:

## Team level

## WG level

nikomatsakis (Feb 21 2019 at 19:30, on Zulip):

The team level ones are things that seem like they are not part of the template but rather across wgs

davidtwco (Feb 21 2019 at 19:30, on Zulip):

I think that README is 90% of the way there. But it would be helpful to have a smaller summary too - i.e. a table with all the active working groups and that information for them.

nikomatsakis (Feb 21 2019 at 19:31, on Zulip):

Of the WG Level:

"Who is in a working group" is not answered really, but as we said, maybe we don't even want to answer it. But having at least some leads would be good.

nikomatsakis (Feb 21 2019 at 19:31, on Zulip):

I feel like it'd be ok to have more than leads, maybe a list like "Contact @so-and-so"

davidtwco (Feb 21 2019 at 19:32, on Zulip):

Of the WG Level:

"Who is in a working group" is not answered really, but as we said, maybe we don't even want to answer it. But having at least some leads would be good.

At this point, I'm leaning toward the idea that not having any sort of membership would be better. Define the leads of a working group and have a way for people to subscribe to relevant groups and notifications but otherwise there's no formal membership.

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

which can be updated periodically

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

so what I think I do want

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

is some way for people to "register themselves"

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

basically subscribing, yes

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

I'm not sure how to achieve that

nikomatsakis (Feb 21 2019 at 19:32, on Zulip):

it does feel like a place where automation is warranted though

davidtwco (Feb 21 2019 at 19:33, on Zulip):

The only benefits to membership I can see are:

nikomatsakis (Feb 21 2019 at 19:33, on Zulip):

@centril I remember there being talk of adopting e.g. a bot that let people leave comments to assign labels and things, right?

Pietro Albini (Feb 21 2019 at 19:33, on Zulip):

yep, and also to sort of assign themselves to issues

centril (Feb 21 2019 at 19:33, on Zulip):

@nikomatsakis ye, https://github.com/rust-lang/triagebot

Pietro Albini (Feb 21 2019 at 19:33, on Zulip):

(hi!)

davidtwco (Feb 21 2019 at 19:33, on Zulip):

@centril I remember there being talk of adopting e.g. a bot that let people leave comments to assign labels and things, right?

Zulip has a bot that adds people to a "contributors" group in their GitHub org and assigns them.

nikomatsakis (Feb 21 2019 at 19:33, on Zulip):

@davidtwco seems right, the second one in particular being the thing I was most interested in

nikomatsakis (Feb 21 2019 at 19:34, on Zulip):

So @Pietro Albini / @centril can that bot add people to the GH org?

Pietro Albini (Feb 21 2019 at 19:34, on Zulip):

no, for security reasons we would prefer not to do that

Pietro Albini (Feb 21 2019 at 19:34, on Zulip):

instead assigning the bot itself as a placeholder if the person is not on the org

nikomatsakis (Feb 21 2019 at 19:34, on Zulip):

hmm

nikomatsakis (Feb 21 2019 at 19:34, on Zulip):

that doesn't seem good enough

davidtwco (Feb 21 2019 at 19:34, on Zulip):

@davidtwco seems right, the second one in particular being the thing I was most interested in

This is somewhat watered down if anyone can just click a button to register their interest and become a member (as much as that is desirable in its own way).

nikomatsakis (Feb 21 2019 at 19:34, on Zulip):

e.g., if you'd like to be cc'd

nikomatsakis (Feb 21 2019 at 19:35, on Zulip):

@davidtwco right, I am giving up on that goal

nikomatsakis (Feb 21 2019 at 19:35, on Zulip):

basically I think I would rather 'recognize' regular people through journeyperson role

Pietro Albini (Feb 21 2019 at 19:35, on Zulip):

well, that's not for WG members

Pietro Albini (Feb 21 2019 at 19:35, on Zulip):

the bot is more for people seeing an E-easy issue and wanting to fix it

nikomatsakis (Feb 21 2019 at 19:35, on Zulip):

@Pietro Albini do you mean "the bot is not meant for that"

Pietro Albini (Feb 21 2019 at 19:36, on Zulip):

yep

nikomatsakis (Feb 21 2019 at 19:36, on Zulip):

yeah, seems fine

Pietro Albini (Feb 21 2019 at 19:36, on Zulip):

WGs can be managed with github and the team repo as usual

centril (Feb 21 2019 at 19:36, on Zulip):

@Pietro Albini does everyone who can assign also have push rights?

nikomatsakis (Feb 21 2019 at 19:36, on Zulip):

WGs can be managed with github and the team repo as usual

I don't know what this means

Pietro Albini (Feb 21 2019 at 19:36, on Zulip):

@centril yep, that's the problem

centril (Feb 21 2019 at 19:36, on Zulip):

@centril yep, that's the problem

but why... :scream:

nikomatsakis (Feb 21 2019 at 19:36, on Zulip):

we're sort of trying to define what the "as usual" even is...

Pietro Albini (Feb 21 2019 at 19:36, on Zulip):

@nikomatsakis adding people to the github team and to the rust-lang/team repo

Pietro Albini (Feb 21 2019 at 19:37, on Zulip):

hopefully in the near future adding them to the team repo will also add them to the org

centril (Feb 21 2019 at 19:37, on Zulip):

@Pietro Albini we could register people in a list manually and cc them each by username

nikomatsakis (Feb 21 2019 at 19:37, on Zulip):

I wanted people to be able to add themselves

centril (Feb 21 2019 at 19:37, on Zulip):

thereby not needing write access

nikomatsakis (Feb 21 2019 at 19:37, on Zulip):

in part because it's tedious and annoying

nikomatsakis (Feb 21 2019 at 19:38, on Zulip):

maybe it's not so important

centril (Feb 21 2019 at 19:38, on Zulip):

ok, we can let people add themselves through a bot

Cem Karan (Feb 21 2019 at 19:38, on Zulip):

@nikomatsakis It is tedious and annoying, but it does slow down bad actors a little.

Pietro Albini (Feb 21 2019 at 19:38, on Zulip):

I still don't like it from a perms point of view

davidtwco (Feb 21 2019 at 19:38, on Zulip):

WGs can be managed with github and the team repo as usual

In this scenario, someone would register their interest in a working group and then get a entry in the rust-lang/team repo so that they appear in that GitHub group (for the purpose of notifications)? Since we want a really easy way to let people get notified and subscribe themselves to a working group (a lightweight membership), this would end up with the "everyone is in the repo, security risk" scenario we want to avoid?

centril (Feb 21 2019 at 19:38, on Zulip):

@triagebot wg-add t-compiler/wg-traits

nikomatsakis (Feb 21 2019 at 19:38, on Zulip):

maybe we can pressure GH to let people make assignments

nikomatsakis (Feb 21 2019 at 19:38, on Zulip):

without write perms

nikomatsakis (Feb 21 2019 at 19:38, on Zulip):

you really need write perms to be assigned?

davidtwco (Feb 21 2019 at 19:38, on Zulip):

I think I've just enumerated the exact conversation that happened while I was typing. :face_palm:

Pietro Albini (Feb 21 2019 at 19:39, on Zulip):

yep

Pietro Albini (Feb 21 2019 at 19:39, on Zulip):

github shrugs

centril (Feb 21 2019 at 19:39, on Zulip):

@nikomatsakis you need write perms to label stuff too... :confused:

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

I don't care so much about labels

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

you can have a bot that adjusts labels

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

but yes, that'd be ncie

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

it's sort of secondary

centril (Feb 21 2019 at 19:39, on Zulip):

@nikomatsakis we can fake assignments by saying "Assigned to: @XYZ" in the top post of an issue

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

Yeah, we could do that

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

it's less nice

nikomatsakis (Feb 21 2019 at 19:39, on Zulip):

it doesn't show up in the issue listing

nikomatsakis (Feb 21 2019 at 19:40, on Zulip):

but it may be "good enough"

nikomatsakis (Feb 21 2019 at 19:40, on Zulip):

often, you are mostly just looking for "unassigned things"

Pietro Albini (Feb 21 2019 at 19:40, on Zulip):

that could be fixed by assigning the bot itself

nikomatsakis (Feb 21 2019 at 19:40, on Zulip):

right

nikomatsakis (Feb 21 2019 at 19:40, on Zulip):

I could certainly live with that

centril (Feb 21 2019 at 19:40, on Zulip):

who knows a lot of people @ github?

Santiago Pastorino (Feb 21 2019 at 19:41, on Zulip):

I know

simulacrum (Feb 21 2019 at 19:41, on Zulip):

We can assign people without giving them write permissions, for the record

davidtwco (Feb 21 2019 at 19:41, on Zulip):

I suppose we'd need to find a good way to notify a group of people too if we want to avoid adding everyone with an interest to the org?

simulacrum (Feb 21 2019 at 19:41, on Zulip):

They only need read permissions to be assigned

Pietro Albini (Feb 21 2019 at 19:41, on Zulip):

oh, TIL

simulacrum (Feb 21 2019 at 19:41, on Zulip):

(basically they need to be in the org)

nikomatsakis (Feb 21 2019 at 19:41, on Zulip):

that seems to change .. everything?

davidtwco (Feb 21 2019 at 19:41, on Zulip):

that seems to change .. everything?

I think so.

nikomatsakis (Feb 21 2019 at 19:42, on Zulip):

i.e., we could have a bot that lets you get added to some group

nikomatsakis (Feb 21 2019 at 19:42, on Zulip):

with read perms

Pietro Albini (Feb 21 2019 at 19:42, on Zulip):

@simulacrum sure about that?

nikomatsakis (Feb 21 2019 at 19:42, on Zulip):

so you can be targetd when somebody does @wg-foo and you can be assigned

simulacrum (Feb 21 2019 at 19:42, on Zulip):

@Pietro Albini 99% yeah, I think I've done it with a temp account

nikomatsakis (Feb 21 2019 at 19:42, on Zulip):

I feel like this is what I remember too

Cem Karan (Feb 21 2019 at 19:42, on Zulip):

@simulacrum would you be willing to test this out? Just to make 100% sure?

simulacrum (Feb 21 2019 at 19:42, on Zulip):

You just can't assign yourself if you don't have write permissions (but the bot can do it for you)

centril (Feb 21 2019 at 19:43, on Zulip):

@Santiago Pastorino can you ask GH to make it possible to label issues w/o write access? :P

Pietro Albini (Feb 21 2019 at 19:43, on Zulip):

I can't assign my bot account to a github issue on rust-lang/rust

Pietro Albini (Feb 21 2019 at 19:43, on Zulip):

and it is on the org

nikomatsakis (Feb 21 2019 at 19:43, on Zulip):

ok, enough on this, let's follow up later

Pietro Albini (Feb 21 2019 at 19:43, on Zulip):

at least not from the UI

nikomatsakis (Feb 21 2019 at 19:43, on Zulip):

or spin it out into a distinct topic ;)

Santiago Pastorino (Feb 21 2019 at 19:43, on Zulip):

@Santiago Pastorino can you ask GH to make it possible to label issues w/o write access? :P

yes :)

Santiago Pastorino (Feb 21 2019 at 19:43, on Zulip):

they may not listen ;)

davidtwco (Feb 21 2019 at 19:43, on Zulip):

So to summarize:

We want to avoid any sort of formal membership to a working group. Instead a wg has leads and interested folk can subscribe to it to receive notifications on GitHub issues and in Zulip.

Santiago Pastorino (Feb 21 2019 at 19:44, on Zulip):

or they probably already know and for some reason is not a priority

simulacrum (Feb 21 2019 at 19:44, on Zulip):

I will follow up after some more testing tonight/this weekend

Santiago Pastorino (Feb 21 2019 at 19:44, on Zulip):

that's what usually happens

centril (Feb 21 2019 at 19:44, on Zulip):

@simulacrum add to release agenda?

nikomatsakis (Feb 21 2019 at 19:45, on Zulip):

I'm looking back at our agenda

nikomatsakis (Feb 21 2019 at 19:45, on Zulip):
davidtwco (Feb 21 2019 at 19:46, on Zulip):

I think we've mostly worked out what we'd need to flesh out a wg template and then start some of them?

nikomatsakis (Feb 21 2019 at 19:46, on Zulip):

I feel... sort of like we've concluded that the WG template I had is pretty good, but we're missing:

nikomatsakis (Feb 21 2019 at 19:46, on Zulip):

Yeah. Something I was thinking about while we were talking also is that

nikomatsakis (Feb 21 2019 at 19:46, on Zulip):

I would sort of like to produce a "guide to leading a WG"

nikomatsakis (Feb 21 2019 at 19:47, on Zulip):

maybe I will try to dump out my current thoughts on this as a starting point

nikomatsakis (Feb 21 2019 at 19:47, on Zulip):

basically like "how to run an effective meeting" etc -- now if only I knew the answer :P

Cem Karan (Feb 21 2019 at 19:47, on Zulip):

@nikomatsakis I think that the table of working groups will need to be generated automatically to keep it up to date.

davidtwco (Feb 21 2019 at 19:47, on Zulip):

I would sort of like to produce a "guide to leading a WG"

In the rustc-guide or the compiler-team repo?

nikomatsakis (Feb 21 2019 at 19:47, on Zulip):

unclear

Cem Karan (Feb 21 2019 at 19:47, on Zulip):

One of the things that we haven't discussed yet is how to wind-down a working group

pnkfelix (Feb 21 2019 at 19:47, on Zulip):

sorry to jump in but I've been skimming

davidtwco (Feb 21 2019 at 19:47, on Zulip):

@nikomatsakis I think that the table of working groups will need to be generated automatically to keep it up to date.

Working groups won't change that much that this will be necessary, IMO.

nikomatsakis (Feb 21 2019 at 19:47, on Zulip):

@nikomatsakis I think that the table of working groups will need to be generated automatically to keep it up to date.

yes, I was thinking that the compiler-team repo should probably build into a gh-pages repository

nikomatsakis (Feb 21 2019 at 19:47, on Zulip):

However, I think we can start with a manually updated table

pnkfelix (Feb 21 2019 at 19:47, on Zulip):

an item similar to "Who is leading the working group?" is "who is the member who acts as liason with T-compiler"

nikomatsakis (Feb 21 2019 at 19:48, on Zulip):

Interesting

nikomatsakis (Feb 21 2019 at 19:48, on Zulip):

I agree

Santiago Pastorino (Feb 21 2019 at 19:48, on Zulip):

basically like "how to run an effective meeting" etc -- now if only I knew the answer :P

I think the most important thing is what do you need to do to lead

pnkfelix (Feb 21 2019 at 19:48, on Zulip):

or rather ... "similar to" as in "these two are often identical, perhaps"; but in case they are not identical, the person organizing the WG-checkin (presumably me, for now) needs to know who the main liason is

nikomatsakis (Feb 21 2019 at 19:48, on Zulip):

yeah; it seems similar to "who should I ping on zulip" -- which might be a fairly broad set of folks

nikomatsakis (Feb 21 2019 at 19:48, on Zulip):

maybe the template should have something like "contact points"

pnkfelix (Feb 21 2019 at 19:48, on Zulip):

well part of my hope was that it would not be broad

davidtwco (Feb 21 2019 at 19:48, on Zulip):

or rather ... "similar to" as in "these two are often identical, perhaps"; but in case they are not identical, the person organizing the WG-checkin needs to know who the main liason is

This might be something that we could get away with being "the leads of the wg, unless specified otherwise (special cases)".

nikomatsakis (Feb 21 2019 at 19:48, on Zulip):

and it can have more or less detail

pnkfelix (Feb 21 2019 at 19:48, on Zulip):

I want someone to blame take responsbility for presenting

nikomatsakis (Feb 21 2019 at 19:49, on Zulip):

well part of my hope was that it would not be broad

well, my point was more like "if I have questions, who might like to answer them", which is distinct from "compiler team liason"

nikomatsakis (Feb 21 2019 at 19:50, on Zulip):

I sort of suspect though that the latter is just "go to the right stream and post something"

Santiago Pastorino (Feb 21 2019 at 19:50, on Zulip):

there's two side of the things, wg members that need something from the leads and the guide @nikomatsakis was talking about may help leads to know what they are supposed to do

nikomatsakis (Feb 21 2019 at 19:50, on Zulip):

and those folks will answer ;)

nikomatsakis (Feb 21 2019 at 19:50, on Zulip):

Right, that's what I meant @Santiago Pastorino -- a kind of "guide to leading a WG"

Santiago Pastorino (Feb 21 2019 at 19:50, on Zulip):

yes

Santiago Pastorino (Feb 21 2019 at 19:50, on Zulip):

it's very important

davidtwco (Feb 21 2019 at 19:50, on Zulip):

The only other unresolved question I've seen asked here was @Cem Karan's "how to wind down a wg".

davidtwco (Feb 21 2019 at 19:51, on Zulip):

And I'm inclined to think we can cross that bridge when we come to it? I guess the main thing is what happens to the stream, GitHub group, labels, etc.

nikomatsakis (Feb 21 2019 at 19:52, on Zulip):

yeah

nikomatsakis (Feb 21 2019 at 19:52, on Zulip):

we are sort of "doing it" with NLL now

nikomatsakis (Feb 21 2019 at 19:52, on Zulip):

but we're not there yet

nikomatsakis (Feb 21 2019 at 19:53, on Zulip):

I prob have to run, this was a good meeting y'all -- I was hoping to leave with some more "action item" like things, but don't know that I know quite what they are yet

nikomatsakis (Feb 21 2019 at 19:53, on Zulip):

I am interested in distributing some of this organizational work around

nikomatsakis (Feb 21 2019 at 19:53, on Zulip):

I guess a clera one is:

nikomatsakis (Feb 21 2019 at 19:53, on Zulip):

can somebody make a "template" for WGs?

davidtwco (Feb 21 2019 at 19:53, on Zulip):

can somebody make a "template" for WGs?

I'm happy to work on this.

nikomatsakis (Feb 21 2019 at 19:53, on Zulip):

based on the existing files, but perhaps comb the logs and add a few things

nikomatsakis (Feb 21 2019 at 19:54, on Zulip):

thanks @davidtwco :heart: !

nikomatsakis (Feb 21 2019 at 19:54, on Zulip):

I am wondering -- does anybody here love jekyll hacking? :P Or something..I'd like to try and convert the compiler-team repository into something that builds into a GitHub pages

Santiago Pastorino (Feb 21 2019 at 19:54, on Zulip):

@davidtwco I can help with that too

nikomatsakis (Feb 21 2019 at 19:54, on Zulip):

this way we could do things like automate the table of working groups etc

nikomatsakis (Feb 21 2019 at 19:54, on Zulip):

fairly easily

nikomatsakis (Feb 21 2019 at 19:55, on Zulip):

that might be another "concrete goal" for this group to work on, though lower in priority

davidtwco (Feb 21 2019 at 19:55, on Zulip):

this way we could do things like automate the table of working groups etc

The table will still need manually updated, whether it is rendered by GitHub as a readme or on a Jekyll page?

davidtwco (Feb 21 2019 at 19:56, on Zulip):

I guess the next step after a template is to work out what the working groups will be and who will lead them?

Cem Karan (Feb 21 2019 at 19:56, on Zulip):

Can Jekyll be hacked to look through all of the organizations repos? If each WG had it's own repo with certain files in place at the root, then this can be semi-automated (e.g., WG.json at the root, with some schema)

Santiago Pastorino (Feb 21 2019 at 19:56, on Zulip):

I guess the automation you might want is to be able to run a command to add somebody and then make a commit and call it a day?

davidtwco (Feb 21 2019 at 19:56, on Zulip):

I guess the next step after a template is to work out what the working groups will be and who will lead them?

Building on the discussions from the All Hands.

davidtwco (Feb 21 2019 at 19:58, on Zulip):

I guess the automation you might want is to be able to run a command to add somebody and then make a commit and call it a day?

I'm not sure who would be being added. We will probably have something like this for the GitHub groups and stuff like that. But the actual working group listing itself will only mention the leads (presumably) and that sort of information won't change all that much.

Santiago Pastorino (Feb 21 2019 at 19:59, on Zulip):

yeah sorry, I got it wrong, anyway, unsure if Jekyll is really needed

davidtwco (Feb 21 2019 at 20:00, on Zulip):

My understanding is that automation will be used to allow people to subscribe themselves to a working group so they can get notifications. But the readme discussing what the working group does, where they meet, etc and the list of active working groups will change so infrequently that I think automation of that wouldn't be worthwhile. Just my 2c.

nikomatsakis (Feb 21 2019 at 21:12, on Zulip):

Hey @centril @Cem Karan @Santiago Pastorino -- I added you to the @WG-compiler-meta user group. (Did you want to be added, @Pietro Albini?)

nikomatsakis (Feb 21 2019 at 21:12, on Zulip):

Also, @WG-compiler-meta, I realize we didn't say explicitly? Should we have another meeting next week, same time, same place?

nikomatsakis (Feb 21 2019 at 21:12, on Zulip):

I'm going to assume yes. I made the calendar event repeat.

Pietro Albini (Feb 21 2019 at 21:12, on Zulip):

why not, mostly for the infra around memberships

Last update: Nov 11 2019 at 22:10UTC