Stream: t-compiler

Topic: design meeting 2020-02-28


nikomatsakis (Feb 28 2020 at 14:59, on Zulip):

Hello @T-compiler/meeting! The design meeting is starting now. Today's today is "Focused and efficient triage" compiler-team#247. Pre-meeting notes can be found in this hackmd. The primary proposal to be discussed is the idea of creating a pre-triage working group -- but more broadly the notes cover some thoughts on the kinds of tasks we need to do, and who should be doing them.

Please add a :wave: emoji to show you are here.

And now, in the leadup to the meeting, any...

Announcements

nikomatsakis (Feb 28 2020 at 14:59, on Zulip):
nikomatsakis (Feb 28 2020 at 15:02, on Zulip):

cc @DPC meeting is here :)

DPC (Feb 28 2020 at 15:02, on Zulip):

Ah lol

nikomatsakis (Feb 28 2020 at 15:04, on Zulip):

OK, shall we get started?

nikomatsakis (Feb 28 2020 at 15:04, on Zulip):

/me caffeinated

pnkfelix (Feb 28 2020 at 15:05, on Zulip):

the document is laid out with a proposal first, but we may want to first review the next section, which summarizes our status today

nikomatsakis (Feb 28 2020 at 15:05, on Zulip):

yeah

pnkfelix (Feb 28 2020 at 15:05, on Zulip):

(the Summary of needs and proposals)

nikomatsakis (Feb 28 2020 at 15:06, on Zulip):

the motivation for this meeting I would say is that (a) a lot of work for our current triage process is falling on @pnkfelix and (b) there is a kind of lack of clarity I think around our goals, how we use our labels, etc

nikomatsakis (Feb 28 2020 at 15:06, on Zulip):

that section @pnkfelix pointed at is an attempt to break down the kinds of things we need to be doing

nikomatsakis (Feb 28 2020 at 15:06, on Zulip):
nikomatsakis (Feb 28 2020 at 15:07, on Zulip):

those were the five things we wrote down, but I guess a first question is whether anyone things there are things missing :)

pnkfelix (Feb 28 2020 at 15:07, on Zulip):

Notably, we actually are pretty good at "Making general quality improvements and enhancements", as already noted in the document.

pnkfelix (Feb 28 2020 at 15:07, on Zulip):

(it basically happens organically)

pnkfelix (Feb 28 2020 at 15:08, on Zulip):

and our proposal for this meeting, now, is to talk about the first bullet above "Monitor and identify "critical bugs" that are not making progress"

nikomatsakis (Feb 28 2020 at 15:08, on Zulip):

and maybe to some extent "Ensuring deferred things are picked up again (e.g., future compatibility warnings)" or some of the other

nikomatsakis (Feb 28 2020 at 15:08, on Zulip):

anyway I can't tell if folks are nodding along or are busy reading

nagisa (Feb 28 2020 at 15:08, on Zulip):

/me busy reading.

Santiago Pastorino (Feb 28 2020 at 15:09, on Zulip):

both, nodding and reading the doc :)

Nell Shamrell-Harrington (Feb 28 2020 at 15:09, on Zulip):
nikomatsakis (Feb 28 2020 at 15:10, on Zulip):

OK, I'll give a minute or two, but I'd like to hear folks' thoughts on the "today" (and of course the "proposed")

pnkfelix (Feb 28 2020 at 15:10, on Zulip):

so, yeah: Our proposal for attacking the "Monitor and identify 'critical bugs' " is to create a WG to replace what I do

simulacrum (Feb 28 2020 at 15:11, on Zulip):

Do we have, uh, people who would be in that? One concern is that these likely need to be pretty experienced and involved I think to have a good sense

nagisa (Feb 28 2020 at 15:11, on Zulip):

@pnkfelix My take on the current state and our failure to do this mostly lies on the fact that we try to do too much synchronously and thus don’t spend enough time evaluating each separate issue we look at. Moving some of the process to be async via a wg sounds like a possible way forward, sure.

nikomatsakis (Feb 28 2020 at 15:11, on Zulip):

ha, a solid question!

pnkfelix (Feb 28 2020 at 15:11, on Zulip):

THat is a good question

pnkfelix (Feb 28 2020 at 15:12, on Zulip):

@nagisa Indeed. One of the results of some brainstorming that is captured in the document is an attempt to revise our prioritization scheme

nagisa (Feb 28 2020 at 15:13, on Zulip):

But then, I feel like we already have an async process – people looking at the issue tracker routinely and sifting through the issues, tagging them etc.

nikomatsakis (Feb 28 2020 at 15:13, on Zulip):

Re: who might do it, I've been talking to @Santiago Pastorino about leading up this effort

nikomatsakis (Feb 28 2020 at 15:14, on Zulip):

I think that @nagisa's point of async is also quite relevant--

pnkfelix (Feb 28 2020 at 15:14, on Zulip):

@nagisa right. I just want to make sure that we have a way to draw attention to the critical bugs that aren't getting addressed.

nikomatsakis (Feb 28 2020 at 15:14, on Zulip):

i.e., I'm not sure how much experience is truly needed in a "sync" way,

nikomatsakis (Feb 28 2020 at 15:14, on Zulip):

but I think we do need some set of folks are dedicated and eager to learn

nikomatsakis (Feb 28 2020 at 15:14, on Zulip):

but I also think it's a good reason for us to be creating and refining guidelines

nikomatsakis (Feb 28 2020 at 15:15, on Zulip):

nikomatsakis said:

but I think we do need some set of folks are dedicated and eager to learn

what I mean by this is ... like maybe at the beginning there are lots of questions, but over time if you've been doing it, I think we'll get a sense for "which are the things we judge to be most important"

nikomatsakis (Feb 28 2020 at 15:15, on Zulip):

and/or who to ask

DPC (Feb 28 2020 at 15:15, on Zulip):

Yep

nikomatsakis (Feb 28 2020 at 15:15, on Zulip):

maybe it'd be good to talk a bit more concretely?

nikomatsakis (Feb 28 2020 at 15:15, on Zulip):

about how group might interface?

simulacrum (Feb 28 2020 at 15:15, on Zulip):

Yes, I think that's likely true, though the initial period needs pretty high dedication

Santiago Pastorino (Feb 28 2020 at 15:16, on Zulip):

I can totally do that :)

DPC (Feb 28 2020 at 15:16, on Zulip):

I have a few people in mind. Just have to reach out to them to see if they are willing to join :slight_smile:

LeSeulArtichaut (Feb 28 2020 at 15:17, on Zulip):

I personnaly would like to join. I think I'm falling in the folks that are dedicated and eager to learn category

LeSeulArtichaut (Feb 28 2020 at 15:17, on Zulip):

If you need and want to accept newcomers

Santiago Pastorino (Feb 28 2020 at 15:17, on Zulip):

@LeSeulArtichaut of course :)

DPC (Feb 28 2020 at 15:18, on Zulip):

We need people who have slight knowledge about the process

nikomatsakis (Feb 28 2020 at 15:18, on Zulip):

I think we should think a bit about the "typical flow" for an issue

nikomatsakis (Feb 28 2020 at 15:18, on Zulip):

i.e., it gets freshly reported

nikomatsakis (Feb 28 2020 at 15:18, on Zulip):

and then what happens?

nikomatsakis (Feb 28 2020 at 15:18, on Zulip):

maybe for 2 or 3 categories of issues :)

nikomatsakis (Feb 28 2020 at 15:18, on Zulip):

I am annoyed I didn't try to write this up before hand

centril (Feb 28 2020 at 15:18, on Zulip):

wg-triage/t-release should apply the appropriate labels at that point

centril (Feb 28 2020 at 15:19, on Zulip):

cc the relevant people

nikomatsakis (Feb 28 2020 at 15:19, on Zulip):

(I'll start editing in the hackmd, btw)

pnkfelix (Feb 28 2020 at 15:19, on Zulip):

right. Today wg-triage tags it with T-compiler and I-nominated

pnkfelix (Feb 28 2020 at 15:19, on Zulip):

which signals it as enqueued for prioritization

nikomatsakis (Feb 28 2020 at 15:19, on Zulip):

typical workflow section

pnkfelix (Feb 28 2020 at 15:19, on Zulip):

and then each week I go through all such issues

pnkfelix (Feb 28 2020 at 15:20, on Zulip):

and (currently) attach prioritizaiton labels: P-high/P-medium/P-low.

nikomatsakis (Feb 28 2020 at 15:20, on Zulip):

centril said:

cc the relevant people

who are the "relevant people"?

simulacrum (Feb 28 2020 at 15:20, on Zulip):

We could push for triage to prioritize as well, though to be honest I have very little sense that our priority labels do... Anything at all. Maybe that's a false impression though.

pnkfelix (Feb 28 2020 at 15:20, on Zulip):

The hackmd proposes some revamping of the priority labels

centril (Feb 28 2020 at 15:20, on Zulip):

@nikomatsakis e.g. matthewjasper if the issue is about the borrow checker

simulacrum (Feb 28 2020 at 15:20, on Zulip):

Yes, I know. I'm just not sure that we can make them meaningful if people aren't looking at them

nikomatsakis (Feb 28 2020 at 15:21, on Zulip):

One of the questions that's coming up already in this discussion I think is "should we have 2 distinct triage groups?" (i.e., release team vs compiler)

simulacrum (Feb 28 2020 at 15:21, on Zulip):

Are people looking at them? Beyond Felix?

nikomatsakis (Feb 28 2020 at 15:21, on Zulip):

but I think we can sidesteep that for a second

DPC (Feb 28 2020 at 15:21, on Zulip):

I would prefer them to be seperate groups

nikomatsakis (Feb 28 2020 at 15:21, on Zulip):

and just think about "triage group" more generally and then see if things fall into one category or the other

nikomatsakis (Feb 28 2020 at 15:21, on Zulip):

in any case it seems like we want some folks to decide "is this a critical bug" or not

nikomatsakis (Feb 28 2020 at 15:22, on Zulip):

and in case people didn't see it, I'm using "critical" to mean "potential release blocker" (it's obviously true that we don't usually block releases, but I think that term gives a good sense of the kind of severity we're talking about)

nikomatsakis (Feb 28 2020 at 15:22, on Zulip):

ideally we don't block them because we always fix all the true blockers

pnkfelix (Feb 28 2020 at 15:22, on Zulip):

right. namely, it is not today's P-high

Santiago Pastorino (Feb 28 2020 at 15:22, on Zulip):

nikomatsakis said:

One of the questions that's coming up already in this discussion I think is "should we have 2 distinct triage groups?" (i.e., release team vs compiler)

I think at least these things are splitable regardless of being tackled by the same group or two different ones

nikomatsakis (Feb 28 2020 at 15:23, on Zulip):

I don't quite know what you mean, @Santiago Pastorino

nikomatsakis (Feb 28 2020 at 15:23, on Zulip):

do you mean "the question of whether to have 2 distinct groups can be split out"?

Santiago Pastorino (Feb 28 2020 at 15:23, on Zulip):

no no, sorry, my message was very confusing, the 2 sets of tasks are clearly splitable

pnkfelix (Feb 28 2020 at 15:23, on Zulip):

Santiago Pastorino said:

I think at least these things are splitable regardless of being tackled by the same group or two different ones

which things are splittable ?

Santiago Pastorino (Feb 28 2020 at 15:24, on Zulip):

the first part that's tackled by release team and what felix does today

centril (Feb 28 2020 at 15:24, on Zulip):

The typical work flow for an ICE right now is:

Santiago Pastorino (Feb 28 2020 at 15:24, on Zulip):

sorry I was stating something obvious because that's today handled by different people so of course tasks are splitable :)

nikomatsakis (Feb 28 2020 at 15:25, on Zulip):

simulacrum said:

Are people looking at them? Beyond Felix?

so @simulacrum, by this, you mean looking at the priority labels? I think I dispute the value of them at all. I think that, beyond critical vs non-criticial, we should leave the judgement of how frequently to review bugs, how important they are to fix, etc, to working groups or folks trying to maintain that area of the compiler (and I think we should separate the discussion of how to ensure that each part of the compiler has known maintainers, to some extent, though I do think that's a good thing to discuss).

DPC (Feb 28 2020 at 15:26, on Zulip):

So you want the working group to nominate the issues?

simulacrum (Feb 28 2020 at 15:26, on Zulip):

Yes, I mean priority labels

nikomatsakis (Feb 28 2020 at 15:26, on Zulip):

so when we said before "release team applies labels", we meant both area labels but also nominated labels

nagisa (Feb 28 2020 at 15:26, on Zulip):

One thing I’m noticing is that the discussion is heading towards establishing a hierarchical process (while we did have one in the past, it wasn’t very well clear cut): somebody does initial tagging that does not require much looking into the task, then somebody else triages the task more, etc.

nikomatsakis (Feb 28 2020 at 15:26, on Zulip):

(and maybe some other, e.g. "is it a regression")

nikomatsakis (Feb 28 2020 at 15:27, on Zulip):

do all new bugs get nominated? or just some that seem "sufficiently important"?

pnkfelix (Feb 28 2020 at 15:27, on Zulip):

@nagisa i don't know, it doesn't have to be hierarchical necessarily.

centril (Feb 28 2020 at 15:27, on Zulip):

@nikomatsakis I don't nominate e.g. requires-nightly I-ICEs

centril (Feb 28 2020 at 15:27, on Zulip):

that's a waste of meeting time

pnkfelix (Feb 28 2020 at 15:28, on Zulip):

@centril so instead such things don't get P-labels in the end, right?

centril (Feb 28 2020 at 15:28, on Zulip):

@pnkfelix I might just apply P-medium immediately

simulacrum (Feb 28 2020 at 15:28, on Zulip):

@nikomatsakis so I think we're on the same page in the sense that there's no real need for p-labels at all, we need to instead push for "this bug needs attention first" basically

simulacrum (Feb 28 2020 at 15:29, on Zulip):

i.e., other than p-critical it's all the same

pnkfelix (Feb 28 2020 at 15:29, on Zulip):

I think I disagree slightly, in that I do think P-labels have meaning

pnkfelix (Feb 28 2020 at 15:29, on Zulip):

but that meaning tends to not be useful enough

centril (Feb 28 2020 at 15:29, on Zulip):

Today: nobody does this “project wide” that we know of

(I do look at forward compat stuff from time to time and try to drive some to completion, but one person doing it isn't enough)

nikomatsakis (Feb 28 2020 at 15:29, on Zulip):

heh I almost wrote "Except for centril"

simulacrum (Feb 28 2020 at 15:29, on Zulip):

to be clear, I'm not saying they don't have meaning, yes, just that we're not using that meaning :)

pnkfelix (Feb 28 2020 at 15:30, on Zulip):

and thus I would prefer a labeling scheme that more directly conveys intent. The document brainstorms the idea of switcing to a "(re)visit frequency" scheme

DPC (Feb 28 2020 at 15:30, on Zulip):

I can help Centril with that if possible

nikomatsakis (Feb 28 2020 at 15:31, on Zulip):

so I've been trying to create a slightly expanded version of the workflow that @centril noted

nikomatsakis (Feb 28 2020 at 15:31, on Zulip):

I would note a few things

nikomatsakis (Feb 28 2020 at 15:31, on Zulip):

we might think of the "canonical" workflow, but with some room for "Shortcuts"

pnkfelix (Feb 28 2020 at 15:31, on Zulip):

time check: we are 31 minutes into the meeting and now I'm not sure what agenda is anymore.

nikomatsakis (Feb 28 2020 at 15:31, on Zulip):

i.e., the canonical workflow is perhaps that "release team" nominates for compiler team triage to process appropriately

nikomatsakis (Feb 28 2020 at 15:32, on Zulip):

but (a) sometimes those are going to be the same folks and (b) the compiler team publishes its guidelines and plans (ok, not yet)

nikomatsakis (Feb 28 2020 at 15:32, on Zulip):

so perhaps we can skip nomination if the route is clear and just take those steps

nikomatsakis (Feb 28 2020 at 15:32, on Zulip):

pnkfelix said:

time check: we are 31 minutes into the meeting and now I'm not sure what agenda is anymore.

heh, I thought we were still trying to work out a bit the "workflow" for bugs

pnkfelix (Feb 28 2020 at 15:32, on Zulip):

as in, our idealized one? Or capturing current practice?

nikomatsakis (Feb 28 2020 at 15:32, on Zulip):

but maybe that's too down in the weeds :)

nikomatsakis (Feb 28 2020 at 15:33, on Zulip):

I was hoping to make the idealized one

nikomatsakis (Feb 28 2020 at 15:33, on Zulip):

but I was starting a bit from what's there today

nikomatsakis (Feb 28 2020 at 15:33, on Zulip):

ok so -- release team nominates for compiler team

nikomatsakis (Feb 28 2020 at 15:34, on Zulip):

I think this is somewhat where things split, right? i.e., we hvae to assess into various "Categories" of bug

nikomatsakis (Feb 28 2020 at 15:34, on Zulip):

I think those categories fit into the "needs" that we listed int he doc earlier, more or less

nikomatsakis (Feb 28 2020 at 15:34, on Zulip):

well, maybe not quite

centril (Feb 28 2020 at 15:35, on Zulip):

T-release/triage does try to "classify" bugs before it goes to t-compiler

nikomatsakis (Feb 28 2020 at 15:35, on Zulip):

but basically there are some things that can be true (I guess non-exclusive):

pnkfelix (Feb 28 2020 at 15:36, on Zulip):

in practice today, I add the P-labels, and for some P-high bugs, I decide that its so high priority (effectively P-critical as proposed here) that I add it to the explicit agenda for the weekly meeting.

pnkfelix (Feb 28 2020 at 15:36, on Zulip):

so I think part of my goal in this proposal is to formalize that

centril (Feb 28 2020 at 15:36, on Zulip):

(I typically join in most meetings with @pnkfelix to help with adding those labels, etc.)

pnkfelix (Feb 28 2020 at 15:37, on Zulip):

via the combination of the P-critical tag and the Hz-freq tag

centril (Feb 28 2020 at 15:37, on Zulip):

One aspect of the proposal that might end up being confusing: if more than one team uses the frequently label, what team does it apply to?

centril (Feb 28 2020 at 15:37, on Zulip):

(same problem with P-high today & I-nominated today; we usually look at context of nomination etc. to see what it means)

nikomatsakis (Feb 28 2020 at 15:38, on Zulip):

I don't thikn it has to be teams even

pnkfelix (Feb 28 2020 at 15:38, on Zulip):

well, I think as long as the bug itself is getting updated with that cadence

nikomatsakis (Feb 28 2020 at 15:38, on Zulip):

i.e., one thing I was thinking about was

pnkfelix (Feb 28 2020 at 15:38, on Zulip):

then does it matter about the team assignment?

nikomatsakis (Feb 28 2020 at 15:38, on Zulip):

the async-await working group has been trying to figure out how to manage our internal triage

nikomatsakis (Feb 28 2020 at 15:38, on Zulip):

and I think that a Hz- like thing might be good, but should that be distinct labels?

pnkfelix (Feb 28 2020 at 15:39, on Zulip):

pnkfelix said:

well, I think as long as the bug itself is getting updated with that cadence

that, combined with the triage team checking in at that same frequency (for P-critical bugs) to double check that they are on track to be fixed

nikomatsakis (Feb 28 2020 at 15:39, on Zulip):

so let's stop one second and revisit what we've said

nikomatsakis (Feb 28 2020 at 15:39, on Zulip):

it seems like there is general agreement that having a group focused on pre-triage is good

nikomatsakis (Feb 28 2020 at 15:39, on Zulip):

instead of having @pnkfelix (and @centril) sort of do this ad-hoc

nikomatsakis (Feb 28 2020 at 15:40, on Zulip):

that is already good because Ithink a lot of this specific stuff we can work out a bit afterwards

pnkfelix (Feb 28 2020 at 15:40, on Zulip):

instead of having pnkfelix (and centril) sort of do this ad-hoc

(and @simulacrum , etc)

nikomatsakis (Feb 28 2020 at 15:40, on Zulip):

we also talked about this idea of critical bugs as a key area of focus -- things that are potential release blockers

nikomatsakis (Feb 28 2020 at 15:41, on Zulip):

and it seems like where there was less clarity was just what we should do with those critical bugs

pnkfelix (Feb 28 2020 at 15:41, on Zulip):

I think establishing the expectations/scope for this group is important

nikomatsakis (Feb 28 2020 at 15:41, on Zulip):

i.e., that's where the Hz- stuff comes in

nikomatsakis (Feb 28 2020 at 15:41, on Zulip):

I agree, @pnkfelix, that's probably a good scope for the meeting, do you think there is clarity on that? do you want to try and re-state?

centril (Feb 28 2020 at 15:43, on Zulip):

I think it's good to ask what's different about this group as compared to wg-triage

nagisa (Feb 28 2020 at 15:43, on Zulip):

I feel that unless we very clearly denote what various priority levels mean and what things can be in there, we’ll just end up filling the category with a ton of stuff again.

pnkfelix (Feb 28 2020 at 15:43, on Zulip):

So lets see. Group needs to: 1. identify new P-critical bugs from amongst the nominated issues, 2. revisit the set of all P-critical bugs to double check that they have logged progress.

pnkfelix (Feb 28 2020 at 15:43, on Zulip):

@nagisa right, as soon as the P-critical set becomes too big to realistically visit on a weekly basis, we have collectively failed.

pnkfelix (Feb 28 2020 at 15:43, on Zulip):

(this is what happened with P-high today.)

pnkfelix (Feb 28 2020 at 15:44, on Zulip):

my hope is that by imposing a higher bar for marking something as P-critical, and a more aggressive demoting of such bugs to non P-critical, will help address that.

nagisa (Feb 28 2020 at 15:44, on Zulip):

I feel that we managed to get this right with the regression- tag – we generally tend to fix them very quickly, or, if we don’t manage to (which happens super rarely) it becomes a regression-from-stable-to-stable and everybody forgets about it.

nikomatsakis (Feb 28 2020 at 15:44, on Zulip):

in terms of expectation, I think there is also (3) of "prepare key highlights to guide the main meeting"

pnkfelix (Feb 28 2020 at 15:44, on Zulip):

centril said:

I think it's good to ask what's different about this group as compared to wg-triage

its possible wg-triage ends up doing this.

nikomatsakis (Feb 28 2020 at 15:45, on Zulip):

nagisa said:

I feel that we managed to get this right with the regression- tag – we generally tend to fix them very quickly, or, if we don’t manage to (which happens super rarely) it becomes a regression-from-stable-to-stable and everybody forgets about it.

yes, and I think we should kind of formalize this concept, basically. The vast majority of critical bugs, I imagine, will be regressions

pnkfelix (Feb 28 2020 at 15:45, on Zulip):

nikomatsakis said:

in terms of expectation, I think there is also (3) of "prepare key highlights to guide the main meeting"

right. In particular, a list of P-critical bugs that have not made progress or are in danger of slipping into release.

pnkfelix (Feb 28 2020 at 15:45, on Zulip):

is a clear candidate

centril (Feb 28 2020 at 15:46, on Zulip):

(this is what happened with P-high today.)

(worth recognizing the sheer amount of incoming bugs, many of them important, that we get... we do have a staffing issue -- but I think it's the point of reorganizing to get more staff to fix things)

nikomatsakis (Feb 28 2020 at 15:46, on Zulip):

@pnkfelix you wrote "identify new P-critical bugs from amongst the nominated issues" but I think there is a slightly larger mandate, to actually just review nominated issues and process them

pnkfelix (Feb 28 2020 at 15:47, on Zulip):

hmm

nikomatsakis (Feb 28 2020 at 15:47, on Zulip):

into finer-grained categories (critical, needs team discussion, delegate)

pnkfelix (Feb 28 2020 at 15:47, on Zulip):

not sure about that, beyond tagging something as triaged and tagged as non-critical in some manner?

nikomatsakis (Feb 28 2020 at 15:47, on Zulip):

that is "processing it"

nikomatsakis (Feb 28 2020 at 15:47, on Zulip):

or that is what I meant by "delegate", I guess

nikomatsakis (Feb 28 2020 at 15:47, on Zulip):

(in particular, we don't expect to review the nominated issues in the meeting)

nikomatsakis (Feb 28 2020 at 15:48, on Zulip):

(unless they're added to the doc of "things worth talking about")

pnkfelix (Feb 28 2020 at 15:48, on Zulip):

the thing I don't want is for the pre-triage working group to get bogged down trying to decide what to do with the non-critical bugs

pnkfelix (Feb 28 2020 at 15:49, on Zulip):

that is something I want to make explicitly not part of scope; it needs to be delegated as quickly as possible to the relevant subteam.

nikomatsakis (Feb 28 2020 at 15:49, on Zulip):

yes, I think they should just decide "critical, must be fixed" or "not critical, cc some folks and move on"

Nell Shamrell-Harrington (Feb 28 2020 at 15:49, on Zulip):

how will they know who to cc for what?

LeSeulArtichaut (Feb 28 2020 at 15:49, on Zulip):

About delegation, maybe do something with the experts map? Increase its visibility or readability?

nikomatsakis (Feb 28 2020 at 15:49, on Zulip):

I think we should defer that

nikomatsakis (Feb 28 2020 at 15:50, on Zulip):

since we have only 10 minutes left :)

pnkfelix (Feb 28 2020 at 15:50, on Zulip):

okay I think I see your point @nikomatsakis again now, your point is that some bugs they won't know whether it needs team discussion or should go direct to P-critical?

simulacrum (Feb 28 2020 at 15:50, on Zulip):

P-critical bugs would presumably fall into (roughly) meeting agenda organizer's hands for that team

simulacrum (Feb 28 2020 at 15:50, on Zulip):

e.g. still @pnkfelix

centril (Feb 28 2020 at 15:50, on Zulip):

knowing those things requires a certain amount of experience, maybe having hacked on the compiler a bit, knowing the right people, etc.

nikomatsakis (Feb 28 2020 at 15:50, on Zulip):

mm, maybe? my point was that from your description, it read to me as if they were "scanning nominated bugs", but I think of it more like "They are processing the nominations"

pnkfelix (Feb 28 2020 at 15:51, on Zulip):

Nell Shamrell-Harrington said:

how will they know who to cc for what?

I think in principle there should be teams focusing on maintance for each area of the compiler

nikomatsakis (Feb 28 2020 at 15:51, on Zulip):

(yes, we have some notes on that in the doc)

centril (Feb 28 2020 at 15:51, on Zulip):

e.g. I know to cc matthewjasper when I see something specific in the backtrace of an ICE

nikomatsakis (Feb 28 2020 at 15:51, on Zulip):

I think we can do better at documenting some of these things but there will still be some element of experience, definitely

nikomatsakis (Feb 28 2020 at 15:51, on Zulip):

ideally you'd not cc matthewjasper, for example, but "borrowck maintainance" or something :)

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

which I guess is sort of @pnkfelix's point

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

but anyway

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

nikomatsakis said:

mm, maybe? my point was that from your description, it read to me as if they were "scanning nominated bugs", but I think of it more like "They are processing the nominations"

maybe we're just saying the same thing here @pnkfelix?

centril (Feb 28 2020 at 15:52, on Zulip):

ideally you'd not cc matthewjasper, for example, but "borrowck maintainance" or something :slight_smile:

Those are the same thing? :D

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

I mainly want to establish that nobody else ever uses a github query for "all nominated things"

pnkfelix (Feb 28 2020 at 15:52, on Zulip):

pnkfelix said:

okay I think I see your point nikomatsakis again now, your point is that some bugs they won't know whether it needs team discussion or should go direct to P-critical?

in practice, I'm not sure we need to distinguish between "pre-triage didn't know how to prioritize" vs "P-critical"

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

(or should have to)

pnkfelix (Feb 28 2020 at 15:52, on Zulip):

in either case, if its not being fixed

pnkfelix (Feb 28 2020 at 15:52, on Zulip):

then it needs to be part of T-Compiler agenda

nikomatsakis (Feb 28 2020 at 15:52, on Zulip):

sure, I think that means "they add it to the meeting agenda"

DPC (Feb 28 2020 at 15:53, on Zulip):

It would be handy if there's a doc with different areas and a person and probably a backup to ping

simulacrum (Feb 28 2020 at 15:53, on Zulip):

"it" = "p-critical bug"?

nikomatsakis (Feb 28 2020 at 15:53, on Zulip):

no

nikomatsakis (Feb 28 2020 at 15:53, on Zulip):

not necessarily, I think

pnkfelix (Feb 28 2020 at 15:53, on Zulip):

"it" == "not non-critical bug that is not making progress"

nikomatsakis (Feb 28 2020 at 15:53, on Zulip):

I think "it" is "some bug where it is unclera what to do"

simulacrum (Feb 28 2020 at 15:53, on Zulip):

pnkfelix said:

in either case, if its not being fixed

A lot of things aren't being fixed is I guess my point

nikomatsakis (Feb 28 2020 at 15:53, on Zulip):

but typically that would be critical bug not making progress

LeSeulArtichaut (Feb 28 2020 at 15:53, on Zulip):

DPC said:

It would be handy if there's a doc with different areas and a person and probably a backup to ping

Isn't that what the experts map is for?

simulacrum (Feb 28 2020 at 15:53, on Zulip):

like, we have 5000+ issues that are mostly not making progress, so I'm confusd

DPC (Feb 28 2020 at 15:54, on Zulip):

Experts map?

simulacrum (Feb 28 2020 at 15:54, on Zulip):

oh, not non-critical

simulacrum (Feb 28 2020 at 15:54, on Zulip):

I see

nikomatsakis (Feb 28 2020 at 15:54, on Zulip):

sorry :)

pnkfelix (Feb 28 2020 at 15:54, on Zulip):

sorry I was trying to be cute

LeSeulArtichaut (Feb 28 2020 at 15:54, on Zulip):

DPC said:

Experts map?

This

DPC (Feb 28 2020 at 15:54, on Zulip):

Ah thanks that if is up to date should be sufficient

pnkfelix (Feb 28 2020 at 15:54, on Zulip):

should have written "potentially or definitely critical, and not being fixed"

nikomatsakis (Feb 28 2020 at 15:54, on Zulip):

the high-level view here:

centril (Feb 28 2020 at 15:54, on Zulip):

like, we have 5000+ issues that are mostly not making progress, so I'm confusd

A lot of those are feature requests, requests for diagnostics improvements, tracking issues, etc.

LeSeulArtichaut (Feb 28 2020 at 15:55, on Zulip):

Oops, didn't mean to write that here

pnkfelix (Feb 28 2020 at 15:55, on Zulip):

Yes, we don't have a ready way to actually tell if our issue database is completely out of control or not today

pnkfelix (Feb 28 2020 at 15:55, on Zulip):

or at least I don't

DPC (Feb 28 2020 at 15:55, on Zulip):

I do often sweep through those issues, not easy to close

nikomatsakis (Feb 28 2020 at 15:56, on Zulip):

I think there is a separable discussion of "the group, perhaps in conjunction with release team wg-triage, also owns the issue tracker" and hence it is totally ins cope for them to adjust labels or practices there

nikomatsakis (Feb 28 2020 at 15:56, on Zulip):

in order to make things more efficient, better, etc

nikomatsakis (Feb 28 2020 at 15:57, on Zulip):

but their main job is to cultivate the meeting and to ensure that critical bugs don't get into a release, with a secondary goal of "relevant folks know about the bugs that they should know about"

nikomatsakis (Feb 28 2020 at 15:57, on Zulip):

ok, 3 mintues left

nikomatsakis (Feb 28 2020 at 15:57, on Zulip):

I am wondering if it is possible to preprae a kind of bullet charter

pnkfelix (Feb 28 2020 at 15:57, on Zulip):

my current plan is to continue doing pre-triage as I have been

nikomatsakis (Feb 28 2020 at 15:57, on Zulip):

probably not in 3 minutes :)

pnkfelix (Feb 28 2020 at 15:58, on Zulip):

We can asynchronously try to establish charter for this group

pnkfelix (Feb 28 2020 at 15:58, on Zulip):

there remains question of whether it should be melded with wg-triage in some way

simulacrum (Feb 28 2020 at 15:59, on Zulip):

I would lean strongly towards no, at least not initially

centril (Feb 28 2020 at 15:59, on Zulip):

(I have noticed a somewhat of an explosion of new issues, getting opened faster than we can close... we were holding below 5000 for a long time and then we couldn't do it anymore)

pnkfelix (Feb 28 2020 at 15:59, on Zulip):

@Santiago Pastorino wasn't wrong that these can be two separate tasks. But there may be efficiency gains from trying to synergize efforts there

simulacrum (Feb 28 2020 at 15:59, on Zulip):

if membership is the same or strongly overlapping that's fine though

DPC (Feb 28 2020 at 15:59, on Zulip):

The triage wg mainly consists of newcomers though

centril (Feb 28 2020 at 15:59, on Zulip):

So we need "wg-advanced-triage"? :D

Santiago Pastorino (Feb 28 2020 at 15:59, on Zulip):

pnkfelix said:

Santiago Pastorino wasn't wrong that these can be two separate tasks. But there may be efficiency gains from trying to synergize efforts there

definitely, I even wonder if I should participate on both efforts

pnkfelix (Feb 28 2020 at 16:00, on Zulip):

triaged-squared

pnkfelix (Feb 28 2020 at 16:00, on Zulip):

cubaged

pnkfelix (Feb 28 2020 at 16:00, on Zulip):

okay that's enough out of me

pnkfelix (Feb 28 2020 at 16:01, on Zulip):

maybe we should make a separate topic

nikomatsakis (Feb 28 2020 at 16:01, on Zulip):

it seems like it makes sense to have triage efforts that are progressively narrower

nikomatsakis (Feb 28 2020 at 16:01, on Zulip):

in terms of expertise

pnkfelix (Feb 28 2020 at 16:01, on Zulip):

for people to join who are interested in participating a drafting of charter?

DPC (Feb 28 2020 at 16:01, on Zulip):

I mean in future we can always merge them if we consider one of them redundant

nikomatsakis (Feb 28 2020 at 16:01, on Zulip):

anyway, I started toying with a mission statement but yes let's end the meeting now ;)

Santiago Pastorino (Feb 28 2020 at 16:02, on Zulip):

pnkfelix said:

for people to join who are interested in participating a drafting of charter?

what does that means?

nikomatsakis (Feb 28 2020 at 16:02, on Zulip):

pnkfelix said:

for people to join who are interested in participating a drafting of charter?

yep, I will spin up a topic in #t-compiler/wg-meta I think

pnkfelix (Feb 28 2020 at 16:02, on Zulip):

Santiago Pastorino said:

pnkfelix said:

for people to join who are interested in participating a drafting of charter?

what does that means?

well, what I was thinking is: If a person is interested in assisting and wants to help write down what the group will do

pnkfelix (Feb 28 2020 at 16:03, on Zulip):

then they could join in that area

Santiago Pastorino (Feb 28 2020 at 16:03, on Zulip):

yeah, I can help

LeSeulArtichaut (Feb 28 2020 at 16:03, on Zulip):

I can, too

centril (Feb 28 2020 at 16:03, on Zulip):

I should probably be in this group but I wonder if I have time...

pnkfelix (Feb 28 2020 at 16:04, on Zulip):

well

DPC (Feb 28 2020 at 16:04, on Zulip):

You have a 25 min break xD

pnkfelix (Feb 28 2020 at 16:04, on Zulip):

my own hope is that by adding people

pnkfelix (Feb 28 2020 at 16:04, on Zulip):

it will end up being less of a burden to each participant

pnkfelix (Feb 28 2020 at 16:04, on Zulip):

or maybe @centril you just meant "do I have time this moment to participate in the charter discussion" ?

centril (Feb 28 2020 at 16:04, on Zulip):

is the goal for the triage group to have meetings?

nikomatsakis (Feb 28 2020 at 16:05, on Zulip):

created #t-compiler/wg-meta > charter for focused and efficient triage

pnkfelix (Feb 28 2020 at 16:05, on Zulip):

Does any group have "have meetings" as a goal (rather than a means to an end) ?

centril (Feb 28 2020 at 16:05, on Zulip):

@pnkfelix not the charter discussion but the group itself

DPC (Feb 28 2020 at 16:05, on Zulip):

I was thinking we could have a meeting to aggregate the efforts and prepare the report which Niko was talking about last week

centril (Feb 28 2020 at 16:06, on Zulip):

@pnkfelix let me rephrase: will the group have regular meetings? :D

nikomatsakis (Feb 28 2020 at 16:06, on Zulip):

I would imagine the group operating in some combination of sync/async, that's to be worked out by group itself imo

nikomatsakis (Feb 28 2020 at 16:06, on Zulip):

DPC said:

I was thinking we could have a meeting to aggregate the efforts and prepare the report which Niko was talking about last week

basically this

pnkfelix (Feb 28 2020 at 16:06, on Zulip):

I suppose its up to the group to decide how it will achieve the visit frequency we hope for

DPC (Feb 28 2020 at 16:06, on Zulip):

Yep

pnkfelix (Feb 28 2020 at 16:07, on Zulip):

I made a choice to try to do pre-triage at roughly the same time, or at least same day, each week

pnkfelix (Feb 28 2020 at 16:07, on Zulip):

Anyway

pnkfelix (Feb 28 2020 at 16:07, on Zulip):

time is up

pnkfelix (Feb 28 2020 at 16:07, on Zulip):

thanks to @T-compiler/meeting for attending!!!

nikomatsakis (Mar 12 2020 at 11:36, on Zulip):

Hey -- can somebody prepare the minutes for this meeting and post a PR to compiler-team (similar to compiler-team#255). I think this would be done by (a) starting with the hackmd we prepared and adding links to notable comments or points, it may not require much editing honestly, I don't remember. @Santiago Pastorino maybe you want to do that?

Santiago Pastorino (Mar 12 2020 at 12:45, on Zulip):

yes, I can do that

Last update: May 29 2020 at 17:55UTC