Stream: t-compiler

Topic: steering meeting 2019.04.12 #58850


nikomatsakis (Apr 12 2019 at 14:02, on Zulip):

Hello @T-compiler/meeting! Welcome to today's steering meeting.

Today's meeting proceeds as follows:

nikomatsakis (Apr 12 2019 at 14:04, on Zulip):

:horn: I opened a PR with a draft RFC on the compiler-team repository. :horn: This RFC describes adding a new level ("compiler team contributors") to recognize people who have been contributing regularly. It also tries to document a few things about members etc. Please give it a read!

nikomatsakis (Apr 12 2019 at 14:05, on Zulip):

(This is what I was calling journeypeople)

nikomatsakis (Apr 12 2019 at 14:06, on Zulip):

Oh, one other thing. As usual, we're going to want a summary from today's meeting. If anybody wants to volunteer to write it, please do -- the idea would be to summarize the key points and conclusions from our discussion, so that people don't have to read the full minutes to know what happened. No need to document every comment. To that end, please use the :point_up: :point_up: emoji to highlight key comments you think (or any other one, I suppose).

nikomatsakis (Apr 12 2019 at 14:09, on Zulip):

No more announcements? :)

nikomatsakis (Apr 12 2019 at 14:09, on Zulip):

@eddyb what's the status of your symbol naming stuff? :)

eddyb (Apr 12 2019 at 14:09, on Zulip):

I'm working as we speak on the CLI integration, and then I'll have to adjust the tests a bit, but I think we can merge soon!

Santiago Pastorino (Apr 12 2019 at 14:10, on Zulip):

Oh, one other thing. As usual, we're going to want a summary from today's meeting. If anybody wants to volunteer to write it, please do -- the idea would be to summarize the key points and conclusions from our discussion, so that people don't have to read the full minutes to know what happened. No need to document every comment. To that end, please use the :point_up: :point_up: emoji to highlight key comments you think (or any other one, I suppose).

I can do it :)

eddyb (Apr 12 2019 at 14:10, on Zulip):

it's been unnecessarily delayed for, ugh, two months, but now I finally have more throughput and I want to get it over with :P

eddyb (Apr 12 2019 at 14:12, on Zulip):

I'm also working on refactoring something in rustc_metadata, where every query had the overhead of decoding like a dozen or so offsets, when it only wanted to read one of them. and, we, just, sort of forgot about it?

eddyb (Apr 12 2019 at 14:12, on Zulip):

so I'm excited about what that perf run is gonna bring

nikomatsakis (Apr 12 2019 at 14:12, on Zulip):

:horn: Another announcement :horn: I think that in the traits meeting on Monday we're going to dig into some refactoring work we've been doing on chalk to try and allow for more code to be shared with rustc, rust-analyzer, etc. This also should allow it to model specialization and handle queries. If you're interested, check it out.

eddyb (Apr 12 2019 at 14:12, on Zulip):

(actually this is a steering meeting, don't let me rant too much :P)

nikomatsakis (Apr 12 2019 at 14:13, on Zulip):

Yeah, all right, people seem quiet today. Let's get started =)

nikomatsakis (Apr 12 2019 at 14:13, on Zulip):

So this was our plan for the main topic today:

nikomatsakis (Apr 12 2019 at 14:13, on Zulip):

I'm not sure if people had a chance to skim the gist already

nikomatsakis (Apr 12 2019 at 14:13, on Zulip):

But let me kind of outline my thinking here

nikomatsakis (Apr 12 2019 at 14:14, on Zulip):

Basically I feel like we don't have a clear place to do and circulate design work

nikomatsakis (Apr 12 2019 at 14:14, on Zulip):

and, I think working groups are great, but sometimes they have to make far-reaching decisions that are probably worth discussing with the wider team

nikomatsakis (Apr 12 2019 at 14:14, on Zulip):

scheduling can often be a pain too

nikomatsakis (Apr 12 2019 at 14:14, on Zulip):

so I thought it might be useful to have a regular team that we try to have these discussions

nikomatsakis (Apr 12 2019 at 14:15, on Zulip):

(in particular, I think the more senior members of the team should make a real effort to hold that time available)

nikomatsakis (Apr 12 2019 at 14:15, on Zulip):

some scenarios I am thinking of:

nikomatsakis (Apr 12 2019 at 14:15, on Zulip):
nikomatsakis (Apr 12 2019 at 14:16, on Zulip):
nikomatsakis (Apr 12 2019 at 14:16, on Zulip):
nikomatsakis (Apr 12 2019 at 14:16, on Zulip):

Each of these cases is maybe slightly different:

nikomatsakis (Apr 12 2019 at 14:16, on Zulip):

some are more exploratory -- e.g., we ahve a problem, want to find a sol'n

nikomatsakis (Apr 12 2019 at 14:17, on Zulip):

others are more of a concrete proposal

eddyb (Apr 12 2019 at 14:17, on Zulip):

one more scenario: I want to redesign the "definition" & HIR layers of the compiler to enable fine-grained incremental parsing, macro expansion and name resolution

nikomatsakis (Apr 12 2019 at 14:17, on Zulip):

Anyway, I've typed enough. So I'll slow down here

eddyb (Apr 12 2019 at 14:17, on Zulip):

I've wanted to start ages ago but better late than never :P

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

The actual mechanism that I was proposing:

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

I want to call out this point:

Each meeting will decide the topic of the next meeting. To be selected, a written proposal must be available for review.

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

I think it's pretty important that we know the agenda in advance

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

so we can ensure the right people are there

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

and I think it'd be useful to have some kind of background stuff for people to start from

nikomatsakis (Apr 12 2019 at 14:18, on Zulip):

even if nobody reads it before the meeting, it'll give some structure :)

mw (Apr 12 2019 at 14:19, on Zulip):

I'd like to propose doing the design meeting at this very time slot, whenever there is no steering meeting

eddyb (Apr 12 2019 at 14:19, on Zulip):

how detailed should the written proposal be? RFC summary, RFC full text, in between?

nikomatsakis (Apr 12 2019 at 14:19, on Zulip):

I'd like to propose doing the design meeting at this very time slot, whenever there is no steering meeting

ooooh

nikomatsakis (Apr 12 2019 at 14:19, on Zulip):

how detailed should the written proposal be? RFC summary, RFC full text, in between?

I think it would depend, but definitely not RFC full text

davidtwco (Apr 12 2019 at 14:19, on Zulip):

Am I right in thinking that a implementation doesn't need to exist to be discussed at one of these meetings? It doesn't have to be "Member Y has this large change, let's discuss it before landing", but "Member Z has these ideas for a change and has written a short summary up" can also work.

nikomatsakis (Apr 12 2019 at 14:19, on Zulip):

In answer to your question @eddyb, I had imagined that it might take the form of like a small gist

nikomatsakis (Apr 12 2019 at 14:20, on Zulip):

but it could be more if there is more to say

nikomatsakis (Apr 12 2019 at 14:20, on Zulip):

right now we have a backlog of design discussion so I bet there are many things with a ot of background

mw (Apr 12 2019 at 14:20, on Zulip):

there has to be enough detail on the important design decisions, I'd say

nikomatsakis (Apr 12 2019 at 14:20, on Zulip):

but I would hope that going forward we'll start sooner, when there's less to say ;)

nikomatsakis (Apr 12 2019 at 14:20, on Zulip):

there has to be enough detail on the important design decisions, I'd say

well, I think if this is a "let's decide if we sohuld do this" sort of meeting, then yes

nikomatsakis (Apr 12 2019 at 14:20, on Zulip):

so maybe that's key

nikomatsakis (Apr 12 2019 at 14:21, on Zulip):

i.e., part of the write-up would be specifying the kind of feedback/decision desired, and that would reflect on the amount of detail expected?

mw (Apr 12 2019 at 14:21, on Zulip):

or maybe: a description of the contentious issues/problem areas

nikomatsakis (Apr 12 2019 at 14:21, on Zulip):

in my ideal world, we'd be able to take these write-ups and move them into rustc-guide as design documents for future ference

eddyb (Apr 12 2019 at 14:21, on Zulip):

I think I went on an issue-opening spree at some point last year, and some of those would qualify, lemme grab a random one...

nikomatsakis (Apr 12 2019 at 14:21, on Zulip):

but I didn't want to add onerous requirements to start

nikomatsakis (Apr 12 2019 at 14:21, on Zulip):

as I said, I hope that going forward, we can do this more organically, so that we are growing a write-up as we go

nikomatsakis (Apr 12 2019 at 14:22, on Zulip):

or maybe: a description of the contentious issues/problem areas

yes -- kind of outlining the space is great

eddyb (Apr 12 2019 at 14:22, on Zulip):

https://github.com/rust-lang/rust/issues/53560

eddyb (Apr 12 2019 at 14:22, on Zulip):

https://github.com/rust-lang/rust/issues/53660

eddyb (Apr 12 2019 at 14:22, on Zulip):

lol I thought I pasted the same link, these are 100 issues apart exactly

nikomatsakis (Apr 12 2019 at 14:23, on Zulip):

So I feel like #53560 is reasonably specific, it might be nice if it dug a bit more into the potential implications, but that could also be the goal of the meeting

eddyb (Apr 12 2019 at 14:23, on Zulip):

so if you have time, @nikomatsakis, I'd be curious to know which of these would qualify as-is, if at all?

nikomatsakis (Apr 12 2019 at 14:23, on Zulip):

https://github.com/rust-lang/rust/issues/53660

this one feels too sparse to me, I at least don't have a clear idea what's being proposed

eddyb (Apr 12 2019 at 14:24, on Zulip):

I think it's literally "rm name: Name fields from HIR", but I agree

nikomatsakis (Apr 12 2019 at 14:24, on Zulip):

I'm curious what others thing though :)

nikomatsakis (Apr 12 2019 at 14:25, on Zulip):

e.g., @mw ;)

nikomatsakis (Apr 12 2019 at 14:25, on Zulip):

it might be nice if it dug a bit more into the potential implications, but that could also be the goal of the meeting

but I want to circle back to this

mw (Apr 12 2019 at 14:25, on Zulip):

I think such a write up should contain a list of things that the author expects are hard to do

mw (Apr 12 2019 at 14:26, on Zulip):

and also a list of things where there is a decision to make

nikomatsakis (Apr 12 2019 at 14:26, on Zulip):

i.e., I imagine a meeting proposal might be something like:

here is a specific idea. i'm not proposing we merge it yet, I'm proposing we talk about what complications might rise up. then I can add this to the write-up for next time and explore those.

mw (Apr 12 2019 at 14:26, on Zulip):

or an open question

davidtwco (Apr 12 2019 at 14:26, on Zulip):

I think it'd be a good way for new contributors to get more familiar with things - being around when big things are changing and knowing why they're changing would be beneficial.

davidtwco (Apr 12 2019 at 14:27, on Zulip):

(I don't think we should make that an explicit goal that takes away from the design discussion though)

nikomatsakis (Apr 12 2019 at 14:27, on Zulip):

I totally agree with that -- I feel like there will be lurkers who just attend and, over time, kind of soak up a lot of knowledge.

nikomatsakis (Apr 12 2019 at 14:27, on Zulip):

it's also one of those things -- usually newbies are nervous to ask questions, but generally 50% of the experts are also not following ;)

eddyb (Apr 12 2019 at 14:27, on Zulip):

anyway I think the reason I brought up these examples is I can end up with a lot of "this is a thing we could do" that just get forgotten, and some of the bigger stuff isn't even written down anywhere, so I think I could triage past design ideas and find the most relevant to short-term development

nikomatsakis (Apr 12 2019 at 14:28, on Zulip):

yes, we could talk a bit more about that part :)

eddyb (Apr 12 2019 at 14:28, on Zulip):

and if presented well then I don't even have to work on them myself :P

nikomatsakis (Apr 12 2019 at 14:28, on Zulip):

part of the proposal was to have a queue of design ideas

mw (Apr 12 2019 at 14:28, on Zulip):

something that would also be very useful is to have some kind of global view somewhere

mw (Apr 12 2019 at 14:28, on Zulip):

in order to coordinate between design efforts

nikomatsakis (Apr 12 2019 at 14:28, on Zulip):

I sort of like the idea of sorting them by what they aim to achieve, and probably kind of having some that are "nice to haves" etc

eddyb (Apr 12 2019 at 14:28, on Zulip):

been wanting a dashboard for years, heh

nikomatsakis (Apr 12 2019 at 14:28, on Zulip):

something that would also be very useful is to have some kind of global view somewhere

global view of what exactly?

mw (Apr 12 2019 at 14:28, on Zulip):

and schedule resources

eddyb (Apr 12 2019 at 14:29, on Zulip):

I wonder if there is any piece of software for something like that, which I don't hate

mw (Apr 12 2019 at 14:29, on Zulip):

of design efforts going on

mw (Apr 12 2019 at 14:29, on Zulip):

how they might interact

nikomatsakis (Apr 12 2019 at 14:29, on Zulip):

Hmm. I kind of feel like those are the working groups.

nikomatsakis (Apr 12 2019 at 14:29, on Zulip):

But maybe that's an invalid assumption ;)

eddyb (Apr 12 2019 at 14:29, on Zulip):

I'd love a waterfall view of everything I've started, and e.g. abandoned because lack of time

nikomatsakis (Apr 12 2019 at 14:29, on Zulip):

i.e., maybe it sometimes makes sense to do design work (e.g., the examples @eddyb gave) that is kind of 'pre working group'

davidtwco (Apr 12 2019 at 14:29, on Zulip):

and if presented well then I don't even have to work on them myself :P

Hopefully a design discussion would give enough context for some intermediate contributors that they could be the ones to work on it (with help from the experts) and these medium-to-large tasks would help them become experts themselves. Though, in a bunch of scenarios, the proposer might want to work on it themselves, which is of course fine.

eddyb (Apr 12 2019 at 14:30, on Zulip):

and this sort of thing for every major contributor? a goldmine of half-baked ideas!

nikomatsakis (Apr 12 2019 at 14:30, on Zulip):

Hopefully a design discussion would give enough context for some intermediate contributors that they could be the ones to work on it (with help from the experts)

(as an aside, this fits into those "mentoring program" ideas we were kicking around, of course)

mw (Apr 12 2019 at 14:30, on Zulip):

I can imagine that one big refactoring of e.g. how definitions/HIR work would affect a lot of other things

Santiago Pastorino (Apr 12 2019 at 14:30, on Zulip):

and if presented well then I don't even have to work on them myself :P

Hopefully a design discussion would give enough context for some intermediate contributors that they could be the ones to work on it (with help from the experts) and these medium-to-large tasks would help them become experts themselves. Though, in a bunch of scenarios, the proposer might want to work on it themselves, which is of course fine.

absolutely agree on this, I'd surely take more stuff if this things happens :)

nikomatsakis (Apr 12 2019 at 14:30, on Zulip):

I can imagine that one big refactoring of e.g. how definitions/HIR work would affect a lot of other things

yes so this is precisely why I wanted to have a "global" Design meeting

nikomatsakis (Apr 12 2019 at 14:30, on Zulip):

and not "per working group" meetings

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

(even if not everybody attends every one)

mw (Apr 12 2019 at 14:31, on Zulip):

ok, then the meeting should try to keep track of things in flight

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

so @mw I sort of liked your idea of using this slot

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

and it makes me wonder

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

if the steering meeting comes every 3 weeks

Santiago Pastorino (Apr 12 2019 at 14:31, on Zulip):

@nikomatsakis to improve rustc_guide is also very important, and that effort could be lead by the docs WG

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

one thing it could do is to be a discussion of what designs we ought to do

nikomatsakis (Apr 12 2019 at 14:31, on Zulip):

i.e., to try and schedule out the next 2 weeks

nikomatsakis (Apr 12 2019 at 14:32, on Zulip):

though that wouldn't leave a clear time for this conversation we're having right now -- but

nikomatsakis (Apr 12 2019 at 14:32, on Zulip):

maybe this could be a valid design meeting topic :)

nikomatsakis (Apr 12 2019 at 14:32, on Zulip):

(so meta!)

nikomatsakis (Apr 12 2019 at 14:32, on Zulip):

i.e., we don't only have to be designing code

nikomatsakis (Apr 12 2019 at 14:33, on Zulip):

ok, then the meeting should try to keep track of things in flight

i'm just imagining that this takes a bit of time

davidtwco (Apr 12 2019 at 14:33, on Zulip):

i.e., we don't only have to be designing code

Isn't the steering meeting in some ways not already a "design meeting for not-code things"?

nikomatsakis (Apr 12 2019 at 14:33, on Zulip):

yes :)

nikomatsakis (Apr 12 2019 at 14:33, on Zulip):

interestingly, this wasn't what I initially had in mind for it

nikomatsakis (Apr 12 2019 at 14:33, on Zulip):

but it's become that and it's clear we need it

nikomatsakis (Apr 12 2019 at 14:34, on Zulip):

initially I thought it would for, well, steering the project :) that is, reviewing what we are working on and checking in on how well that is going

mw (Apr 12 2019 at 14:34, on Zulip):

ok, then the meeting should try to keep track of things in flight
i'm just imagining that this takes a bit of time

I'm trying to avoid the situation we have with RFCs now: lots of designs "done" but no one implementing them

nikomatsakis (Apr 12 2019 at 14:34, on Zulip):

yes

nikomatsakis (Apr 12 2019 at 14:34, on Zulip):

so, let me just say one bit about the "mentoring program" ideas that we've been kicking around

nikomatsakis (Apr 12 2019 at 14:34, on Zulip):

because I think it's related

davidtwco (Apr 12 2019 at 14:35, on Zulip):

(different from existing mentoring instructions on issues)

nikomatsakis (Apr 12 2019 at 14:35, on Zulip):

as I announced earlier, I am proposing to add an official "contributor" level for people who've been making contributions for some time -- roughly what r+ rights are now, but more official and nicer

nikomatsakis (Apr 12 2019 at 14:35, on Zulip):

and I was thinking that it'd be great to have a "member in training" program, where basically a full team member takes on one (or more, perhaps) of these contributors as a kind of 'intern'

nikomatsakis (Apr 12 2019 at 14:36, on Zulip):

working to realize some sort of ambitious design

nikomatsakis (Apr 12 2019 at 14:36, on Zulip):

I think in practice @mw you are roughly doing this with @Wesley Wiser, for example

mw (Apr 12 2019 at 14:36, on Zulip):

yes

nikomatsakis (Apr 12 2019 at 14:36, on Zulip):

the idea being that at the end of the project, they become a team member

nikomatsakis (Apr 12 2019 at 14:36, on Zulip):

this has a natural kind of .. "rate limiting" factor

nikomatsakis (Apr 12 2019 at 14:36, on Zulip):

in that one person can only "mentor" so many things at a time

nikomatsakis (Apr 12 2019 at 14:37, on Zulip):

maybe we just put a limit of 1 :)

nikomatsakis (Apr 12 2019 at 14:37, on Zulip):

but I think that when we are selecting the agenda for the design meeting,

nikomatsakis (Apr 12 2019 at 14:37, on Zulip):

I would want to take into account "is somebody lined up to implement this"?

nikomatsakis (Apr 12 2019 at 14:37, on Zulip):

(and I feel like that could fit very well with the "intern" concept)

mw (Apr 12 2019 at 14:38, on Zulip):

from the perspective of an "external" contributor like me (i.e. someone who is not part of Rust team managed by Niko and who cannot entirely freely decide what they spend their time on) it's important that any mentoring is part of the work they are paid for, I think. Just so that the time is available.

nikomatsakis (Apr 12 2019 at 14:38, on Zulip):

seems correct

mw (Apr 12 2019 at 14:39, on Zulip):

to some extent this applies to everything that is time-consuming :)

nikomatsakis (Apr 12 2019 at 14:39, on Zulip):

I would also say that -- and this is a very complex topic so take this with a huge grain of salt -- I'd like to see if we can find ways to connect contracting or other funding to well-defined projects like these.

nikomatsakis (Apr 12 2019 at 14:39, on Zulip):

But I want to tread carefully there :)

nikomatsakis (Apr 12 2019 at 14:40, on Zulip):

anyway, to circle back to the design meeting... i'm not sure if we've arrived at a consensus, maybe we can sort of summarize where we are?

nikomatsakis (Apr 12 2019 at 14:41, on Zulip):

it seems like @mw felt that we

nikomatsakis (Apr 12 2019 at 14:42, on Zulip):

we had proposed to use this same slot for design meetings, but it wasn't clear to me if:

or take a more "sprint like" approach

nikomatsakis (Apr 12 2019 at 14:43, on Zulip):

(I also think this slot is not great for people on the west coast, but not sure what to do about that, maybe we can sometimes do a different time period)

mw (Apr 12 2019 at 14:43, on Zulip):

I proposed the first version

nikomatsakis (Apr 12 2019 at 14:43, on Zulip):

do you still favor it? I like that it gives us a kind of enforced ratio

nikomatsakis (Apr 12 2019 at 14:44, on Zulip):

(I also think this slot is not great for people on the west coast, but not sure what to do about that, maybe we can sometimes do a different time period)

in particular I know it doesn't work for @Aaron Turon at the moment

mw (Apr 12 2019 at 14:44, on Zulip):

not sure which I like better

mw (Apr 12 2019 at 14:44, on Zulip):

any later time of day is hard for me to fit in at a regular basis

nikomatsakis (Apr 12 2019 at 14:44, on Zulip):

this is another reason though to pick topic of meetings in advance -- if we have e.g. a trait-related thing, we could try to select a different day/time

nikomatsakis (Apr 12 2019 at 14:44, on Zulip):

yeah I mean I think we're very Europe-based, and I'm not inclined to change the "normal time" yet

pnkfelix (Apr 12 2019 at 14:45, on Zulip):

Later times are basically unworkable for me too

pnkfelix (Apr 12 2019 at 14:45, on Zulip):

Especially on Fridays

nikomatsakis (Apr 12 2019 at 14:45, on Zulip):

(let's not worry about that, suffice to say that for specific topics we might pick different times)

nikomatsakis (Apr 12 2019 at 14:46, on Zulip):

not sure which I like better

me either :)

mw (Apr 12 2019 at 14:46, on Zulip):

in general I'm not a fan of things with "sprint" in the name

mw (Apr 12 2019 at 14:46, on Zulip):

it conveys a sense of urgency that should be limited to exceptional circumstances

nikomatsakis (Apr 12 2019 at 14:47, on Zulip):

I use the word very loosely and am happy to pick another. I basically mean alternating a "select the things to do" with "do the things"

nikomatsakis (Apr 12 2019 at 14:47, on Zulip):

I agree sprint makes it sound like we are .. well .. running. :)

Santiago Pastorino (Apr 12 2019 at 14:48, on Zulip):

I guess the important thing is define a period of time where we work towards some specific goals with no rush like running :)

mw (Apr 12 2019 at 14:48, on Zulip):

yes, I want us jogging at most because we have a long distance to cover :)

nikomatsakis (Apr 12 2019 at 14:49, on Zulip):

This:

![Calm down, I got this](https://nikomatsakis.github.io/rust-latam-2019/content/images/obama-got-this.gif)

not this:

![spongebob](https://nikomatsakis.github.io/rust-latam-2019/content/images/multitasking-spongebob.gif)

davidtwco (Apr 12 2019 at 14:49, on Zulip):

I understood @nikomatsakis's suggestion as something like the following: At the start of the month, selecting what the design meeting topics will be for the month ahead and then doing those before deciding again at the start of the next month.

nikomatsakis (Apr 12 2019 at 14:49, on Zulip):

oh come on zulip!

nikomatsakis (Apr 12 2019 at 14:49, on Zulip):

ok, it worked eventually ;)

nikomatsakis (Apr 12 2019 at 14:50, on Zulip):

it basically comes down to whether

nikomatsakis (Apr 12 2019 at 14:50, on Zulip):

when selecting the things

nikomatsakis (Apr 12 2019 at 14:50, on Zulip):

there is value in doing multiple at once

nikomatsakis (Apr 12 2019 at 14:50, on Zulip):

to help maintain global balance

nikomatsakis (Apr 12 2019 at 14:50, on Zulip):

and amortize the cost of that sort of thinking

pnkfelix (Apr 12 2019 at 14:50, on Zulip):

I hadn’t considered planning out more than one meeting in advance, but it sounds like something to try doing

mw (Apr 12 2019 at 14:51, on Zulip):

I don't have a strong opinion

nikomatsakis (Apr 12 2019 at 14:51, on Zulip):

well, if we were going to try that, it raises the question of when to do the first "selection" meeting :)

mw (Apr 12 2019 at 14:51, on Zulip):

let's just try one of the approaches and see how it goes

nikomatsakis (Apr 12 2019 at 14:51, on Zulip):

yes, that

nikomatsakis (Apr 12 2019 at 14:52, on Zulip):

ok, it seems like everybod is basically in favor of the plan

nikomatsakis (Apr 12 2019 at 14:52, on Zulip):

that is, the general idea

nikomatsakis (Apr 12 2019 at 14:52, on Zulip):

so let's briefly (in our last 8 minutes) talk about how we might actually get this underway

nikomatsakis (Apr 12 2019 at 14:52, on Zulip):

a key part of the plan was that we needed meeting proposals

nikomatsakis (Apr 12 2019 at 14:52, on Zulip):

I feel like we have no shortage of ideas

nikomatsakis (Apr 12 2019 at 14:53, on Zulip):

maybe we should give a bit of time for people to write proposals (say, next week? 2 weeks?) and then try a selection meeting? (if we start with the "alternating" approach)

mw (Apr 12 2019 at 14:53, on Zulip):

prioritization and idea selection might actually a big part of the work

eddyb (Apr 12 2019 at 14:53, on Zulip):

yeah we have almost 5 years of backlog :P

nikomatsakis (Apr 12 2019 at 14:53, on Zulip):

we could basically just say that we will start with the alternating approach, and that we'll start on the next steering meeting :)

nikomatsakis (Apr 12 2019 at 14:54, on Zulip):

that gives 2 weeks I think for people to make proposals

mw (Apr 12 2019 at 14:54, on Zulip):

taking stock of the things we are doing already might be a good idea

nikomatsakis (Apr 12 2019 at 14:54, on Zulip):

that gives 2 weeks I think for people to make proposals

well, 3 I guess

mw (Apr 12 2019 at 14:54, on Zulip):

i.e. to get the initial "global view"

nikomatsakis (Apr 12 2019 at 14:54, on Zulip):

to some extent, isn't that what we've been doing with the working groups?

mw (Apr 12 2019 at 14:54, on Zulip):

true

nikomatsakis (Apr 12 2019 at 14:54, on Zulip):

well

nikomatsakis (Apr 12 2019 at 14:55, on Zulip):

if we started with alternating, then the meeting might begin with a kind of review

nikomatsakis (Apr 12 2019 at 14:55, on Zulip):

taking stock of the things we are doing already might be a good idea

can you say a bit more about what shape you think this should take

nikomatsakis (Apr 12 2019 at 14:55, on Zulip):

I think it's probably a good idea but i'm trying to understand what it means more concretely :)

mw (Apr 12 2019 at 14:56, on Zulip):

not sure. I mostly want a way to see where we actually have capacity to do things that are not already part of a working group

nikomatsakis (Apr 12 2019 at 14:57, on Zulip):

we could basically just say that we will start with the alternating approach, and that we'll start on the next steering meeting :)

Incidentally, I'm leaning more and more this way. It seems .. obvious. And it gives us some time to get ready. =) My main concern would be if waiting 3 weeks is too long. (i.e., we'd start on May 3rd)

nikomatsakis (Apr 12 2019 at 14:57, on Zulip):

not sure. I mostly want a way to see where we actually have capacity to do things that are not already part of a working group

(I am imagining a lot of the design questions will originate from the working groups)

nikomatsakis (Apr 12 2019 at 14:57, on Zulip):

but it also seems like we might decide "you know, that WG should spin down, it's not worth it, let's do XXX isntead"

mw (Apr 12 2019 at 14:58, on Zulip):

as an example, where would the 3 topics you mentioned initially fit in here?

nikomatsakis (Apr 12 2019 at 14:58, on Zulip):

A good question. So @davidtwco's design question came from the wg-async-await WG

mw (Apr 12 2019 at 14:58, on Zulip):

i.e. end-to-end queries, HIR, definitions

nikomatsakis (Apr 12 2019 at 14:59, on Zulip):

(that was HIR)

nikomatsakis (Apr 12 2019 at 14:59, on Zulip):

the trait working group questions are similar

nikomatsakis (Apr 12 2019 at 14:59, on Zulip):

end-to-end queries -- well, we don't have a clear WG for that, unless you could wg-parallel-rustc, but arguably we should

nikomatsakis (Apr 12 2019 at 14:59, on Zulip):

so I think some of the design meetings might be exploring topics that aren't yet, and I guess part of that should be "Who will do this"

nikomatsakis (Apr 12 2019 at 15:00, on Zulip):

(I do think that the 'tcx lifetime stuff that @Zoxc was pursuing is actually something quite realted to trait wg as well though, as chalk really wants to have inference state that outlives a particular fn call)

nikomatsakis (Apr 12 2019 at 15:01, on Zulip):

I don't know, is that a satisfactory answer? I'm not sure if I'm fully addressing your concerns.

nikomatsakis (Apr 12 2019 at 15:01, on Zulip):

(But: time check, it's been 1 hour)

mw (Apr 12 2019 at 15:02, on Zulip):

I'm 100% OK with the design meetings trying to solve existing problems that came out of a working group

mw (Apr 12 2019 at 15:03, on Zulip):

I'd be wary of coming up with lots of "wouldn't it be nice if we did this" designs that we don't have the bandwidth to actually implement

mw (Apr 12 2019 at 15:04, on Zulip):

so it seems to me that we might want to try to fit design topics that are being discussed into existing working groups

nikomatsakis (Apr 12 2019 at 15:04, on Zulip):

I think most of the time that is good. And if we don't have one, but we do have people, we should just create a WG.

mw (Apr 12 2019 at 15:04, on Zulip):

or decide to create a new working group for a new topic

mw (Apr 12 2019 at 15:05, on Zulip):

i.e. WGs are the means to schedule resources

mw (Apr 12 2019 at 15:05, on Zulip):

and the design meeting is there to do technical coordination between them

nikomatsakis (Apr 12 2019 at 15:05, on Zulip):

Let's see if we can summarize:

I'm leaning towards the alternating approach. It feels like deciding what to do will be hard and will take time, and we shouldn't try to squeeze it in.

mw (Apr 12 2019 at 15:06, on Zulip):

I'm leaning towards the alternating approach. It feels like deciding what to do will be hard and will take time, and we shouldn't try to squeeze it in.

I agree

nikomatsakis (Apr 12 2019 at 15:07, on Zulip):

In that case, we need to pick a start date -- May 3rd? (next steering meeting)

nikomatsakis (Apr 12 2019 at 15:07, on Zulip):

OK :tada:

nikomatsakis (Apr 12 2019 at 15:08, on Zulip):

Thanks all

mw (Apr 12 2019 at 15:08, on Zulip):

:wave:

nikomatsakis (Apr 12 2019 at 15:11, on Zulip):

@Santiago Pastorino -- did you say you were going to try and write a summary? Let's discuss. I might also be ok with trying to write something here, since this seems of like it merits an "official announcement"

nikomatsakis (Apr 12 2019 at 15:11, on Zulip):

but that is perhaps separate from a summary, which is more like "we decided to do something like this"

Santiago Pastorino (Apr 12 2019 at 15:35, on Zulip):

sorry @nikomatsakis I left for a bit after the meeting, go ahead if you want :)

Santiago Pastorino (Apr 12 2019 at 15:36, on Zulip):

it took some notes at the beginning but at some point I stopped and decided to revisit later

Santiago Pastorino (Apr 12 2019 at 15:36, on Zulip):

can share that if you think it's useful

Santiago Pastorino (Apr 12 2019 at 15:36, on Zulip):

I'm also ok with finishing this

Santiago Pastorino (Apr 12 2019 at 15:37, on Zulip):

whatever you prefer, but if you want to do the write up that's better for me :)

nikomatsakis (Apr 12 2019 at 21:01, on Zulip):

Minutes posted

Last update: Nov 16 2019 at 01:40UTC