Stream: t-compiler

Topic: pre-meeting triage 2019-06-13 #54818


pnkfelix (Jun 13 2019 at 12:34, on Zulip):

I will be doing pre-triage in this channel.

pnkfelix (Jun 13 2019 at 12:40, on Zulip):

first pre-pass: unprioritized nominated T-compiler issues

pnkfelix (Jun 13 2019 at 12:41, on Zulip):

lets sort by oldest first

pnkfelix (Jun 13 2019 at 12:41, on Zulip):

first up: " single_use_lifetimes warns when deriving a struct" #53738

oli (Jun 13 2019 at 12:46, on Zulip):

does this just need a macro check?

pnkfelix (Jun 13 2019 at 12:46, on Zulip):

I'm trying to even understand the state of this lint with respect to the rest of the compiler

pnkfelix (Jun 13 2019 at 12:46, on Zulip):

what code are we supposed to be replacing that with in the current stable language?

pnkfelix (Jun 13 2019 at 12:47, on Zulip):

(is '_ supposed to work within that struct declaration?)

oli (Jun 13 2019 at 12:47, on Zulip):

impl Debug for Foo<'a> { ... } never mentioning 'a again

pnkfelix (Jun 13 2019 at 12:47, on Zulip):

no I mean

pnkfelix (Jun 13 2019 at 12:47, on Zulip):

struct Foo { bar: &u32 } does not compile

pnkfelix (Jun 13 2019 at 12:48, on Zulip):

nor does struct Foo { bar: &'_ u32, }

pnkfelix (Jun 13 2019 at 12:48, on Zulip):

(this is without the #[derive] ; like I said above, I'm just trying to understand whether it makes sense to try to "fix" the lint right now)

pnkfelix (Jun 13 2019 at 12:49, on Zulip):

ah I think I see

pnkfelix (Jun 13 2019 at 12:49, on Zulip):

you are supposed to write struct Foo<'a> { bar: &'a u32, }

pnkfelix (Jun 13 2019 at 12:50, on Zulip):

and then the derive causes #[warn(single_use_lifetimes)] to complain. Which is exactly what the bug says. Okay.

pnkfelix (Jun 13 2019 at 12:57, on Zulip):

triaging as P-medium (since I don't think it matters much to our end users currently) but leaving I-nominated label on so that we can hopefully discuss whether to still prioritize it just for easing progress in compiler itself.

pnkfelix (Jun 13 2019 at 12:57, on Zulip):

next: "ICE when macro produces multiple functions with "wrong number of type arguments" errors" #55873

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

well this seems like a pretty niche scenario

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

but definitely annoying

pnkfelix (Jun 13 2019 at 13:05, on Zulip):

I'm torn between P-medium and P-high on this one.

pnkfelix (Jun 13 2019 at 13:26, on Zulip):

triage: P-medium; removing nomination label.

pnkfelix (Jun 13 2019 at 13:28, on Zulip):

next: "ICE while running cargo test on embedded-sdmmc (uses 2018 edition)" #56445

pnkfelix (Jun 13 2019 at 13:28, on Zulip):

can't really tell exactly how bad that is from context.

pnkfelix (Jun 13 2019 at 13:28, on Zulip):

triage: P-high, assigning to self, removing nomination label.

pnkfelix (Jun 13 2019 at 13:30, on Zulip):

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

pnkfelix (Jun 13 2019 at 13:31, on Zulip):

hmm. This seems both important to solve but also subtle and complicated

pnkfelix (Jun 13 2019 at 13:31, on Zulip):

triage: P-high. Leaving nomination label. Not assigning yet.

pnkfelix (Jun 13 2019 at 13:32, on Zulip):

next: "Use const generics for array impls" #61415

pnkfelix (Jun 13 2019 at 13:32, on Zulip):

ah as I commented last week, I'm not sure this is meant to be nominated for T-compiler.

pnkfelix (Jun 13 2019 at 13:32, on Zulip):

Nonetheless, it would be nice to prioritize it...

centril (Jun 13 2019 at 13:33, on Zulip):

Don't think it is relevant for t-compiler

pnkfelix (Jun 13 2019 at 13:33, on Zulip):

anyway, leaving #61415 nominated and unprioritized.

pnkfelix (Jun 13 2019 at 13:33, on Zulip):

next: "1.30 -> 1.31 dylib late-binding regression with less recent Linux distro toolchains." #61539

pnkfelix (Jun 13 2019 at 13:34, on Zulip):

this is a bug that @eddyb had hoped we'd discuss last week. I don't know how to prioritize it, it seems pretty nasty.

pnkfelix (Jun 13 2019 at 13:34, on Zulip):

leaving #61539 unprioritized and nominated. (We really need to make sure we get around to the nominated issues list today.)

pnkfelix (Jun 13 2019 at 13:35, on Zulip):

next: "TypeId differs between builds" #61553

pnkfelix (Jun 13 2019 at 13:35, on Zulip):

I believe this is not-a-bug; just a user misunderstanding how "global" a given TypeId might be.

pnkfelix (Jun 13 2019 at 13:36, on Zulip):

leaving #61553 unprioritized and nominated (under assumption that lang team would like to confirm the above belief).

pnkfelix (Jun 13 2019 at 13:38, on Zulip):

next: "ICE on nightly: "Forcing query with already existing DepNode."" #61594

pnkfelix (Jun 13 2019 at 13:44, on Zulip):

there's some rumbling about the relationship of this with #61530, but overall I cannot tell how bad either of these bugs are

pnkfelix (Jun 13 2019 at 13:45, on Zulip):

#61594 triage: P-high, assigning to self for initial investigation. Leaving nomination label so that we can try to discuss relationship with #61530 (which @oli self-assigned, yay!)

pnkfelix (Jun 13 2019 at 13:46, on Zulip):

next: "ICE with index out of bounds const_err" #61595

pnkfelix (Jun 13 2019 at 13:47, on Zulip):

seems like @Esteban Küber has a PR for this. Triage, P-high, assigning to @Esteban Küber

pnkfelix (Jun 13 2019 at 13:48, on Zulip):

next: " rustc panic due to reference passed instead of mut reference" #61623

pnkfelix (Jun 13 2019 at 13:48, on Zulip):

@Matthew Jasper has done some initial investigation here.

pnkfelix (Jun 13 2019 at 13:49, on Zulip):

triage: P-high. @Matthew Jasper , I'll assign to you for now; tell me (here or on issue) if it needs reassignment.

pnkfelix (Jun 13 2019 at 13:50, on Zulip):

next: "ICE using a type alias to a const generic struct across crates" #61624

pnkfelix (Jun 13 2019 at 13:50, on Zulip):

no one's looked at this yet.

pnkfelix (Jun 13 2019 at 13:50, on Zulip):

nothing really to discuss from what's written, IMO.

centril (Jun 13 2019 at 13:50, on Zulip):

Seems like something for @varkor :slight_smile:

pnkfelix (Jun 13 2019 at 13:50, on Zulip):

triage: P-high, assigning to self for initial investigation. Removing nomination label. I'll cc @varkor as well.

pnkfelix (Jun 13 2019 at 13:52, on Zulip):

actually I'll also assign @varkor

pnkfelix (Jun 13 2019 at 13:52, on Zulip):

next: "Self as default type isnt typechecked" #61631

pnkfelix (Jun 13 2019 at 13:53, on Zulip):

whoa, for some strange reason I thought we hadn't even stabilized type parameter defaults ...?

pnkfelix (Jun 13 2019 at 13:54, on Zulip):

definitely needs further investigation

pnkfelix (Jun 13 2019 at 13:54, on Zulip):

triage: P-high. leaving nominated tag. Leaving unassigned for now.

pnkfelix (Jun 13 2019 at 13:55, on Zulip):

next: "Panic with Layout mismatch when copying! from macro" #61663

pnkfelix (Jun 13 2019 at 13:56, on Zulip):

dialogue makes things sound pretty strange.

pnkfelix (Jun 13 2019 at 13:56, on Zulip):

stable-to-beta regression

pnkfelix (Jun 13 2019 at 13:57, on Zulip):

triage: P-high. Not sure if i should assign to self or to @oli

pnkfelix (Jun 13 2019 at 13:57, on Zulip):

I'll leave it nominated

pnkfelix (Jun 13 2019 at 13:57, on Zulip):

and unassigned

pnkfelix (Jun 13 2019 at 13:58, on Zulip):

next: "Soundness hole in pattern matching on enums with an uninhabited variant" #61696

pnkfelix (Jun 13 2019 at 13:59, on Zulip):

@centril by the way, if possible, do try to produce "minimized" reductions that do not use features

pnkfelix (Jun 13 2019 at 13:59, on Zulip):

(as illustrated on the comment thread on #61696), just to make it easier to e.g. compare against stable behavior, for example.

centril (Jun 13 2019 at 13:59, on Zulip):

@pnkfelix it would be less minimal if I did

pnkfelix (Jun 13 2019 at 13:59, on Zulip):

there are different kinds of minimalism

centril (Jun 13 2019 at 13:59, on Zulip):

and the MIR and LLVM would be substantially more complicated

centril (Jun 13 2019 at 14:00, on Zulip):

@pnkfelix the minimalism I'm going for here is "easier to find the source of the bug"

pnkfelix (Jun 13 2019 at 14:00, on Zulip):

for example, you can have one-liners that rely on libstd defintions

centril (Jun 13 2019 at 14:00, on Zulip):

relying on libstd definitions is cheating; you need to inline their definitions =)

pnkfelix (Jun 13 2019 at 14:00, on Zulip):

yes so then we just have a debate about how best to ease finding sources of bugs

pnkfelix (Jun 13 2019 at 14:01, on Zulip):

(yes, I think relying on libstd definitions is to be eschewed. Glad to see we agree on that.)

pnkfelix (Jun 13 2019 at 14:01, on Zulip):

Anywa

pnkfelix (Jun 13 2019 at 14:01, on Zulip):

looks P-high

pnkfelix (Jun 13 2019 at 14:01, on Zulip):

dammit time's up

centril (Jun 13 2019 at 14:01, on Zulip):

I think you need to start pre-triage earlier :P

pnkfelix (Jun 13 2019 at 14:03, on Zulip):

that or hold my bladder during it

Last update: Nov 22 2019 at 04:50UTC