Stream: t-compiler

Topic: pre-meeting triage 2019-04-04 #54818


pnkfelix (Apr 04 2019 at 11:27, on Zulip):

This week's triage will be tracked here.

pnkfelix (Apr 04 2019 at 11:34, on Zulip):

first pre-pass: nominated unprioritized T-compiler issues

pnkfelix (Apr 04 2019 at 11:34, on Zulip):

wow, 7 issues. Okay

pnkfelix (Apr 04 2019 at 11:35, on Zulip):

first: "Compiler panic with generic-typed nested closures" #59494

pnkfelix (Apr 04 2019 at 11:35, on Zulip):

Its an ICE, a stable-to-stable regression.

pnkfelix (Apr 04 2019 at 11:36, on Zulip):

I'm going to initially prioritize this as P-high, since that's our default for ICE's.

pnkfelix (Apr 04 2019 at 11:36, on Zulip):

and I'm going to remove the nominated tag because I don't think there's anything to discuss beyond the prioirization.

pnkfelix (Apr 04 2019 at 11:37, on Zulip):

next: "LNK1189 "library limit of 65535 obj exceeded" building rustc" #53014

pnkfelix (Apr 04 2019 at 11:39, on Zulip):

seems like a big problem, at least for Windows developers using incremental mode.

pnkfelix (Apr 04 2019 at 11:40, on Zulip):

oh wait, some people report this occurs for them without incremental ...? What ...?

pnkfelix (Apr 04 2019 at 11:42, on Zulip):

anyway, triage: P-high. Leaving nominated.

pnkfelix (Apr 04 2019 at 11:42, on Zulip):

next: "ICE in librustc_codegen_llvm when building kernel" #59548

pnkfelix (Apr 04 2019 at 11:45, on Zulip):

triage: P-high, removing I-nominated.

pnkfelix (Apr 04 2019 at 11:45, on Zulip):

next: "Incremental compilation error building rustc (-i --keep-stage 0)" #59105

pnkfelix (Apr 04 2019 at 11:47, on Zulip):

@centril nominated this presumably to get it prioritized.

centril (Apr 04 2019 at 11:48, on Zulip):

yep

pnkfelix (Apr 04 2019 at 11:48, on Zulip):

it strikes me as an unfortunate known bug. Obviously I'd be happy if someone fixed it as part of scratching this nasty itch.

centril (Apr 04 2019 at 11:48, on Zulip):

(also this has happened to me many times)

pnkfelix (Apr 04 2019 at 11:48, on Zulip):

but I don't know if I can give it high priority based on that alone

pnkfelix (Apr 04 2019 at 11:48, on Zulip):

I did fix something similar recently because it was my own personal itch

pnkfelix (Apr 04 2019 at 11:49, on Zulip):

but in this case its hypothesized to be a rustbuild issue, while the thing I addressed was about needing to whitelist rustc attributes in the compiler itself...

pnkfelix (Apr 04 2019 at 11:51, on Zulip):

I'll leave it nominated, add a note that my initial inclination is to prioritize as P-medium but I'll wait to see if others disagree that the T-compiler meeting itself. Its possible that just giving the issue visibility there will be enough to get a volunteer to fix it.

pnkfelix (Apr 04 2019 at 11:53, on Zulip):

next: "Unused arguments to async fn are dropped too early" #54716

pnkfelix (Apr 04 2019 at 11:53, on Zulip):

ah this is nominated for T-lang discussion, not T-compiler.

pnkfelix (Apr 04 2019 at 11:55, on Zulip):

I don't know how to prioritize it either

pnkfelix (Apr 04 2019 at 11:56, on Zulip):

its blocking AsyncAwait, but that's already reflected in its tags.

pnkfelix (Apr 04 2019 at 11:56, on Zulip):

Leaving unprioritized for now. It sounds like there are stakeholders moving to push it forward without a P-label anyway

pnkfelix (Apr 04 2019 at 11:56, on Zulip):

next: "Implement "pipelined" rustc compilation (compiler side)" #58465

pnkfelix (Apr 04 2019 at 11:57, on Zulip):

yep, another week with this nominated.

pnkfelix (Apr 04 2019 at 11:57, on Zulip):

ah but now there are comments on it

pnkfelix (Apr 04 2019 at 12:01, on Zulip):

okay well its nice to see progress in the form of debate and also in the form of volunteers to work

pnkfelix (Apr 04 2019 at 12:02, on Zulip):

I still don't have a clue as to prioritization of this work

pnkfelix (Apr 04 2019 at 12:02, on Zulip):

so I'm going to leave it unprioritized and still nominated once again.

pnkfelix (Apr 04 2019 at 12:20, on Zulip):

(@nikomatsakis is on PTO this week so I suspect we probably wont prioritize #58465 today. but we will see.)

pnkfelix (Apr 04 2019 at 12:21, on Zulip):

next: "[firefox] error: relocation refers to local symbol "" [12], which is defined in a discarded section" #59652

pnkfelix (Apr 04 2019 at 12:22, on Zulip):

yikes, the commit that is bisected to seems like it really shouldn't be causing that bug

pnkfelix (Apr 04 2019 at 12:22, on Zulip):

marking P-high

pnkfelix (Apr 04 2019 at 12:24, on Zulip):

okay that's all the nominated unprioritized issues.

pnkfelix (Apr 04 2019 at 12:24, on Zulip):

there are zero issues that are nominated with no team at all assigned, so nothing to look at there

pnkfelix (Apr 04 2019 at 12:38, on Zulip):

there are likewise zero beta nominations without P-label

pnkfelix (Apr 04 2019 at 12:38, on Zulip):

and zero nightly regressions without P-label

pnkfelix (Apr 04 2019 at 12:38, on Zulip):

so that's all the pre-passes. Now on to the real triage

pnkfelix (Apr 04 2019 at 12:39, on Zulip):

first up, the P-high issues

pnkfelix (Apr 04 2019 at 12:40, on Zulip):

first: "two-phase-borrows need a specification" #46901

pnkfelix (Apr 04 2019 at 12:40, on Zulip):

(i really should dedicate some time to that...)

pnkfelix (Apr 04 2019 at 12:41, on Zulip):

next: "beta 1.33 seems to break tarpaulin on multithreading" #58104

pnkfelix (Apr 04 2019 at 12:41, on Zulip):

closing as non-actionable.

pnkfelix (Apr 04 2019 at 12:41, on Zulip):

(which I threatened to do over a month ago.)

pnkfelix (Apr 04 2019 at 12:43, on Zulip):

next: "user type annotations are captured post normalization" #54940

pnkfelix (Apr 04 2019 at 12:44, on Zulip):

still no updates. Maybe I should try to broaden target audience to @WG-traits ?

pnkfelix (Apr 04 2019 at 12:44, on Zulip):

next: "Nightly rustc crashes with "unexpected region in query response" #57464

pnkfelix (Apr 04 2019 at 12:44, on Zulip):

i continue to poke at this during idle time

pnkfelix (Apr 04 2019 at 12:46, on Zulip):

i am going to at least add A-impl-trait label to it, since I'm pretty sure this issue only arises from uses of impl Trait, and thus anyone interested in that topic may likewise be interested in this bug

pnkfelix (Apr 04 2019 at 12:48, on Zulip):

next: "rustc 1.32.0 produces faulty wasm code" #58548

pnkfelix (Apr 04 2019 at 12:48, on Zulip):

ooh shiny pictures

pnkfelix (Apr 04 2019 at 13:02, on Zulip):

sounds like it is fixed but no one has confirmed. The narrowed example requires something called "yarn" to run, and I do not want to take the time now to learn what that is.

pnkfelix (Apr 04 2019 at 13:02, on Zulip):

next: "Encountered error Unimplemented selecting <...> during codegen" #58375

pnkfelix (Apr 04 2019 at 13:02, on Zulip):

no updates since it was assigned two weeks ago

pnkfelix (Apr 04 2019 at 13:03, on Zulip):

@Aaron Turon if you have any notes on #58375, feel free to jot them in that issue.

pnkfelix (Apr 04 2019 at 13:03, on Zulip):

(and if you need me to reassign, let me know.)

pnkfelix (Apr 04 2019 at 13:05, on Zulip):

next: "Compiler panic when using a slice pattern" #59016

pnkfelix (Apr 04 2019 at 13:06, on Zulip):

@oli wrote [this note}(https://github.com/rust-lang/rust/issues/59016#issuecomment-475523277) 13 days ago:

I believe that the problem is that structurally_resolved_type doesn't normalize its types. So while it resolves Self to [(); 0] in the above example, the 0 is still an unevaluated constant.

We basically have two options now:

1. remove unwrap_usize and instead create an evaluate_usize method that takes a tcx and tries to evaluate the constant
2. normalize types in more places (I'm not sure if that will cause unintended side effects or slowdowns)

oli (Apr 04 2019 at 13:06, on Zulip):

I have a PR open on that

oli (Apr 04 2019 at 13:07, on Zulip):

https://github.com/rust-lang/rust/pull/59369#discussion_r269218244

pnkfelix (Apr 04 2019 at 13:07, on Zulip):

oh yes you do

oli (Apr 04 2019 at 13:07, on Zulip):

I'll speed it up so we'll get it in soonish

pnkfelix (Apr 04 2019 at 13:07, on Zulip):

okay thanks @oli . PR #59369 still tagged as S-waiting-on-review; is that label inacurate?

eddyb (Apr 04 2019 at 13:08, on Zulip):

is that the one I need to review?

pnkfelix (Apr 04 2019 at 13:08, on Zulip):

You (@eddyb) are assigned to it, but you've written comments

eddyb (Apr 04 2019 at 13:08, on Zulip):

also, how available is @nikomatsakis for discussing internals of the trait/inference system?

pnkfelix (Apr 04 2019 at 13:09, on Zulip):

@nikomatsakis is on PTO this week

eddyb (Apr 04 2019 at 13:09, on Zulip):

both const generics and this PR need some input

eddyb (Apr 04 2019 at 13:09, on Zulip):

OH

pnkfelix (Apr 04 2019 at 13:09, on Zulip):

so they are not available right now

pnkfelix (Apr 04 2019 at 13:10, on Zulip):

well in any case it looks like #59016 is well in hand, in any case

eddyb (Apr 04 2019 at 13:10, on Zulip):

what about other {traits,infer}-system people?

pnkfelix (Apr 04 2019 at 13:10, on Zulip):

dunno, maybe lets ask at T-compiler meeting itself?

eddyb (Apr 04 2019 at 13:10, on Zulip):

sounds good, just wanted to mention it in case I forget later

pnkfelix (Apr 04 2019 at 13:11, on Zulip):

next: "miri no longer builds after rust-lang/rust#59471" #59477

centril (Apr 04 2019 at 13:12, on Zulip):

@eddyb maybe ariel?

pnkfelix (Apr 04 2019 at 13:13, on Zulip):

#59477 seems like it is being dealt with

pnkfelix (Apr 04 2019 at 13:14, on Zulip):

based on comments I saw ... last week? ... from @RalfJ , I'm going to treat miri tool breakage as P-medium, not P-high, in this case and in future instances

pnkfelix (Apr 04 2019 at 13:16, on Zulip):

next: "Function pointer docs may need updating" #46989

pnkfelix (Apr 04 2019 at 13:20, on Zulip):

I'm going to downgrade #46989 to P-medium but also mark it as blocking the re-remove the leak check issue.

pnkfelix (Apr 04 2019 at 13:21, on Zulip):

next: " Future-incompatible warnings should always print a warning, even if lints are allowed" #34596

pnkfelix (Apr 04 2019 at 13:21, on Zulip):

this has a PR (#59658), yay

pnkfelix (Apr 04 2019 at 13:21, on Zulip):

next: "Incorporate RLS bug tracking into compiler team triage" #58858

pnkfelix (Apr 04 2019 at 13:22, on Zulip):

as you might guess, I still haven't made progress here either.

pnkfelix (Apr 04 2019 at 13:22, on Zulip):

I probably should at least talk with @WG-rls2.0 about it

pnkfelix (Apr 04 2019 at 13:22, on Zulip):

I'll try to bring it up maybe next week

pnkfelix (Apr 04 2019 at 13:23, on Zulip):

next: "Specific code layout can cause compiler panic with lto=true" #59137

pnkfelix (Apr 04 2019 at 13:24, on Zulip):

Hey @mw , if you have any progress to note regarding #59137, feel free to jot a comment on the issue. If you need me to re-assign it, let me know.

pnkfelix (Apr 04 2019 at 13:25, on Zulip):

next: "match arm bindings have weird lifetimes" #46525

pnkfelix (Apr 04 2019 at 13:25, on Zulip):

assigned to @Matthew Jasper a week ago. I'm actually skeptical whether this needs to be P-high ...

pnkfelix (Apr 04 2019 at 13:26, on Zulip):

next: "Wrapping simd call results in compiler panic (nightly)" #59469

pnkfelix (Apr 04 2019 at 13:27, on Zulip):

On the surface, this sounds like it might be easy to fix. Assigning to self to either fix or mentor.

pnkfelix (Apr 04 2019 at 13:28, on Zulip):

(since its marked regression-from-stable-to-beta, it does sound like this is super high priority to address, the beta is being promoted to stable next week.)

pnkfelix (Apr 04 2019 at 13:29, on Zulip):

I guess all stable-to-beta regressions are in that state.

pnkfelix (Apr 04 2019 at 13:29, on Zulip):

we'll review them soon enough.

pnkfelix (Apr 04 2019 at 13:30, on Zulip):

next: "Rustc 1.33 stable panicked on a closure inside a closure" #58840

mw (Apr 04 2019 at 13:31, on Zulip):

@pnkfelix Just opened a PR that should fix #59137

pnkfelix (Apr 04 2019 at 13:32, on Zulip):

@mw fantastic, thanks

pnkfelix (Apr 04 2019 at 13:32, on Zulip):

regarding #58840, i haven't seen any response to my query asking for a re-posting of the sample project demo'ing the bug

pnkfelix (Apr 04 2019 at 13:33, on Zulip):

but meanwhile, other bugs have been filed that have been closed as duplicates of this one?

pnkfelix (Apr 04 2019 at 13:33, on Zulip):

seems non-ideal

pnkfelix (Apr 04 2019 at 13:38, on Zulip):

oh well I guess I will also adopt the assumption that those bugs are duplicates and thus one of their examples will suffice for making a reproduction.

pnkfelix (Apr 04 2019 at 13:42, on Zulip):

so... that's sort of like "progress" ...

pnkfelix (Apr 04 2019 at 13:43, on Zulip):

next: "[NLL] prohibit "two-phase borrows" with existing borrows?" #56254

pnkfelix (Apr 04 2019 at 13:43, on Zulip):

this will be resolved once PR #58739 lands

pnkfelix (Apr 04 2019 at 13:44, on Zulip):

next: "[nll] change how MIR represents places" #52708

pnkfelix (Apr 04 2019 at 13:49, on Zulip):

discussion and presumably progress continues

pnkfelix (Apr 04 2019 at 13:50, on Zulip):

next: "Compiler panic with generic-typed nested closures" #59494

pnkfelix (Apr 04 2019 at 13:51, on Zulip):

posted request-for-volunteer in main meeting topic area

pnkfelix (Apr 04 2019 at 13:52, on Zulip):

next: "LNK1189 "library limit of 65535 obj exceeded" building rustc" #53014

pnkfelix (Apr 04 2019 at 13:52, on Zulip):

I remember this one from the pre-pass, we'll need to talk about that at the meeting itself.

pnkfelix (Apr 04 2019 at 13:52, on Zulip):

next: "ICE in librustc_codegen_llvm when building kernel" #59548

pnkfelix (Apr 04 2019 at 13:54, on Zulip):

posted request-for-volunteer in main meeting topic area

pnkfelix (Apr 04 2019 at 13:54, on Zulip):

last: "[firefox] error: relocation refers to local symbol "" [12], which is defined in a discarded section" #59652

pnkfelix (Apr 04 2019 at 13:54, on Zulip):

this one looks goofy but maybe its easy. Anyway, I'll wait and see if japaric jumps in with anything

pnkfelix (Apr 04 2019 at 13:55, on Zulip):

(the fact that its killing attempts to build firefox is definitely worrisome)

pnkfelix (Apr 04 2019 at 13:55, on Zulip):

alright, that's all the P-high's, and I have 5 minutes before the meeting itself

pnkfelix (Apr 04 2019 at 13:55, on Zulip):

we'll look at the two beta-nominations in the meeting itself.

pnkfelix (Apr 04 2019 at 13:56, on Zulip):

there are no stable-nominations

pnkfelix (Apr 04 2019 at 13:56, on Zulip):

I've assigned the two stable-to-beta regressions to myself, so I'll try to fix them ... tomorrow I guess.

pnkfelix (Apr 04 2019 at 13:57, on Zulip):

oh wait, one of those is P-medium

pnkfelix (Apr 04 2019 at 13:58, on Zulip):

right, "compiler error: lifetimes in associated types leading to a compiler error." #58694 officially became a stable-to-stable regression at some point

pnkfelix (Apr 04 2019 at 13:59, on Zulip):

the only stable-to-nightly regression is the firefox bug mentioned above.

pnkfelix (Apr 04 2019 at 13:59, on Zulip):

"[do not merge] Measure performance impact of local interners" #57214 is still S-waiting-on-team (and nominated)

pnkfelix (Apr 04 2019 at 14:01, on Zulip):

it does seem like most team members have given assent here

pnkfelix (Apr 04 2019 at 14:01, on Zulip):

and there are six nominated issues

pnkfelix (Apr 04 2019 at 14:01, on Zulip):

so we'll see if we get to them during the meeting itself, starting now

Last update: Nov 16 2019 at 01:05UTC