Stream: t-compiler/wg-meta

Topic: meeting areas of the compiler


Santiago Pastorino (May 07 2020 at 17:59, on Zulip):

hey @nikomatsakis pinging you here

Santiago Pastorino (May 07 2020 at 18:00, on Zulip):

it wasn't clear to me if we should have the meeting today or we should move it to next week

Santiago Pastorino (May 07 2020 at 18:00, on Zulip):

we are supposed to be starting right now and we had scheduled 1h

nikomatsakis (May 07 2020 at 18:01, on Zulip):

hello

nikomatsakis (May 07 2020 at 18:01, on Zulip):

good question!

nikomatsakis (May 07 2020 at 18:01, on Zulip):

we've been trying to do this forever and it's always sort of "low-ish" priority

nikomatsakis (May 07 2020 at 18:01, on Zulip):

let's try to take a stab at it I say

nikomatsakis (May 07 2020 at 18:01, on Zulip):

I was thinking about how to start though

nikomatsakis (May 07 2020 at 18:02, on Zulip):

you'll have to send me the hackmds or whatever that you had so far

Santiago Pastorino (May 07 2020 at 18:02, on Zulip):

sure

nikomatsakis (May 07 2020 at 18:02, on Zulip):

I wanted to back-up -- did we agree a good idea to try and start with a smaller number of areas, perhaps non-exhaustive?

Santiago Pastorino (May 07 2020 at 18:02, on Zulip):

https://hackmd.io/jdF3GQCkQRaR7VsmkigFag

nikomatsakis (May 07 2020 at 18:03, on Zulip):

thanks!

Santiago Pastorino (May 07 2020 at 18:03, on Zulip):

I think we should base this on specific needs

Santiago Pastorino (May 07 2020 at 18:03, on Zulip):

for instance, if you look at this from the issues perspective, today I was seeing a bunch of incr comp issues

nikomatsakis (May 07 2020 at 18:04, on Zulip):

hmm

Santiago Pastorino (May 07 2020 at 18:04, on Zulip):

and I guess we should know who are the knows that know about this, we should have a github group and be able to organize with them how to tackle those

nikomatsakis (May 07 2020 at 18:04, on Zulip):

so I was starting at the top a list of "areas" that seem to me to be logically related

Santiago Pastorino (May 07 2020 at 18:04, on Zulip):

probably incr comp is a different problem which may need a wg already :)

nikomatsakis (May 07 2020 at 18:04, on Zulip):

but it loses some precision that way I guess

nikomatsakis (May 07 2020 at 18:04, on Zulip):

but I think that's probably ok

nikomatsakis (May 07 2020 at 18:04, on Zulip):

we're not that many people

Santiago Pastorino (May 07 2020 at 18:05, on Zulip):

yeah I didn't know how to call the thing you called "plumbing" :slight_smile:

Santiago Pastorino (May 07 2020 at 18:06, on Zulip):

I'm watching you editing the doc live :)

nikomatsakis (May 07 2020 at 18:07, on Zulip):

I'm looking through the existing map

nikomatsakis (May 07 2020 at 18:07, on Zulip):

these distinctions are meaningful but so fine-grained

Santiago Pastorino (May 07 2020 at 18:07, on Zulip):

yeah

nikomatsakis (May 07 2020 at 18:07, on Zulip):

it might be useful to have a two-level partitioning

nikomatsakis (May 07 2020 at 18:08, on Zulip):

i.e., it's true that if you have a question about exhaustiveness code, I'm not the best one to answer it

nikomatsakis (May 07 2020 at 18:08, on Zulip):

even though I would consider that fairly clearly in the "type system" area

nikomatsakis (May 07 2020 at 18:08, on Zulip):

and I feel like I'm not the worst person to ask

nikomatsakis (May 07 2020 at 18:08, on Zulip):

anyway

nikomatsakis (May 07 2020 at 18:08, on Zulip):

I think it's better to be too coarse

nikomatsakis (May 07 2020 at 18:08, on Zulip):

and to have to refine

nikomatsakis (May 07 2020 at 18:08, on Zulip):

than the opposite :)

Santiago Pastorino (May 07 2020 at 18:08, on Zulip):

yes

Santiago Pastorino (May 07 2020 at 18:09, on Zulip):

my feeling is that the top level areas you're using would work

Santiago Pastorino (May 07 2020 at 18:09, on Zulip):

BUT ...

Santiago Pastorino (May 07 2020 at 18:09, on Zulip):

unsure about Type System and Plumbing

Santiago Pastorino (May 07 2020 at 18:09, on Zulip):

the rest :+1:

nikomatsakis (May 07 2020 at 18:09, on Zulip):

if you look

nikomatsakis (May 07 2020 at 18:10, on Zulip):

the people who know the query system well and incremental

nikomatsakis (May 07 2020 at 18:10, on Zulip):

are basically the same people

nikomatsakis (May 07 2020 at 18:10, on Zulip):

the two are sort of intimiately connected, even if they're distinct

nikomatsakis (May 07 2020 at 18:10, on Zulip):

metadata...somewhat less so :)

Santiago Pastorino (May 07 2020 at 18:10, on Zulip):

in particular in Type System unsure if I'd join borrowck and trait system

nikomatsakis (May 07 2020 at 18:10, on Zulip):

though I think we'd like that to be true

nikomatsakis (May 07 2020 at 18:10, on Zulip):

yeah, type system might be too broad

Santiago Pastorino (May 07 2020 at 18:10, on Zulip):

well let's be plumbing then :)

Santiago Pastorino (May 07 2020 at 18:11, on Zulip):

unsure if the name is one you would use :)

nikomatsakis (May 07 2020 at 18:11, on Zulip):

I would ;)

Santiago Pastorino (May 07 2020 at 18:11, on Zulip):

great then :)

nikomatsakis (May 07 2020 at 18:13, on Zulip):

I wonder if it's worth breaking out the trait system

nikomatsakis (May 07 2020 at 18:13, on Zulip):

probably

Santiago Pastorino (May 07 2020 at 18:13, on Zulip):

interested in knowing why Type Layout is where it is

nikomatsakis (May 07 2020 at 18:13, on Zulip):

you mean in the "hmm" category? :)

nikomatsakis (May 07 2020 at 18:13, on Zulip):

that last list is stuff that I don't know where to fit

Santiago Pastorino (May 07 2020 at 18:13, on Zulip):

I liked the type checker / borrow checker split you've done

nikomatsakis (May 07 2020 at 18:13, on Zulip):

type layout could be codegen

nikomatsakis (May 07 2020 at 18:13, on Zulip):

or could be its own thing

nikomatsakis (May 07 2020 at 18:13, on Zulip):

I think codegen is probably right though

Santiago Pastorino (May 07 2020 at 18:13, on Zulip):

:+1:

nikomatsakis (May 07 2020 at 18:13, on Zulip):

it interacts with ABI etc pretty deeply

nikomatsakis (May 07 2020 at 18:14, on Zulip):

I don't know where "generators" goes

nikomatsakis (May 07 2020 at 18:15, on Zulip):

maybe in MIR

nikomatsakis (May 07 2020 at 18:15, on Zulip):

or maybe it's a separate thing, but it doesn't quite seem big enough

nikomatsakis (May 07 2020 at 18:15, on Zulip):

I think MIR has a number of these 'larger' items

nikomatsakis (May 07 2020 at 18:15, on Zulip):

e.g. drop elaboration, which might actually be sort of borrow check

Santiago Pastorino (May 07 2020 at 18:15, on Zulip):

so I'd say that if we consider something separate but doesn't justify for now, I'd wait to add those

nikomatsakis (May 07 2020 at 18:15, on Zulip):

I'm moving drop elaboration anyway

nikomatsakis (May 07 2020 at 18:15, on Zulip):

I'm going to put generators under MIR

nikomatsakis (May 07 2020 at 18:15, on Zulip):

I think the right structure here is probably:

nikomatsakis (May 07 2020 at 18:15, on Zulip):

big areas

nikomatsakis (May 07 2020 at 18:16, on Zulip):

and then a (non-exhaustive) list of "subprojects" with some specific names

Santiago Pastorino (May 07 2020 at 18:16, on Zulip):

:+1:

nikomatsakis (May 07 2020 at 18:16, on Zulip):

i.e., "oh but if you're question is generators, probably ping tmandry"

Santiago Pastorino (May 07 2020 at 18:16, on Zulip):

yes

Santiago Pastorino (May 07 2020 at 18:16, on Zulip):

so that's the main question

Santiago Pastorino (May 07 2020 at 18:16, on Zulip):

what do we want this for?

nikomatsakis (May 07 2020 at 18:16, on Zulip):

ha a very good question

nikomatsakis (May 07 2020 at 18:16, on Zulip):

so

nikomatsakis (May 07 2020 at 18:16, on Zulip):

I think we want:

Santiago Pastorino (May 07 2020 at 18:16, on Zulip):

there's 2 use cases in my opinion

nikomatsakis (May 07 2020 at 18:16, on Zulip):
nikomatsakis (May 07 2020 at 18:17, on Zulip):
nikomatsakis (May 07 2020 at 18:17, on Zulip):

i.e., i'm making an MCP, now I have some idea who might be interested

nikomatsakis (May 07 2020 at 18:17, on Zulip):

or I need advice on an MCP :)

nikomatsakis (May 07 2020 at 18:17, on Zulip):
nikomatsakis (May 07 2020 at 18:17, on Zulip):

Santiago Pastorino said:

there's 2 use cases in my opinion

sorry, let me hear your version :)

Santiago Pastorino (May 07 2020 at 18:17, on Zulip):

you kind of included those

Santiago Pastorino (May 07 2020 at 18:17, on Zulip):

but let me add something

Santiago Pastorino (May 07 2020 at 18:18, on Zulip):

nikomatsakis said:

I'd go a bit deeper here

Santiago Pastorino (May 07 2020 at 18:18, on Zulip):

I'd integrate this idea into rust-lang/team

Santiago Pastorino (May 07 2020 at 18:18, on Zulip):

we should have github groups for the areas with the maintainers included there

nikomatsakis (May 07 2020 at 18:18, on Zulip):

yeah

Santiago Pastorino (May 07 2020 at 18:18, on Zulip):

and also

nikomatsakis (May 07 2020 at 18:18, on Zulip):

and I think github groups correspond to the "top-level" bullets, right?

nikomatsakis (May 07 2020 at 18:18, on Zulip):

not the "subgroups"

Santiago Pastorino (May 07 2020 at 18:19, on Zulip):

once you label somethign with ... let's say MIR, the MIR github group is pinged

nikomatsakis (May 07 2020 at 18:19, on Zulip):

nikomatsakis said:

not the "subgroups"

I guess this is not obvious, the main thing is that not everything merits a "subgroup"

Santiago Pastorino (May 07 2020 at 18:19, on Zulip):

yeah at least I'd start with the main group

Santiago Pastorino (May 07 2020 at 18:20, on Zulip):

continuing with the MIR example

Santiago Pastorino (May 07 2020 at 18:20, on Zulip):

we can just have MIR

Santiago Pastorino (May 07 2020 at 18:20, on Zulip):

but if at some point we see the need of spliting we can have MIR but also MIR-building, MIR-generators and so on when specificity is needed

nikomatsakis (May 07 2020 at 18:21, on Zulip):

right

nikomatsakis (May 07 2020 at 18:21, on Zulip):

definitely less is more to start

Santiago Pastorino (May 07 2020 at 18:21, on Zulip):

yes yes yes :)

Santiago Pastorino (May 07 2020 at 18:21, on Zulip):

this is going to collide a bit with some wgs setups

Santiago Pastorino (May 07 2020 at 18:21, on Zulip):

like we have MIR-opt github group and things like that

Santiago Pastorino (May 07 2020 at 18:22, on Zulip):

unsure if it worth really, I'd remove and prefer this idea and just call it MIR

nikomatsakis (May 07 2020 at 18:22, on Zulip):

I think the collisions mostly indicate working groups that should be retired

nikomatsakis (May 07 2020 at 18:22, on Zulip):

in favor of this system

Santiago Pastorino (May 07 2020 at 18:22, on Zulip):

exactly

Santiago Pastorino (May 07 2020 at 18:22, on Zulip):

agree completely

nikomatsakis (May 07 2020 at 18:22, on Zulip):

I would probably remove all "working groups" and replace them "project groups" that have defined focus, or an "area" like this

Santiago Pastorino (May 07 2020 at 18:22, on Zulip):

would we need to do an RFC or something like that for this?

nikomatsakis (May 07 2020 at 18:22, on Zulip):

no, but it is probably good to make a meeting proposal

nikomatsakis (May 07 2020 at 18:22, on Zulip):

steering meeting

Santiago Pastorino (May 07 2020 at 18:23, on Zulip):

:+1:

nikomatsakis (May 07 2020 at 18:23, on Zulip):

I mean the working groups were just something I dreamed up, no RFC :P

nikomatsakis (May 07 2020 at 18:23, on Zulip):

we should open the proposal now

nikomatsakis (May 07 2020 at 18:23, on Zulip):

it'd be good to dress up that hackmd with a bit more details

nikomatsakis (May 07 2020 at 18:23, on Zulip):

an interesting question

nikomatsakis (May 07 2020 at 18:24, on Zulip):

is what happens to #t-compiler/wg-prioritization and #t-compiler/wg-meta

nikomatsakis (May 07 2020 at 18:24, on Zulip):

these two groups are somewhat distinct

nikomatsakis (May 07 2020 at 18:24, on Zulip):

the former I think is a good example of a "standing committee" that isn't expected to ever be "done"

nikomatsakis (May 07 2020 at 18:24, on Zulip):

it could be considered an "area" actually

nikomatsakis (May 07 2020 at 18:24, on Zulip):

sort of a "meta-area" :)

Santiago Pastorino (May 07 2020 at 18:24, on Zulip):

I see what you mean

Santiago Pastorino (May 07 2020 at 18:24, on Zulip):

yes

nikomatsakis (May 07 2020 at 18:25, on Zulip):

to some extent I feel the same about wg-meta, actually, but it's a bit different

nikomatsakis (May 07 2020 at 18:25, on Zulip):

that is, it's not like we have some 'end goal' that we're pursuing exactly

nikomatsakis (May 07 2020 at 18:25, on Zulip):

we're just tweaking the way the group runs

nikomatsakis (May 07 2020 at 18:25, on Zulip):

maybe there is a 'meta area' and prioritization is a subbullet, along with 'procedures', I don't know

nikomatsakis (May 07 2020 at 18:25, on Zulip):

but maybe we don't have to answer this question yet

Santiago Pastorino (May 07 2020 at 18:26, on Zulip):

this group seems to me like rust compiler organizational work but in a very tiny way

nikomatsakis (May 07 2020 at 18:26, on Zulip):

define "thin"

Santiago Pastorino (May 07 2020 at 18:27, on Zulip):

I meant, there are zillions of things that you and Felix do that do not run throw this group

nikomatsakis (May 07 2020 at 18:27, on Zulip):

I see

Santiago Pastorino (May 07 2020 at 18:27, on Zulip):

and organizational work is way huger than what we do here

nikomatsakis (May 07 2020 at 18:27, on Zulip):

not sure about zillions but

Santiago Pastorino (May 07 2020 at 18:27, on Zulip):

it's just sprinkles of organizational work :)

Santiago Pastorino (May 07 2020 at 18:27, on Zulip):

in particular the form, we didn't do that here

nikomatsakis (May 07 2020 at 18:28, on Zulip):

I have thought about trying to grow this gruop to be more "doing the work of running the team" and not so much "tweaking procedures"

nikomatsakis (May 07 2020 at 18:28, on Zulip):

anyway

Santiago Pastorino (May 07 2020 at 18:28, on Zulip):

this is why we called it meta I guess :slight_smile:

Santiago Pastorino (May 07 2020 at 18:28, on Zulip):

it may be a bit in a grey area

nikomatsakis (May 07 2020 at 18:28, on Zulip):

one thing I see as a problem btw

nikomatsakis (May 07 2020 at 18:29, on Zulip):

is that the membership lists for these groups will quickly fall out of date

nikomatsakis (May 07 2020 at 18:29, on Zulip):

it's going to take some work to try and maintain them in a meaningful way

nikomatsakis (May 07 2020 at 18:29, on Zulip):

frankly, it's already kind of hard to keep compiler contributors up to date ;)

nikomatsakis (May 07 2020 at 18:29, on Zulip):

i.e., to notice when people are active and take action

Santiago Pastorino (May 07 2020 at 18:29, on Zulip):

hehe

nikomatsakis (May 07 2020 at 18:29, on Zulip):

(obviously a place we should be using automated metrics)

Santiago Pastorino (May 07 2020 at 18:29, on Zulip):

I think this is why is important to move to team

Santiago Pastorino (May 07 2020 at 18:30, on Zulip):

but that doesn't solve the problem neither, I agree

nikomatsakis (May 07 2020 at 18:30, on Zulip):

in a way it makes it worse

Santiago Pastorino (May 07 2020 at 18:30, on Zulip):

it's also hard to keep up wgs members

nikomatsakis (May 07 2020 at 18:30, on Zulip):

since editing team is a bit heavyweight

nikomatsakis (May 07 2020 at 18:30, on Zulip):

but

nikomatsakis (May 07 2020 at 18:30, on Zulip):

I think it is a good idea nonetheless

nikomatsakis (May 07 2020 at 18:30, on Zulip):

I wonder actually if we could use mtrics

nikomatsakis (May 07 2020 at 18:30, on Zulip):

i.e., what if we like scraped git annotate

Santiago Pastorino (May 07 2020 at 18:30, on Zulip):

the key I guess is giving everyone ownership about this

nikomatsakis (May 07 2020 at 18:30, on Zulip):

to look for "whose names are in these files"

Santiago Pastorino (May 07 2020 at 18:30, on Zulip):

so people can just change it

nikomatsakis (May 07 2020 at 18:30, on Zulip):

and compare to the list of contributors etc

nikomatsakis (May 07 2020 at 18:30, on Zulip):

right, we should make an official policy that

nikomatsakis (May 07 2020 at 18:31, on Zulip):

any compiler-team member or contributor can "just add themselves" to any group they want -- or any person?

nikomatsakis (May 07 2020 at 18:31, on Zulip):

i.e., we strongly encourage you to add yourself

Santiago Pastorino (May 07 2020 at 18:31, on Zulip):

also ... in my opinion it may worth to have a leader per area

nikomatsakis (May 07 2020 at 18:31, on Zulip):

it still means somebody has to merge PRs on team promptly :)

nikomatsakis (May 07 2020 at 18:31, on Zulip):

but that's ok

nikomatsakis (May 07 2020 at 18:31, on Zulip):

Santiago Pastorino said:

also ... in my opinion it may worth to have a lider per area

yes, I was assuming that, but we should talk about what it means to be a leader

Santiago Pastorino (May 07 2020 at 18:32, on Zulip):

the meaning is that it's the person that keeps this list up to date :joy:

nikomatsakis (May 07 2020 at 18:32, on Zulip):

lol

nikomatsakis (May 07 2020 at 18:32, on Zulip):

that...is wishful thinking :P

Santiago Pastorino (May 07 2020 at 18:32, on Zulip):

:P

nikomatsakis (May 07 2020 at 18:32, on Zulip):

but a nice dream

nikomatsakis (May 07 2020 at 18:32, on Zulip):

oh btw

nikomatsakis (May 07 2020 at 18:32, on Zulip):

where does rustc-dev-guide fall in this :)

nikomatsakis (May 07 2020 at 18:32, on Zulip):

I think it's clearly an area

nikomatsakis (May 07 2020 at 18:33, on Zulip):

or at least it's not a "project group" in the sense of "something that can be easily done"

Santiago Pastorino (May 07 2020 at 18:33, on Zulip):

I guess there are a bunch of miscellaneous areas like rustc-dev-guide

Santiago Pastorino (May 07 2020 at 18:33, on Zulip):

which doesn't map to a portion of the compiler source code :)

nikomatsakis (May 07 2020 at 18:33, on Zulip):

rustc-dev-guide at least maps to a repository

Santiago Pastorino (May 07 2020 at 18:33, on Zulip):

yep

Santiago Pastorino (May 07 2020 at 18:34, on Zulip):

I guess there are 2 different things and depends on what are we talking about

Santiago Pastorino (May 07 2020 at 18:34, on Zulip):

I meant

Santiago Pastorino (May 07 2020 at 18:34, on Zulip):

there's a need for areas related to the source code

Santiago Pastorino (May 07 2020 at 18:35, on Zulip):

hmmm actually that's not true :)

Santiago Pastorino (May 07 2020 at 18:36, on Zulip):

forget about what I've said

Santiago Pastorino (May 07 2020 at 18:36, on Zulip):

I'm thinking :)

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

so for instance

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

do we want to have a github group of rustc-dev-guide in rust repo?

nikomatsakis (May 07 2020 at 18:37, on Zulip):

don't we?

nikomatsakis (May 07 2020 at 18:37, on Zulip):

I thought we already have one

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

today I don't think we have one

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

unsure

nikomatsakis (May 07 2020 at 18:37, on Zulip):

I guess I think the answer is yes, because

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

but if so it's not used

Santiago Pastorino (May 07 2020 at 18:37, on Zulip):

because rustc-dev-guide has it's own repo

nikomatsakis (May 07 2020 at 18:38, on Zulip):

(1) we have a policy that repo permissions are managed by github teams

nikomatsakis (May 07 2020 at 18:38, on Zulip):

and not individuals

nikomatsakis (May 07 2020 at 18:38, on Zulip):

so we want to have the folks who merge PRs and so forth

nikomatsakis (May 07 2020 at 18:38, on Zulip):

added to some GH team

Santiago Pastorino (May 07 2020 at 18:38, on Zulip):

it could happen that in an issue someone decides that the thing needs documentation in the guide and may want to ping us?

nikomatsakis (May 07 2020 at 18:38, on Zulip):

I don't really think the role of that group should be to document things

nikomatsakis (May 07 2020 at 18:38, on Zulip):

although that would be fine

nikomatsakis (May 07 2020 at 18:38, on Zulip):

but to support people in documenting things?

Santiago Pastorino (May 07 2020 at 18:39, on Zulip):

yes

Santiago Pastorino (May 07 2020 at 18:39, on Zulip):

I just meant as an example

nikomatsakis (May 07 2020 at 18:39, on Zulip):

i.e., provide feedback, help them edit, ensure consistent linking style is maintained or whatever

nikomatsakis (May 07 2020 at 18:39, on Zulip):

side note that

nikomatsakis (May 07 2020 at 18:39, on Zulip):

if library-ification works the way I hope it will

nikomatsakis (May 07 2020 at 18:39, on Zulip):

then I think that these areas would maybe all become crates

Santiago Pastorino (May 07 2020 at 18:39, on Zulip):

or in an issue someone ask is this documented in the guide? where should it fit? how should we write? and want to ping the group

nikomatsakis (May 07 2020 at 18:39, on Zulip):

or at least crates and repos would be the level that we manage this at

nikomatsakis (May 07 2020 at 18:40, on Zulip):

Santiago Pastorino said:

or in an issue someone ask is this documented in the guide? where should it fit? how should we write? and want to ping the group

sounds plausible?

nikomatsakis (May 07 2020 at 18:40, on Zulip):

I guess I think github teams are good for more than pinging, as they also drive access control

Santiago Pastorino (May 07 2020 at 18:40, on Zulip):

nikomatsakis said:

I guess I think the answer is yes, because

to answer this, we don't.

nikomatsakis (May 07 2020 at 18:40, on Zulip):

not to mention just a "sense of belonging"

Santiago Pastorino (May 07 2020 at 18:40, on Zulip):

https://github.com/rust-lang/rust/labels?q=guide

Santiago Pastorino (May 07 2020 at 18:41, on Zulip):

nikomatsakis said:

I guess I think github teams are good for more than pinging, as they also drive access control

yeah, :+1:

nikomatsakis (May 07 2020 at 18:41, on Zulip):

I would imagine the critera for being a "member" of an area is probably similar to the criteria for being a member of a project group -- you've done a few PRs

Santiago Pastorino (May 07 2020 at 18:41, on Zulip):

Santiago Pastorino said:

nikomatsakis said:

I guess I think github teams are good for more than pinging, as they also drive access control

yeah, :+1:

the main thing about all this is that rustc-dev-guide is a different repo

Santiago Pastorino (May 07 2020 at 18:42, on Zulip):

so access control and all that is just in rust-lang/rustc-dev-guide

Santiago Pastorino (May 07 2020 at 18:42, on Zulip):

anyway ...

Santiago Pastorino (May 07 2020 at 18:42, on Zulip):

going back to the list for a bit

Santiago Pastorino (May 07 2020 at 18:43, on Zulip):

I think is totally fine to have a more complete list in this HackMD document

Santiago Pastorino (May 07 2020 at 18:43, on Zulip):

but I'd also like to choose a few to try this idea out

Santiago Pastorino (May 07 2020 at 18:43, on Zulip):

maybe all that map to some part of the code in the compiler

Santiago Pastorino (May 07 2020 at 18:43, on Zulip):

that are less debatable

nikomatsakis (May 07 2020 at 18:44, on Zulip):

hmm

nikomatsakis (May 07 2020 at 18:44, on Zulip):

i.e., we don't have to do this "all at once"

nikomatsakis (May 07 2020 at 18:44, on Zulip):

but we can say that we propose to start by creating these groups

Santiago Pastorino (May 07 2020 at 18:44, on Zulip):

let's skip for now rustc-dev-guide, meta, prioritization and all these groups and focus on the ones related to code maybe?

nikomatsakis (May 07 2020 at 18:44, on Zulip):

and here is a list of other groups we would add if this makes sense (but we can also tweak)

Santiago Pastorino (May 07 2020 at 18:44, on Zulip):

yes

Santiago Pastorino (May 07 2020 at 18:44, on Zulip):

but we can leave these ones for the very end

Santiago Pastorino (May 07 2020 at 18:44, on Zulip):

that's what I meant

Santiago Pastorino (May 07 2020 at 18:45, on Zulip):

let's try first with MIR, LLVM and things like that, that're going to be easier

nikomatsakis (May 07 2020 at 18:46, on Zulip):

I think starting with 2 or 3

nikomatsakis (May 07 2020 at 18:46, on Zulip):

is a good idea

Santiago Pastorino (May 07 2020 at 18:46, on Zulip):

I was going to write, should I start a proposal but switched to the tab and you already did :)

nikomatsakis (May 07 2020 at 18:46, on Zulip):

regardless

Santiago Pastorino (May 07 2020 at 18:46, on Zulip):

:heart:

nikomatsakis (May 07 2020 at 18:48, on Zulip):

I started making it a bit more like a proposal

nikomatsakis (May 07 2020 at 18:48, on Zulip):

but I ahve to run soon

nikomatsakis (May 07 2020 at 18:48, on Zulip):

I'd appreciate any further edits, e.g., I think saying a bit more about the motivation/use-cases is a good idea

nikomatsakis (May 07 2020 at 18:49, on Zulip):

I think it's also worth talking a bit about tooling

nikomatsakis (May 07 2020 at 18:49, on Zulip):

one random thing

nikomatsakis (May 07 2020 at 18:49, on Zulip):

meh, nm

nikomatsakis (May 07 2020 at 18:49, on Zulip):

I remember that in some group I was talking to they used a OWNERS file in each directory

nikomatsakis (May 07 2020 at 18:50, on Zulip):

i.e., the repo had a OWNERS file, and then subdirectories could have add'l OWNERS

nikomatsakis (May 07 2020 at 18:50, on Zulip):

it's an interesting scheme

nikomatsakis (May 07 2020 at 18:50, on Zulip):

but I don't think we want to do that if we plan to tie it to github

Santiago Pastorino (May 07 2020 at 18:51, on Zulip):

yeah

nikomatsakis (May 07 2020 at 18:52, on Zulip):

ok, take a look at the doc now

nikomatsakis (May 07 2020 at 18:52, on Zulip):

I calle our 5 initial cases

nikomatsakis (May 07 2020 at 18:52, on Zulip):

and converted to a table :)

Santiago Pastorino (May 07 2020 at 18:52, on Zulip):

yeah I was seeing the doc meanwhile you were editing and nodding along :)

nikomatsakis (May 07 2020 at 18:53, on Zulip):

put in some random thoughts for leads

nikomatsakis (May 07 2020 at 18:53, on Zulip):

not sure if they are right :)

Santiago Pastorino (May 07 2020 at 18:53, on Zulip):

const_eval is oli and Ralfj?

nikomatsakis (May 07 2020 at 18:53, on Zulip):

I think a good thing to add would be a bit more about the role of a lead

Santiago Pastorino (May 07 2020 at 18:53, on Zulip):

anyway, people can fix this

nikomatsakis (May 07 2020 at 18:53, on Zulip):

my guess is that the role of a lead doesn't have to be especially active

nikomatsakis (May 07 2020 at 18:54, on Zulip):

like, leads and areas can be coordinating projects

nikomatsakis (May 07 2020 at 18:54, on Zulip):

(and those can become a dedicated project group, if it seems appropriate)

nikomatsakis (May 07 2020 at 18:54, on Zulip):

but they can also just be hanging out, answering questions

nikomatsakis (May 07 2020 at 18:54, on Zulip):

the main role is to be a contact point who kind of knows what's going on

nikomatsakis (May 07 2020 at 18:54, on Zulip):

and can help assess the severity of an issue or whatever

nikomatsakis (May 07 2020 at 18:54, on Zulip):

another question: "area check-ins"?

Santiago Pastorino (May 07 2020 at 18:55, on Zulip):

nikomatsakis said:

like, leads and areas can be coordinating projects

yes

Santiago Pastorino (May 07 2020 at 18:56, on Zulip):

nikomatsakis said:

another question: "area check-ins"?

I think this makes more sense that some of the check-ins we have :)

Santiago Pastorino (May 07 2020 at 18:56, on Zulip):

because some check-ins are about groups that are included in areas if you want to see it that way, and often don't have updates

Santiago Pastorino (May 07 2020 at 18:57, on Zulip):

so those updates could be included as part of an area checkin

nikomatsakis (May 07 2020 at 18:57, on Zulip):

Leading an area should not be a large commitment. It mostly means that you are aware of what's going on and can help figure out the right person to answer a question. At minimum, it means you should be generally monitoring the corresponding Zulip stream(s).

However, areas can also be more active. For example, areas can coordinate projects (which may eventually be spun out into dedicated project groups, if they're big enough) or have a wishlist of ideas.

nikomatsakis (May 07 2020 at 18:57, on Zulip):

?

nikomatsakis (May 07 2020 at 18:57, on Zulip):

ok, I gotta go to lang team mtg

Santiago Pastorino (May 07 2020 at 18:57, on Zulip):

sounds a good start

nikomatsakis (May 07 2020 at 18:58, on Zulip):

Do you want to create steering meeting proposal, or should I?

Santiago Pastorino (May 07 2020 at 18:58, on Zulip):

I can do it

Santiago Pastorino (May 07 2020 at 18:58, on Zulip):

probably not today

nikomatsakis (May 07 2020 at 18:58, on Zulip):

well the planning meeting is tomorrow

nikomatsakis (May 07 2020 at 18:58, on Zulip):

I'll make the issue at least

Santiago Pastorino (May 07 2020 at 18:58, on Zulip):

ouch

nikomatsakis (May 07 2020 at 18:58, on Zulip):

it doesn't have to be more than paragraph :)

Santiago Pastorino (May 07 2020 at 18:58, on Zulip):

ok, I will do it later then

nikomatsakis (May 07 2020 at 18:58, on Zulip):

the idea is just to sketch what we will talk about

nikomatsakis (May 07 2020 at 18:58, on Zulip):

we can continue to develop the proposal in the meantime

Santiago Pastorino (May 07 2020 at 18:58, on Zulip):

yeah, I'm also jumping into a 2hs meeting now :)

Santiago Pastorino (May 07 2020 at 18:59, on Zulip):

sure, I can create the proposal

Santiago Pastorino (May 07 2020 at 21:34, on Zulip):

@nikomatsakis compiler-team#288

nikomatsakis (May 07 2020 at 21:49, on Zulip):

yay!

Last update: Sep 28 2020 at 15:15UTC