Stream: t-compiler

Topic: weekly meeting 2018-08-02


pnkfelix (Aug 02 2018 at 13:57, on Zulip):

@T-compiler hey everyone (but not @everyone), just a note that we'll be holding the meeting under this topic, starting in about 3 minutes or so

oli (Aug 02 2018 at 13:58, on Zulip):

O/

pnkfelix (Aug 02 2018 at 13:58, on Zulip):

(Did that first @ notification work? I hope so)

pnkfelix (Aug 02 2018 at 13:58, on Zulip):

@T-compiler

nagisa (Aug 02 2018 at 13:58, on Zulip):

@pnkfelix we need ISO dates :P

nagisa (Aug 02 2018 at 13:58, on Zulip):

yes it very much works.

pnkfelix (Aug 02 2018 at 13:58, on Zulip):

@nagisa feel free to edit the topic! :)

pnkfelix (Aug 02 2018 at 13:58, on Zulip):

or lets see

nagisa (Aug 02 2018 at 13:59, on Zulip):

no permission

pnkfelix (Aug 02 2018 at 13:59, on Zulip):

you can do it on your own msg

pnkfelix (Aug 02 2018 at 13:59, on Zulip):

and then have it pervasively change all previous and following

nagisa (Aug 02 2018 at 13:59, on Zulip):

but it forks?

pnkfelix (Aug 02 2018 at 13:59, on Zulip):

not if you select the option in the combo box that pops up

nagisa (Aug 02 2018 at 14:00, on Zulip):

no combo on web, at least not here

pnkfelix (Aug 02 2018 at 14:00, on Zulip):

(you may need to refresh in your browser if you're using one to see that I just merged the forks)

nagisa (Aug 02 2018 at 14:00, on Zulip):

ok, tis is 17 my time now

mw (Aug 02 2018 at 14:00, on Zulip):

test test

pnkfelix (Aug 02 2018 at 14:00, on Zulip):

it should pop up once you edit the topic in an existing message

oli (Aug 02 2018 at 14:00, on Zulip):

test

oli (Aug 02 2018 at 14:00, on Zulip):

this is neat

nagisa (Aug 02 2018 at 14:01, on Zulip):

wait when you see #1 linking magic

pnkfelix (Aug 02 2018 at 14:01, on Zulip):

okay so lets see, gotta get used to working here. :)

pnkfelix (Aug 02 2018 at 14:02, on Zulip):

sort of amazing that we managed to get you all over here

pnkfelix (Aug 02 2018 at 14:02, on Zulip):

thanks to @nagisa

pnkfelix (Aug 02 2018 at 14:02, on Zulip):

clearly we need that same sort of initiative to get us moved off of the etherpad

pnkfelix (Aug 02 2018 at 14:02, on Zulip):

speaking of which, here's the link for our agenda: etherpad

pnkfelix (Aug 02 2018 at 14:03, on Zulip):

(if anyone has UI issues during the meeting and cannot be heard here, feel free to ping me on IRC)

pnkfelix (Aug 02 2018 at 14:04, on Zulip):

Having said that, lets get started

pnkfelix (Aug 02 2018 at 14:04, on Zulip):

First up, P-high T-compiler

pnkfelix (Aug 02 2018 at 14:04, on Zulip):

First issue #52390

pnkfelix (Aug 02 2018 at 14:05, on Zulip):

(yes, unfortunately the github integration does not, yet, print the issue title (nor include it in the popup). It just gives you a smart link)

pnkfelix (Aug 02 2018 at 14:06, on Zulip):

Does anyone have anything to add beyond the comment I put on this issue last week ?

oli (Aug 02 2018 at 14:06, on Zulip):

Nah, let's close after 1.28

pnkfelix (Aug 02 2018 at 14:06, on Zulip):

Honestly if its stayed open this long it won't hurt to leave it open until the release ... which is ... when now?

oli (Aug 02 2018 at 14:07, on Zulip):

tomorrow

pnkfelix (Aug 02 2018 at 14:07, on Zulip):

fun

nagisa (Aug 02 2018 at 14:07, on Zulip):

We are releasing very soon anyway, I don't see the point. (although some rust users have expressed concerns with this strategy with another such issue)

pnkfelix (Aug 02 2018 at 14:07, on Zulip):

okay next up #52363

oli (Aug 02 2018 at 14:08, on Zulip):

compile error -> ICE regression

oli (Aug 02 2018 at 14:08, on Zulip):

lower prio?

pnkfelix (Aug 02 2018 at 14:08, on Zulip):

@Vadim Petrochenkov reported ... (hey I think I forgot to put them onto the T-compiler group here)

oli (Aug 02 2018 at 14:08, on Zulip):

@Vadim Petrochenkov seems to think so

pnkfelix (Aug 02 2018 at 14:09, on Zulip):

okay so should we actually just make this P-medium?

pnkfelix (Aug 02 2018 at 14:10, on Zulip):

we'll visit it anyway when we go through regressions each week

pnkfelix (Aug 02 2018 at 14:10, on Zulip):

next #52092

pnkfelix (Aug 02 2018 at 14:10, on Zulip):

i haven't had a chance to look at this yet

pnkfelix (Aug 02 2018 at 14:11, on Zulip):

I might have a chance tonight or tomorrow

pnkfelix (Aug 02 2018 at 14:11, on Zulip):

but I'll unassign myself now

pnkfelix (Aug 02 2018 at 14:11, on Zulip):

(since I go on PTO after tomorrow.)

pnkfelix (Aug 02 2018 at 14:12, on Zulip):

I don't know if @nikomatsakis will be able to look at it when he pauses his PTO, because if his remote Linux machine is still inaccessible then he won't be able to do useful profiling.

pnkfelix (Aug 02 2018 at 14:13, on Zulip):

but I'll let you all figure that out next week

pnkfelix (Aug 02 2018 at 14:13, on Zulip):

next #52053

pnkfelix (Aug 02 2018 at 14:13, on Zulip):

ah great it looks like @varkor finished their PR #52644 for this ?

oli (Aug 02 2018 at 14:13, on Zulip):

yea, needs a rebase and another review

pnkfelix (Aug 02 2018 at 14:14, on Zulip):

@mw if you don't think you'll have time to review, definitely speak up. :)

varkor (Aug 02 2018 at 14:14, on Zulip):

ah, hadn't spotted it needed rebasing

varkor (Aug 02 2018 at 14:14, on Zulip):

I'll do that now

pnkfelix (Aug 02 2018 at 14:14, on Zulip):

next up: #51608

mw (Aug 02 2018 at 14:14, on Zulip):

I did a review of the part I feel comfortable with

mw (Aug 02 2018 at 14:14, on Zulip):

I should re-assign probably

pnkfelix (Aug 02 2018 at 14:15, on Zulip):

I still don't know whether to leave #51608 at P-high

pnkfelix (Aug 02 2018 at 14:15, on Zulip):

I guess I should have CC'ed someone on the servo team about it

pnkfelix (Aug 02 2018 at 14:15, on Zulip):

but I figure if one of the participants on the issue cares, they'll speak up

pnkfelix (Aug 02 2018 at 14:15, on Zulip):

I'll leave another note

nagisa (Aug 02 2018 at 14:15, on Zulip):

pcwalton seems like a good person to ping, given that they investigated this already

pnkfelix (Aug 02 2018 at 14:16, on Zulip):

okay, done

pnkfelix (Aug 02 2018 at 14:16, on Zulip):

at least, ping'ed via github

pnkfelix (Aug 02 2018 at 14:17, on Zulip):

though I know from personal experience that such pings are not the best

pnkfelix (Aug 02 2018 at 14:17, on Zulip):

but for now I'll live with it

pnkfelix (Aug 02 2018 at 14:17, on Zulip):

next: #50865

pnkfelix (Aug 02 2018 at 14:18, on Zulip):

@Vadim Petrochenkov @eddyb I'm still curious if either of you have an answer for #50865, namely the question posed here. (Its possible that eddyb already responded via IRC and I just forgot...)

oli (Aug 02 2018 at 14:19, on Zulip):

should this issue be on the edition milestone?

oli (Aug 02 2018 at 14:19, on Zulip):

seems to be an impl Trait adoption blocker

pnkfelix (Aug 02 2018 at 14:20, on Zulip):

I ... don't have an objection to doing that.

pnkfelix (Aug 02 2018 at 14:20, on Zulip):

Put it on "Rust 2018 RC" milestone, then ?

pnkfelix (Aug 02 2018 at 14:21, on Zulip):

I'm going to do that, feel free to message me with outcries/objections/etc

pnkfelix (Aug 02 2018 at 14:22, on Zulip):

next: #49482

pnkfelix (Aug 02 2018 at 14:22, on Zulip):

huh I wonder why I didn't write any comments here last week

mw (Aug 02 2018 at 14:23, on Zulip):

no progress here :/

pnkfelix (Aug 02 2018 at 14:23, on Zulip):

okay

pnkfelix (Aug 02 2018 at 14:23, on Zulip):

@mw do you suspect you wont have time to investigate this?

pnkfelix (Aug 02 2018 at 14:24, on Zulip):

I guess give it another week but also think about potentially reassigning next week to someone else if no progress has been made.

mw (Aug 02 2018 at 14:24, on Zulip):

depends on how urgent it really is

pnkfelix (Aug 02 2018 at 14:26, on Zulip):

yeah okay I wrote a comment trying to reflect that.

pnkfelix (Aug 02 2018 at 14:26, on Zulip):

next: #49300

pnkfelix (Aug 02 2018 at 14:26, on Zulip):

same deal there I imagine

pnkfelix (Aug 02 2018 at 14:26, on Zulip):

I think I'll write a similar comment there.

pnkfelix (Aug 02 2018 at 14:27, on Zulip):

okay that's all the P-high issues

pnkfelix (Aug 02 2018 at 14:27, on Zulip):

next, stable-to-beta regressions, including non- T-compiler ones, as usual

pnkfelix (Aug 02 2018 at 14:27, on Zulip):

first #52966

pnkfelix (Aug 02 2018 at 14:29, on Zulip):

@Oli I'm trying to make sure I understand your comment there

pnkfelix (Aug 02 2018 at 14:29, on Zulip):

You're saying that you think we used to warn on that example in the past, then at some point we stopped, but now we've started warning again? Is that right?

oli (Aug 02 2018 at 14:29, on Zulip):

so we used to produce this warning to the best of my knowledge

oli (Aug 02 2018 at 14:29, on Zulip):

yea

oli (Aug 02 2018 at 14:29, on Zulip):

that

pnkfelix (Aug 02 2018 at 14:29, on Zulip):

hmm okay

pnkfelix (Aug 02 2018 at 14:30, on Zulip):

this sort of thing is just fact of life, right?

oli (Aug 02 2018 at 14:30, on Zulip):

we don't have static if... the same issue arises when you index an array with a constant after checking that the constant is in bounds

oli (Aug 02 2018 at 14:30, on Zulip):

you get a message about out of bounds indexing in dead code

pnkfelix (Aug 02 2018 at 14:30, on Zulip):

as in, we don't have a goal to try to warn about portability-issues due to usize size dependence and casts and what not, right?

oli (Aug 02 2018 at 14:31, on Zulip):

"dead" only in the sense that it's unreachable due to constants, but not that rustc knows about the deadness

oli (Aug 02 2018 at 14:31, on Zulip):

yea I'm not sure how to act on these issues, we can close it as a dup, I'll find the one that complains about the same issue

pnkfelix (Aug 02 2018 at 14:32, on Zulip):

in any case, there's two things I see here. 1. What priority label to assign to this, and 2. If this is low or medium priority, do we agree with the issue filer that it should get a release note?

oli (Aug 02 2018 at 14:32, on Zulip):

it's not just pointer sizes, it's literally anything that is dead code due to a condition

pnkfelix (Aug 02 2018 at 14:32, on Zulip):

ah I see

oli (Aug 02 2018 at 14:32, on Zulip):

if you do if false { 1 / 0; } we will still complain about div by zero

varkor (Aug 02 2018 at 14:33, on Zulip):

it seems likely that future changes to the const evaluator are going to keep producing these refinements of what's acceptable or not

pnkfelix (Aug 02 2018 at 14:33, on Zulip):

okay I'm going to assign this P-high, not because I think it is a bug that needs fixing per se, but just to note that we need to decide how much to say about this in the docs/release notes

varkor (Aug 02 2018 at 14:33, on Zulip):

unless there's a specific decision not to allow them

pnkfelix (Aug 02 2018 at 14:33, on Zulip):

i'm going to put I-needs-decision on it too

pnkfelix (Aug 02 2018 at 14:34, on Zulip):

(though I guess that label is more often used for less fuzzy decisions...)

varkor (Aug 02 2018 at 14:34, on Zulip):

in which case, it would be common for release notes to mention "more constant expressions will fail to compile"

varkor (Aug 02 2018 at 14:34, on Zulip):

which seems superfluous

oli (Aug 02 2018 at 14:35, on Zulip):

also they don't fail to compile. it's just a lint

pnkfelix (Aug 02 2018 at 14:35, on Zulip):

@Oli I'm deliberately not assigning to you since I know you're pretty busy

oli (Aug 02 2018 at 14:36, on Zulip):

thx

pnkfelix (Aug 02 2018 at 14:36, on Zulip):

next #52849

pnkfelix (Aug 02 2018 at 14:37, on Zulip):

seems like there's a lot of investigatory activity here, that's good

pnkfelix (Aug 02 2018 at 14:37, on Zulip):

I ... guess I'll make it P-high?

oli (Aug 02 2018 at 14:38, on Zulip):

yea ralf has been very successful in eliminating a large part of the regression

pnkfelix (Aug 02 2018 at 14:38, on Zulip):

not really sure if there's any person here to assign it to. @RalfJ has been pretty active on it.

oli (Aug 02 2018 at 14:38, on Zulip):

there's a smaller one left that might be unavoidable since checking a 50k array simply will take some time

pnkfelix (Aug 02 2018 at 14:39, on Zulip):

I privmsg'ed Ralf

pnkfelix (Aug 02 2018 at 14:39, on Zulip):

so we'll see what he says

pnkfelix (Aug 02 2018 at 14:39, on Zulip):

assigning P-high in any case so that we look at it again next week ASAP

pnkfelix (Aug 02 2018 at 14:40, on Zulip):

and we already discussed #52363 and #52092

pnkfelix (Aug 02 2018 at 14:40, on Zulip):

so that's all the stable-to-beta regressions for us

pnkfelix (Aug 02 2018 at 14:40, on Zulip):

next, stable-to-nightly regressions

pnkfelix (Aug 02 2018 at 14:41, on Zulip):

first up #52900

nagisa (Aug 02 2018 at 14:42, on Zulip):

Seems to be undefined behaviour in some of the X bindings

pnkfelix (Aug 02 2018 at 14:42, on Zulip):

okay so @nagisa hypothesizes that #52900 is a dupe of #52875 (which is next on the list)

pnkfelix (Aug 02 2018 at 14:42, on Zulip):

and as @nagisa says, #52875 has been blamed on UB in some unsafe code in X bindings

nagisa (Aug 02 2018 at 14:42, on Zulip):

namely they do something along the lines of

let x: ManuallyDrop<SomethingContaining<fn()>> = mem::zeroed()
nagisa (Aug 02 2018 at 14:44, on Zulip):

What they wanted instead was not ManuallyDrop, but MaybeUninitialized.

pnkfelix (Aug 02 2018 at 14:44, on Zulip):

because they want &uninit and dont have it

pnkfelix (Aug 02 2018 at 14:44, on Zulip):

yeah

pnkfelix (Aug 02 2018 at 14:44, on Zulip):

what is our plan here?

pnkfelix (Aug 02 2018 at 14:44, on Zulip):

just close as non-actionable?

nagisa (Aug 02 2018 at 14:45, on Zulip):

well, I’m surprised this ends up in sigill at creation time, but other than that I think we should just close

pnkfelix (Aug 02 2018 at 14:45, on Zulip):

i feel okay closing #52875 as non-actionable

pnkfelix (Aug 02 2018 at 14:45, on Zulip):

yes that is odd

pnkfelix (Aug 02 2018 at 14:45, on Zulip):

well okay I'm going to close them

pnkfelix (Aug 02 2018 at 14:45, on Zulip):

we have enough work on our plate

pnkfelix (Aug 02 2018 at 14:46, on Zulip):

if someone wants to spend time investigating how it happens at creation time, that's fine. but I don't want us revisiting this every week.

pnkfelix (Aug 02 2018 at 14:47, on Zulip):

okay then that's all the stable-to-nightly regressions that have been filed

pnkfelix (Aug 02 2018 at 14:48, on Zulip):

next, I-beta-nominated

pnkfelix (Aug 02 2018 at 14:48, on Zulip):

nothing there for us i think

pnkfelix (Aug 02 2018 at 14:49, on Zulip):

next, waiting on T-compiler

pnkfelix (Aug 02 2018 at 14:49, on Zulip):

zaaaroh

pnkfelix (Aug 02 2018 at 14:49, on Zulip):

Next, i-nominated issues

pnkfelix (Aug 02 2018 at 14:49, on Zulip):

first up #52545

pnkfelix (Aug 02 2018 at 14:50, on Zulip):

I'm looking at this and I think I have something plausible, thanks to input from @eddyb

pnkfelix (Aug 02 2018 at 14:50, on Zulip):

I hope to have a PR up today

pnkfelix (Aug 02 2018 at 14:51, on Zulip):

(or rather, my thing is a change to -Z unpretty=everybody-loops, which quite surprisingly has been co-opted by rustdoc for some nefarious purpose)

pnkfelix (Aug 02 2018 at 14:51, on Zulip):

next #52141

nagisa (Aug 02 2018 at 14:52, on Zulip):

I remember glob imports breaking from all sorts of things, wouldn’t be surprised if this is a duplicate of some issue from 2 or 3 years ago :slight_smile:

pnkfelix (Aug 02 2018 at 14:52, on Zulip):

This should probably be assigned to someone

pnkfelix (Aug 02 2018 at 14:52, on Zulip):

aturon already pinged @Vadim Petrochenkov

pnkfelix (Aug 02 2018 at 14:53, on Zulip):

is this P-high?

pnkfelix (Aug 02 2018 at 14:54, on Zulip):

Im not quite clear on why this has been tagged with 2018 edition stuff

pnkfelix (Aug 02 2018 at 14:54, on Zulip):

oh, use crate::...

pnkfelix (Aug 02 2018 at 14:54, on Zulip):

okay

pnkfelix (Aug 02 2018 at 14:54, on Zulip):

I guess I'll tag as P-high.

pnkfelix (Aug 02 2018 at 14:54, on Zulip):

though I'm not really happy doing so without knowing whom is going to work on it.

pnkfelix (Aug 02 2018 at 14:55, on Zulip):

next #52140

pnkfelix (Aug 02 2018 at 14:56, on Zulip):

next: #51456

pnkfelix (Aug 02 2018 at 14:57, on Zulip):

(oh sorry, I didn't speak here about #52140; I tagged that with same message at #52141)

pnkfelix (Aug 02 2018 at 14:58, on Zulip):

egad @simulacrum wants this this week?

pnkfelix (Aug 02 2018 at 14:58, on Zulip):

I'll assign it to myself, see if I can assist @qmx

pnkfelix (Aug 02 2018 at 14:59, on Zulip):

I'm going to leave the I-nominated tag, and add a P-high to it too

pnkfelix (Aug 02 2018 at 14:59, on Zulip):

(not that I know if the P-* labels are all that meaningful when applied to PR's...)

pnkfelix (Aug 02 2018 at 15:00, on Zulip):

next: #51131

pnkfelix (Aug 02 2018 at 15:02, on Zulip):

@eddyb do you know if this has now been rebased atop #52461 ?

eddyb (Aug 02 2018 at 15:02, on Zulip):

it looks like it

eddyb (Aug 02 2018 at 15:02, on Zulip):

I just didn't get a ping

pnkfelix (Aug 02 2018 at 15:02, on Zulip):

and if so, were there other outstanding changes you were hoping to see? I know there was ongoing conversation in the comment thread about whether having some form of &own could help things, but that shouldn't block this PR, right?

pnkfelix (Aug 02 2018 at 15:03, on Zulip):

okay

pnkfelix (Aug 02 2018 at 15:03, on Zulip):

does this (#51131) need to stay on I-nominated ?

pnkfelix (Aug 02 2018 at 15:05, on Zulip):

@eddyb ^

pnkfelix (Aug 02 2018 at 15:05, on Zulip):

I guess I'll leave the I-nominated tag so that this doesn't fall by the wayside (at least, not any more than it already has...)

pnkfelix (Aug 02 2018 at 15:06, on Zulip):

We're 6 minutes over time but there's only one nominated issue left, #44489

pnkfelix (Aug 02 2018 at 15:07, on Zulip):

seems like there's active discussion on #44489 at least

pnkfelix (Aug 02 2018 at 15:07, on Zulip):

but it also sounds like some participants really want it in the edition

pnkfelix (Aug 02 2018 at 15:08, on Zulip):

but certainly it cant hope to be in EP2, right?

pnkfelix (Aug 02 2018 at 15:08, on Zulip):

I guess I'll leave the I-nominated tag on this too

pnkfelix (Aug 02 2018 at 15:10, on Zulip):

okay so now we're 10 minutes over time, but that's everything I'd hope to cover

pnkfelix (Aug 02 2018 at 15:10, on Zulip):

thanks everyone in @T-compiler for attending!

oli (Aug 02 2018 at 15:10, on Zulip):

is it possible to archive discussions?

pnkfelix (Aug 02 2018 at 15:11, on Zulip):

you mean, keep them from being edited or something?

pnkfelix (Aug 02 2018 at 15:11, on Zulip):

after ten minutes you can't edit anymore

eddyb (Aug 02 2018 at 15:11, on Zulip):

uhh I forgot to mention why #51131 was even nominated

oli (Aug 02 2018 at 15:11, on Zulip):

or further commented on

pnkfelix (Aug 02 2018 at 15:11, on Zulip):

and there are links

eddyb (Aug 02 2018 at 15:11, on Zulip):

https://github.com/rust-lang/rust/pull/51131#issuecomment-404775018

pnkfelix (Aug 02 2018 at 15:11, on Zulip):

i.e. https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/subject/weekly.20meeting.202018-08-02/near/130779492

oli (Aug 02 2018 at 15:11, on Zulip):

neat

eddyb (Aug 02 2018 at 15:13, on Zulip):

by this point, however, I think I had a pretty good strategy https://github.com/rust-lang/rust/pull/51131#issuecomment-405031609 (effectively taking by-ref/indirect cases and adding an optional metadata besides them, to reduce duplication, and just handle it in the backend)

pnkfelix (Aug 02 2018 at 15:15, on Zulip):

okay so I interpret that as meaning "there may be interesting debates to have here w.r.t. current and future design, but this particular PR does not need to remain I-nominated", right?

eddyb (Aug 02 2018 at 15:15, on Zulip):

yeah I think we can handle it

Last update: Nov 20 2019 at 01:35UTC