Stream: t-compiler/wg-meta

Topic: meeting 2020.03.26


Santiago Pastorino (Mar 26 2020 at 18:00, on Zulip):

@T-compiler/WG-meta :wave:

Santiago Pastorino (Mar 26 2020 at 18:02, on Zulip):

pretty sure that @nikomatsakis would like to talk about draft process that we should post as an RFC

Santiago Pastorino (Mar 26 2020 at 18:02, on Zulip):

and if we have time I'd like to talk about the ideas stated in #t-compiler > active working groups

Santiago Pastorino (Mar 26 2020 at 18:02, on Zulip):

in particular I've opened compiler-team#264

nikomatsakis (Mar 26 2020 at 18:08, on Zulip):

gah

nikomatsakis (Mar 26 2020 at 18:08, on Zulip):

I'm here :)

nikomatsakis (Mar 26 2020 at 18:09, on Zulip):

mostly I'd like to move that process forward

nikomatsakis (Mar 26 2020 at 18:09, on Zulip):

I'm actually feeling pretty good about the whole "find someone who likes the idea and post it during the announcements at a triage meeting" part of it

nikomatsakis (Mar 26 2020 at 18:10, on Zulip):

I think we should probably specify that we are going to create a #t-compiler/major-change-proposals stream

nikomatsakis (Mar 26 2020 at 18:10, on Zulip):

and then the process is:

nikomatsakis (Mar 26 2020 at 18:10, on Zulip):
nikomatsakis (Mar 26 2020 at 18:10, on Zulip):
DPC (Mar 26 2020 at 18:11, on Zulip):

:wave:

nikomatsakis (Mar 26 2020 at 18:11, on Zulip):
nikomatsakis (Mar 26 2020 at 18:11, on Zulip):
nikomatsakis (Mar 26 2020 at 18:12, on Zulip):

I guess that's what we already wrote

Santiago Pastorino (Mar 26 2020 at 18:13, on Zulip):

makes sense :+1:

nikomatsakis (Mar 26 2020 at 18:13, on Zulip):

I think we should discuss "blocking objections" a bit

nikomatsakis (Mar 26 2020 at 18:13, on Zulip):

I don't want to wind up in the situation of "indefinite blocks" that we get elsewhere

nikomatsakis (Mar 26 2020 at 18:13, on Zulip):

at the same time, I don't want us to just barrel forward

nikomatsakis (Mar 26 2020 at 18:14, on Zulip):

there are two pathologies and it's hard to navigate between them

It all kind of comes down I guess to overall bandwidth.

Santiago Pastorino (Mar 26 2020 at 18:14, on Zulip):

yeah

Santiago Pastorino (Mar 26 2020 at 18:15, on Zulip):

yeah, so I think the bandwidth and energy factor is really important

Santiago Pastorino (Mar 26 2020 at 18:15, on Zulip):

if you don't have time or energy you may overlook some stuff and end with proposals that wouldn't have been accepted

Santiago Pastorino (Mar 26 2020 at 18:16, on Zulip):

or something like that

Santiago Pastorino (Mar 26 2020 at 18:16, on Zulip):

well that's the too hard to block, your summary was perfect :)

pnkfelix (Mar 26 2020 at 18:18, on Zulip):

if someone blocks, can we just deal with it at a design meeting?

nikomatsakis (Mar 26 2020 at 18:18, on Zulip):

I think so

nikomatsakis (Mar 26 2020 at 18:18, on Zulip):

This actually connects to the "consitution" question

pnkfelix (Mar 26 2020 at 18:18, on Zulip):

obviously that could still cause problems, if someone is being troublesome and that ends up filling up design meeting slots

pnkfelix (Mar 26 2020 at 18:19, on Zulip):

but at least we shouldn't end up with "indefinite blocking"; progress should be made eventually

nikomatsakis (Mar 26 2020 at 18:19, on Zulip):

i.e., I'd ultimately like us to have a clearer set of "deciders", @pnkfelix and I were talking about the idea of a kind of "technical steering committee" that is a few folks who are making the final decisions

nikomatsakis (Mar 26 2020 at 18:19, on Zulip):

(at least in complex cases...)

nikomatsakis (Mar 26 2020 at 18:20, on Zulip):

but I don't think that has to be part of this specific ida

nikomatsakis (Mar 26 2020 at 18:20, on Zulip):

I think it suffices for now to say "when an objection is raised, you should discuss, and if it you want to create space for the discussion, file a design meeting"

nikomatsakis (Mar 26 2020 at 18:21, on Zulip):

I maybe wouldn't even phrase as 'objection', maybe there's a better term

nikomatsakis (Mar 26 2020 at 18:21, on Zulip):

I guess we call them "concerns" elsewhere

Santiago Pastorino (Mar 26 2020 at 18:21, on Zulip):

nikomatsakis said:

i.e., I'd ultimately like us to have a clearer set of "deciders", pnkfelix and I were talking about the idea of a kind of "technical steering committee" that is a few folks who are making the final decisions

is this really needed? isn't enough having you and @pnkfelix as compiler team leaders giving the final call?

nikomatsakis (Mar 26 2020 at 18:22, on Zulip):

probably

Santiago Pastorino (Mar 26 2020 at 18:22, on Zulip):

in the cases where that's needed I guess we will mostly have consensus :)

nikomatsakis (Mar 26 2020 at 18:22, on Zulip):

technicaly, though we've rarely exercised this,

nikomatsakis (Mar 26 2020 at 18:22, on Zulip):

the original team rfc did state that "in the case where consensus cannot be reached, the team lead decides"

nikomatsakis (Mar 26 2020 at 18:22, on Zulip):

(something I saw recently when re-reading)

Santiago Pastorino (Mar 26 2020 at 18:23, on Zulip):

nikomatsakis said:

I maybe wouldn't even phrase as 'objection', maybe there's a better term

yeah more positive words are better than object and concern :)

nikomatsakis (Mar 26 2020 at 18:23, on Zulip):

I have a question

nikomatsakis (Mar 26 2020 at 18:23, on Zulip):
nikomatsakis (Mar 26 2020 at 18:23, on Zulip):

this is something I just wrote -- it's a minor detail -- but I'm basically just wondering about

nikomatsakis (Mar 26 2020 at 18:23, on Zulip):

I think that FCP should always "begin" at some synchronous point

nikomatsakis (Mar 26 2020 at 18:23, on Zulip):

so that folks can skim meeting blog posts or whatever and see what's going on

nikomatsakis (Mar 26 2020 at 18:23, on Zulip):

does that make sense?

nikomatsakis (Mar 26 2020 at 18:24, on Zulip):

I'm imagining then that part of the "pre-triage" procss for prioritization wg

nikomatsakis (Mar 26 2020 at 18:24, on Zulip):

would be to skim a list of "major change candidates" or something?

nikomatsakis (Mar 26 2020 at 18:24, on Zulip):

or maybe people add it to the list themselves when they "second it"

Santiago Pastorino (Mar 26 2020 at 18:24, on Zulip):

yeah I think it makes sense

pnkfelix (Mar 26 2020 at 18:24, on Zulip):

I think we want to enable a structure where the decision making power/responsibility is rotated through the membership

pnkfelix (Mar 26 2020 at 18:25, on Zulip):

(sorry I was responding to a message way way back up there...)

nikomatsakis (Mar 26 2020 at 18:25, on Zulip):

You mean relating to the TSC? That is true, that was a motivation too

nikomatsakis (Mar 26 2020 at 18:25, on Zulip):

I also think a slightly larger TSC would be better than just the two leads

nikomatsakis (Mar 26 2020 at 18:25, on Zulip):

but for now it's ok

pnkfelix (Mar 26 2020 at 18:25, on Zulip):

yes the Technical Steering Committee

pnkfelix (Mar 26 2020 at 18:26, on Zulip):

anyway it doesn't matter for the short term

nikomatsakis (Mar 26 2020 at 18:26, on Zulip):

thinking a bit more...

pnkfelix (Mar 26 2020 at 18:26, on Zulip):

its more a question of how best to govern things in the long run

nikomatsakis (Mar 26 2020 at 18:26, on Zulip):

if part of the role of the TSC is to select the things to do, as well

pnkfelix (Mar 26 2020 at 18:26, on Zulip):

nowadays I have to assume that we all need to be a little more serious about bus factors

nikomatsakis (Mar 26 2020 at 18:26, on Zulip):

...you could imagine having a kind of election early on to elect TSC members, which might help to set our agenda for the year...

nikomatsakis (Mar 26 2020 at 18:26, on Zulip):

but yeah I'd rather get this MCP thing through first :)

pnkfelix (Mar 26 2020 at 18:27, on Zulip):

okay, yeah, sorry, I didn't mean to forcibly continue an abandoned digression. :slight_smile:

nikomatsakis (Mar 26 2020 at 18:27, on Zulip):

side note @pnkfelix that I still want to raise the question of surveying our ongoing projects, project groups, etc

Santiago Pastorino (Mar 26 2020 at 18:28, on Zulip):

pnkfelix said:

yes the Technical Steering Committee

well another way to see it is if you have the leads (as I know you're) always listening and trying to have people reach consensus, rather than having a vertical style, I guess something like that is not needed :)

Santiago Pastorino (Mar 26 2020 at 18:28, on Zulip):

but you know better than I do, so ... :)

nikomatsakis (Mar 26 2020 at 18:29, on Zulip):

btw

nikomatsakis (Mar 26 2020 at 18:29, on Zulip):

we don't have anything in the process to reject a MCP

nikomatsakis (Mar 26 2020 at 18:29, on Zulip):

how do we decide to "close" one of these things? :)

nikomatsakis (Mar 26 2020 at 18:29, on Zulip):

seems important to discuss

pnkfelix (Mar 26 2020 at 18:29, on Zulip):

(I think something @nikomatsakis left unsaid is that there is a distinction between technical steering and social steering.)

pnkfelix (Mar 26 2020 at 18:30, on Zulip):

(team leads may need to do both. Not everyone is good at both.)

nikomatsakis (Mar 26 2020 at 18:30, on Zulip):

yes, although I do think the role of the TSC would also be to do a lot of listening

nikomatsakis (Mar 26 2020 at 18:30, on Zulip):

this comes back a little bit to "how do we decide"

nikomatsakis (Mar 26 2020 at 18:30, on Zulip):

nikomatsakis said:

how do we decide to "close" one of these things? :)

that is, this does

nikomatsakis (Mar 26 2020 at 18:30, on Zulip):

in paricular, it's easier to judge "consensus",

nikomatsakis (Mar 26 2020 at 18:31, on Zulip):

but I can see that if we decide "no we are not doing this"

nikomatsakis (Mar 26 2020 at 18:31, on Zulip):

that may never get a "consensus"

nikomatsakis (Mar 26 2020 at 18:31, on Zulip):

maybe they just time out

Santiago Pastorino (Mar 26 2020 at 18:31, on Zulip):

right

nikomatsakis (Mar 26 2020 at 18:31, on Zulip):

(and/or the leads can decide)

nikomatsakis (Mar 26 2020 at 18:32, on Zulip):

nikomatsakis said:

maybe they just time out

how much time?

nikomatsakis (Mar 26 2020 at 18:32, on Zulip):

"after 3 months of inactivity, an issue can be closed", something like that?

pnkfelix (Mar 26 2020 at 18:32, on Zulip):

ah

pnkfelix (Mar 26 2020 at 18:32, on Zulip):

hah

pnkfelix (Mar 26 2020 at 18:33, on Zulip):

aren't we the project that has the Postponed label

pnkfelix (Mar 26 2020 at 18:33, on Zulip):

that just leaves things to die

pnkfelix (Mar 26 2020 at 18:33, on Zulip):

sigh

nikomatsakis (Mar 26 2020 at 18:34, on Zulip):

I think I would say something like

nikomatsakis (Mar 26 2020 at 18:34, on Zulip):
nikomatsakis (Mar 26 2020 at 18:34, on Zulip):
nikomatsakis (Mar 26 2020 at 18:34, on Zulip):
nikomatsakis (Mar 26 2020 at 18:35, on Zulip):

not sure how imp't the first bullet is

nikomatsakis (Mar 26 2020 at 18:35, on Zulip):

maybe we don't need to specify a time, but I think it'd be kind of useful

nikomatsakis (Mar 26 2020 at 18:36, on Zulip):

but basically I imagine what will happen in practice is that every once in a while we'll sweep the list and "gc" old ideas

nikomatsakis (Mar 26 2020 at 18:36, on Zulip):

that never quite got off the ground

nikomatsakis (Mar 26 2020 at 18:36, on Zulip):

I'm thinking though about some of the trickier cases we've been confronted with

Santiago Pastorino (Mar 26 2020 at 18:36, on Zulip):

sorry but I need to leave for a bit, going to read in a while

nikomatsakis (Mar 26 2020 at 18:37, on Zulip):

e.g., the "end to end" query PRs that led to an inconclusive design meeting

nikomatsakis (Mar 26 2020 at 18:37, on Zulip):

I guess that it would be valid after some time to close that proposal

nikomatsakis (Mar 26 2020 at 18:37, on Zulip):

i.e., mw and I had unresolved objections, we had the design meeting, it doesn't seem like we were going to reach a consensus on the existing approach

nikomatsakis (Mar 26 2020 at 18:40, on Zulip):

gah, I guess it's "lang pre-triage" time

Last update: Jan 22 2021 at 13:30UTC