Stream: t-compiler/wg-meta

Topic: meeting 2020.02.13


Santiago Pastorino (Feb 13 2020 at 19:02, on Zulip):

@T-compiler/WG-meta :wave:

DPC (Feb 13 2020 at 19:02, on Zulip):

:wave:

Santiago Pastorino (Feb 13 2020 at 19:03, on Zulip):

@nikomatsakis you around?

Santiago Pastorino (Feb 13 2020 at 19:03, on Zulip):

during last meeting we landed Cleanup Crew ICE Breakers

Santiago Pastorino (Feb 13 2020 at 19:03, on Zulip):

we needed to syncronize a bit with @DPC about this

Santiago Pastorino (Feb 13 2020 at 19:04, on Zulip):

and also I saw some discussions about a Diagnostic ICE breaker group

Santiago Pastorino (Feb 13 2020 at 19:04, on Zulip):

I guess those are some of the things we could talk about?

nikomatsakis (Feb 13 2020 at 19:05, on Zulip):

Hey y'all

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

So, yes, we landed cleanup crew , and we started putting it into practice

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

exciting

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

one thing I did want to say is that I want to create a meeting request

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

to discuss how we're doing triage

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

i.e. a "compiler team design meeting"

davidtwco (Feb 13 2020 at 19:06, on Zulip):

:wave:

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

I was hoping @DPC that we can also get you involved in said discussion

nikomatsakis (Feb 13 2020 at 19:06, on Zulip):

or perhaps pre-discussion

nikomatsakis (Feb 13 2020 at 19:07, on Zulip):

because I suspect we should be coordinating better!

DPC (Feb 13 2020 at 19:07, on Zulip):

Sure

Santiago Pastorino (Feb 13 2020 at 19:07, on Zulip):

:+1:

nikomatsakis (Feb 13 2020 at 19:07, on Zulip):

I was planning on trying to focus the meeting on the question of:

nikomatsakis (Feb 13 2020 at 19:08, on Zulip):

and, more to the point, I sort of think our goals ought to be somewhat narrower: we should be focusing our energy on tracking the "most cricitical" bugs (regressions, super bad bugs, things that break a lot of the ecosystem)

Santiago Pastorino (Feb 13 2020 at 19:08, on Zulip):

those are nice questions

nikomatsakis (Feb 13 2020 at 19:08, on Zulip):

I may just copy and paste that into the design meeting request

Santiago Pastorino (Feb 13 2020 at 19:08, on Zulip):

:+1:

DPC (Feb 13 2020 at 19:08, on Zulip):

I agree. The problem generally is the triagers do not know about what's important and what's not

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

interesting, so I think one of the things we should be doing

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

is trying to come up with more specific guidelines on this

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

unfortunately I think @pnkfelix is out today

DPC (Feb 13 2020 at 19:09, on Zulip):

My personal focus is try to and get the older prs merged as much as possible

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

but I think they'd be best positioned to try and write out the "de facto" rules we use

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

My personal focus is try to and get the older prs merged as much as possible

interesting

nikomatsakis (Feb 13 2020 at 19:09, on Zulip):

this is definitely a different sort of triage

nikomatsakis (Feb 13 2020 at 19:10, on Zulip):

though a laudable goal

Santiago Pastorino (Feb 13 2020 at 19:10, on Zulip):

I was going to say that

DPC (Feb 13 2020 at 19:10, on Zulip):

Ye I understand

nikomatsakis (Feb 13 2020 at 19:10, on Zulip):

i.e., I'm thinking primarily about "how do we avoid shipping horrible and embarassing bugs" :)

Santiago Pastorino (Feb 13 2020 at 19:10, on Zulip):

I think we have different goals

nikomatsakis (Feb 13 2020 at 19:10, on Zulip):

note that I've lowered the goal from "how do we ensure the general quality of rustc" ;)

nikomatsakis (Feb 13 2020 at 19:11, on Zulip):

I think that's a good goal too but I think (a) it often works out and (b) I'm not sure that we have enough dedicated person power to devote to sort of "guaranteeing" that outcome :)

Santiago Pastorino (Feb 13 2020 at 19:11, on Zulip):

my understanding is that triage is trying to keep the issue tracker in a sane and manageable way, which is a nice thing to persue

DPC (Feb 13 2020 at 19:11, on Zulip):

Just to be clear, I meant by personal focus I meant what I'm generally looking at, not what I expect from this :slight_smile:

Santiago Pastorino (Feb 13 2020 at 19:11, on Zulip):

but what Niko is saying means more like prioritizing some bad bugs and first care about them :)

nikomatsakis (Feb 13 2020 at 19:12, on Zulip):

I think it's pretty useful @DPC to think about what the "role" of (e.g.) release team triage vs compiler team triage is

nikomatsakis (Feb 13 2020 at 19:12, on Zulip):

and I could imagine that it would indeed be different focus

nikomatsakis (Feb 13 2020 at 19:12, on Zulip):

but I also am pretty unconvinced that the way we're approaching things now is the best use of time

DPC (Feb 13 2020 at 19:12, on Zulip):

Yes I get that

nikomatsakis (Feb 13 2020 at 19:12, on Zulip):

(that is, the way compiler team is approaching things)

DPC (Feb 13 2020 at 19:12, on Zulip):

I agree

nikomatsakis (Feb 13 2020 at 19:12, on Zulip):

I'm not even sure that a general triage meeting (where whole team theoretically attends) makes sense

nikomatsakis (Feb 13 2020 at 19:13, on Zulip):

it seems like it might be better to have a smaller group that is generally watching over things and just highlighting cases where "horrible bug is not making progress!"

DPC (Feb 13 2020 at 19:13, on Zulip):

Yes

DPC (Feb 13 2020 at 19:14, on Zulip):

So I'm assuming we are tracking ICEs and checking if they are fixed or not?

Santiago Pastorino (Feb 13 2020 at 19:15, on Zulip):

I guess not necessarily just ICEs

Santiago Pastorino (Feb 13 2020 at 19:16, on Zulip):

it's worse a bug report about vec.push (to say something) performing very badly than an ICE on a very edge case use of the language

DPC (Feb 13 2020 at 19:16, on Zulip):

Yeah a smaller group makes sense especially people who have some background of how things work

nikomatsakis (Feb 13 2020 at 19:16, on Zulip):

I would say

nikomatsakis (Feb 13 2020 at 19:16, on Zulip):

yes, ICE is definitely too broad

nikomatsakis (Feb 13 2020 at 19:17, on Zulip):

like, I just think we don't have enough people to try and track and guarantee every ICE gets fixed

nikomatsakis (Feb 13 2020 at 19:17, on Zulip):

but it is interesting to try and narrow down when an ICE is important

Santiago Pastorino (Feb 13 2020 at 19:17, on Zulip):

yep

nikomatsakis (Feb 13 2020 at 19:17, on Zulip):

e.g., @pnkfelix has mentioned stuff like "if there is a useful error reported before the ICE, maybe it's not so bad"

Santiago Pastorino (Feb 13 2020 at 19:17, on Zulip):

I'd generally say, it depends how much a problem is hitten by users and how bad is it

nikomatsakis (Feb 13 2020 at 19:17, on Zulip):

and I guess it matters how easy it is to encounter etc

Santiago Pastorino (Feb 13 2020 at 19:18, on Zulip):

e.g., pnkfelix has mentioned stuff like "if there is a useful error reported before the ICE, maybe it's not so bad"

agreed

DPC (Feb 13 2020 at 19:18, on Zulip):

True

nikomatsakis (Feb 13 2020 at 19:19, on Zulip):

this has been kind of useful :)

Santiago Pastorino (Feb 13 2020 at 19:19, on Zulip):

and I guess it matters how easy it is to encounter etc

yep, this is why I'm saying I guess it depends on how easily people will hit the issue + how bad is the issue and another factor to consider is how easy will be to fix it

nikomatsakis (Feb 13 2020 at 19:19, on Zulip):

maybe I'll make a hackmd with some notes

nikomatsakis (Feb 13 2020 at 19:19, on Zulip):

hackmd

nikomatsakis (Feb 13 2020 at 19:20, on Zulip):

PS, separate from this topic,

Santiago Pastorino (Feb 13 2020 at 19:20, on Zulip):

it may not be very easy for triggers to figure out how easy it is to fix it but maybe you could guess a bit by looking at the error :)

nikomatsakis (Feb 13 2020 at 19:20, on Zulip):

it's probably a good idea for us -- maybe at another meeting? in a zulip topic? -- to try and figure out a bit what our "next few roadmap" items will be

nikomatsakis (Feb 13 2020 at 19:20, on Zulip):

I think there is some stuff for us to follow up on

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

there is of course the diagnostic ICE-breaker idea, but I think at this point we know the "script" there

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

mm we should probably make a forge page

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

that describes "how to propose and create an ICE-breaker group"

Santiago Pastorino (Feb 13 2020 at 19:21, on Zulip):

yes to everything

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

but the other thing is

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

we had that "working group retro" a bit back

nikomatsakis (Feb 13 2020 at 19:21, on Zulip):

and there were some good lessons there

nikomatsakis (Feb 13 2020 at 19:22, on Zulip):

e.g. changes to how we do the wg updates

nikomatsakis (Feb 13 2020 at 19:22, on Zulip):

and I would kind of like to go back over that and see if we can find some concrete work items in there

Santiago Pastorino (Feb 13 2020 at 19:23, on Zulip):

did we do some writings that contains the outputs of the retro?

nikomatsakis (Feb 13 2020 at 19:24, on Zulip):

yes, I think so

nikomatsakis (Feb 13 2020 at 19:24, on Zulip):

let me check

nikomatsakis (Feb 13 2020 at 19:24, on Zulip):

oh taht reminds me

nikomatsakis (Feb 13 2020 at 19:24, on Zulip):

that I need to upload minutes from the last design meeting

nikomatsakis (Feb 13 2020 at 19:25, on Zulip):

did we do some writings that contains the outputs of the retro?

design meeting minutes are here

nikomatsakis (Feb 13 2020 at 19:26, on Zulip):

that I need to upload minutes from the last design meeting

that is, the one discussed in #t-compiler > design meeting 2020.02.07

Santiago Pastorino (Feb 13 2020 at 19:30, on Zulip):

read most of it

Santiago Pastorino (Feb 13 2020 at 19:30, on Zulip):

@nikomatsakis how should we continue?

nikomatsakis (Feb 13 2020 at 19:30, on Zulip):

continue?

nikomatsakis (Feb 13 2020 at 19:30, on Zulip):

well, I think this meeting is roughly at an end :)

nikomatsakis (Feb 13 2020 at 19:31, on Zulip):

I guess the question is

Santiago Pastorino (Feb 13 2020 at 19:31, on Zulip):

yeah, that's what I meant

Santiago Pastorino (Feb 13 2020 at 19:31, on Zulip):

what are the next steps?

nikomatsakis (Feb 13 2020 at 19:31, on Zulip):

well one thing -- next week I don't think I can make this meeting slot

DPC (Feb 13 2020 at 19:31, on Zulip):

My question is how do we ensure we don't ship an embarrassing feature

nikomatsakis (Feb 13 2020 at 19:31, on Zulip):

we could try to have some async conversation

Santiago Pastorino (Feb 13 2020 at 19:31, on Zulip):

async conversation makes sense

DPC (Feb 13 2020 at 19:32, on Zulip):

Works for me

nikomatsakis (Feb 13 2020 at 19:32, on Zulip):

My question is how do we ensure we don't ship an embarrassing feature

can you elaboate a bit on this, @DPC?

nikomatsakis (Feb 13 2020 at 19:32, on Zulip):

I'm not sure what you mean by "how do we ensure"

nikomatsakis (Feb 13 2020 at 19:32, on Zulip):

(but I do have to run, afk)

DPC (Feb 13 2020 at 19:33, on Zulip):

I mean our goal is to avoid catastrophic bugs

DPC (Feb 13 2020 at 19:35, on Zulip):

Nevermind I think I got my answer

nikomatsakis (Feb 13 2020 at 23:00, on Zulip):

opened compiler team meeting proposal

nikomatsakis (Feb 13 2020 at 23:01, on Zulip):

cc @pnkfelix on that :point_up:

Santiago Pastorino (Feb 14 2020 at 13:57, on Zulip):

@nikomatsakis isn't it better to get rid of the hackmd document and just embed the questions to answer in the proposal?

Santiago Pastorino (Feb 14 2020 at 13:57, on Zulip):

or are you aiming for people to easily add stuff in there?

Santiago Pastorino (Feb 14 2020 at 13:58, on Zulip):

because other than the questions to answer in the hackmd the rest is repeated from what's in the github proposal

Santiago Pastorino (Feb 14 2020 at 13:59, on Zulip):

Santiago Pastorino said:

or are you aiming for people to easily add stuff in there?

if the intention is that one, why don't we put "questions to answer for an efficient tracking" title and get rid of the previous parts

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

/me shrugs

Santiago Pastorino (Feb 14 2020 at 13:59, on Zulip):

yeah, not really important I guess :)

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

good questions :)

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

I guess my intent was

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

that we would refine the details in hackmd

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

(potentially)

nikomatsakis (Feb 14 2020 at 13:59, on Zulip):

in general we moved from "write proposal in a hackmd" to "write a few details in the issue"

nikomatsakis (Feb 14 2020 at 14:00, on Zulip):

so I was thinking of the issue as an "advertisement" for the hackmd, which is the full thing

nikomatsakis (Feb 14 2020 at 14:00, on Zulip):

but I'm good with whatever structure you think best :)

Santiago Pastorino (Feb 14 2020 at 14:00, on Zulip):

it's fine :P

Santiago Pastorino (Feb 14 2020 at 14:00, on Zulip):

now that I have commented I'm thinking that it doesn't worth to spending more time :)

nikomatsakis (Feb 14 2020 at 14:01, on Zulip):

<3

Last update: Sep 27 2020 at 13:15UTC