Stream: t-compiler

Topic: weekly meeting 2019-10-24 #54818


pnkfelix (Oct 24 2019 at 09:27, on Zulip):

Hi @T-compiler/meeting ; the triage meeting will be starting in 4 hours 33 minutes

pnkfelix (Oct 24 2019 at 09:28, on Zulip):

I have a conflict this afternoon, so @nagisa will be running the meeting today, and I will be doing the pre-triage a little earlier than usual. In addition, since I won't be able to attend the meeting, I may put a few more notes here than usual as I do the pre-triage.

pnkfelix (Oct 24 2019 at 09:29, on Zulip):

I will be doing pre-triage in a parallel topic

pnkfelix (Oct 24 2019 at 09:30, on Zulip):

For this weeks WG-checkins, we have scheduled NLL and Parallel Rustc

pnkfelix (Oct 24 2019 at 09:32, on Zulip):

I suspect NLL has nothing to report, although skimming over NLL stream I am reminded that we (I?) need to write up a blog post about migration being "complete", all the way to stable...

pnkfelix (Oct 24 2019 at 09:33, on Zulip):

for Parallel Rustc: hey @Zoxc , would you like to give an update at the end of today's meeting (in a little over 5 hours I suppose...)

pnkfelix (Oct 24 2019 at 10:16, on Zulip):

regarding issue "Non-'static lifetime inferred for return position impl Trait (fixed in beta)" #65582, it would be good to survey the cases that are affected. I wrote some comments on the issue. The main point is that there are a number of long-standing issues here, and it seems like PR #62849 might be saying "okay lets just error in this cases rather than commit to some fix that we might dislike in the long-term." I don't object to that on its face, but I want to make sure everyone's aware of it.

pnkfelix (Oct 24 2019 at 10:16, on Zulip):

(because it seems to me like we explicitly decided not to disallow this in the past...)

pnkfelix (Oct 24 2019 at 10:29, on Zulip):

:exclamation: regarding issue "Rust 1.38 regressions weren't fully triaged" #65577 : during discussion of this nomination, it would be good if we could dig into why the triage of the regressions wasn't completed before the release. (Was it a process failure, or was it due to lack of resources aka ran out of time)

pnkfelix (Oct 24 2019 at 11:43, on Zulip):

:exclamation: I left this one nominated: "Reduce the places where stable annotations are needed" #65515

pnkfelix (Oct 24 2019 at 11:43, on Zulip):

it sort of left me speechless

pnkfelix (Oct 24 2019 at 11:44, on Zulip):

next, beta-regressions without priority labells

pnkfelix (Oct 24 2019 at 11:45, on Zulip):

there are four

pnkfelix (Oct 24 2019 at 11:45, on Zulip):

beta regr 1/4: "non-empty glob statement regression 1.39" #65523

pnkfelix (Oct 24 2019 at 11:45, on Zulip):

triage: P-high

pnkfelix (Oct 24 2019 at 11:46, on Zulip):

beta regr 2/4: "array impls <= 32 1.39 regression" #65522

pnkfelix (Oct 24 2019 at 11:46, on Zulip):

comments say possibly not a true regression

pnkfelix (Oct 24 2019 at 11:47, on Zulip):

triage: P-high (to get to the bottom of what the bug is)

pnkfelix (Oct 24 2019 at 11:47, on Zulip):

beta regr 3/4: "derive structs (inert attribute?) regression 1.39" #65521

pnkfelix (Oct 24 2019 at 11:48, on Zulip):

seems like this was anticipated fallout from PR #63468

pnkfelix (Oct 24 2019 at 11:49, on Zulip):

triage: P-high, I-nominated for team to decide whether to close with no deprecation lint.

pnkfelix (Oct 24 2019 at 11:50, on Zulip):

beta regr 4/4: "bindgen changes in 1.39" #65520

pnkfelix (Oct 24 2019 at 11:50, on Zulip):

triage: P-high to get to bottom of what happened.

pnkfelix (Oct 24 2019 at 11:51, on Zulip):

next nightly regressions

pnkfelix (Oct 24 2019 at 11:51, on Zulip):

we have one

pnkfelix (Oct 24 2019 at 11:51, on Zulip):

nightly regr: "Compiler doesn't work when using Docker" #65662

pnkfelix (Oct 24 2019 at 11:52, on Zulip):

looks like the T-libs team is taking care of this

pnkfelix (Oct 24 2019 at 11:52, on Zulip):

(I must have followed the wrong link for the regressions list...) no, its the right link; just doesn't filter T-libs tagged out

pnkfelix (Oct 24 2019 at 11:53, on Zulip):

I have to go now

pnkfelix (Oct 24 2019 at 11:54, on Zulip):

(we have 32 open P-high issues. I'll try to walk over them either later tonight, or tomorrow)

nagisa (Oct 24 2019 at 13:09, on Zulip):

Meeting’s in 50 mins

nagisa (Oct 24 2019 at 13:56, on Zulip):

@nikomatsakis I rubbed something wrong into my eye and I don’t think it will appreciate me trying to look at a bright screen for an hour. Can you please take over?

nikomatsakis (Oct 24 2019 at 13:56, on Zulip):

can do @nagisa

nagisa (Oct 24 2019 at 13:56, on Zulip):

Thanks!

nikomatsakis (Oct 24 2019 at 14:02, on Zulip):

Dear @T-compiler/meeting, triage meeting is now!

To begin, please leave any

Announcements

nikomatsakis (Oct 24 2019 at 14:03, on Zulip):

Today we will have WG check-ins from:

simulacrum (Oct 24 2019 at 14:03, on Zulip):

I am, yes -- not sure how much of an update I can give but a quick summary probably

simulacrum (Oct 24 2019 at 14:03, on Zulip):

We _just_ landed the split out of rustc-dev today

simulacrum (Oct 24 2019 at 14:04, on Zulip):

I think it'll stick this time so if you're testing with nightly you'll need to add the rustc-dev component if you want extern crate rustc; and such to work

nikomatsakis (Oct 24 2019 at 14:04, on Zulip):

(Not sure if @Zoxc is here, they didn't answer @pnkfelix's ping earlier, so I'm assuming not)

nikomatsakis (Oct 24 2019 at 14:05, on Zulip):
nikomatsakis (Oct 24 2019 at 14:05, on Zulip):

OK, I guess we'll get going.

nikomatsakis (Oct 24 2019 at 14:06, on Zulip):

Heh, I'm trying to remember the order @pnkfelix normally uses

nikomatsakis (Oct 24 2019 at 14:06, on Zulip):

/me looks at #54818

nikomatsakis (Oct 24 2019 at 14:06, on Zulip):

next, beta-regressions without priority labells

let's start with these

nikomatsakis (Oct 24 2019 at 14:06, on Zulip):

beta regr 1/4: "non-empty glob statement regression 1.39" #65523

nikomatsakis (Oct 24 2019 at 14:07, on Zulip):

@Vadim Petrochenkov writes:

Similarly to #65090 this is very likely a combination of the recent changes in expansion order (#63248, #63867) and globs not fetching macro expanded names properly (#56593).

In this case, however, the fix may be to turn the "non-empty glob must import something" error into a lint, we have an issue about that - #62334.

nikomatsakis (Oct 24 2019 at 14:08, on Zulip):

I guess this seems P-high

nikomatsakis (Oct 24 2019 at 14:10, on Zulip):

beta regr 2/4: "array impls <= 32 1.39 regression" #65522

nikomatsakis (Oct 24 2019 at 14:10, on Zulip):

Seems like this needs someone to try and get to the bottom of it

nikomatsakis (Oct 24 2019 at 14:10, on Zulip):

(OK, I guess @pnkfelix already categorized these as P-high, I see, I was confused)

nikomatsakis (Oct 24 2019 at 14:10, on Zulip):

I'll just highlight them for now

nikomatsakis (Oct 24 2019 at 14:11, on Zulip):

beta regr 3/4: "derive structs (inert attribute?) regression 1.39" #65521

this was anticipated, the question is to decide whether we want a future compat warning

nikomatsakis (Oct 24 2019 at 14:11, on Zulip):

beta regr 4/4: "bindgen changes in 1.39" #65520

needs investigation

nikomatsakis (Oct 24 2019 at 14:12, on Zulip):

let's try to decide about #65521 --

nikomatsakis (Oct 24 2019 at 14:12, on Zulip):

@centril writes:

The regressions do not seem numerous or serious enough to warrant that work imo.

there are four I think

nikomatsakis (Oct 24 2019 at 14:12, on Zulip):

@Vadim Petrochenkov wrote:

This is a consequence of #63468.
Attributes on fields, variants and other places are now properly passed through name resolution.
This means we now detect attempts to apply derives to them, while previously they were just ignored.

The new behavior is correct, but we can make a deprecation lint for this if necessary.

mw (Oct 24 2019 at 14:14, on Zulip):

what's an "inert attribute"?

nikomatsakis (Oct 24 2019 at 14:14, on Zulip):

I think it means "an attribute that doesn't have any meaning"

mw (Oct 24 2019 at 14:14, on Zulip):

beside that I think it's fine that code that's obviously wrong breaks now

nikomatsakis (Oct 24 2019 at 14:14, on Zulip):

I'm seeing errors for cases like

#[derive(Foo)] struct Bar { #[derive(...)] f: u32 }
nikomatsakis (Oct 24 2019 at 14:15, on Zulip):

previously, derive was kind of ignored here; now it's getting an error

nikomatsakis (Oct 24 2019 at 14:15, on Zulip):

ok, let's leave it

nikomatsakis (Oct 24 2019 at 14:15, on Zulip):

Sorry, I was skimming the examples above. I think I agree the scope of the fallout is acceptable.

nikomatsakis (Oct 24 2019 at 14:16, on Zulip):

ok, beta backports

nikomatsakis (Oct 24 2019 at 14:17, on Zulip):

Avoid ICE when adjusting bad self ty #65755

nikomatsakis (Oct 24 2019 at 14:18, on Zulip):

(Seems harmless to me, i'm going to move to the next)

nikomatsakis (Oct 24 2019 at 14:18, on Zulip):

(but feel free to object)

nikomatsakis (Oct 24 2019 at 14:18, on Zulip):

Avoid ICE when checking Destination of break inside a closure #65518

nikomatsakis (Oct 24 2019 at 14:19, on Zulip):

(same)

nikomatsakis (Oct 24 2019 at 14:19, on Zulip):

save-analysis: Nest tables when processing impl block definitions #65511

nikomatsakis (Oct 24 2019 at 14:20, on Zulip):

(I believe @Igor Matuszewski said this was the cause of a number of problems for RLS users)

nikomatsakis (Oct 24 2019 at 14:20, on Zulip):

OK, no votes, but I'm going to go ahead and mark all as accepted unless any appear soon :)

nikomatsakis (Oct 24 2019 at 14:21, on Zulip):

Very good, all accepted. :)

mw (Oct 24 2019 at 14:21, on Zulip):

#65755 seems fine but hasn't landed yet

mw (Oct 24 2019 at 14:22, on Zulip):

it's also nominated for a stable backport?

nikomatsakis (Oct 24 2019 at 14:23, on Zulip):

yeah, I forgot what the procedure is on that (stable backpor)

nikomatsakis (Oct 24 2019 at 14:23, on Zulip):

also, I found one backport nomination that had no team assigned:

save-analysis: Don't ICE when resolving qualified type paths in struct members #65353

nikomatsakis (Oct 24 2019 at 14:23, on Zulip):

#65755 seems fine but hasn't landed yet

as far as that goes, I figured since it was small etc seemed ok to "pre-approve"

mw (Oct 24 2019 at 14:24, on Zulip):

Those are some great fixes by @Igor Matuszewski :)

nikomatsakis (Oct 24 2019 at 14:25, on Zulip):

OK -- I'm not sure what to do about stable backports, I think we're just supposed to evaluate the "technical merits" thereof

nikomatsakis (Oct 24 2019 at 14:25, on Zulip):

there are two cases, both ones we already saw

mw (Oct 24 2019 at 14:25, on Zulip):

I'm fine with accepting #65755, just want to mention that it hadn't landed yet

nikomatsakis (Oct 24 2019 at 14:25, on Zulip):
nikomatsakis (Oct 24 2019 at 14:26, on Zulip):

(in both cases I would be comfortable with a stable backport from a technical perspective)

nikomatsakis (Oct 24 2019 at 14:27, on Zulip):

OK, i'm going to comment to that effect and move on

nikomatsakis (Oct 24 2019 at 14:28, on Zulip):

S-waiting-on-our-team

nikomatsakis (Oct 24 2019 at 14:28, on Zulip):

I believe both are old-ish -- I'll remove the tag from #65408, and I think #59064 (HIR indexing) is a pending design meeting proposal

nikomatsakis (Oct 24 2019 at 14:29, on Zulip):

Nominated issues (link)

nikomatsakis (Oct 24 2019 at 14:29, on Zulip):

@pnkfelix kind of highlighted some of these above

nikomatsakis (Oct 24 2019 at 14:30, on Zulip):

[Quoting…]

nikomatsakis (Oct 24 2019 at 14:30, on Zulip):

:exclamation: regarding issue "Non-'static lifetime inferred for return position impl Trait (fixed in beta)" #65582, it would be good to survey the cases that are affected. I wrote some comments on the issue. The main point is that there are a number of long-standing issues here, and it seems like PR #62849 might be saying "okay lets just error in this cases rather than commit to some fix that we might dislike in the long-term." I don't object to that on its face, but I want to make sure everyone's aware of it.

nikomatsakis (Oct 24 2019 at 14:31, on Zulip):

So, what happened here is that in 1.39 we fixed some old behavior that was quite questionable (in PR https://github.com/rust-lang/rust/pull/62849)

nikomatsakis (Oct 24 2019 at 14:31, on Zulip):

We did a crater run

nikomatsakis (Oct 24 2019 at 14:32, on Zulip):

I think there's really no action needed here

nikomatsakis (Oct 24 2019 at 14:32, on Zulip):

though I'm not clear on much fallout we wound up observing later

nikomatsakis (Oct 24 2019 at 14:33, on Zulip):

skimming #64962, I don't see any, but @ecstatic-morse reported that they observed it?

nikomatsakis (Oct 24 2019 at 14:33, on Zulip):

anyway, I guess I'll move on

nikomatsakis (Oct 24 2019 at 14:34, on Zulip):

Rust 1.38 regressions weren't fully triaged #65577

nikomatsakis (Oct 24 2019 at 14:35, on Zulip):

It looks like @Matthew Jasper did a lot of work here, but there remain 5 cases untriaged?

nikomatsakis (Oct 24 2019 at 14:35, on Zulip):

(ugh, that's too ugly, removing)

nikomatsakis (Oct 24 2019 at 14:35, on Zulip):

seems like we need someone to do some of that investigation? @simulacrum do you know whether it occurred? I'm a bit unclear

simulacrum (Oct 24 2019 at 14:36, on Zulip):

I am unclear as well how much investigation has occurred

simulacrum (Oct 24 2019 at 14:36, on Zulip):

I also don't think this is super high priority given stable/stable regressions at this point

simulacrum (Oct 24 2019 at 14:36, on Zulip):

but would be good to dig in, probably

nikomatsakis (Oct 24 2019 at 14:37, on Zulip):

OK, I see, these are regressions from 1.38 that got overlooked, and we're now focused on 1.39 (currently in beta)

nikomatsakis (Oct 24 2019 at 14:37, on Zulip):

does seem like we should investigate

nikomatsakis (Oct 24 2019 at 14:38, on Zulip):

let's look at other nominated issues, maybe we'll find someone for that

nikomatsakis (Oct 24 2019 at 14:38, on Zulip):

ICE when using 'impl Trait' with an infinity recursive function #65561

nikomatsakis (Oct 24 2019 at 14:38, on Zulip):

tagged as P-high by @pnkfelix

nikomatsakis (Oct 24 2019 at 14:39, on Zulip):

first step is probably to bisect

nikomatsakis (Oct 24 2019 at 14:39, on Zulip):

(speaking of which, that's probably another ICE-breakers like group we could create: compiler bisectors)

nikomatsakis (Oct 24 2019 at 14:40, on Zulip):

anybody volunteer to run a cargo-bisect?

varkor (Oct 24 2019 at 14:40, on Zulip):

it emits a proper warning message before ICEing, so it's possible we can just do an early return somewhere to avoid this

nikomatsakis (Oct 24 2019 at 14:40, on Zulip):

I may have the setup somewhere

nikomatsakis (Oct 24 2019 at 14:40, on Zulip):

it emits a proper warning message before ICEing, so it's possible we can just do an early return somewhere to avoid this

or delay_span_bug cal

nikomatsakis (Oct 24 2019 at 14:40, on Zulip):

likely an easy fix

nikomatsakis (Oct 24 2019 at 14:40, on Zulip):

or, it emits a warning, sorry

varkor (Oct 24 2019 at 14:41, on Zulip):

well, it's a warning

davidtwco (Oct 24 2019 at 14:41, on Zulip):

I could look into this?

nikomatsakis (Oct 24 2019 at 14:41, on Zulip):

that'd be great @davidtwco :)

nikomatsakis (Oct 24 2019 at 14:41, on Zulip):

Reduce the places where stable annotations are needed #65515

nikomatsakis (Oct 24 2019 at 14:41, on Zulip):

I'm going to un-nominate, I think that @pnkfelix mostly wanted to draw attention to this.

nikomatsakis (Oct 24 2019 at 14:42, on Zulip):

Oh, I see, the other thing is that

This is normally not an issue, but with the new output in #65421 we will start showing this code to end users in errors, and it'd be nice if we could eliminate unnecessary boilerplate.

nikomatsakis (Oct 24 2019 at 14:42, on Zulip):

people will start to see the "under the covers" source code :)

nikomatsakis (Oct 24 2019 at 14:43, on Zulip):

linking of libtest failed #64872

nikomatsakis (Oct 24 2019 at 14:43, on Zulip):

remains nominated, but looks like maybe it doesn't have to be

mw (Oct 24 2019 at 14:44, on Zulip):

yeah, I think @Alex Crichton just needs to confirm the fix

ecstatic-morse (Oct 24 2019 at 14:44, on Zulip):

skimming #64962, I don't see any, but ecstatic-morse reported that they observed it?

Breakage occurred socketto-0.2. Two projects in crater depended on this version.

nikomatsakis (Oct 24 2019 at 14:44, on Zulip):

OK, that's all the nominated issues, beta backports, etc

mw (Oct 24 2019 at 14:44, on Zulip):

once there is a PR

varkor (Oct 24 2019 at 14:45, on Zulip):

there are also 2 nominated PRs

nikomatsakis (Oct 24 2019 at 14:45, on Zulip):

ok, thanks

nikomatsakis (Oct 24 2019 at 14:45, on Zulip):

Allow foreign exceptions to unwind through Rust code #65646

nikomatsakis (Oct 24 2019 at 14:45, on Zulip):

this is under discussion also in the #wg-ffi-unwind channel (which I need to catch up on)

nikomatsakis (Oct 24 2019 at 14:46, on Zulip):

I think the role for us in this team is more technical review, if anything

nikomatsakis (Oct 24 2019 at 14:46, on Zulip):

though if you have opinions about what our behavior ought to be here, please go to #wg-ffi-unwind !

nikomatsakis (Oct 24 2019 at 14:46, on Zulip):

Lint ignored #[inline] on function prototypes #65294

nikomatsakis (Oct 24 2019 at 14:46, on Zulip):

This can be a breaking change in some cases:

As a consequence, foreign items, impl items and trait items now have their attributes checked, which could cause some code to no longer compile (it was previously erroneously ignored).

nikomatsakis (Oct 24 2019 at 14:47, on Zulip):

In particular, @varkor notes that this affects the old bitflags crate

nikomatsakis (Oct 24 2019 at 14:47, on Zulip):

@varkor do you want to discuss this here?

nikomatsakis (Oct 24 2019 at 14:48, on Zulip):

I guess the question is: how hard to error here, and do we want to have some sort of special treatment whereby we don't error for cases injected from dependencies?

varkor (Oct 24 2019 at 14:48, on Zulip):

the question I had was whether it was reasonable to emit an error that acted like a lint, in that it was ignored when compiling external crates

varkor (Oct 24 2019 at 14:48, on Zulip):

because as far as I know, there's not precedent for that now

nikomatsakis (Oct 24 2019 at 14:49, on Zulip):

I'm tempted to say we should make this a future-compatibility warning (or just a warn-by-default lint)

nikomatsakis (Oct 24 2019 at 14:49, on Zulip):

(and improve our tooling around this, as has been discussed, to help prod users into upgrading)

varkor (Oct 24 2019 at 14:49, on Zulip):

the confusing thing about making it a warning is that the same error will be emitted as a proper error and as a warning depending on where the attribute occurs

nikomatsakis (Oct 24 2019 at 14:49, on Zulip):

in general, is there a strong concern about #[inline] being used in wrong places? like, do we expect to want to give it some semantics that would be inappropriate?

varkor (Oct 24 2019 at 14:50, on Zulip):

and also that it could then be allowed

varkor (Oct 24 2019 at 14:51, on Zulip):

I think there's concern that it may look like it has an effect on function prototypes, but is ignored

varkor (Oct 24 2019 at 14:51, on Zulip):

we could special case just inline on consts specifically, essentially special-casing bitflags

nikomatsakis (Oct 24 2019 at 14:51, on Zulip):

yeah, that seems...not that serious to me

nikomatsakis (Oct 24 2019 at 14:51, on Zulip):

esp. if we are giving warnings

nikomatsakis (Oct 24 2019 at 14:51, on Zulip):

I guess I just don't feel strongly motivated to add a lot of special casing

varkor (Oct 24 2019 at 14:52, on Zulip):

well, we either special case items in traits and extern, or we special case consts in traits and extern

varkor (Oct 24 2019 at 14:52, on Zulip):

items used anywhere else already throw the error

varkor (Oct 24 2019 at 14:53, on Zulip):

but I'll just make those cases a future-incompatibility lint :+1:

nikomatsakis (Oct 24 2019 at 14:53, on Zulip):

ok

nikomatsakis (Oct 24 2019 at 14:53, on Zulip):

sounds good

nikomatsakis (Oct 24 2019 at 14:53, on Zulip):

ok, we have a few minutes left, brief wg check-ins maybe?

nikomatsakis (Oct 24 2019 at 14:53, on Zulip):

NLL

nikomatsakis (Oct 24 2019 at 14:54, on Zulip):

Not sure who to tag for this -- but the big news is that the next release will see the first stable release without the HIR borrow checker

nikomatsakis (Oct 24 2019 at 14:54, on Zulip):

we need to do a blog post warning people about this

nikomatsakis (Oct 24 2019 at 14:54, on Zulip):

I think at that point the NLL WG will basically disband

nikomatsakis (Oct 24 2019 at 14:55, on Zulip):

( but #t-compiler/wg-polonius carries the torch! :runner: )

nikomatsakis (Oct 24 2019 at 14:55, on Zulip):

Any other notes on this? maybe cc @centril, @Matthew Jasper, @simulacrum?

(I think I will try to get a 1st draft of blog post going, maybe hand it off to y'all for tips and further advice)

nikomatsakis (Oct 24 2019 at 14:55, on Zulip):

wg-parallel-rustc

nikomatsakis (Oct 24 2019 at 14:56, on Zulip):

@simulacrum you still around?

simulacrum (Oct 24 2019 at 14:56, on Zulip):

next next release fwiw, 1.39 still has HIR IIRC

simulacrum (Oct 24 2019 at 14:56, on Zulip):

Yeah

simulacrum (Oct 24 2019 at 14:56, on Zulip):

So we've been going fairly slow but steady

simulacrum (Oct 24 2019 at 14:56, on Zulip):

We have regular triage meetings on Mondays (in compiler calendar if you're interested)

simulacrum (Oct 24 2019 at 14:57, on Zulip):

So far the general goal is to refactor/audit compiler locks and other parallel pieces, as well as to investigate improving performance.

nikomatsakis (Oct 24 2019 at 14:57, on Zulip):

(oh, that reminds me, I need to post video from the last meeting)

simulacrum (Oct 24 2019 at 14:57, on Zulip):

Our plan is to have the MVP be to do high-level parallelization in a few critical places

simulacrum (Oct 24 2019 at 14:58, on Zulip):

(LLVM -- already done, but we can do better; typeck, I believe; linting, etc)

simulacrum (Oct 24 2019 at 14:58, on Zulip):

I think that's all

nikomatsakis (Oct 24 2019 at 14:58, on Zulip):

it seems like we've determined that the use of locks is fairly ok. We've still got some cases that need to be looked at and cleaned-up and/or documented (cough Session cough), but overall not a huge problem.

nikomatsakis (Oct 24 2019 at 14:58, on Zulip):

Not that this is that surprising :)

nikomatsakis (Oct 24 2019 at 14:59, on Zulip):

But the bigger concern is that the parallel performance isn't as good as we might like, and in particular the jobserver integration and timing for big cargo builds is not great, right? Thus it's not quite at the point we would like to justify the slowdown of adding locks

nikomatsakis (Oct 24 2019 at 14:59, on Zulip):

(Thus we were investigating how things are working now and what experiments we might do to see what needs to be improved)

simulacrum (Oct 24 2019 at 15:00, on Zulip):

Ah, yes -- we're looking to investigate jobserver performance in particular as the current belief of the source of our problems, but we don't yet know if that's actually the problem :)

mw (Oct 24 2019 at 15:01, on Zulip):

freakin' jobserver :)

nikomatsakis (Oct 24 2019 at 15:01, on Zulip):

video is uploading now

nikomatsakis (Oct 24 2019 at 15:01, on Zulip):

OK, I guess that is the end of our time here folks

nikomatsakis (Oct 24 2019 at 15:01, on Zulip):

Fin

centril (Oct 24 2019 at 16:16, on Zulip):

@nikomatsakis I think the wg-nll stream should probably disband after the leak-check / migrate-mode and possibly x @ Some(y) things have been fixed

centril (Oct 24 2019 at 16:16, on Zulip):

so there's still a tad bit of work left

nikomatsakis (Oct 24 2019 at 17:53, on Zulip):

Makes sense, though I think in practice that work is more taking place in / around #wg-traits. Anyway, not a big distinction.

centril (Oct 24 2019 at 18:17, on Zulip):

Fair point :slight_smile:

centril (Oct 24 2019 at 18:18, on Zulip):

x @ Some(y) can probably be moved to the t-compiler stream

Last update: Nov 22 2019 at 04:35UTC