Stream: t-compiler

Topic: weekly meeting 2020-04-02 #54818


Santiago Pastorino (Apr 01 2020 at 18:14, on Zulip):

Hi @T-compiler/meeting ; the triage meeting will be starting in 19 hours 46 minutes

Santiago Pastorino (Apr 01 2020 at 18:16, on Zulip):

The @WG-prioritization will be doing pre-triage in #t-compiler/wg-prioritization > pre-meeting triage 2020-04-02 #54818

Santiago Pastorino (Apr 01 2020 at 18:16, on Zulip):

we will have @T-compiler/WG-meta and @WG-mir-opt checkins

Santiago Pastorino (Apr 01 2020 at 18:17, on Zulip):

I can provide the Meta WG update

Santiago Pastorino (Apr 01 2020 at 18:17, on Zulip):

@oli can you do the Mir opt one?

Santiago Pastorino (Apr 01 2020 at 18:20, on Zulip):

here we will be filing the agenda for the meeting

oli (Apr 02 2020 at 04:08, on Zulip):

sure

Santiago Pastorino (Apr 02 2020 at 13:35, on Zulip):

@T-compiler/meeting; the triage meeting will be starting in 25 minutes

pnkfelix (Apr 02 2020 at 14:00, on Zulip):

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

pnkfelix (Apr 02 2020 at 14:01, on Zulip):

We'll be starting off with five minutes for ...

pnkfelix (Apr 02 2020 at 14:01, on Zulip):

Announcements

pnkfelix (Apr 02 2020 at 14:02, on Zulip):

During pre-triage I noticed some discussion about the difficulty in identifying performance deltas in the face of roll-ups.

centril (Apr 02 2020 at 14:02, on Zulip):

Just a reminder: don't forget to use rollup=never on perf or otherwise sensitive PRs

oli (Apr 02 2020 at 14:02, on Zulip):

:wave: everyone, I'm attempting to get rid of our submodules (at least for tools). Lmk in https://github.com/rust-lang/compiler-team/issues/266 if you have opinions on the topic (e.g. on our future usage of git subtree)

eddyb (Apr 02 2020 at 14:02, on Zulip):

oh and @centril renamed librustc to librustc_middle, I think that happened since the last meeting?

nikomatsakis (Apr 02 2020 at 14:03, on Zulip):
centril (Apr 02 2020 at 14:03, on Zulip):

@nikomatsakis hah you're late to this :slight_smile:

centril (Apr 02 2020 at 14:03, on Zulip):

Many have been using rust-analyzer to hack on rustc and it works great ^^

nikomatsakis (Apr 02 2020 at 14:04, on Zulip):

I'm aware, but you must not have been getting cargo watch integration

nikomatsakis (Apr 02 2020 at 14:04, on Zulip):

Or else you have a trick I don't know :)

centril (Apr 02 2020 at 14:04, on Zulip):

o.O

centril (Apr 02 2020 at 14:05, on Zulip):

@nikomatsakis fancy! =P

pnkfelix (Apr 02 2020 at 14:05, on Zulip):

we have a lot on our agenda for today, so lets dive in.

Santiago Pastorino (Apr 02 2020 at 14:05, on Zulip):

@pnkfelix you didn't paste news from the agenda :)

pnkfelix (Apr 02 2020 at 14:05, on Zulip):

oh, there was one item there for the announcements

Santiago Pastorino (Apr 02 2020 at 14:05, on Zulip):
pnkfelix (Apr 02 2020 at 14:05, on Zulip):

or two

pnkfelix (Apr 02 2020 at 14:05, on Zulip):

so yeah, the back-story on I-prioritize

pnkfelix (Apr 02 2020 at 14:05, on Zulip):

we've had I-nominated for ages

pnkfelix (Apr 02 2020 at 14:06, on Zulip):

initially, it was used for "this needs to be brought to the attention of the relevant team"

Santiago Pastorino (Apr 02 2020 at 14:06, on Zulip):

I-nominated current description says: Indicates that an issue has been nominated for prioritizing at the next team triage meeting.

pnkfelix (Apr 02 2020 at 14:06, on Zulip):

and over time, as bugs were filed that needed to be triaged, it was used for that too

Wesley Wiser (Apr 02 2020 at 14:06, on Zulip):
pnkfelix (Apr 02 2020 at 14:07, on Zulip):

Santiago Pastorino said:

I-nominated current description says: Indicates that an issue has been nominated for prioritizing at the next team triage meeting.

hmm well I don't recall the original intent being that specific

Wesley Wiser (Apr 02 2020 at 14:07, on Zulip):

/me oops, sorry, I was getting some links for that

centril (Apr 02 2020 at 14:07, on Zulip):

Basically I-nominated is somewhat confusing... is it nominated for triage, or for discussion?

pnkfelix (Apr 02 2020 at 14:07, on Zulip):

in any case, the proposal is to attempt to split off a more specific label

centril (Apr 02 2020 at 14:07, on Zulip):

e.g., if it is I-nominated T-compiler T-lang

pnkfelix (Apr 02 2020 at 14:08, on Zulip):

The only misgiving I currently have about splitting off a new label, is that, at least at first, WG-prioritization is still going to need to go over the I-nominated issues and prioritize them

nikomatsakis (Apr 02 2020 at 14:08, on Zulip):

Seems good to have a more specific label

pnkfelix (Apr 02 2020 at 14:08, on Zulip):

since not everyone would switch immediately

centril (Apr 02 2020 at 14:08, on Zulip):

I think the switch will go fast

pnkfelix (Apr 02 2020 at 14:09, on Zulip):

and I'm not sure whether they would be expected to actually write feedback on the issue saying "please use I-prioritize for this in the future"

centril (Apr 02 2020 at 14:09, on Zulip):

we can tell the triagists to use the new label

pnkfelix (Apr 02 2020 at 14:09, on Zulip):

but yes, I think if the WG-triage uses the new label

pnkfelix (Apr 02 2020 at 14:09, on Zulip):

then it should go very fast, because I think WG-triage is the primary source of issues being tagged with I-nominated for purposes of prioritization

centril (Apr 02 2020 at 14:09, on Zulip):

(and a lot of the adding of that label is done by release team members anyways ^^)

pnkfelix (Apr 02 2020 at 14:09, on Zulip):

(or maybe its WG-release, not sure)

pnkfelix (Apr 02 2020 at 14:10, on Zulip):

(T-release? Anyway...)

nikomatsakis (Apr 02 2020 at 14:10, on Zulip):

One final announcement:

There are 3 pending T-compiler FCP requests

Check your boxes :)

centril (Apr 02 2020 at 14:10, on Zulip):

(T-release, wg-triage is a WG of that team)

pnkfelix (Apr 02 2020 at 14:11, on Zulip):

Anyway I don't anticipate any real problems. If anyone objects to adding I-prioritize label, or feels strongly that it should have a different name, or a different "category" than I-, please let @WG-prioritization know.

nikomatsakis (Apr 02 2020 at 14:11, on Zulip):
pnkfelix (Apr 02 2020 at 14:12, on Zulip):

(and then, yes, as @Santiago Pastorino said, there is a T-compiler design meeting tomorrow on integrating a cranelift backend.)

pnkfelix (Apr 02 2020 at 14:12, on Zulip):

Now, as previously noted, lets make headway on this big agenda

pnkfelix (Apr 02 2020 at 14:13, on Zulip):

beta nominations: 2

pnkfelix (Apr 02 2020 at 14:13, on Zulip):

beta-nom 1/2: "Fix "since" field for Once::is_complete's #[stable] attribute" #70018

centril (Apr 02 2020 at 14:13, on Zulip):

(I added t-compiler to this cause I don't think t-libs has regular meetings)

pnkfelix (Apr 02 2020 at 14:14, on Zulip):

this certainly seems harmless

pnkfelix (Apr 02 2020 at 14:14, on Zulip):

I'm curious: Does it have an affect on anything beyond the docs?

centril (Apr 02 2020 at 14:14, on Zulip):

Nope

pnkfelix (Apr 02 2020 at 14:14, on Zulip):

okay. I guess beta-accepted.

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

beta-nom 2/2: "parse_and_disallow_postfix_after_cast: account for ExprKind::Err." #70556

pnkfelix (Apr 02 2020 at 14:16, on Zulip):

seems fine to me too. beta-accepted.

pnkfelix (Apr 02 2020 at 14:16, on Zulip):

We don't have any stable nominations this week

pnkfelix (Apr 02 2020 at 14:17, on Zulip):

There was one PR marked waiting on team. (But I'm wondering whether FCP requests should be similarly marked, based on @nikomatsakis 's note above...)

centril (Apr 02 2020 at 14:17, on Zulip):

We have S-waiting-on-fcp and S-waiting-on-team

pnkfelix (Apr 02 2020 at 14:18, on Zulip):

The PR marked S-waiting-on-team is "Remove -Z no-landing-pads flag" #70175

centril (Apr 02 2020 at 14:18, on Zulip):

the former for things already in FCP (not PFCP)

nikomatsakis (Apr 02 2020 at 14:18, on Zulip):

currently waiting on @Zoxc, @Esteban Küber, @Vadim Petrochenkov, and @varkor

pnkfelix (Apr 02 2020 at 14:19, on Zulip):

okay, but two of the three issues that @nikomatsakis noted at the top have neither of the labels @centril mentioned. One of them is marked S-waiting-on-review, so that makes sense

centril (Apr 02 2020 at 14:19, on Zulip):

I adjusted those labels

pnkfelix (Apr 02 2020 at 14:19, on Zulip):

I'm basically wondering if the rfcbot should be auto-adding the waiting-on-team label if the issue is not already marked as waiting on something else.

pnkfelix (Apr 02 2020 at 14:20, on Zulip):

anyway...

centril (Apr 02 2020 at 14:20, on Zulip):

Sure, but no one is hacking on rfcbot :P

DPC (Apr 02 2020 at 14:21, on Zulip):

you want someone to? :P

pnkfelix (Apr 02 2020 at 14:21, on Zulip):

@nikomatsakis already mentioned the three issues: "Add Option to Force Unwind Tables" (rust-lang/rust#69984), "Remove -Z no-landing-pads flag" (rust-lang/rust#70175), "Move LLVM bitcode destination" (rust-lang/rust#70458). Just want to make sure at least mentioned them again.

centril (Apr 02 2020 at 14:21, on Zulip):

@DPC that would be nice, there are a bunch of feature requests

DPC (Apr 02 2020 at 14:21, on Zulip):

sure, let's talk after this

pnkfelix (Apr 02 2020 at 14:21, on Zulip):

okay, so issues of note

pnkfelix (Apr 02 2020 at 14:21, on Zulip):

There are 51 P-high issues (1 more than last week)

pnkfelix (Apr 02 2020 at 14:22, on Zulip):

28 of those are unassigned (2 less than last week)

pnkfelix (Apr 02 2020 at 14:22, on Zulip):

(this is all data as of yesterday, so don't be surprised if the numbers have changed slightly)

pnkfelix (Apr 02 2020 at 14:22, on Zulip):

2 P-high and 2 P-medium regression-from-stable-to-beta

pnkfelix (Apr 02 2020 at 14:22, on Zulip):

and 2 P-high and 4 P-medium regression-from-stable-to-nightly

Santiago Pastorino (Apr 02 2020 at 14:22, on Zulip):

pnkfelix said:

(this is all data as of yesterday, so don't be surprised if the numbers have changed slightly)

I've adjusted those before the meeting

pnkfelix (Apr 02 2020 at 14:23, on Zulip):

/msg @Santiago Pastorino we probably should encode in the pre-triage procedure to call out the unassigned regressions

pnkfelix (Apr 02 2020 at 14:23, on Zulip):

doh

pnkfelix (Apr 02 2020 at 14:23, on Zulip):

What, IRC commands no work here?! :angry_devil:

Santiago Pastorino (Apr 02 2020 at 14:23, on Zulip):

yeah it's below :)

pnkfelix (Apr 02 2020 at 14:23, on Zulip):

oh okay

pnkfelix (Apr 02 2020 at 14:23, on Zulip):

I see now

pnkfelix (Apr 02 2020 at 14:24, on Zulip):

there is one unassigned stable-to-beta regression: “no_mangle causes compilation errors with async-await on armv7-linux-androideabi and aarch64-linux-android targets” #70098

pnkfelix (Apr 02 2020 at 14:24, on Zulip):

this sounds like you don't actually need the hardware to test this, you just need to have the cross-compiling toolchain installed (and the patience to work with it)

nikomatsakis (Apr 02 2020 at 14:25, on Zulip):

Is this a regression, do we know?

pnkfelix (Apr 02 2020 at 14:25, on Zulip):

@nikomatsakis do you think the @WG-async-foundations is going to look at this? Or should we be asking here?

nikomatsakis (Apr 02 2020 at 14:25, on Zulip):

It's categorized as that, but what PR regressed it?

nikomatsakis (Apr 02 2020 at 14:25, on Zulip):

I guess we don't know yet

eddyb (Apr 02 2020 at 14:25, on Zulip):

this is weird

nikomatsakis (Apr 02 2020 at 14:25, on Zulip):

Well, it's not presently tagged as "async-await",

eddyb (Apr 02 2020 at 14:26, on Zulip):

does the async block inherit the #[no_mangle] accidentally?!

nikomatsakis (Apr 02 2020 at 14:26, on Zulip):

and it seems like it's not necessarily an async-await error

nikomatsakis (Apr 02 2020 at 14:26, on Zulip):

But maybe it is..?

centril (Apr 02 2020 at 14:26, on Zulip):

(This is P-medium btw)

pnkfelix (Apr 02 2020 at 14:27, on Zulip):

True, it is P-medium

pnkfelix (Apr 02 2020 at 14:27, on Zulip):

well I've tagged with A-async-await

nikomatsakis (Apr 02 2020 at 14:28, on Zulip):

OK. @eddyb this reproduction suggests that the problem is not really the async block inheriting no-mangle

pnkfelix (Apr 02 2020 at 14:28, on Zulip):

it would be nice if someone could poke at it after the meeting.

pnkfelix (Apr 02 2020 at 14:28, on Zulip):

anyway, lets keep moving

nikomatsakis (Apr 02 2020 at 14:28, on Zulip):

but maybe that's a separate thing

nikomatsakis (Apr 02 2020 at 14:28, on Zulip):

that has more to do with putting no-mangle on main

eddyb (Apr 02 2020 at 14:28, on Zulip):

what does the #[no_mangle] even do

eddyb (Apr 02 2020 at 14:28, on Zulip):

are we sure it's not just equivalent to pub?

pnkfelix (Apr 02 2020 at 14:28, on Zulip):

there are five stable-to-nightly regressions that are unassigned.

centril (Apr 02 2020 at 14:29, on Zulip):

@eddyb best to investigate after meeting :slight_smile:

pnkfelix (Apr 02 2020 at 14:29, on Zulip):

unassigned regression 1/5: “internal compiler error: cannot relate region: LUB(ReErased, ReErased)” #70608

eddyb (Apr 02 2020 at 14:29, on Zulip):

@centril sure just noting for whomever will

centril (Apr 02 2020 at 14:29, on Zulip):

Is @Matthew Jasper here?

nikomatsakis (Apr 02 2020 at 14:30, on Zulip):

curious, though the fix might just be to modify LUB of two erased to yield erased

nikomatsakis (Apr 02 2020 at 14:30, on Zulip):

I could see that it's an artifact of erasure flowing into some other pipeline

pnkfelix (Apr 02 2020 at 14:30, on Zulip):

should we just assign to @Matthew Jasper ?

eddyb (Apr 02 2020 at 14:31, on Zulip):

seems a bit dubious, trait system doesn't use subtyping, does it?

eddyb (Apr 02 2020 at 14:31, on Zulip):

oh we're missing the backtrace

pnkfelix (Apr 02 2020 at 14:31, on Zulip):

pnkfelix said:

should we just assign to Matthew Jasper ?

(on the assumption that this is indeed due to PR #69189 )

nikomatsakis (Apr 02 2020 at 14:31, on Zulip):

( it can, to deal with equality and higher-ranked things, but I'm not sure this is coming out of the trait system )

nikomatsakis (Apr 02 2020 at 14:31, on Zulip):

if @Matthew Jasper can look good, else I can do some quick investigation perhaps

eddyb (Apr 02 2020 at 14:31, on Zulip):

@Esteban Küber @centril do we have some label that could mean "needs backtrace"? should we add one?

pnkfelix (Apr 02 2020 at 14:32, on Zulip):

I'm going to assign to @Matthew Jasper for now

eddyb (Apr 02 2020 at 14:32, on Zulip):

oh wait I didn't look at the example, I know what this is

centril (Apr 02 2020 at 14:32, on Zulip):

@eddyb added a backtrace

pnkfelix (Apr 02 2020 at 14:33, on Zulip):

unassigned regression 2/5: "Some macro errors now include file names into the standard library (JSON)." #70396

centril (Apr 02 2020 at 14:33, on Zulip):

Curious, this doesn't repro with:

const e: _ = & & 39;
centril (Apr 02 2020 at 14:33, on Zulip):

so this is probably my fault :D

eddyb (Apr 02 2020 at 14:34, on Zulip):

yeah left a comment

eddyb (Apr 02 2020 at 14:34, on Zulip):

@centril actually that's probably buggy in a different way

pnkfelix (Apr 02 2020 at 14:35, on Zulip):

I'm not exactly clear on what our options for #70396 even are

pnkfelix (Apr 02 2020 at 14:35, on Zulip):

is the bug filer saying we should be omitting the filename: field in the JSON for this case?

centril (Apr 02 2020 at 14:36, on Zulip):

@eddyb did some work related to #70396?

eddyb (Apr 02 2020 at 14:36, on Zulip):

sorry I was still on the last issue

eddyb (Apr 02 2020 at 14:36, on Zulip):

why is this nominated?

centril (Apr 02 2020 at 14:36, on Zulip):

(https://github.com/rust-lang/rust/pull/70642)

pnkfelix (Apr 02 2020 at 14:36, on Zulip):

Its not nominated. Its an unassigned regression

nikomatsakis (Apr 02 2020 at 14:36, on Zulip):

is it breaking tooling?

pnkfelix (Apr 02 2020 at 14:36, on Zulip):

so I want us to either assign it or talk about it

nikomatsakis (Apr 02 2020 at 14:37, on Zulip):

I couldn't quite tell from the comments :)

pnkfelix (Apr 02 2020 at 14:37, on Zulip):

(or explicitly close it, or mark it as P-medium, etc)

nikomatsakis (Apr 02 2020 at 14:37, on Zulip):

in any case I too am a bit unclear on what behavior we think is "desired" here

eddyb (Apr 02 2020 at 14:37, on Zulip):

it's not exactly a regression in the strictest sense, except maybe for tools that were assuming far too much, I haven't seen a good example yet

eddyb (Apr 02 2020 at 14:37, on Zulip):

it would've been possible to break those tools before, just not with macros

pnkfelix (Apr 02 2020 at 14:37, on Zulip):

(Oh it already is marked as P-medium)

nikomatsakis (Apr 02 2020 at 14:37, on Zulip):

I think it's perfectly valid to say "working as expected" and guide tools in what they ought to do; I think your final comment goes in that direction, I didn't quite understand exactly what you were suggesting though, but maybe IDE authors did :)

eddyb (Apr 02 2020 at 14:37, on Zulip):

anyway, once #70642 lands, most of this should be addressed via rust-src

eddyb (Apr 02 2020 at 14:38, on Zulip):

if you have it, the paths will point to a local copy of libstd

pnkfelix (Apr 02 2020 at 14:38, on Zulip):

okay gotta keep going

pnkfelix (Apr 02 2020 at 14:38, on Zulip):

@eddyb are you willing to take assignment on this?

eddyb (Apr 02 2020 at 14:38, on Zulip):

sure, I guess

pnkfelix (Apr 02 2020 at 14:38, on Zulip):

unassigned regressions 3/5: “file not found for module” #70314

centril (Apr 02 2020 at 14:39, on Zulip):

I was hoping #70314 would have a reasonably sized reproducer by now

centril (Apr 02 2020 at 14:39, on Zulip):

I fixed a related bug which had a repro, but I can't fix it (assuming it's a bug) if I ain't got one

nikomatsakis (Apr 02 2020 at 14:40, on Zulip):

Well, we know it regressed due to commit https://github.com/rust-lang/rust/commit/57e1da59cd0761330b4ea8d47b16340a78eeafa9, but that's a rollup

pnkfelix (Apr 02 2020 at 14:40, on Zulip):

yes it doesn't seem like #70314 is quite actionable on our end yet

nikomatsakis (Apr 02 2020 at 14:40, on Zulip):

A rollup of a rollup?

nikomatsakis (Apr 02 2020 at 14:41, on Zulip):

Probably #69838

centril (Apr 02 2020 at 14:41, on Zulip):

@nikomatsakis I know, ain't it glorious? :D

nikomatsakis (Apr 02 2020 at 14:41, on Zulip):

Yeah, agreed, I guess we can leave some comments encouraging folks to make a more minimal example

eddyb (Apr 02 2020 at 14:41, on Zulip):

isn't that the module loading changes?

nikomatsakis (Apr 02 2020 at 14:41, on Zulip):

Yes

centril (Apr 02 2020 at 14:41, on Zulip):

Yeah I would be shocked if it wasn't https://github.com/rust-lang/rust/pull/69838

eddyb (Apr 02 2020 at 14:41, on Zulip):

https://github.com/rust-lang/rust/issues/70314#issuecomment-604471059 suggests it's not the exact same thing that got fixed

pnkfelix (Apr 02 2020 at 14:42, on Zulip):

hmm rollups of rollups are going to complicate any attempt we make to have tooling descend into rollups

eddyb (Apr 02 2020 at 14:42, on Zulip):

the rollup-of-rollup was done for a build-unbreaking reason but yeah :(

pnkfelix (Apr 02 2020 at 14:43, on Zulip):

lets give the cleanup crew another week to try to make an MCVE

pnkfelix (Apr 02 2020 at 14:43, on Zulip):

maybe I'll tweet about it and/or offer to mentor the process

pnkfelix (Apr 02 2020 at 14:43, on Zulip):

unassigned regression 4/5: “Regression in error message quality for macro_rules involving $:ident” #69604

pnkfelix (Apr 02 2020 at 14:44, on Zulip):

this is also P-medium

pnkfelix (Apr 02 2020 at 14:44, on Zulip):

for reasons documented on the issue

centril (Apr 02 2020 at 14:44, on Zulip):

@Vadim Petrochenkov asked whether the other metadata PR improved things in https://github.com/rust-lang/rust/issues/69604#issuecomment-594737564

centril (Apr 02 2020 at 14:45, on Zulip):

no developments since then

centril (Apr 02 2020 at 14:45, on Zulip):

it could possibly be fixed already

pnkfelix (Apr 02 2020 at 14:45, on Zulip):

I'll leave #69604 unassigned

pnkfelix (Apr 02 2020 at 14:45, on Zulip):

would be good, between now and next week, to double check if this replicates.

pnkfelix (Apr 02 2020 at 14:45, on Zulip):

unassigned regression 5/5: “Compiler error while compiling Winrt” #66402

centril (Apr 02 2020 at 14:45, on Zulip):

@pnkfelix maybe let's leave a comment asking David to recheck :slight_smile:

pnkfelix (Apr 02 2020 at 14:46, on Zulip):

oh lordy never mind about #66402

pnkfelix (Apr 02 2020 at 14:46, on Zulip):

(its just E-needstest)

pnkfelix (Apr 02 2020 at 14:46, on Zulip):

we have five nominated issues

pnkfelix (Apr 02 2020 at 14:47, on Zulip):

(luckily its effectively four because we already discussed one of them)

pnkfelix (Apr 02 2020 at 14:47, on Zulip):

nom 1/5: “Box<dyn FnOnce> doesn’t respect self alignment” #68304

centril (Apr 02 2020 at 14:47, on Zulip):

I just checked and, a function call r = f(a, b, c) turns into roughly this (touched up for clarity):
rust _1 = move a; _2 = move b; _3 = move c; _4 = f(move _1, move _2, move _3); r = move _4;
Some of my previous comments assumed the r = move _4 wasn't there and the call would use the destination directly, but given that there is a move, that simplifies my suggestion.

<hr/>

Summary of my proposal

For virtual by-value self calls, we can skip one move when building the MIR, specifically the move out of the dereference of a Box<dyn Trait> for the self argument.
Given that we already move the result of the call, there should be no aliasing caused by this, but if we decide to change anything about how we build MIR calls, we'd need to be extra careful.

We can also make this pattern be gated by a separate feature-gate (rather than unsized_locals), and make sure the standard library only uses that feature-gate and not unsized_locals.

This should solve any alignment issues Box<dyn FnOnce(...) -> _> might have, because we'd not be using dynamic alloca unless the broader unsized_locals feature is used.

pnkfelix (Apr 02 2020 at 14:47, on Zulip):

@eddyb has put forth a proposal for #68304 on the issue itself

bjorn3 (Apr 02 2020 at 14:48, on Zulip):

I vote in favor of the proposal

nikomatsakis (Apr 02 2020 at 14:48, on Zulip):

@eddyb -- is this basically the 'original unsized locals' proposal? ie., we only permit unsized locals for cases where we can point to the original storage, and don't require a move?

nikomatsakis (Apr 02 2020 at 14:49, on Zulip):

or a subset thereof, in some sense

eddyb (Apr 02 2020 at 14:49, on Zulip):

right

nikomatsakis (Apr 02 2020 at 14:49, on Zulip):

excerpted and given a distinct feature gate?

pnkfelix (Apr 02 2020 at 14:49, on Zulip):

The proposal has some components that seem ... optional? Like gating the pattern under a feature gate ...?

nikomatsakis (Apr 02 2020 at 14:49, on Zulip):

OK :+1: I'm in favor too, I'll leave a comment

centril (Apr 02 2020 at 14:49, on Zulip):

Seems like a good idea to restrict libstd's usage of this

centril (Apr 02 2020 at 14:49, on Zulip):

to prevent accidents

pnkfelix (Apr 02 2020 at 14:50, on Zulip):

but I think we're all in favor of such gating, right? Any objections?

eddyb (Apr 02 2020 at 14:50, on Zulip):

somehow we ended up with calling Box<dyn FnOnce()> doing a dynamic alloca (and copying into it) before the virtual call, I'm not sure we ever intended for this (Box<dyn FnBox()> didn't have this problem)

pnkfelix (Apr 02 2020 at 14:50, on Zulip):

okay, I think there's no one objecting, at least

pnkfelix (Apr 02 2020 at 14:50, on Zulip):

no one is assigned to this ticket

centril (Apr 02 2020 at 14:50, on Zulip):

@eddyb got time to tackle this?

eddyb (Apr 02 2020 at 14:50, on Zulip):

I wonder if anyone from @WG-mir-opt wants to do some parts of this

pnkfelix (Apr 02 2020 at 14:51, on Zulip):

so I suspect we'll just touch on it every week until someone gets fed up with me mentioning it

eddyb (Apr 02 2020 at 14:51, on Zulip):

I don't think I'll be able to spend much time on it in the next month or two

eddyb (Apr 02 2020 at 14:51, on Zulip):

(but I could mentor)

centril (Apr 02 2020 at 14:51, on Zulip):

perhaps @Wesley Wiser @oli or @bjorn3 ?

nikomatsakis (Apr 02 2020 at 14:51, on Zulip):

@Santiago Pastorino might be interested

pnkfelix (Apr 02 2020 at 14:51, on Zulip):

I don't want to attempt to find an assignee right now, but I do think querying @WG-mir-opt is a good idea

pnkfelix (Apr 02 2020 at 14:51, on Zulip):

nominated 2/5: “internal compiler error: cannot relate region: LUB(ReErased, ReErased)” #70608

nikomatsakis (Apr 02 2020 at 14:51, on Zulip):

(that said, I remember a possible complication, going to do some experiments and leave a comment)

Santiago Pastorino (Apr 02 2020 at 14:51, on Zulip):

nikomatsakis said:

Santiago Pastorino might be interested

:+1:

pnkfelix (Apr 02 2020 at 14:52, on Zulip):

we already discussed #70608

pnkfelix (Apr 02 2020 at 14:52, on Zulip):

nominated 3/5: “regression: assertion failed: data.is_empty()” #70445

bjorn3 (Apr 02 2020 at 14:52, on Zulip):

@centril I don't think I will be able work on rustc itself for now.

centril (Apr 02 2020 at 14:52, on Zulip):

Matthew says its already fixed?

pnkfelix (Apr 02 2020 at 14:53, on Zulip):

@Matthew Jasper has stated they think #70445 is fixed. and this appears like it was a prioritization request, anyway?

centril (Apr 02 2020 at 14:53, on Zulip):

@simulacrum do you want to check?

centril (Apr 02 2020 at 14:53, on Zulip):

or should we just close?

pnkfelix (Apr 02 2020 at 14:53, on Zulip):

or, no the priority was addeed at the same time as the I-nominated ?

centril (Apr 02 2020 at 14:53, on Zulip):

@pnkfelix that's a suggested priority

pnkfelix (Apr 02 2020 at 14:53, on Zulip):

anyway we should check if it is indeed fixed

pnkfelix (Apr 02 2020 at 14:53, on Zulip):

Suggested priority?

pnkfelix (Apr 02 2020 at 14:53, on Zulip):

Okay

centril (Apr 02 2020 at 14:54, on Zulip):

@pnkfelix a priority suggested by me to be rejected or accepted at prioritization

pnkfelix (Apr 02 2020 at 14:55, on Zulip):

anyway it sounds like it is in hand, one way or another

pnkfelix (Apr 02 2020 at 14:55, on Zulip):

nominated 4/5: “internal compiler error: no type for local variable” #70594

centril (Apr 02 2020 at 14:55, on Zulip):

Hmm; I'm not so sure it is

centril (Apr 02 2020 at 14:55, on Zulip):

It's not clear to me who will confirm whether it was fixed

centril (Apr 02 2020 at 14:55, on Zulip):

unless we say that Matthew has already confirmed

nikomatsakis (Apr 02 2020 at 14:56, on Zulip):

I'm confused by that comment

centril (Apr 02 2020 at 14:56, on Zulip):

mine or Matthew's?

nikomatsakis (Apr 02 2020 at 14:56, on Zulip):

Are you referring to #70594, @centril ?

centril (Apr 02 2020 at 14:56, on Zulip):

@nikomatsakis no, https://github.com/rust-lang/rust/issues/70445

nikomatsakis (Apr 02 2020 at 14:57, on Zulip):

Oh, ok

nikomatsakis (Apr 02 2020 at 14:57, on Zulip):

It seems like #70594 has a pending fix by @Jonas Schievink

nikomatsakis (Apr 02 2020 at 14:57, on Zulip):

It's a fuzzer regression with useful-ish diagnostics, so medium is ok I thnk

centril (Apr 02 2020 at 14:57, on Zulip):

Usually when I have a fix, I remove I-nominated from an issue so it need not end up in a meeting :slight_smile:

simulacrum (Apr 02 2020 at 14:57, on Zulip):

sorry, had to step away for a moment, I think we should keep https://github.com/rust-lang/rust/issues/70445 open until we confirm it's fixed

simulacrum (Apr 02 2020 at 14:58, on Zulip):

I plan to do beta backports today

pnkfelix (Apr 02 2020 at 14:58, on Zulip):

I certainly wasn't suggesting that we close #70445

pnkfelix (Apr 02 2020 at 14:58, on Zulip):

I just meant that it is assigned

pnkfelix (Apr 02 2020 at 14:58, on Zulip):

the fact that it is nominated, sounds like @centril was asking WG-prioritization to confirm their suggested priority

pnkfelix (Apr 02 2020 at 14:58, on Zulip):

but hell, I'm not a mind reader so who knows

centril (Apr 02 2020 at 14:59, on Zulip):

@pnkfelix yep

Esteban Küber (Apr 02 2020 at 14:59, on Zulip):

In the last minute I wanted to ask, what should we do about joke PRs that should be harmless?

pnkfelix (Apr 02 2020 at 14:59, on Zulip):

wait wait

pnkfelix (Apr 02 2020 at 14:59, on Zulip):

we have one last nominatino

pnkfelix (Apr 02 2020 at 14:59, on Zulip):

lets try to cover it

pnkfelix (Apr 02 2020 at 14:59, on Zulip):

nominated 5/5: “Should enum discriminants have generics in scope?” #70453

Esteban Küber (Apr 02 2020 at 14:59, on Zulip):

Sure

pnkfelix (Apr 02 2020 at 15:00, on Zulip):

though #70453 may not be the kind of issue we can resolve in one minute of discussion

pnkfelix (Apr 02 2020 at 15:00, on Zulip):

so we could just agree to leave it for next week? Okay?

Esteban Küber (Apr 02 2020 at 15:00, on Zulip):

I feel we should have led with it :sweat_smile:

eddyb (Apr 02 2020 at 15:00, on Zulip):

lol

eddyb (Apr 02 2020 at 15:00, on Zulip):

it's not important though

eddyb (Apr 02 2020 at 15:01, on Zulip):

feel free to leave your own comments on the issue if you have thoughts

nikomatsakis (Apr 02 2020 at 15:01, on Zulip):

seems good to defer

nikomatsakis (Apr 02 2020 at 15:01, on Zulip):

interesting problem :)

pnkfelix (Apr 02 2020 at 15:01, on Zulip):

okay lets defer

centril (Apr 02 2020 at 15:01, on Zulip):

hmm; #70453 might get resolved at tonight's meeting anyways :slight_smile:

pnkfelix (Apr 02 2020 at 15:01, on Zulip):

so back to @Esteban Küber 's question

nikomatsakis (Apr 02 2020 at 15:01, on Zulip):

( I have a hard stop )

eddyb (Apr 02 2020 at 15:01, on Zulip):

oh yeah I should try to attend lang meetings even passively

pnkfelix (Apr 02 2020 at 15:01, on Zulip):

@Esteban Küber are we talking about April 1'st PR's or something?

Esteban Küber (Apr 02 2020 at 15:01, on Zulip):

Yep https://github.com/rust-lang/rust/pull/70645

eddyb (Apr 02 2020 at 15:02, on Zulip):

/me has voice/video call bandwidth now because WFH

centril (Apr 02 2020 at 15:02, on Zulip):

:pineapple: on :pizza: is clearly immoral :slight_smile:

Esteban Küber (Apr 02 2020 at 15:02, on Zulip):

That was closed by somebody else, but honestly it seemed harmless enough

nikomatsakis (Apr 02 2020 at 15:02, on Zulip):

I think closing is correct :)

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

I think we close joke PR's quickly. April 1st ones can wait until April 2nd.

nikomatsakis (Apr 02 2020 at 15:02, on Zulip):

Though I enjoyed the PR

Esteban Küber (Apr 02 2020 at 15:03, on Zulip):

Beyond the culinary ethics, should we accept code to handle things like break 'rust

nikomatsakis (Apr 02 2020 at 15:03, on Zulip):

are you asking, should we add easter eggs?

centril (Apr 02 2020 at 15:03, on Zulip):

@Esteban Küber break rust; has a technical purpose :slight_smile:

nikomatsakis (Apr 02 2020 at 15:03, on Zulip):

I'm inclined not to :)

pnkfelix (Apr 02 2020 at 15:03, on Zulip):

Hey @Santiago Pastorino and @oli : I have to go

Esteban Küber (Apr 02 2020 at 15:03, on Zulip):

Or accept them if people come with them

centril (Apr 02 2020 at 15:03, on Zulip):

says @RalfJ https://github.com/rust-lang/rust/issues/70661

pnkfelix (Apr 02 2020 at 15:03, on Zulip):

but can you each post your WG's updates in turn? Maybe @Santiago Pastorino can let @oli go first?

Esteban Küber (Apr 02 2020 at 15:04, on Zulip):

Ok, that settles it, no Easter eggs :egg:

pnkfelix (Apr 02 2020 at 15:04, on Zulip):

(that is, @oli first, then @Santiago Pastorino after with WG-meta)

centril (Apr 02 2020 at 15:04, on Zulip):

Esteban Küber said:

Or accept them if people come with them

Depends on how funny they are? :P

DPC (Apr 02 2020 at 15:04, on Zulip):

close it else you might get a ping from triage :D

oli (Apr 02 2020 at 15:04, on Zulip):

wg-mir-opt checking

Santiago Pastorino (Apr 02 2020 at 15:05, on Zulip):

T-compiler/meta

eddyb (Apr 02 2020 at 15:11, on Zulip):

do you mean EMIT_MIR? :P

centril (Apr 02 2020 at 15:12, on Zulip):

@pnkfelix btw, don't forget to add beta-accepted to the PRs :slight_smile:

eddyb (Apr 02 2020 at 15:12, on Zulip):

(also, @anyska is on it, idk how many mir-opt tests she'll change though)

oli (Apr 02 2020 at 15:13, on Zulip):

oops

oli (Apr 02 2020 at 15:13, on Zulip):

yea EMIT_MIR

centril (Apr 02 2020 at 15:13, on Zulip):

@simulacrum should I assign you to https://github.com/rust-lang/rust/issues/70445 to check whether it was fixed?

simulacrum (Apr 02 2020 at 15:14, on Zulip):

I guess? I mean anyone can do that work

eddyb (Apr 02 2020 at 16:34, on Zulip):

@nikomatsakis wait, are we doing FCP or anything like that for the git subtree switch?

eddyb (Apr 02 2020 at 16:34, on Zulip):

I recall something along the lines of we'll discuss it in the meeting, but did we?

pnkfelix (Apr 02 2020 at 17:32, on Zulip):

we didn't discuss it in this meeting

pnkfelix (Apr 02 2020 at 17:32, on Zulip):

if its high priority, I suppose we could allocate time for it at the beginning of the design meeting tomorrow

pnkfelix (Apr 02 2020 at 17:33, on Zulip):

otherwise, it could wait until next thursday

eddyb (Apr 02 2020 at 17:33, on Zulip):

I think the hope was that we'd do it relatively soon, but I'm not sure

simulacrum (Apr 02 2020 at 17:37, on Zulip):

I mean as with all things we want to do them sooner rather than later

simulacrum (Apr 02 2020 at 17:37, on Zulip):

But I think it can be async FCP, personally

simulacrum (Apr 02 2020 at 17:37, on Zulip):

OTOH, I don't think oli has kicked that off (and I think they can?)

simulacrum (Apr 02 2020 at 17:37, on Zulip):

cc @oli

pnkfelix (Apr 02 2020 at 17:40, on Zulip):

lets at least mention it in the design meeting tomorrow. That will hopefully help get/keep the ball rolling.

simulacrum (Apr 02 2020 at 17:40, on Zulip):

https://github.com/rust-lang/compiler-team/issues/266 is the major change issue

pnkfelix (Apr 02 2020 at 17:41, on Zulip):

Ah, that's something else we need to add to the standing agenda. :)

oli (Apr 02 2020 at 17:50, on Zulip):

I haven't done an FCP, because I have no clue what the next step on the major changes proposal is

oli (Apr 02 2020 at 17:50, on Zulip):

(except mention it in the announcements of the compiler mtg)

eddyb (Apr 02 2020 at 21:27, on Zulip):

eddyb said:

(also, @anyska is on it, idk how many mir-opt tests she'll change though)

the answer is "all of them" turns out: #70721

Last update: Jun 04 2020 at 17:45UTC