Stream: t-compiler/meetings

Topic: [weekly meeting] 2020-10-15 #54818


Santiago Pastorino (Oct 14 2020 at 20:05, on Zulip):

Hi @T-compiler/meeting; the triage meeting will happen tomorrow at

Santiago Pastorino (Oct 14 2020 at 20:06, on Zulip):

WG-prioritization has done pre-triage in #t-compiler/wg-prioritization/alerts

Santiago Pastorino (Oct 14 2020 at 20:06, on Zulip):

@WG-prioritization has prepared the meeting agenda

Santiago Pastorino (Oct 14 2020 at 20:06, on Zulip):

We will have one checkin this time from WG-async-foundations

Santiago Pastorino (Oct 14 2020 at 20:07, on Zulip):

@tmandry @nikomatsakis do you have something you want to share about @WG-async-foundations?

tmandry (Oct 15 2020 at 03:57, on Zulip):

Most activity has been in RFCs and improving the async book lately. We've fixed some implementation issues lately and have a large backlog of others.

There haven't been a ton of people picking up new issues to work on, but I'd still say the main bottlenecks have been reviews and mentoring. I'm shifting priorities to focus more time on this and would love to grow more reviewers from the people who are contributing. One (minor) challenge here is it's unclear if we want to grant r+ to individual wg members, but I think we do.

tmandry (Oct 15 2020 at 03:58, on Zulip):

@Santiago Pastorino ^

Santiago Pastorino (Oct 15 2020 at 13:34, on Zulip):

Hi @T-compiler/meeting, triage meeting will be starting in ~ 26 minutes

Santiago Pastorino (Oct 15 2020 at 13:34, on Zulip):

Check out the meeting agenda

pnkfelix (Oct 15 2020 at 14:01, on Zulip):

Hi @T-compiler/meeting! Add a :wave: emoji to show you're here :)

pnkfelix (Oct 15 2020 at 14:02, on Zulip):

we will start off with 5 minutes for ...

Announcements

pnkfelix (Oct 15 2020 at 14:02, on Zulip):
pnkfelix (Oct 15 2020 at 14:02, on Zulip):
pnkfelix (Oct 15 2020 at 14:02, on Zulip):
pnkfelix (Oct 15 2020 at 14:02, on Zulip):
pnkfelix (Oct 15 2020 at 14:03, on Zulip):
bjorn3 (Oct 15 2020 at 14:03, on Zulip):

I just opened #77975 for cg_clif.

pnkfelix (Oct 15 2020 at 14:04, on Zulip):
nikomatsakis (Oct 15 2020 at 14:04, on Zulip):

Regarding compiler-team#370

nikomatsakis (Oct 15 2020 at 14:04, on Zulip):

("inherit stable annotations")

nikomatsakis (Oct 15 2020 at 14:04, on Zulip):

@simulacrum suggested that might be a good candidate for a full fcp

nikomatsakis (Oct 15 2020 at 14:05, on Zulip):

given that some folks had expressed doubt in the past

nikomatsakis (Oct 15 2020 at 14:05, on Zulip):

I think that's a good point

pnkfelix (Oct 15 2020 at 14:05, on Zulip):

okay, lets do that then

pnkfelix (Oct 15 2020 at 14:05, on Zulip):

what's the protocol for such a switch? Just write @rfcbot fcp merge ?

pnkfelix (Oct 15 2020 at 14:05, on Zulip):

I guess I'll try that

nikomatsakis (Oct 15 2020 at 14:06, on Zulip):

I can do it, but yes

nikomatsakis (Oct 15 2020 at 14:06, on Zulip):

I don't know how it will respond with the labeling and all

pnkfelix (Oct 15 2020 at 14:07, on Zulip):

hup, data race on the issue!

pnkfelix (Oct 15 2020 at 14:07, on Zulip):

or rather, just a race!

nikomatsakis (Oct 15 2020 at 14:07, on Zulip):

yep...

nikomatsakis (Oct 15 2020 at 14:07, on Zulip):

I think we've got it fixed now ;)

pnkfelix (Oct 15 2020 at 14:07, on Zulip):

okay, moving on

pnkfelix (Oct 15 2020 at 14:07, on Zulip):

WG checkins

@WG-async-foundations checkin by @tmandry:

Most activity has been in RFCs and improving the async book lately. We've fixed some implementation issues lately and have a large backlog of others.

There haven't been a ton of people picking up new issues to work on, but I'd still say the main bottlenecks have been reviews and mentoring. I'm shifting priorities to focus more time on this and would love to grow more reviewers from the people who are contributing. One (minor) challenge here is it's unclear if we want to grant r+ to individual wg members, but I think we do.

nikomatsakis (Oct 15 2020 at 14:08, on Zulip):

@tmandry let's discuss, I think some of those regular committers are prob good candidates for compiler team contributors

pnkfelix (Oct 15 2020 at 14:08, on Zulip):

yeah, I think a good pattern here might be

pnkfelix (Oct 15 2020 at 14:09, on Zulip):

"if a wg member is contributing enough to consider giving them r+ privs, then probably best to nominate them for outright compiler contributor status"

nikomatsakis (Oct 15 2020 at 14:09, on Zulip):

yeah

pnkfelix (Oct 15 2020 at 14:09, on Zulip):

@WG-diagnostics checkin by @estebank:

As usual, there has been a flurry of unstructured activity, lots of small and medium sized improvements. The following are a loose cathegorization of some of the salient work since the last update, but I'm sure I'm missing some (mainly PRs I'm fogetting about that didn't close an existing A-diagnostics ticket)

Upcoming change needing eyes: * Provide appropriate types in turbofish suggestions rust#76043

pnkfelix (Oct 15 2020 at 14:09, on Zulip):

hmm

pnkfelix (Oct 15 2020 at 14:09, on Zulip):

let me fix that

nikomatsakis (Oct 15 2020 at 14:10, on Zulip):

short path in error messages for unique identifiers :tada:

:tada: :tada: indeed

pnkfelix (Oct 15 2020 at 14:11, on Zulip):

that's PR #73996, right?

pnkfelix (Oct 15 2020 at 14:12, on Zulip):

(or is that too old...)

pnkfelix (Oct 15 2020 at 14:12, on Zulip):

still, seems right to me

pnkfelix (Oct 15 2020 at 14:13, on Zulip):

okay, on to beta nominations

pnkfelix (Oct 15 2020 at 14:13, on Zulip):

Beta-nominations

T-compiler

pnkfelix (Oct 15 2020 at 14:13, on Zulip):
nikomatsakis (Oct 15 2020 at 14:15, on Zulip):

I don't quite understand what this PR does

pnkfelix (Oct 15 2020 at 14:15, on Zulip):

/me squints at rustfmt panicking due to a looser span

nikomatsakis (Oct 15 2020 at 14:15, on Zulip):

but I agree that the rustfmt bug is worth fixing

pnkfelix (Oct 15 2020 at 14:15, on Zulip):

I personally think this is a sign that rustfmt should not be relying on the spans in this way

pnkfelix (Oct 15 2020 at 14:15, on Zulip):

but what do I know

nikomatsakis (Oct 15 2020 at 14:15, on Zulip):

I see, it only tweaks the span..

pnkfelix (Oct 15 2020 at 14:16, on Zulip):

anyway I guess it seems fine to backport, assuming it did indeed fix the issue

pnkfelix (Oct 15 2020 at 14:16, on Zulip):

(okay, the fix was confirmed)

pnkfelix (Oct 15 2020 at 14:17, on Zulip):

so, I guess backport approved

pnkfelix (Oct 15 2020 at 14:17, on Zulip):

next

pnkfelix (Oct 15 2020 at 14:17, on Zulip):
pnkfelix (Oct 15 2020 at 14:18, on Zulip):

@Joshua Nelson just so you know, in the specific context of beta backport nominations

pnkfelix (Oct 15 2020 at 14:18, on Zulip):

we use the emojis as a sort of informal "voting" system

Joshua Nelson (Oct 15 2020 at 14:18, on Zulip):

I'll make the :wave: separate then

pnkfelix (Oct 15 2020 at 14:18, on Zulip):

(okay, thanks for removing the :wave: ; hopefully that will keep things from getting too confused)

pnkfelix (Oct 15 2020 at 14:19, on Zulip):

This use of emojis is an artifact, in part, of Zulips polling system not working on mobile devices. :laughter_tears:

nikomatsakis (Oct 15 2020 at 14:20, on Zulip):

(although setting up a poll would be annoying too, unless there's a non-interactive way I don't know about)

nikomatsakis (Oct 15 2020 at 14:20, on Zulip):

we don't have the diff from crossbeam-channel

pnkfelix (Oct 15 2020 at 14:20, on Zulip):

do we need to backport rust-lang/rls#1703 as well here?

nikomatsakis (Oct 15 2020 at 14:20, on Zulip):

but .. I trust it ;)

pnkfelix (Oct 15 2020 at 14:20, on Zulip):

rust-lang/rls#1703

pnkfelix (Oct 15 2020 at 14:21, on Zulip):

or is it enough to just backport the change to the .lock file for the purposes of the release channels?

pnkfelix (Oct 15 2020 at 14:21, on Zulip):

cc @simulacrum ^ ?

simulacrum (Oct 15 2020 at 14:21, on Zulip):

.lock should be fine for backport purposes

pnkfelix (Oct 15 2020 at 14:22, on Zulip):

(btw, https://github.com/RustSec/advisory-db/pull/425 is pretty interesting to read, in terms of being aware of this kind of bug occurring elsewhere)

pnkfelix (Oct 15 2020 at 14:23, on Zulip):

anyway, backport approved

pnkfelix (Oct 15 2020 at 14:23, on Zulip):
pnkfelix (Oct 15 2020 at 14:25, on Zulip):

adding delay_span_bug in this manner seems pretty safe to me

pnkfelix (Oct 15 2020 at 14:25, on Zulip):

so I'm good with approving this backport too

pnkfelix (Oct 15 2020 at 14:25, on Zulip):

libs-impl

T-rustdoc

pnkfelix (Oct 15 2020 at 14:25, on Zulip):

Stable-nominations

T-compiler

pnkfelix (Oct 15 2020 at 14:25, on Zulip):
oli (Oct 15 2020 at 14:26, on Zulip):

this is teh one we just approved above

pnkfelix (Oct 15 2020 at 14:26, on Zulip):

right

oli (Oct 15 2020 at 14:26, on Zulip):

ah, stable backport

pnkfelix (Oct 15 2020 at 14:26, on Zulip):

anyway I think its ...f ine

simulacrum (Oct 15 2020 at 14:26, on Zulip):

I am inclined to reject it, P-medium and 1.36

oli (Oct 15 2020 at 14:26, on Zulip):

simple enough, but not very important to backport

pnkfelix (Oct 15 2020 at 14:26, on Zulip):

though I'm also fine with rejecting it

simulacrum (Oct 15 2020 at 14:27, on Zulip):

I think stable backports need to be justified more than just "Stable to stable" regression

pnkfelix (Oct 15 2020 at 14:27, on Zulip):

in some ways, stable backports should probably be reserved for ... right

simulacrum (Oct 15 2020 at 14:27, on Zulip):

after all, they're fairly high risk

pnkfelix (Oct 15 2020 at 14:27, on Zulip):

like things that are critical problems

pnkfelix (Oct 15 2020 at 14:27, on Zulip):

but an ICE on code is otherwise an error

pnkfelix (Oct 15 2020 at 14:27, on Zulip):

is unlikely to warrant a stable backport

nikomatsakis (Oct 15 2020 at 14:27, on Zulip):

so

nikomatsakis (Oct 15 2020 at 14:27, on Zulip):

I thought that our role was more to judge the technical merits

pnkfelix (Oct 15 2020 at 14:27, on Zulip):

unless there's some very severe UI issue that the backport resolves

nikomatsakis (Oct 15 2020 at 14:27, on Zulip):

and less to judge the importance

simulacrum (Oct 15 2020 at 14:27, on Zulip):

I think that's true yes

nikomatsakis (Oct 15 2020 at 14:27, on Zulip):

though I do agree that the bar is higher from a technical perspective as well

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

well, to my mind we're not here to judge whether to have a point release

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

but we can still say "this is not worth the development effort

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

of producing a backport

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

i.e., even if we do have a point release, not worth backporting?

simulacrum (Oct 15 2020 at 14:28, on Zulip):

Right, yes

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

anyway I think this is a very safe PR

pnkfelix (Oct 15 2020 at 14:29, on Zulip):

yeah I could honestly go either way

nikomatsakis (Oct 15 2020 at 14:29, on Zulip):

I'm...not sure about prioritity, I think it's maybe non-zero, I guess it depends

pnkfelix (Oct 15 2020 at 14:29, on Zulip):

the PR is both safe and simple

pnkfelix (Oct 15 2020 at 14:30, on Zulip):

what I cannot tell is if the user experience is significantly improved in any case

nikomatsakis (Oct 15 2020 at 14:30, on Zulip):

ok, I tinkered a bit with the example

pnkfelix (Oct 15 2020 at 14:30, on Zulip):

in the sense that, is there a variant where the diagnostic feedback is useless without this PR in place

nikomatsakis (Oct 15 2020 at 14:31, on Zulip):

it does seem fairly sensitive to the particular example, so I agree low priority

nikomatsakis (Oct 15 2020 at 14:31, on Zulip):

(I was a bit afraid it might occur for all kinds of unresolved symbols or something...)

pnkfelix (Oct 15 2020 at 14:31, on Zulip):

lets decline for stable backport

pnkfelix (Oct 15 2020 at 14:32, on Zulip):

if someone can come up with an example where the UX is awful without this, then we can reconsider

pnkfelix (Oct 15 2020 at 14:32, on Zulip):

(also, I will freely admit that we might be a bit wishy-washy in our policy regarding stable backports)

pnkfelix (Oct 15 2020 at 14:33, on Zulip):

(it might be a good idea to actually keep stats from week to week about how many things are accepted vs rejected, and what kinds of PR's they were, so that we might form a collective set of values here that isn't reconstructed on-the-fly each week)

pnkfelix (Oct 15 2020 at 14:33, on Zulip):

libs-impl

T-rustdoc

pnkfelix (Oct 15 2020 at 14:33, on Zulip):

PRs S-waiting-on-team

T-compiler

pnkfelix (Oct 15 2020 at 14:34, on Zulip):
Joshua Nelson (Oct 15 2020 at 14:34, on Zulip):

this has not gone through MCP but I think it would be a good change

Joshua Nelson (Oct 15 2020 at 14:34, on Zulip):

right now all the crates are named rustc/rustc_xxx which seems repetitive

pnkfelix (Oct 15 2020 at 14:34, on Zulip):

um

pnkfelix (Oct 15 2020 at 14:34, on Zulip):

there is an associated MCP

nikomatsakis (Oct 15 2020 at 14:34, on Zulip):

given past controversies, we probably want this to be FCP'd?

pnkfelix (Oct 15 2020 at 14:35, on Zulip):

was that not approved?

Joshua Nelson (Oct 15 2020 at 14:35, on Zulip):

pnkfelix said:

there is an associated MCP

yes, but the MCP doesn't mention this change I don't think

nikomatsakis (Oct 15 2020 at 14:35, on Zulip):

there is going to be the challenge that usually occurs

Joshua Nelson (Oct 15 2020 at 14:35, on Zulip):

this is an extension

nikomatsakis (Oct 15 2020 at 14:35, on Zulip):

where people check their boxes and then the proposal changes a bit

pnkfelix (Oct 15 2020 at 14:35, on Zulip):

these things need to be made clearer

nikomatsakis (Oct 15 2020 at 14:35, on Zulip):

(I wonder if we should ask folks who have particular concerns to write them out)

Joshua Nelson (Oct 15 2020 at 14:35, on Zulip):

Eric Huss and Vadim Petrochenkov had concerns this would make the crates less readable

Joshua Nelson (Oct 15 2020 at 14:36, on Zulip):

I think if we only renamed the directories and kept the imports as rustc_ast, etc. that would be pretty readable, but of course I can't speak for them

pnkfelix (Oct 15 2020 at 14:36, on Zulip):

right, this comment in particular was the main one that @Eric Huss and @Vadim Petrochenkov agreed on:

I'm a bit confused, where was it decided that the rustc_ prefix should be removed? I personally find it helpful, as it very clearly indicates which parts are crates from the compiler "proper", and which are from external crates. Also, a lot of these names seem like they could introduce ambiguity.

nikomatsakis (Oct 15 2020 at 14:37, on Zulip):

do these crates go into sysroot or something?

nikomatsakis (Oct 15 2020 at 14:37, on Zulip):

I guess that doesn't matter

pnkfelix (Oct 15 2020 at 14:37, on Zulip):

I think the intention is to keep the crates within compiler/

nikomatsakis (Oct 15 2020 at 14:38, on Zulip):

(sorry, I meant in the distribution; I think they do, but I don't think it matters very much)

nikomatsakis (Oct 15 2020 at 14:38, on Zulip):

(I was wondering if having more generic names would induce more conflicts)

pnkfelix (Oct 15 2020 at 14:38, on Zulip):

years ago we had big arguments about line lengths, and people said "we have wide monitors, we don't need an 80 character limit"

nikomatsakis (Oct 15 2020 at 14:38, on Zulip):

anyway there is definitely objection, I don't know the best way to resolve this discussion, but this triage meeting is probably not it

pnkfelix (Oct 15 2020 at 14:39, on Zulip):

in that spriit, I'm mostly in the camp of "I don't understand the benefit here"

Joshua Nelson (Oct 15 2020 at 14:39, on Zulip):

I guess I don't really like that the paths are so much longer now

pnkfelix (Oct 15 2020 at 14:39, on Zulip):

anyway this is waiting on team

pnkfelix (Oct 15 2020 at 14:39, on Zulip):

so we need to figure out where the discussion is going to happen

DPC (Oct 15 2020 at 14:39, on Zulip):

it feels redundant to have rustc prefixes everywhere since now all the crates have it

pnkfelix (Oct 15 2020 at 14:39, on Zulip):

if it happens at all

pnkfelix (Oct 15 2020 at 14:40, on Zulip):

maybe we need a design meeting about this

Joshua Nelson (Oct 15 2020 at 14:40, on Zulip):

there was a zulip stream in #t-compiler/performance a while back

pnkfelix (Oct 15 2020 at 14:40, on Zulip):

when's our next planning meeting?

nikomatsakis (Oct 15 2020 at 14:40, on Zulip):

I feel like a design meeting proposal is not a bad idea. It may also be that somebody should kind of go person to person and gather feedback and report it, maybe that's a big burden :)

nikomatsakis (Oct 15 2020 at 14:41, on Zulip):

I'd be curious to get a kind of survey/straw-poll

pnkfelix (Oct 15 2020 at 14:41, on Zulip):

okay next planning meeting is a week from tomorrow

pnkfelix (Oct 15 2020 at 14:41, on Zulip):

so maybe lets post a comment on the issue

nikomatsakis (Oct 15 2020 at 14:41, on Zulip):

pnkfelix said:

when's our next planning meeting?

Oct 23

pnkfelix (Oct 15 2020 at 14:41, on Zulip):

saying "we think this deserves a planning meeting proposal. Get one in before next week!"

nikomatsakis (Oct 15 2020 at 14:41, on Zulip):

nikomatsakis said:

I'd be curious to get a kind of survey/straw-poll

like -- what are the reasons to do it, what are the reasons not to, and where do people roughly fall

pnkfelix (Oct 15 2020 at 14:41, on Zulip):

next

pnkfelix (Oct 15 2020 at 14:42, on Zulip):

(sorry, I figure we should cut discussion short)

nikomatsakis (Oct 15 2020 at 14:42, on Zulip):

in prtaicular though I think I wouldn't want to schedule the meeting without someone agreeing to do that prep, or something to give it more structure

pnkfelix (Oct 15 2020 at 14:42, on Zulip):

libs-impl

pnkfelix (Oct 15 2020 at 14:42, on Zulip):

Issues of Note

Short Summary

pnkfelix (Oct 15 2020 at 14:42, on Zulip):

P-critical

T-compiler

pnkfelix (Oct 15 2020 at 14:42, on Zulip):
pnkfelix (Oct 15 2020 at 14:42, on Zulip):

okay we definitely discussed this last week. :)

pnkfelix (Oct 15 2020 at 14:43, on Zulip):

I think we decided last week that wg-llvm should decide whether to cherry-pick, right?

pnkfelix (Oct 15 2020 at 14:43, on Zulip):

Or at least, we wanted to give them authority to make a decision there. If wg-llvm doesn't have time/resources to evaluate whether to cherry-pick, then I guess it falls back on T-compiler itself.

pnkfelix (Oct 15 2020 at 14:44, on Zulip):

/me remembers that he wants to try to get in habit of linking from issues to the relevant point in the zulip-archive after each discussion like this, to make it easier to track backwards

pnkfelix (Oct 15 2020 at 14:45, on Zulip):

anyway, we did discuss this last week, and I still think its under control

pnkfelix (Oct 15 2020 at 14:45, on Zulip):

but it would be good to make forward progress on deciding whether to cherry-pick

pnkfelix (Oct 15 2020 at 14:45, on Zulip):

next

pnkfelix (Oct 15 2020 at 14:46, on Zulip):
pnkfelix (Oct 15 2020 at 14:46, on Zulip):

Oooh I didn't actually know that LLVM 11 final release was done

pnkfelix (Oct 15 2020 at 14:46, on Zulip):

we should have included that in the annoucements. :smile:

pnkfelix (Oct 15 2020 at 14:47, on Zulip):

and we even finished merging it ... 6 hours ago ...

pnkfelix (Oct 15 2020 at 14:47, on Zulip):

so cool

Santiago Pastorino (Oct 15 2020 at 14:47, on Zulip):

you're right, that was worth an announcement :)

pnkfelix (Oct 15 2020 at 14:47, on Zulip):

so okay, that to me means that rust#77382 is also under control

pnkfelix (Oct 15 2020 at 14:47, on Zulip):

@cuviper already said that they would take care of backporting the patch

pnkfelix (Oct 15 2020 at 14:47, on Zulip):

next

Santiago Pastorino (Oct 15 2020 at 14:47, on Zulip):

this one in particular was not clear to me in which LLVM version was going to be included

pnkfelix (Oct 15 2020 at 14:48, on Zulip):

well I'm confident we'll get the fix one way or another

pnkfelix (Oct 15 2020 at 14:48, on Zulip):
pnkfelix (Oct 15 2020 at 14:48, on Zulip):

okay, the PR here is just awaiting review I think

Santiago Pastorino (Oct 15 2020 at 14:48, on Zulip):

we need to review @Matthew Jasper's PR

pnkfelix (Oct 15 2020 at 14:48, on Zulip):

@nikomatsakis do you think you'll have time, or do you want to delegate?

Santiago Pastorino (Oct 15 2020 at 14:49, on Zulip):

for this P-critical and for the following one

pnkfelix (Oct 15 2020 at 14:49, on Zulip):

the next one is this

Santiago Pastorino (Oct 15 2020 at 14:49, on Zulip):

same PR fixes this issue :)

nikomatsakis (Oct 15 2020 at 14:49, on Zulip):

pnkfelix said:

nikomatsakis do you think you'll have time, or do you want to delegate?

sorry, I was reading the issue thread

pnkfelix (Oct 15 2020 at 14:49, on Zulip):

(to be clear, the same PR fixes both issues; its not that @Matthew Jasper has two distinct PR's here awaiting review)

nikomatsakis (Oct 15 2020 at 14:49, on Zulip):

is the question about reviewing https://github.com/rust-lang/rust/pull/77720 ?

nikomatsakis (Oct 15 2020 at 14:50, on Zulip):

if so, I will definitely do that today

pnkfelix (Oct 15 2020 at 14:50, on Zulip):

@nikomatsakis yep, that was the question

pnkfelix (Oct 15 2020 at 14:50, on Zulip):

great!

pnkfelix (Oct 15 2020 at 14:50, on Zulip):

libs-impl

T-rustdoc

pnkfelix (Oct 15 2020 at 14:50, on Zulip):

P-high regressions

P-high beta regressions

pnkfelix (Oct 15 2020 at 14:50, on Zulip):
Matthew Jasper (Oct 15 2020 at 14:50, on Zulip):

I'm not entirely happy with how the PR fixes rust#77656 , which is an issue around how we handle cycles in normalization/trait solving, so that might require some more discussion.

pnkfelix (Oct 15 2020 at 14:52, on Zulip):

Looks to me like rust#75982 needs an MCVE, no?

nikomatsakis (Oct 15 2020 at 14:52, on Zulip):

looks that wy

nikomatsakis (Oct 15 2020 at 14:53, on Zulip):

it also needs bisection

nikomatsakis (Oct 15 2020 at 14:53, on Zulip):

(right?)

nikomatsakis (Oct 15 2020 at 14:53, on Zulip):

seems like we should ping cleanup crew

pnkfelix (Oct 15 2020 at 14:53, on Zulip):

/me is unhappy that its releatively difficult to google for the syntax to invoke the ice breakers

Santiago Pastorino (Oct 15 2020 at 14:53, on Zulip):

yes, this probably went unnoticed because it was originally a P-medium issue

nikomatsakis (Oct 15 2020 at 14:53, on Zulip):

@pnkfelix rustbot ping cleanup ?

nikomatsakis (Oct 15 2020 at 14:53, on Zulip):

I think

pnkfelix (Oct 15 2020 at 14:53, on Zulip):

e.g. links here are not useful: https://blog.rust-lang.org/inside-rust/2020/02/06/Cleanup-Crew-ICE-breakers.html

pnkfelix (Oct 15 2020 at 14:54, on Zulip):

@nikomatsakis right, my point is that we have a lot of pages talking about these groups

pnkfelix (Oct 15 2020 at 14:54, on Zulip):

but very few of them include the syntax for actually doing the ping

nikomatsakis (Oct 15 2020 at 14:55, on Zulip):

should probably update https://rustc-dev-guide.rust-lang.org/notification-groups/cleanup-crew.html

nikomatsakis (Oct 15 2020 at 14:55, on Zulip):

to include that info

pnkfelix (Oct 15 2020 at 14:55, on Zulip):

and so googling is less useful than it could/should be. (That, and/or the page with the ping syntax needs better SEO)

Santiago Pastorino (Oct 15 2020 at 14:55, on Zulip):

I think in general you want https://rustc-dev-guide.rust-lang.org/notification-groups/about.html

pnkfelix (Oct 15 2020 at 14:55, on Zulip):

and then this is a dead link: https://rustc-dev-guide.rust-lang.org/notifiation-groups/about.html#tagging-an-issue-for-a-notification-group

pnkfelix (Oct 15 2020 at 14:56, on Zulip):

hmm why did my url fail but @Santiago Pastorino 's worked

Santiago Pastorino (Oct 15 2020 at 14:56, on Zulip):

notifiation

Santiago Pastorino (Oct 15 2020 at 14:56, on Zulip):

there's a typo

Santiago Pastorino (Oct 15 2020 at 14:56, on Zulip):

the right link is https://rustc-dev-guide.rust-lang.org/notification-groups/about.html#tagging-an-issue-for-a-notification-group

pnkfelix (Oct 15 2020 at 14:56, on Zulip):

omg the blog post had a typo

pnkfelix (Oct 15 2020 at 14:56, on Zulip):

this blog post, I mean: https://blog.rust-lang.org/inside-rust/2020/02/06/Cleanup-Crew-ICE-breakers.html

pnkfelix (Oct 15 2020 at 14:56, on Zulip):

has a typo in its url

pnkfelix (Oct 15 2020 at 14:56, on Zulip):

agh gotta love the web

pnkfelix (Oct 15 2020 at 14:57, on Zulip):

anywya

pnkfelix (Oct 15 2020 at 14:57, on Zulip):

there's work to be done on all sorts of levels here

pnkfelix (Oct 15 2020 at 14:57, on Zulip):

but I'm not too worried

pnkfelix (Oct 15 2020 at 14:57, on Zulip):

lets move along

pnkfelix (Oct 15 2020 at 14:57, on Zulip):
pnkfelix (Oct 15 2020 at 14:57, on Zulip):

(omg and we're almost out of time)

Pietro Albini (Oct 15 2020 at 14:58, on Zulip):

the workaround works

Pietro Albini (Oct 15 2020 at 14:58, on Zulip):

or at least, nobody reported this is affecting 1.47 stable

pnkfelix (Oct 15 2020 at 14:58, on Zulip):

:tada:

nikomatsakis (Oct 15 2020 at 14:58, on Zulip):

that's great

simulacrum (Oct 15 2020 at 14:58, on Zulip):

I plan to backport the same fix to 1.48, btw -- just haven't gotten around to it. I think there's some exploration on the issue about proper fixes though

pnkfelix (Oct 15 2020 at 14:59, on Zulip):

but we probably should figure out a better fix

pnkfelix (Oct 15 2020 at 14:59, on Zulip):

namely, the workaround was to disable ThinLTO, right?

simulacrum (Oct 15 2020 at 14:59, on Zulip):

yes

pnkfelix (Oct 15 2020 at 14:59, on Zulip):

okay

Santiago Pastorino (Oct 15 2020 at 14:59, on Zulip):

pnkfelix said:

omg the blog post had a typo

https://github.com/rust-lang/blog.rust-lang.org/pull/711

simulacrum (Oct 15 2020 at 14:59, on Zulip):

(well shared linking which is on by default with thinlto)

pnkfelix (Oct 15 2020 at 14:59, on Zulip):

main question is how high priority is this

pnkfelix (Oct 15 2020 at 15:00, on Zulip):

i.e. is it P-high or P-critical

simulacrum (Oct 15 2020 at 15:00, on Zulip):

We can keep backporting the fix to beta

pnkfelix (Oct 15 2020 at 15:00, on Zulip):

I'm inclined to leave it P-high

nikomatsakis (Oct 15 2020 at 15:00, on Zulip):

can we .. open an issue for LLVM?

pnkfelix (Oct 15 2020 at 15:00, on Zulip):

it seems like we're likely to keep discussing it

pnkfelix (Oct 15 2020 at 15:00, on Zulip):

either way

nikomatsakis (Oct 15 2020 at 15:00, on Zulip):

(do we believe this is an LLVM issue?)

simulacrum (Oct 15 2020 at 15:00, on Zulip):

I don't think it's an llvm bug

simulacrum (Oct 15 2020 at 15:00, on Zulip):

linkers, if anything

nikomatsakis (Oct 15 2020 at 15:00, on Zulip):

oh, ok

pnkfelix (Oct 15 2020 at 15:01, on Zulip):

so we are out of time

pnkfelix (Oct 15 2020 at 15:01, on Zulip):

:sad:

pnkfelix (Oct 15 2020 at 15:01, on Zulip):

there's one unassigned P-high nightly regression listed in agenda

pnkfelix (Oct 15 2020 at 15:02, on Zulip):

and a performance log that reports a busy week without any major regressions

pnkfelix (Oct 15 2020 at 15:02, on Zulip):

(and a nice performance boost from @Matthew Jasper in "Separate projection bounds and predicates" #73905 )

pnkfelix (Oct 15 2020 at 15:02, on Zulip):

and one nominated issue that we're going to leave for next week

pnkfelix (Oct 15 2020 at 15:03, on Zulip):

so that's all folks. Thanks to everyone in @T-compiler/meeting for attending; sorry for letting the earlier topics slow us down so much!

pnkfelix (Oct 15 2020 at 15:04, on Zulip):

(by the way, #75982 has got to be a stable-to-stable regression by now, right?)

Santiago Pastorino (Oct 15 2020 at 15:05, on Zulip):

ohh I think so

Santiago Pastorino (Oct 15 2020 at 15:05, on Zulip):

question, after releasing a new version, do we have something that changes nightly regressions tags to beta and beta to stable?

Santiago Pastorino (Oct 15 2020 at 15:06, on Zulip):

I was assuming that something or some group does that, but maybe @WG-prioritization should pay extra attention to that?

LeSeulArtichaut (Oct 15 2020 at 15:06, on Zulip):

I think it’s done manually by @simulacrum?

LeSeulArtichaut (Oct 15 2020 at 15:06, on Zulip):

There was an issue on triagebot to add a new command to do that IIRC

Santiago Pastorino (Oct 15 2020 at 15:11, on Zulip):

have created next week's agenda https://hackmd.io/yszFneboQA2gJ0qaG6pllw?both

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

we will have checkins from @WG-rustc-dev-guide (which @Joshua Nelson and I need to fill) and from @WG-llvm (cc @nagisa)

apiraino (Oct 15 2020 at 15:26, on Zulip):

tmandry said:

There haven't been a ton of people picking up new issues to work on [...]

@tmandry regarding this part: do you need some new people working on issues? Just a random thought: would something as a blog post help? Although I'm not sure, I see the stream #wg-async-foundations already has a lot of people subscribed

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

(its certainly possible for a channel to have a large number of subscribers and yet still need more active contributors ...)

simulacrum (Oct 15 2020 at 16:37, on Zulip):

No one switches labels actively today

simulacrum (Oct 15 2020 at 16:38, on Zulip):

It's an active pain point but I've yet to find time to fix it

Jack Huey (Oct 15 2020 at 16:49, on Zulip):

pnkfelix said:

Sorry, don't want to spam. But was there a reason I was cced on this? :rolling_eyes:

apiraino (Oct 15 2020 at 16:56, on Zulip):

Jack Huey said:

Sorry, don't want to spam. But was there a reason I was cced on this? :rolling_eyes:

:thinking: I believe as suggested reviewer of that PR

Jack Huey (Oct 15 2020 at 16:59, on Zulip):

But I'm not?

apiraino (Oct 15 2020 at 17:00, on Zulip):

right, you're not. so I guess just a copy&paste error by our side (sorry about that) :sweat_smile:

Jack Huey (Oct 15 2020 at 17:01, on Zulip):

No problem. I was just super confused

tmandry (Oct 15 2020 at 18:54, on Zulip):

apiraino said:

tmandry regarding this part: do you need some new people working on issues? Just a random thought: would something as a blog post help? Although I'm not sure, I see the stream #wg-async-foundations already has a lot of people subscribed

it might help, yeah. my focus right now is on getting reviews and mentoring up to par before adding more contributors. but maybe adding more contributors would be a good forcing function :)

mark-i-m (Oct 18 2020 at 19:09, on Zulip):

nikomatsakis said:

nikomatsakis said:

I'd be curious to get a kind of survey/straw-poll

like -- what are the reasons to do it, what are the reasons not to, and where do people roughly fall

Sorry I don't know where else to leave this comment, but I think the main arguments that I have seen currently are:

In favor:

Against:

Joshua Nelson (Oct 18 2020 at 19:13, on Zulip):

I had a different idea in mind

Joshua Nelson (Oct 18 2020 at 19:13, on Zulip):

I was hoping we could only rename the directories

Joshua Nelson (Oct 18 2020 at 19:13, on Zulip):

and keep the use the same

Joshua Nelson (Oct 18 2020 at 19:13, on Zulip):

I like use rustc_ast, I think that's exactly as clear as it needs to be

Aaron Hill (Oct 18 2020 at 19:22, on Zulip):

@Joshua Nelson Would that mean having compiler/ast, but importing it as rustc_ast?

Joshua Nelson (Oct 18 2020 at 19:22, on Zulip):

yes, that's my idea: rustc_ast = { path = "../ast" }

Joshua Nelson (Oct 18 2020 at 19:26, on Zulip):

summary to compare to mark-i-m 's writeup:

In favor:

Against:

Aaron Hill (Oct 18 2020 at 19:34, on Zulip):

imho, it just seems like making things inconsistent for very little benefit

Aaron Hill (Oct 18 2020 at 19:34, on Zulip):

i think it's very useful to have the crate and directory name match

Jack Huey (Oct 18 2020 at 19:39, on Zulip):

It's another big renaming, which github doesn't follow well, causing merge conflicts

This is actually a pretty big concern for me. It doesn't seem like any of the benefits really outweigh this (for me at least)

nikomatsakis (Oct 19 2020 at 14:23, on Zulip):

I think it would be great to pull the above text into a hackmd

nikomatsakis (Oct 19 2020 at 14:28, on Zulip):

Done: hackmd link

nikomatsakis (Oct 19 2020 at 14:28, on Zulip):

I'll file a design meeting proposal I think

DPC (Oct 21 2020 at 14:05, on Zulip):

i think using separate names for the crate and for the import is going to cause more confusion than the current system we have

Last update: Nov 25 2020 at 02:00UTC