Stream: t-compiler

Topic: pre-meeting triage 2019-10-10 #54818


pnkfelix (Oct 10 2019 at 12:31, on Zulip):

I will be doing pre-triage in this channel.

pnkfelix (Oct 10 2019 at 12:35, on Zulip):

there are 8 nominated unprioritized issues for T-compiler

pnkfelix (Oct 10 2019 at 12:35, on Zulip):

nom unpri: "check_mod_item_types query times regressed bigly" #65147

pnkfelix (Oct 10 2019 at 12:37, on Zulip):

the async await folks seem to have triaged this, so I'll mark it P-high.

pnkfelix (Oct 10 2019 at 12:39, on Zulip):

nom unpri: "Star Import Precedence Regression in Beta" #65090

pnkfelix (Oct 10 2019 at 12:41, on Zulip):

so @Vadim Petrochenkov has done some analysis in the comments on this issue

pnkfelix (Oct 10 2019 at 12:42, on Zulip):

from those comments, it sounds like what happened here is that a pre-existing bug was exposed by a change to expansion order.

pnkfelix (Oct 10 2019 at 12:42, on Zulip):

even the bug filer notes that the behavior in question may reflect a bug fix

pnkfelix (Oct 10 2019 at 12:44, on Zulip):

@Vadim Petrochenkov also notes that the main fix they can imagine for this is to "implement globs properly"

pnkfelix (Oct 10 2019 at 12:47, on Zulip):

@Vadim Petrochenkov has also supplied a workaround to the bug filer

pnkfelix (Oct 10 2019 at 12:48, on Zulip):

I'm going to mark as P-medium and leave nominated, so that we can double-check that priority assignment in the triage meeting.

pnkfelix (Oct 10 2019 at 12:49, on Zulip):

nom unpri: " rustfmt no longer builds after rust-lang/rust#65076" #65085

pnkfelix (Oct 10 2019 at 12:50, on Zulip):

we probably need to fix that

pnkfelix (Oct 10 2019 at 12:51, on Zulip):

(the discussion indicates that the fix is blocked due to rules forbidding tools from depending on "multiple rustc-ap-* crates at once", and RLS is also not building.

pnkfelix (Oct 10 2019 at 12:51, on Zulip):

P-high, removing nomination label.

pnkfelix (Oct 10 2019 at 12:52, on Zulip):

nom unpri: "rls no longer builds after rust-lang/rust#65076" #65084

pnkfelix (Oct 10 2019 at 12:52, on Zulip):

P-high, removing nomination label.

pnkfelix (Oct 10 2019 at 12:54, on Zulip):

nom unpri: "Add an unstable 'extra randomization' compile-time flag for hashers" #65042

centril (Oct 10 2019 at 12:54, on Zulip):

we wanted a design meeting possibly around determinism?

centril (Oct 10 2019 at 12:55, on Zulip):

I think if the compiler team says yes to the request it will be implemented by some random person soon-ish

pnkfelix (Oct 10 2019 at 12:55, on Zulip):

@RalfJ had a comment and link therein saying that FxHashMap is fully deterministic for a fixed platform and fixed order of insertions and deletions

pnkfelix (Oct 10 2019 at 12:55, on Zulip):

which to me indicates that adding extra randomness isn't going to actually address the core desire here, no?

centril (Oct 10 2019 at 12:56, on Zulip):

but if you change the seed then it is no longer deterministic?

centril (Oct 10 2019 at 12:57, on Zulip):

Anyways; let's continue with pre-triage here and maybe continue the discussion on-issue in the interest of completing pre-triage on time

pnkfelix (Oct 10 2019 at 12:57, on Zulip):

I guess re-reading the issue, adding extra randomness might achieve a goal of exposing bugs

pnkfelix (Oct 10 2019 at 12:58, on Zulip):

but I tend to agree with @simulacrum and @nikomatsakis that it seems like there would be more directed way of avoiding the bugs in the first place.

pnkfelix (Oct 10 2019 at 12:58, on Zulip):

anyway I don't think it needs to be nominated for further discussion at the weekly meeting

centril (Oct 10 2019 at 12:59, on Zulip):

@pnkfelix possibly yes, but I think one can attack this from multiple angles at once

centril (Oct 10 2019 at 12:59, on Zulip):

(I don't think tweaking the seed is a large project; I would suspect not more than a 50 LOC PR or something...)

pnkfelix (Oct 10 2019 at 12:59, on Zulip):

anyway lets try to table this and couple it with a broader discussion of determinism

centril (Oct 10 2019 at 12:59, on Zulip):

(so there's not a lot of time investment)

pnkfelix (Oct 10 2019 at 13:00, on Zulip):

is there an issue or meeting proposal page around determinism somewhere? I want to link this before I remove the nomination tag

centril (Oct 10 2019 at 13:01, on Zulip):

doesn't seem so

pnkfelix (Oct 10 2019 at 13:01, on Zulip):

ah well I'm going to tag this as P-low and let people complain about it in the aftermath.

centril (Oct 10 2019 at 13:02, on Zulip):

fair

pnkfelix (Oct 10 2019 at 13:03, on Zulip):

unpri nom: "Rustdoc support API in rustc" #59790

centril (Oct 10 2019 at 13:05, on Zulip):

sounds vague

pnkfelix (Oct 10 2019 at 13:05, on Zulip):

i nominated this last week. I assume I nominated it after the meeting, since I see no mention of this issue in the log from last week.

pnkfelix (Oct 10 2019 at 13:06, on Zulip):

so i'll just leave it nominated for discussion today. No need to prioritize.

centril (Oct 10 2019 at 13:06, on Zulip):

ah; @oli is talking about methods named rustdoc_something_something

pnkfelix (Oct 10 2019 at 13:06, on Zulip):

okay that's all the unprioritized nominated issues.

pnkfelix (Oct 10 2019 at 13:07, on Zulip):

there are zero unprioritized beta regressions

pnkfelix (Oct 10 2019 at 13:07, on Zulip):

there are zero unprioritized nightly regressions

pnkfelix (Oct 10 2019 at 13:08, on Zulip):

doing prepass over current nominations ... unnominating #64153 since @mw says they will look at it now

mw (Oct 10 2019 at 13:08, on Zulip):

:+1:

pnkfelix (Oct 10 2019 at 13:10, on Zulip):

doing prepass over all beta-nominations to make sure that all ones relevant to rustc are tagged T-compiler ...

pnkfelix (Oct 10 2019 at 13:10, on Zulip):

adding T-compiler tag to #65074

pnkfelix (Oct 10 2019 at 13:11, on Zulip):

there are no stable-nominations

centril (Oct 10 2019 at 13:11, on Zulip):

@pnkfelix can we send out an email clarifying the labeling system re beta noms?

pnkfelix (Oct 10 2019 at 13:11, on Zulip):

in what sense? the meaning of beta-nominated/beta-accepted? Or the necessity of actually adding a T-label?

centril (Oct 10 2019 at 13:12, on Zulip):

the latter

pnkfelix (Oct 10 2019 at 13:12, on Zulip):

there's this: https://forge.rust-lang.org/release/beta-backporting.html

centril (Oct 10 2019 at 13:12, on Zulip):

I see this mistake often

centril (Oct 10 2019 at 13:12, on Zulip):

I don't think that page is read by people other than the release team

pnkfelix (Oct 10 2019 at 13:12, on Zulip):

Whom are you talking about emailing?

centril (Oct 10 2019 at 13:12, on Zulip):

compiler team + contributors

pnkfelix (Oct 10 2019 at 13:13, on Zulip):

okay, so we're just trying remind people to include T-labels when they beta-nominate to ensure that the relevant teams actually see the nominations? Sure I can do that.

pnkfelix (Oct 10 2019 at 13:13, on Zulip):

/me doesn't know if we have an email list of all contributors, but he'll find out.

nagisa (Oct 10 2019 at 13:14, on Zulip):

Do we actually spend enough effort looking through the full list of no-T nominations and tagging with the team during triage to warrant a mail?

pnkfelix (Oct 10 2019 at 13:14, on Zulip):

probably not

pnkfelix (Oct 10 2019 at 13:18, on Zulip):

(moved forked conversation)

pnkfelix (Oct 10 2019 at 13:19, on Zulip):

okay so 41 minutes left, what P-high issues are there

pnkfelix (Oct 10 2019 at 13:19, on Zulip):

32 open p-high issues

pnkfelix (Oct 10 2019 at 13:20, on Zulip):

we've already discusseed #65147 #65085 and #65084 so I'll skip them

pnkfelix (Oct 10 2019 at 13:20, on Zulip):

P-high: "UI Tests that check ICEing thread names break with parallel-compiler = true" #65047

pnkfelix (Oct 10 2019 at 13:21, on Zulip):

ah I agreed to mentor this issue and then overlooked the efforts on it

pnkfelix (Oct 10 2019 at 13:21, on Zulip):

@cuviper argues it should not be P-high

nagisa (Oct 10 2019 at 13:22, on Zulip):

I… probably agree.

pnkfelix (Oct 10 2019 at 13:22, on Zulip):

but I suspect it remains a high-impact item for @WG-parallel-rustc ... would anyone from that team working group care to comment on that question?

nikomatsakis (Oct 10 2019 at 13:22, on Zulip):

it might be fixed now

cuviper (Oct 10 2019 at 13:22, on Zulip):

Oh, my rustc-rayon pr also landed, which included the thread name

pnkfelix (Oct 10 2019 at 13:22, on Zulip):

heh

nikomatsakis (Oct 10 2019 at 13:22, on Zulip):

(because of that)

pnkfelix (Oct 10 2019 at 13:22, on Zulip):

okay then.

pnkfelix (Oct 10 2019 at 13:23, on Zulip):

I'll close as fixed.

pnkfelix (Oct 10 2019 at 13:23, on Zulip):

P-high: "linking of libtest failed" #64872

pnkfelix (Oct 10 2019 at 13:24, on Zulip):

I failed to look into this because I was having too much fun with const_to_pat code...

pnkfelix (Oct 10 2019 at 13:25, on Zulip):

but progress has been made, of a sort, since bjorn3 has posted updates that include a repro method that does not require rustc patching

pnkfelix (Oct 10 2019 at 13:25, on Zulip):

P-high: "Internal compiler error but only when running under rls" #64821

pnkfelix (Oct 10 2019 at 13:26, on Zulip):

peku33 added a simpler way to reproduce

pnkfelix (Oct 10 2019 at 13:26, on Zulip):

and @Igor Matuszewski self-assigned

pnkfelix (Oct 10 2019 at 13:27, on Zulip):

I'll assume @Igor Matuszewski will post update when they have one

pnkfelix (Oct 10 2019 at 13:27, on Zulip):

P-high: "error: internal compiler error: unexpected panic: inconsistent resolution for a macro" #64803

pnkfelix (Oct 10 2019 at 13:27, on Zulip):

fixed by PR #65140 that s being tested by bors now

pnkfelix (Oct 10 2019 at 13:28, on Zulip):

P-high: "Rustc panics while compiling gstreamer in RLS" #64659

pnkfelix (Oct 10 2019 at 13:28, on Zulip):

seems like multiple people are asking if they can help @Igor Matuszewski on this

pnkfelix (Oct 10 2019 at 13:29, on Zulip):

I'll add self to assignee list so I remember to try to ping them about this

pnkfelix (Oct 10 2019 at 13:29, on Zulip):

P-high: "Drop implementations not being called on panic with Fat LTO." #64655

pnkfelix (Oct 10 2019 at 13:30, on Zulip):

sigh, I need to update that PR #65020

pnkfelix (Oct 10 2019 at 13:30, on Zulip):

(or rather, I need to figure out why my tests are failing when the other similar ones I based them on do not seem to require // ignore-asmjs or whatever...)

pnkfelix (Oct 10 2019 at 13:31, on Zulip):

P-high: "debuginfo/pretty-uninitialized-vec fails with Cannot access memory at address 0x7fffff7fe000" #64343

pnkfelix (Oct 10 2019 at 13:31, on Zulip):

we'll hopefully talk about this at Oct 18th planning meeting.

pnkfelix (Oct 10 2019 at 13:31, on Zulip):

P-high: " ThinLTO and -o compiler option can lead to duplicate object file inclusion in staticlib" #64153

pnkfelix (Oct 10 2019 at 13:31, on Zulip):

@mw is looking into this now

pnkfelix (Oct 10 2019 at 13:32, on Zulip):

P-high: "STATUS_ACCESS_VIOLATION and STATUS_HEAP_CORRUPTION during compilation" #63959

pnkfelix (Oct 10 2019 at 13:33, on Zulip):

hmm. local build for @Mateusz Mikuła does not reproduce the crash, but "can still reproduce on nightly" (which I infer means the nightly produced by our CI ) ...

pnkfelix (Oct 10 2019 at 13:33, on Zulip):

could this somehow be a CI build artifact ?

pnkfelix (Oct 10 2019 at 13:34, on Zulip):

anyway nothing to report there beyond that. Lets see if @Mateusz Mikuła reports anything more over the next week.

pnkfelix (Oct 10 2019 at 13:34, on Zulip):

P-high: "Associated types, impl traits and closures; oh my, an ICE. #63154

pnkfelix (Oct 10 2019 at 13:35, on Zulip):

posted fix in PR #65099 (review ping to @nikomatsakis )

pnkfelix (Oct 10 2019 at 13:35, on Zulip):

P-high: "Bug running cargo check" #63150

pnkfelix (Oct 10 2019 at 13:35, on Zulip):

I posted a note last week pointing out the difficulty on acting on an (incremental) bug like this

pnkfelix (Oct 10 2019 at 13:35, on Zulip):

not sure if there's much we can do here.

pnkfelix (Oct 10 2019 at 13:36, on Zulip):

I'll at least change the bug title. :)

pnkfelix (Oct 10 2019 at 13:38, on Zulip):

the bug filer did say they were willing to try to extract more information, by e.g. running rustc with different arguments. (But they are definitely cagey about exposing info about their internal product that is being built here...)

pnkfelix (Oct 10 2019 at 13:38, on Zulip):

but I'm not sure what I can tell them

pnkfelix (Oct 10 2019 at 13:38, on Zulip):

P-high: "Segfault compiling libc on armv7-unknown-linux-gnueabihf" #62896

pnkfelix (Oct 10 2019 at 13:39, on Zulip):

ah there's been much recent activity here

pnkfelix (Oct 10 2019 at 13:39, on Zulip):

bug filer says the CI images may have something wrong with them?

pnkfelix (Oct 10 2019 at 13:39, on Zulip):

^ @centril is there someone on the release team I should ping about that?

pnkfelix (Oct 10 2019 at 13:40, on Zulip):

or maybe the infrastructure team, rather?

pnkfelix (Oct 10 2019 at 13:40, on Zulip):

P-high: "Undefined symbol _fltused when compiling to x86_64-unknown-uefi" #62785

pnkfelix (Oct 10 2019 at 13:41, on Zulip):

okay it seems like @oli has been trying to address this and keeps having to move their PR up the chain

pnkfelix (Oct 10 2019 at 13:42, on Zulip):

P-high: "ICE: Generic type alias to invalid type crashes during type check on latest stable" #62742

pnkfelix (Oct 10 2019 at 13:42, on Zulip):

niko's been looking at this and has posted a big update to it. :)

pnkfelix (Oct 10 2019 at 13:43, on Zulip):

P-high: "internal compiler error: src/librustc/dep_graph/graph.rs:688: DepNode Hir(...) should have been pre-allocated but wasn't." #62649

pnkfelix (Oct 10 2019 at 13:43, on Zulip):

I reassigned this to myself last week and haven't had a chance to look at it.

pnkfelix (Oct 10 2019 at 13:43, on Zulip):

P-high: "couldn't load codegen backend #61561

pnkfelix (Oct 10 2019 at 13:44, on Zulip):

I'm going to change title to make windows-gnu-specific nature clear

pnkfelix (Oct 10 2019 at 13:45, on Zulip):

though I do find the dialogue there somewhat confusing. The bug filer was on windows .. but then later ubuntu?

pnkfelix (Oct 10 2019 at 13:46, on Zulip):

P-high: " 1.30 -> 1.31 dylib late-binding regression with GNU binutils 2.28 or older." #61539

pnkfelix (Oct 10 2019 at 13:46, on Zulip):

I'd love to know if we should bother attacking this problem.

pnkfelix (Oct 10 2019 at 13:47, on Zulip):

I'll leave it P-high in the hopes that I'll find time for it once I finish with my const_to_pat nonsense.

pnkfelix (Oct 10 2019 at 13:47, on Zulip):

speaking of which ...

nagisa (Oct 10 2019 at 13:47, on Zulip):

I feel that there is nothing really actionable to do with that issue.

pnkfelix (Oct 10 2019 at 13:47, on Zulip):

P-high: "ICE when trying to match on non-PartialEq slice." #61188

pnkfelix (Oct 10 2019 at 13:48, on Zulip):

I feel that there is nothing really actionable to do with that issue.

(I suspect you're right. But I did want to try to explore it a little bit before ... what, closing? or just downgrading? its still a real bug, and worth keeping open just so people can google for it? We might also consider disabling PLT, maybe...)

pnkfelix (Oct 10 2019 at 13:48, on Zulip):

so

pnkfelix (Oct 10 2019 at 13:48, on Zulip):

P-high: "ICE when trying to match on non-PartialEq slice." #61188

pnkfelix (Oct 10 2019 at 13:48, on Zulip):

I've been spending much of the last week on this

pnkfelix (Oct 10 2019 at 13:50, on Zulip):

I've got a branch in progress that addresses it.

pnkfelix (Oct 10 2019 at 13:50, on Zulip):

P-high: "Compiler panic at Box<Any>" #60363

pnkfelix (Oct 10 2019 at 13:50, on Zulip):

haven't looked at this yet

pnkfelix (Oct 10 2019 at 13:51, on Zulip):

needs a better title (but probably needs a minimal test case first?)

centril (Oct 10 2019 at 13:51, on Zulip):

renamed the issue title

pnkfelix (Oct 10 2019 at 13:51, on Zulip):

thanks

pnkfelix (Oct 10 2019 at 13:51, on Zulip):

P-high: "Incremental compilation results in linker error when method use is removed" #59535

pnkfelix (Oct 10 2019 at 13:51, on Zulip):

oh wow this totally fell off my radar

pnkfelix (Oct 10 2019 at 13:52, on Zulip):

like, the other issues I say "didn't get time for this", but I don't even remember seeing this on my assigned bugs list ...

pnkfelix (Oct 10 2019 at 13:52, on Zulip):

hopefully I'll get to it soonish

pnkfelix (Oct 10 2019 at 13:52, on Zulip):

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

centril (Oct 10 2019 at 13:52, on Zulip):

oh wow this totally fell off my radar

you're probably assigned to too much?

pnkfelix (Oct 10 2019 at 13:53, on Zulip):

oh wow this totally fell off my radar

you're probably assigned to too much?

that would be a reasonable inference

pnkfelix (Oct 10 2019 at 13:53, on Zulip):

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

pnkfelix (Oct 10 2019 at 13:53, on Zulip):

I'm going to downgrade this to P-medium

pnkfelix (Oct 10 2019 at 13:53, on Zulip):

it doesn't warrant revisiting every week

pnkfelix (Oct 10 2019 at 13:54, on Zulip):

P-high: "Better error message when attempting to instantiate tuple structs with private fields" #58017

pnkfelix (Oct 10 2019 at 13:54, on Zulip):

hey, look at that, has a proposed PR in #65153

pnkfelix (Oct 10 2019 at 13:55, on Zulip):

thanks to da-x, and also @Vadim Petrochenkov for some excellent reviewing+mentoring+work-stealing

pnkfelix (Oct 10 2019 at 13:56, on Zulip):

P-high: "Coherence can be bypassed by an indirect impl for a trait object #57893

centril (Oct 10 2019 at 13:56, on Zulip):

I guess we should all be a bit more like rayon ;)

pnkfelix (Oct 10 2019 at 13:56, on Zulip):

I've heard work is ongoing here

pnkfelix (Oct 10 2019 at 13:57, on Zulip):

@nikomatsakis is that true? (regarding #57893, that is)

pnkfelix (Oct 10 2019 at 13:57, on Zulip):

P-high: " ICE from const item in lifetime-parametric impl" #56445

centril (Oct 10 2019 at 13:57, on Zulip):

yep, https://rust-lang.zulipchat.com/#narrow/stream/144729-wg-traits/topic/coherence.20bypasses.20for.20traits.20.2357893

pnkfelix (Oct 10 2019 at 13:58, on Zulip):

not sure what to do about #56445

pnkfelix (Oct 10 2019 at 13:58, on Zulip):

probably needs to stay P-high

pnkfelix (Oct 10 2019 at 13:59, on Zulip):

P-high: " ICE in macro: doc meta with expr on an item, string concat, stringify!(...)" #55414

pnkfelix (Oct 10 2019 at 13:59, on Zulip):

hmm dormant since Aug 30th

pnkfelix (Oct 10 2019 at 13:59, on Zulip):

mismatch between Visitor and MutVisitor definitely sounds goofy there

pnkfelix (Oct 10 2019 at 14:00, on Zulip):

@Vadim Petrochenkov do you want the above (#55414) reassigned?

pnkfelix (Oct 10 2019 at 14:00, on Zulip):

ah, so close! four more P-high bugs!

pnkfelix (Oct 10 2019 at 14:00, on Zulip):

oh well.

Vadim Petrochenkov (Oct 10 2019 at 14:01, on Zulip):

do you want the above (#55414) reassigned?

We've been discussing possible ways to proceed with that issue with @Caio .
I wouldn't say it's P-high, it's not a regression or anything.

Vadim Petrochenkov (Oct 10 2019 at 14:04, on Zulip):

It's assigned to me as a mentoring issue right now.

pnkfelix (Oct 10 2019 at 14:04, on Zulip):

okay thanks!

DPC (Oct 10 2019 at 14:04, on Zulip):

should we mark it as e-mentor?

Mateusz Mikuła (Oct 14 2019 at 11:09, on Zulip):

hmm. local build for Mateusz Mikuła does not reproduce the crash, but "can still reproduce on nightly" (which I infer means the nightly produced by our CI ) ...

I guess it has something to do with one of configuration options (maybe debuginfo?), I can assign some time to do one more build (if somebody guides me what to change) but I don't have time to investigate it myself.

Last update: Nov 22 2019 at 05:50UTC