Stream: t-compiler

Topic: pre-meeting triage 2019-09-26 #54818


pnkfelix (Sep 26 2019 at 11:28, on Zulip):

I will be doing pre-triage in this channel.

pnkfelix (Sep 26 2019 at 11:33, on Zulip):

We have 10 unprioritized nominated T-compiler issues

pnkfelix (Sep 26 2019 at 11:34, on Zulip):

unpri nom: "clippy-driver no longer builds after rust-lang/rust#64513" #64808

pnkfelix (Sep 26 2019 at 11:35, on Zulip):

has PR in #64810

pnkfelix (Sep 26 2019 at 11:35, on Zulip):

going to tag as P-medium just to move along

pnkfelix (Sep 26 2019 at 11:37, on Zulip):

unpri nom: "error: internal compiler error: unexpected panic: inconsistent resolution for a macro" #64803

pnkfelix (Sep 26 2019 at 11:39, on Zulip):

going to tag as P-high for initial investigation as to severity. Doesn't seem like it warrants discussion at T-compiler meeting. Removing nomination tag.

pnkfelix (Sep 26 2019 at 11:40, on Zulip):

unpri nom: "miri no longer builds after rust-lang/rust#64785" #64798

pnkfelix (Sep 26 2019 at 11:40, on Zulip):

triage: P-medium. Removing nomination label.

pnkfelix (Sep 26 2019 at 11:41, on Zulip):

unpri nom: "ICE when combining wrong syntax with Unicode string literal #64792

pnkfelix (Sep 26 2019 at 11:42, on Zulip):

@centril investigated. Apparently does not repro on beta or nightly.

pnkfelix (Sep 26 2019 at 11:43, on Zulip):

In an ideal world, I'd like to bisect to the point where it was fixed.

pnkfelix (Sep 26 2019 at 11:44, on Zulip):

presumably its fixed in the release scheduled for today

pnkfelix (Sep 26 2019 at 11:44, on Zulip):

(and thus the value of such a bisection is slightly reduced, since its not like that would identify a commit that we would then backport.)

pnkfelix (Sep 26 2019 at 11:45, on Zulip):

triage: P-medium. Would be good to identify point where it was fixed by bisection, and maybe add a test if one doesn't already exist. Removing nomination.

pnkfelix (Sep 26 2019 at 11:46, on Zulip):

unpri nom: "ICE when using target_feature with a const" #64768

pnkfelix (Sep 26 2019 at 11:47, on Zulip):

has PR (thanks @davidtwco !)

pnkfelix (Sep 26 2019 at 11:48, on Zulip):

triage: P-high. Assigning to @davidtwco . Removing nomination label.

pnkfelix (Sep 26 2019 at 11:49, on Zulip):

unpri nom: "proc-macro param attrs dropping first attrs in impl fns" #64682

pnkfelix (Sep 26 2019 at 11:51, on Zulip):

seems pretty bad

pnkfelix (Sep 26 2019 at 11:51, on Zulip):

triage: P-high. Leaving nominated in hopes to find someone to look into it, since @Caio won't have time for next few days.

pnkfelix (Sep 26 2019 at 11:56, on Zulip):

unpri nom: "Rustc panics while compiling gstreamer in RLS" #64659

pnkfelix (Sep 26 2019 at 11:57, on Zulip):

I'm torn about whether to prioritize this as P-high.

pnkfelix (Sep 26 2019 at 11:58, on Zulip):

I'm going to just leave nominated and hopefully we'll discuss at the meeting.

pnkfelix (Sep 26 2019 at 11:59, on Zulip):

unpri nom: "Drop implementations not being called on panic with Fat LTO." #64655

pnkfelix (Sep 26 2019 at 12:04, on Zulip):

Lets call this P-high for now. (As we all know, you're not supposed to rely on destructors always being called. But I'd like to know why they're not being called in this scenario.) I'll assign it to myself.

pnkfelix (Sep 26 2019 at 12:07, on Zulip):

unpri nom: "ICE with return outside of function body" #64638

pnkfelix (Sep 26 2019 at 12:09, on Zulip):

triage: P-medium. has fix in PR #64649 (thanks @Esteban Küber !). Assigning to latter and removing nomination label.

pnkfelix (Sep 26 2019 at 12:10, on Zulip):

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

pnkfelix (Sep 26 2019 at 12:15, on Zulip):

again I'm not sure how to prioritize this

pnkfelix (Sep 26 2019 at 12:15, on Zulip):

I'm tempted to do P-medium, but I'll just leave it nominated and we'll discuss at the meeting.

pnkfelix (Sep 26 2019 at 12:19, on Zulip):

so that's all 10

pnkfelix (Sep 26 2019 at 12:23, on Zulip):

there are no unprioritized beta-regressions

pnkfelix (Sep 26 2019 at 12:23, on Zulip):

there are no unprioritized nightly-regressions

pnkfelix (Sep 26 2019 at 12:24, on Zulip):

skimming over all the nominations now to check if any of those labels can be removed

pnkfelix (Sep 26 2019 at 12:25, on Zulip):

only possibility is maybe we discussed #64426 ? Need to go check.

pnkfelix (Sep 26 2019 at 12:26, on Zulip):

okay, yes, #64426 was discussed

pnkfelix (Sep 26 2019 at 12:33, on Zulip):

there are two beta-nominations; we'll talk about them at the meeting.

pnkfelix (Sep 26 2019 at 12:35, on Zulip):

skimming over all nominated PRs, found one that was missing a T-label: #63416

pnkfelix (Sep 26 2019 at 12:35, on Zulip):

tagged it T-compiler; we'll catch it in the meeting

pnkfelix (Sep 26 2019 at 12:36, on Zulip):

(personally I'm with eddyb on this matter, given the nature of the patch. But we'll discuss at the meeting.)

pnkfelix (Sep 26 2019 at 12:45, on Zulip):

so I think "all" that is left is iterating through the P-high bugs

pnkfelix (Sep 26 2019 at 12:45, on Zulip):

we have 33 open P-high bugs; of those, 10 are unassigned

pnkfelix (Sep 26 2019 at 12:46, on Zulip):

lets go through the P-high's.

pnkfelix (Sep 26 2019 at 12:46, on Zulip):

P-high 1/33: "error: internal compiler error: unexpected panic: inconsistent resolution for a macro" #64803

pnkfelix (Sep 26 2019 at 12:47, on Zulip):

needs assignee. I'll point it out in the main meeting channel.

pnkfelix (Sep 26 2019 at 12:48, on Zulip):

P-high 2/33: "ICE when using target_feature with a const" #64768

pnkfelix (Sep 26 2019 at 12:48, on Zulip):

has PR. (Is there github metadata that I could use to keep these out of the search?)

pnkfelix (Sep 26 2019 at 12:51, on Zulip):

P-high 3/33: "proc-macro param attrs dropping first attrs in impl fns" #64682

pnkfelix (Sep 26 2019 at 12:51, on Zulip):

needs assignee. Already pointed out in main meeting channel.

pnkfelix (Sep 26 2019 at 12:51, on Zulip):

P-high 4/33: " Drop implementations not being called on panic with Fat LTO." #64655

pnkfelix (Sep 26 2019 at 12:51, on Zulip):

assigned to self already. nothing to report yet :wink:

pnkfelix (Sep 26 2019 at 12:52, on Zulip):

P-high 5/33: "ThinLTO and -o compiler option can lead to duplicate object file inclusion in staticlib" #64153

pnkfelix (Sep 26 2019 at 12:52, on Zulip):

@mw Do you think you'll have time to look into #64153? Or just mentor someone else?

pnkfelix (Sep 26 2019 at 12:53, on Zulip):

P-high 6/33: "redundant_semicolon and clippy::no_effect tripped without emitting line numbers" #63967

pnkfelix (Sep 26 2019 at 12:54, on Zulip):

has PR (thanks nathan whitaker!)

pnkfelix (Sep 26 2019 at 12:54, on Zulip):

but, ah, apparently it failed a rollup. I'll make a note in #63967.

pnkfelix (Sep 26 2019 at 12:57, on Zulip):

P-high 7/33: "STATUS_ACCESS_VIOLATION and STATUS_HEAP_CORRUPTION during compilation" #63959

pnkfelix (Sep 26 2019 at 12:59, on Zulip):

hard to evaluate status based on comment thread

pnkfelix (Sep 26 2019 at 13:03, on Zulip):

left comment

pnkfelix (Sep 26 2019 at 13:03, on Zulip):

P-high 8/33: "Associated types, impl traits and closures; oh my, an ICE." #63154

pnkfelix (Sep 26 2019 at 13:05, on Zulip):

assigned to self last week but haven't have chance to look at it yet

pnkfelix (Sep 26 2019 at 13:06, on Zulip):

P-high 9/33: "Bug running cargo check" #63150

pnkfelix (Sep 26 2019 at 13:07, on Zulip):

@mw has noted that #63150 and #62649 are probably duplicates of each other.

pnkfelix (Sep 26 2019 at 13:07, on Zulip):

but the latter has potentially more interesting/informative discussion on its comment thread. Maybe.

mw (Sep 26 2019 at 13:08, on Zulip):

Regarding #64153, I'll try to make time to take a closer look next week. If somebody else wants to take a look before that, don't be shy :)

pnkfelix (Sep 26 2019 at 13:08, on Zulip):

@mw can I assign #63150 and #62649 to you (to either investigate or delegate to WG-incr-comp ?)

pnkfelix (Sep 26 2019 at 13:09, on Zulip):

(That was just coincidence that I happened to post that message right after you chimed in regarding #64153 ...)

pnkfelix (Sep 26 2019 at 13:09, on Zulip):

(... I don't want to scare people from speaking up for fear that they'll get bugs assigned to them...)

mw (Sep 26 2019 at 13:09, on Zulip):

It's p-high, isn't it :/

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

should it not be P-high ?

mw (Sep 26 2019 at 13:10, on Zulip):

It probably should be, yes. It just looks like debugging it could be quite the time sink

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

yeah I concur

nagisa (Sep 26 2019 at 13:11, on Zulip):

should assign somebody else for ICE: Rust spins when referencing associated types in where clause #62430 and regression: rustdoc resolution changes #61560

pnkfelix (Sep 26 2019 at 13:11, on Zulip):

(we can leave #63150 unassigned for now, and revisit next week, since it sounds like @mw does not have time to look at it in short term anyway)

pnkfelix (Sep 26 2019 at 13:11, on Zulip):

@nagisa okay thanks for letting me know

mw (Sep 26 2019 at 13:12, on Zulip):

@pnkfelix yes, it's unlikely that I'd get to it before next week's meeting

pnkfelix (Sep 26 2019 at 13:14, on Zulip):

reassigned #62430 and #61560 to self. (And prioritized #62430 as P-medium.)

pnkfelix (Sep 26 2019 at 13:16, on Zulip):

P-high 10/33: " Segfault compiling libc on armv7-unknown-linux-gnueabihf" #62896

pnkfelix (Sep 26 2019 at 13:16, on Zulip):

ah there's been activity here in comment thread of #62896 over past week

pnkfelix (Sep 26 2019 at 13:17, on Zulip):

Interesting detail: "Looking at the distribution configuration, rustc is compiled against a very old glibc version (2.16) which may have bugs affecting LLVM 9."

pnkfelix (Sep 26 2019 at 13:18, on Zulip):

Current theory is that #62896 is injected by ThinLTO

nagisa (Sep 26 2019 at 13:19, on Zulip):

We know that old versions of libc are buggy

nagisa (Sep 26 2019 at 13:19, on Zulip):

e.g. there is that one where dynamic linking does not work right…

nagisa (Sep 26 2019 at 13:19, on Zulip):

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

pnkfelix (Sep 26 2019 at 13:44, on Zulip):

well anyway not much we can do about #62896 at this point. Just going to leak to stable I think.

pnkfelix (Sep 26 2019 at 13:45, on Zulip):

P-high 11/33: "Undefined symbol _fltused when compiling to x86_64-unknown-uefi" #62785

pnkfelix (Sep 26 2019 at 13:45, on Zulip):

there's been activity in past two weeks.

pnkfelix (Sep 26 2019 at 13:46, on Zulip):

@Josh Triplett do you think you might be able to take point on this issue (#62785)? I'm not sure if @oli has bandwidth for it.

pnkfelix (Sep 26 2019 at 13:46, on Zulip):

P-high 12/33: "ICE: Generic type alias to invalid type crashes during type check on latest stable" #62742

pnkfelix (Sep 26 2019 at 13:47, on Zulip):

posted note there for @nikomatsakis (to whom #62742 is assigned)

pnkfelix (Sep 26 2019 at 13:48, on Zulip):

P-high 13/33: "internal compiler error: src/librustc/dep_graph/graph.rs:688: DepNode Hir(...) should have been pre-allocated but wasn't. #62649

pnkfelix (Sep 26 2019 at 13:48, on Zulip):

this was mentioned earlier in our discussion of #63150

pnkfelix (Sep 26 2019 at 13:49, on Zulip):

P-high 14/33: "Forgone caching in cycles caused much overflow in trait solving" #61960

pnkfelix (Sep 26 2019 at 13:49, on Zulip):

it does not make sense for this to be P-high anymore

pnkfelix (Sep 26 2019 at 13:49, on Zulip):

I don't think the problem itself has been resolved, though I do not know if it will ever be resolved...

pnkfelix (Sep 26 2019 at 13:50, on Zulip):

going to downgrade to P-medium and reassign back to self.

pnkfelix (Sep 26 2019 at 13:50, on Zulip):

P-high 15/33: "Self as default type isnt typechecked" #61631

pnkfelix (Sep 26 2019 at 13:51, on Zulip):

ah I was intending to look more at this and forgot

pnkfelix (Sep 26 2019 at 13:52, on Zulip):

(seems to be a pattern for me each week. :sad: )

pnkfelix (Sep 26 2019 at 13:52, on Zulip):

anyway I hope to look at it.

pnkfelix (Sep 26 2019 at 13:52, on Zulip):

(it probably needs to stay P-high too...)

pnkfelix (Sep 26 2019 at 13:53, on Zulip):

P-high 16/33: "couldn't load codegen backend" #61561

pnkfelix (Sep 26 2019 at 13:53, on Zulip):

I made #61561 P-high 2 weeks ago, based on it blocking a developer's progress on #63959.

pnkfelix (Sep 26 2019 at 13:55, on Zulip):

I've got to allocate time to look into these bugs on my windows machine

pnkfelix (Sep 26 2019 at 13:56, on Zulip):

P-high 17/33: " regression: rustdoc resolution changes" #61560

pnkfelix (Sep 26 2019 at 13:56, on Zulip):

(I reassigned this from nagisa to myself earlier in this pre-triage session. nothing to report.)

pnkfelix (Sep 26 2019 at 13:57, on Zulip):

P-high 18/33: "1.30 -> 1.31 dylib late-binding regression with GNU binutils 2.28 or older." #61539

pnkfelix (Sep 26 2019 at 13:57, on Zulip):

I tried to push some discussion forward of this bug last week, but do not really have anything new to report beyond what we said then.

pnkfelix (Sep 26 2019 at 13:58, on Zulip):

P-high 19/33: "Creating a recursive type with infinite size leads to ICE" #61323

pnkfelix (Sep 26 2019 at 13:58, on Zulip):

ugh another bug that I haven't looked at (self-assigned two weeks ago)

pnkfelix (Sep 26 2019 at 13:59, on Zulip):

P-high 20/33: " ICE when trying to match on non-PartialEq slice." #61188

pnkfelix (Sep 26 2019 at 14:00, on Zulip):

ibid another task I need to follow-up on.

pnkfelix (Sep 26 2019 at 14:00, on Zulip):

P-high 21/33: "Compiler panic at Box<Any>" #60363

pnkfelix (Sep 26 2019 at 14:01, on Zulip):

we got feedback from people claiming this bug is still happening

nikomatsakis (Sep 26 2019 at 14:01, on Zulip):

(As we all know, you're not supposed to rely on destructors always being called. But I'd like to know why they're not being called in this scenario.)

I'd just like to dispute this characterization. That's true if you give away ownership of a value, but we should definitely invoke destructors from things on stack frames and so forth (I've not looked into the issue itself)

pnkfelix (Sep 26 2019 at 14:01, on Zulip):

though I cannot immediately tell from the issue whether the reports are true duplicates or not of #60363

pnkfelix (Sep 26 2019 at 14:02, on Zulip):

I'd just like to dispute this characterization. That's true if you give away ownership of a value, but we should definitely invoke destructors from things on stack frames and so forth (I've not looked into the issue itself)

Okay yes I was being flippant.

nikomatsakis (Sep 26 2019 at 14:03, on Zulip):

I figured, it was more for the bystanders :) I've seen people say "not a bug, dtor is not guaranteed" before , and that is definitely not categorically true

Last update: Nov 21 2019 at 13:10UTC