Stream: t-compiler

Topic: pre-meeting triage 2020-01-23 #54818


pnkfelix (Jan 23 2020 at 13:27, on Zulip):

I will be doing pre-triage in this channel.

pnkfelix (Jan 23 2020 at 13:30, on Zulip):

there are six nominated unprioritized issues

pnkfelix (Jan 23 2020 at 13:30, on Zulip):

nom unpri 1/6: "ICE cause by tokio::spawn a async function return future opaque type" #68477

pnkfelix (Jan 23 2020 at 13:32, on Zulip):

#68477: @Aaron Hill says it should be fixed by PR #67717, which is blocked, I think, on perf results?

centril (Jan 23 2020 at 13:33, on Zulip):

and some Miri stuff?

pnkfelix (Jan 23 2020 at 13:33, on Zulip):

Lets go with P-high here. Assigning to @Aaron Hill

pnkfelix (Jan 23 2020 at 13:34, on Zulip):

and some Miri stuff?

maybe, though I don't know if that work needs to block this PR, vs being follow-on effort.

centril (Jan 23 2020 at 13:35, on Zulip):

just going by comments :slight_smile:

pnkfelix (Jan 23 2020 at 13:35, on Zulip):

(hmm was assigning to @Aaron Hill premature? I just realized that the author of PR #67717 is a different contributor.)

centril (Jan 23 2020 at 13:35, on Zulip):

assign to both maybe?

pnkfelix (Jan 23 2020 at 13:35, on Zulip):

well at the very least, @Aaron Hill can act as a proponent for pushing PR #67717 forward.

centril (Jan 23 2020 at 13:36, on Zulip):

nom unpri 2/6: " ICE: 'rustc' panicked at 'Box<Any>', src/librustc_errors/lib.rs:931:9" #68372

pnkfelix (Jan 23 2020 at 13:36, on Zulip):

/me is always nervous about assigning bugs to multiple people, since responsibility becomes ambiguous at thZT point

pnkfelix (Jan 23 2020 at 13:37, on Zulip):

ugh #68372 looks yet another incremental bug

centril (Jan 23 2020 at 13:37, on Zulip):

Lovely, added the label

pnkfelix (Jan 23 2020 at 13:37, on Zulip):

/me again muses about making incremental compilation take snapshot of source code

pnkfelix (Jan 23 2020 at 13:38, on Zulip):

(on successful compiles)

centril (Jan 23 2020 at 13:38, on Zulip):

unlikely to be fixed with just that backtrace and no mcve

centril (Jan 23 2020 at 13:38, on Zulip):

p-medium?

centril (Jan 23 2020 at 13:39, on Zulip):

here's an idea: grep the backtrace for mentions of relevant methods to incremental, and suggest cargo clean if it occurs

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

actually

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

this might be fixed

centril (Jan 23 2020 at 13:40, on Zulip):

You may want to try cargo clean

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

by a PR I filed recently

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

let me check

mw (Jan 23 2020 at 13:40, on Zulip):

#68372 might be a duplicate of #62649

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

because there's an assert that's, AFAICT, just wrong

pnkfelix (Jan 23 2020 at 13:40, on Zulip):

PR #68289 I think

pnkfelix (Jan 23 2020 at 13:41, on Zulip):

well, maybe; that message occurs at different places

pnkfelix (Jan 23 2020 at 13:41, on Zulip):

but yeah, PR #68289 is removing that assertion.

pnkfelix (Jan 23 2020 at 13:42, on Zulip):

here's an idea: grep the backtrace for mentions of relevant methods to incremental, and suggest cargo clean if it occurs

but this just papers over incremental bugs

pnkfelix (Jan 23 2020 at 13:42, on Zulip):

there's an adage about this: "Fault-tolerant systems aren't", because faults that are tolerated don't get fixed.

pnkfelix (Jan 23 2020 at 13:43, on Zulip):

still, suggesting cargo clean might at least help people get over initial roadblock

centril (Jan 23 2020 at 13:43, on Zulip):

seems like a reasonable concern

pnkfelix (Jan 23 2020 at 13:43, on Zulip):

as long as we still suggest that they file a bug, it seems reasonable.

centril (Jan 23 2020 at 13:43, on Zulip):

yeah I only meant it as a "note"

mw (Jan 23 2020 at 13:43, on Zulip):

how many other such bugs are there?

centril (Jan 23 2020 at 13:43, on Zulip):

at the start/end of the ICE message

pnkfelix (Jan 23 2020 at 13:44, on Zulip):

Lets go P-high for #68372. Removing nomination.

pnkfelix (Jan 23 2020 at 13:44, on Zulip):

(and assigning to self since its my PR that should fix it.)

centril (Jan 23 2020 at 13:44, on Zulip):

nom unpri 3/6: Exponential compile time for the amount of impls. #68324

pnkfelix (Jan 23 2020 at 13:45, on Zulip):

how many other such bugs are there?

(good question but not for right now)

pnkfelix (Jan 23 2020 at 13:45, on Zulip):

nom unpri 3/6: Exponential compile time for the amount of impls. #68324

seems like this needs bisection

pnkfelix (Jan 23 2020 at 13:45, on Zulip):

though we are not even sure its a regression?

centril (Jan 23 2020 at 13:46, on Zulip):

yeah, but oh hey cool, this is for https://www.luna-lang.org/

centril (Jan 23 2020 at 13:46, on Zulip):

seems we are not sure

pnkfelix (Jan 23 2020 at 13:46, on Zulip):

Also not sure if its P-high or P-medium...

pnkfelix (Jan 23 2020 at 13:46, on Zulip):

Perhaps P-high to determine if its a regression or not

pnkfelix (Jan 23 2020 at 13:46, on Zulip):

and then we revisit after determining that.

centril (Jan 23 2020 at 13:46, on Zulip):

I suspect medium

centril (Jan 23 2020 at 13:47, on Zulip):

sgtm though

pnkfelix (Jan 23 2020 at 13:47, on Zulip):

(removing nomination label too)

pnkfelix (Jan 23 2020 at 13:48, on Zulip):

((unless someone wants to argue that we should discuss at meeting))

centril (Jan 23 2020 at 13:48, on Zulip):

nom unpri 4/6: -- ICE in librustc::traits::select::confirm_param_candidate A-traits C-bug E-needs-mcve I-ICE I-nominated T-compiler #68295 opened 7 days ago by aweinstock314

pnkfelix (Jan 23 2020 at 13:50, on Zulip):

potential soundness issues in our trait system caching

pnkfelix (Jan 23 2020 at 13:50, on Zulip):

sounds P-high to me

centril (Jan 23 2020 at 13:50, on Zulip):

..ouch, and yes, P-high

centril (Jan 23 2020 at 13:51, on Zulip):

cced wg-traits

pnkfelix (Jan 23 2020 at 13:51, on Zulip):

nom unpri 5/6: "[spurious] thread 'rustc' panicked at 'slice index starts at 24722962 but ends at 13279232', src/libcore/slice/mod.rs:2680:5" #68132

pnkfelix (Jan 23 2020 at 13:51, on Zulip):

hmm I guess I didn't unnominate this last week

pnkfelix (Jan 23 2020 at 13:51, on Zulip):

nor prioritize it

pnkfelix (Jan 23 2020 at 13:52, on Zulip):

we don't even have an MCVE, right? We know the input crate is hashbrown from the error diagnostic

pnkfelix (Jan 23 2020 at 13:53, on Zulip):

I'll add it to an explicit list for discussion today

centril (Jan 23 2020 at 13:53, on Zulip):

https://github.com/rust-lang/rust/pull/66364 could possibly be relevant

centril (Jan 23 2020 at 13:54, on Zulip):

though probably not

pnkfelix (Jan 23 2020 at 13:54, on Zulip):

hackmd for today: https://hackmd.io/YBpyTmQ7S_uPhScJWDwBag?both

centril (Jan 23 2020 at 13:55, on Zulip):

last ICE: Another macro ICE #68060

pnkfelix (Jan 23 2020 at 13:56, on Zulip):

heh

pnkfelix (Jan 23 2020 at 13:56, on Zulip):

this one might be telling us how to fix the bug

centril (Jan 23 2020 at 13:56, on Zulip):

(I'll reduce the input a bit while we're triaging)

pnkfelix (Jan 23 2020 at 13:57, on Zulip):

anyway lets call this P-high

centril (Jan 23 2020 at 13:58, on Zulip):

sgtm

pnkfelix (Jan 23 2020 at 13:59, on Zulip):

there are zero unprioritrized beta-regressions

centril (Jan 23 2020 at 13:59, on Zulip):

what; impossible! :D

pnkfelix (Jan 23 2020 at 13:59, on Zulip):

there are two unprioritized nightly regressions

pnkfelix (Jan 23 2020 at 14:00, on Zulip):

unpri regr: "ICE when using a reference incorrectly in quick_error" #68449

pnkfelix (Jan 23 2020 at 14:01, on Zulip):

P-high. Assigning to @Esteban Küber

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

(It would be good to confirm it is indeed due to the CC'ed PR via bisection...)

pnkfelix (Jan 23 2020 at 14:03, on Zulip):

unpri regr: "'Conflicting representations' warning became an error" #68428

pnkfelix (Jan 23 2020 at 14:05, on Zulip):

Lets nominate this for discussion at the meeting. I want to get our policy here straight.

centril (Jan 23 2020 at 14:05, on Zulip):

Re. the specific issue, I don't have a problem with turning this back into a warning (as I was originally going to make this into a silenceable lint)

centril (Jan 23 2020 at 14:05, on Zulip):

(but I do mind the assertions re. "you didn't follow the process")

pnkfelix (Jan 23 2020 at 14:06, on Zulip):

I don't want to take the time to review the thread right now

pnkfelix (Jan 23 2020 at 14:06, on Zulip):

(though obviously I'll need to do so before the meeting)

pnkfelix (Jan 23 2020 at 14:07, on Zulip):

anyway, @centril , can I task you with making a PR to turn this back into a warning? (even a deny-by-default one)?

centril (Jan 23 2020 at 14:07, on Zulip):

sure

centril (Jan 23 2020 at 14:08, on Zulip):

(@pnkfelix btw, it wasn't a lint originally, it was a hardcoded warning)

centril (Jan 23 2020 at 14:08, on Zulip):

(re. "back into")

pnkfelix (Jan 23 2020 at 14:10, on Zulip):

okay I revised my comment

pnkfelix (Jan 23 2020 at 14:11, on Zulip):

so next up is my review of the nominated issues

pnkfelix (Jan 23 2020 at 14:12, on Zulip):

there are, :sad: , fifteen nominations

pnkfelix (Jan 23 2020 at 14:12, on Zulip):

so lets see

pnkfelix (Jan 23 2020 at 14:12, on Zulip):

nom: "Rust 1.26.0 fails to build via homebrew on OS X 10.10.5" #50776

pnkfelix (Jan 23 2020 at 14:13, on Zulip):

I nominated this 9 days ago to discuss closing

pnkfelix (Jan 23 2020 at 14:13, on Zulip):

ah that was for the release team. Sounds like @simulacrum says "close wontfix"

pnkfelix (Jan 23 2020 at 14:13, on Zulip):

lets go for it. Or rather, lets stop.

centril (Jan 23 2020 at 14:13, on Zulip):

I think the release team collectively thinks so as well

centril (Jan 23 2020 at 14:14, on Zulip):

at least from the discussion in https://github.com/rust-lang/release-team/issues/2

simulacrum (Jan 23 2020 at 14:14, on Zulip):

There is separate discussion happening on some broader questions raised within the release team, too

simulacrum (Jan 23 2020 at 14:14, on Zulip):

I'm not sure we should close quite yet

centril (Jan 23 2020 at 14:14, on Zulip):

ah, maybe not

pnkfelix (Jan 23 2020 at 14:14, on Zulip):

@simulacrum can you open fresh bug regarding broader Q's ?

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

the one that's there seems somewhat narrow in its scope

simulacrum (Jan 23 2020 at 14:15, on Zulip):

(see release team repo issue)

centril (Jan 23 2020 at 14:15, on Zulip):

@simulacrum should we continue the broader Q in our team issue?

centril (Jan 23 2020 at 14:15, on Zulip):

and leave the other one closed?

simulacrum (Jan 23 2020 at 14:15, on Zulip):

Yeah - but I think we've not settled on this one

centril (Jan 23 2020 at 14:15, on Zulip):

I see; @pnkfelix let's reopen?

pnkfelix (Jan 23 2020 at 14:16, on Zulip):

okay I'll reopen

simulacrum (Jan 23 2020 at 14:16, on Zulip):

i.e. some don't share my opinion of closing

simulacrum (Jan 23 2020 at 14:16, on Zulip):

@Pietro Albini in particular I think

pnkfelix (Jan 23 2020 at 14:17, on Zulip):

I'm guessing I'd have to read https://github.com/rust-lang/release-team/issues/2 to find out what particular details lead them to want to keep #50776 open

pnkfelix (Jan 23 2020 at 14:17, on Zulip):

anyway I'm removing nomination tag from it

centril (Jan 23 2020 at 14:17, on Zulip):

Next up: Add GitHub issue templates
C-enhancement I-nominated S-waiting-on-review T-cargo T-compiler T-release
#68413 opened 2 days ago by XAMPPRocky

Pietro Albini (Jan 23 2020 at 14:17, on Zulip):

my opinion is that I'm fine with closing this, as long as the project decides on a support policy that excludes macOS 10.10

Pietro Albini (Jan 23 2020 at 14:18, on Zulip):

because at the moment this is a version that we're supporting as tier 1

pnkfelix (Jan 23 2020 at 14:18, on Zulip):

Next up: Add GitHub issue templates
C-enhancement I-nominated S-waiting-on-review T-cargo T-compiler T-release
#68413 opened 2 days ago by XAMPPRocky

oh I was going bottom up in this case

centril (Jan 23 2020 at 14:18, on Zulip):

oh :slight_smile:

Pietro Albini (Jan 23 2020 at 14:18, on Zulip):

https://forge.rust-lang.org/release/platform-support.html
"x86_64-apple-darwin ✓ ✓ ✓ 64-bit OSX (10.7+, Lion+)"

pnkfelix (Jan 23 2020 at 14:18, on Zulip):

my opinion is that I'm fine with closing this, as long as the project decides on a support policy that excludes macOS 10.10

(but its not just that, its also the homebrew recipe aspect, right?)

pnkfelix (Jan 23 2020 at 14:19, on Zulip):

I'm going bottom-up in the hopes of being more likely to find stuff quickly that we can unnominate or close.

pnkfelix (Jan 23 2020 at 14:19, on Zulip):

nom: "[nll] _ patterns should not count as borrows" #53114

centril (Jan 23 2020 at 14:20, on Zulip):

So I nominated this due to @pnkfelix's question on the slice patterns stabilization

centril (Jan 23 2020 at 14:20, on Zulip):

feels like it has found new relevance in being fixed?

centril (Jan 23 2020 at 14:20, on Zulip):

but we also have a lot of other stuff to contend with, and this is strictly about allowing more code

centril (Jan 23 2020 at 14:21, on Zulip):

..but this is also not a polonius issue

pnkfelix (Jan 23 2020 at 14:21, on Zulip):

funnily enough #53114 itself points out that let _ = ... was untested: https://github.com/rust-lang/rust/issues/53114#issuecomment-450747337

centril (Jan 23 2020 at 14:22, on Zulip):

I added some new tests for match with _ and slice pats btw

pnkfelix (Jan 23 2020 at 14:22, on Zulip):

@centril you've probably looked at the set of tests most recently. Do you have any idea whether we are coverinvg the rows in this table?

centril (Jan 23 2020 at 14:22, on Zulip):

but those document the current incorrect behavior

pnkfelix (Jan 23 2020 at 14:23, on Zulip):

(the table in the description, that is)

pnkfelix (Jan 23 2020 at 14:23, on Zulip):

but those document the current incorrect behavior

sure, that's better than nothing, IMO

centril (Jan 23 2020 at 14:23, on Zulip):

@pnkfelix sorry, didn't look at the table

centril (Jan 23 2020 at 14:23, on Zulip):

was pretty strictly focused on my narrow slice ;)

pnkfelix (Jan 23 2020 at 14:23, on Zulip):

sure, I understand that

pnkfelix (Jan 23 2020 at 14:24, on Zulip):

this testing stuff all stems from a conversation from literally over a year ago

centril (Jan 23 2020 at 14:24, on Zulip):

re. priority, should we raise to p-high or keep p-medium?

centril (Jan 23 2020 at 14:24, on Zulip):

(I don't personally feel competent to fix this atm)

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

well, there's the testing, and there's the "fixing"

centril (Jan 23 2020 at 14:25, on Zulip):

yea, I was thinking about the behavior change for match

pnkfelix (Jan 23 2020 at 14:26, on Zulip):

At this point, I think we should put P-high priority on ensuring the tests are present (or add them); its embarrassing that sat so long.

centril (Jan 23 2020 at 14:26, on Zulip):

sounds reasonable

pnkfelix (Jan 23 2020 at 14:26, on Zulip):

as for the behavior change.... I probably figure that's still P-medium.

pnkfelix (Jan 23 2020 at 14:27, on Zulip):

marking #53114 P-high, removing nomination, and assigning to self.

pnkfelix (Jan 23 2020 at 14:28, on Zulip):

and @centril

centril (Jan 23 2020 at 14:28, on Zulip):

(we probably want to avoid closing the issue when adding tests)

pnkfelix (Jan 23 2020 at 14:28, on Zulip):

Unless you'd prefer I take care of the tests?

centril (Jan 23 2020 at 14:28, on Zulip):

I'm not sure when I'll get to it but I can be assigned

pnkfelix (Jan 23 2020 at 14:29, on Zulip):

you and I can race for it. If I get to it first, I'll unassign you and ping you.

centril (Jan 23 2020 at 14:29, on Zulip):

sgtm :slight_smile:

pnkfelix (Jan 23 2020 at 14:30, on Zulip):

nom: "./x.py check failed if incremental builds enabled" #58633

pnkfelix (Jan 23 2020 at 14:31, on Zulip):

maybe I should have @mw spell this out to me more

centril (Jan 23 2020 at 14:32, on Zulip):

there was discussion re. removing unused_attributes wholesale

centril (Jan 23 2020 at 14:32, on Zulip):

let's maybe nominate that for t-lang?

pnkfelix (Jan 23 2020 at 14:32, on Zulip):

where was that?

centril (Jan 23 2020 at 14:32, on Zulip):

@pnkfelix on some t-compiler triage; I think you raised it

centril (Jan 23 2020 at 14:32, on Zulip):

as a suggestion

pnkfelix (Jan 23 2020 at 14:33, on Zulip):

I'm going to remove nomination from #58633.

centril (Jan 23 2020 at 14:34, on Zulip):

but let's add t-lang maybe? or do we need a separate issue for removal of the lint?

pnkfelix (Jan 23 2020 at 14:34, on Zulip):

I'd say its a separate issue

pnkfelix (Jan 23 2020 at 14:35, on Zulip):

nom: "COPYRIGHT file is wildly out of date" #63232

pnkfelix (Jan 23 2020 at 14:36, on Zulip):

while #63232 is still high priority, I don't think further discussion of it at the triage meeting is necessary nor useful

pnkfelix (Jan 23 2020 at 14:38, on Zulip):

but oh, maybe I need to keep the nominated label because its tagged with T-core as well.

pnkfelix (Jan 23 2020 at 14:38, on Zulip):

Maybe I remove T-compiler instead then...

pnkfelix (Jan 23 2020 at 14:38, on Zulip):

(and really, its not a compiler thing. its a project thing.)

pnkfelix (Jan 23 2020 at 14:39, on Zulip):

nom: "Pin is unsound due to rogue Deref/DerefMut implementations" #66544

pnkfelix (Jan 23 2020 at 14:40, on Zulip):

I think work here is ongoing in PR #68004

centril (Jan 23 2020 at 14:40, on Zulip):

waiting on lang team stuff

pnkfelix (Jan 23 2020 at 14:40, on Zulip):

okay. I'm going to switch gears now

pnkfelix (Jan 23 2020 at 14:41, on Zulip):

because we are 20 minutes until the meeting starts and I want to go through the rest of the agenda items (e.g. making the list of beta backports)

pnkfelix (Jan 23 2020 at 14:45, on Zulip):

(and also I need to eat something or I will turn into an ogre)

Last update: Jun 07 2020 at 09:50UTC