Stream: t-compiler/wg-nll

Topic: weekly meeting December 19


pnkfelix (Dec 19 2018 at 20:31, on Zulip):

Hi @WG-compiler-nll

pnkfelix (Dec 19 2018 at 20:31, on Zulip):

so lets see

pnkfelix (Dec 19 2018 at 20:31, on Zulip):

i totally failed to do my normal pre-meeting triage

pnkfelix (Dec 19 2018 at 20:32, on Zulip):

(though I did do some triage of the NLL-deferred issues)

pnkfelix (Dec 19 2018 at 20:32, on Zulip):

before I open the meeting up to general discussion

pnkfelix (Dec 19 2018 at 20:32, on Zulip):

lets at least go through the nominated issues

pnkfelix (Dec 19 2018 at 20:32, on Zulip):

nominated issues

pnkfelix (Dec 19 2018 at 20:33, on Zulip):

oldest first: "figure out how to integrate constants and the MIR type checker" #46702

pnkfelix (Dec 19 2018 at 20:33, on Zulip):

I nominated this today while doing re-triage of NLL-deferred

pnkfelix (Dec 19 2018 at 20:33, on Zulip):

basically I wanted to see if anyone here would be willing to investigate whether this is even a problem

pnkfelix (Dec 19 2018 at 20:34, on Zulip):

(as in: You're not signing up to fix anything. Yet.)

nikomatsakis (Dec 19 2018 at 20:34, on Zulip):

I would be

nikomatsakis (Dec 19 2018 at 20:34, on Zulip):

I opened the dang bug :P

nikomatsakis (Dec 19 2018 at 20:34, on Zulip):

I think the relevant miri pieces now exist

nikomatsakis (Dec 19 2018 at 20:35, on Zulip):

I sort of suspect there's no "there there" anymore

pnkfelix (Dec 19 2018 at 20:35, on Zulip):

I'm going to assign it to niko then

pnkfelix (Dec 19 2018 at 20:36, on Zulip):

oh, the other nominated issue is one I closed

nikomatsakis (Dec 19 2018 at 20:36, on Zulip):

which issue was that?

pnkfelix (Dec 19 2018 at 20:36, on Zulip):

(I include closed ones in the search to ensure that e.g. merged PR's or issues closed due to having associated PR's that were merged are still on list of nominated things)

pnkfelix (Dec 19 2018 at 20:37, on Zulip):

it was "should let _: T = RHS be synonymous with let _ = RHS: T?" #56715

nikomatsakis (Dec 19 2018 at 20:37, on Zulip):

ah, you closed it :)

pnkfelix (Dec 19 2018 at 20:37, on Zulip):

basically niko and centril convinced me that let _: T = RHS; and let _ = RHS: T; need not necessarily be synonymous

pnkfelix (Dec 19 2018 at 20:38, on Zulip):

(we may make them synonymous in the future, but if we did, it would probably be due to changing the semantics of RHS: T, not other stuff)

pnkfelix (Dec 19 2018 at 20:38, on Zulip):

okay

pnkfelix (Dec 19 2018 at 20:38, on Zulip):

we don't have any unassigned P-high issues that I know of

nikomatsakis (Dec 19 2018 at 20:38, on Zulip):

let _ = RHS;

I think you mean let _ = RHS: T;

pnkfelix (Dec 19 2018 at 20:39, on Zulip):

thank you. fixed.

pnkfelix (Dec 19 2018 at 20:39, on Zulip):

by the way

nikomatsakis (Dec 19 2018 at 20:39, on Zulip):

I just realized I forgot to update the tracking paper doc

pnkfelix (Dec 19 2018 at 20:39, on Zulip):

can/should we get rid of the NLL-priority label?

nikomatsakis (Dec 19 2018 at 20:39, on Zulip):

( are we still doing that? :) is it useful? )

pnkfelix (Dec 19 2018 at 20:39, on Zulip):

I would think I-nominated and/or P-high could suffice for it

nikomatsakis (Dec 19 2018 at 20:39, on Zulip):

seems fine

nikomatsakis (Dec 19 2018 at 20:39, on Zulip):

wait

pnkfelix (Dec 19 2018 at 20:40, on Zulip):

the only reason to have a separate NLL-priority tag

nikomatsakis (Dec 19 2018 at 20:40, on Zulip):

like, literally a label called NLL-priority?

pnkfelix (Dec 19 2018 at 20:40, on Zulip):

would be if I-nominated and/or P-high is causing noise for other groups

nikomatsakis (Dec 19 2018 at 20:40, on Zulip):

(that is, priority is not a metavariable here..?)

pnkfelix (Dec 19 2018 at 20:40, on Zulip):

there is a label called NLL-priority: https://github.com/rust-lang/rust/issues?utf8=%E2%9C%93&q=+label%3ANLL-priority+

nikomatsakis (Dec 19 2018 at 20:40, on Zulip):

ok

nikomatsakis (Dec 19 2018 at 20:40, on Zulip):

seems fine then

pnkfelix (Dec 19 2018 at 20:40, on Zulip):

but I think we should kill it.

pnkfelix (Dec 19 2018 at 20:41, on Zulip):

okay then

Santiago Pastorino (Dec 19 2018 at 20:41, on Zulip):

unrelated question, when's the retrospective happening?

nikomatsakis (Dec 19 2018 at 20:41, on Zulip):

egads

nikomatsakis (Dec 19 2018 at 20:41, on Zulip):

I forgot to try and schedule that

pnkfelix (Dec 19 2018 at 20:41, on Zulip):

that hasn't been scheduled yet, has it?

nikomatsakis (Dec 19 2018 at 20:41, on Zulip):

I will add a todo item :)

Santiago Pastorino (Dec 19 2018 at 20:41, on Zulip):

ok

Santiago Pastorino (Dec 19 2018 at 20:41, on Zulip):

:)

nikomatsakis (Dec 19 2018 at 20:41, on Zulip):

but no it has not

nikomatsakis (Dec 19 2018 at 20:42, on Zulip):

I am assuming early Jan

nikomatsakis (Dec 19 2018 at 20:42, on Zulip):

seeing as holidays are coming up

Santiago Pastorino (Dec 19 2018 at 20:42, on Zulip):

I'm sorry but I have been failing a bit to be back to work on rustc

Santiago Pastorino (Dec 19 2018 at 20:42, on Zulip):

will do soon

Santiago Pastorino (Dec 19 2018 at 20:42, on Zulip):

next week probably

nikomatsakis (Dec 19 2018 at 20:42, on Zulip):

just in time for everyone to disappear on vacation :)

Santiago Pastorino (Dec 19 2018 at 20:42, on Zulip):

hehe :)

Santiago Pastorino (Dec 19 2018 at 20:42, on Zulip):

when everyone is back I can say ... guys I have been working a lot and just spend all the time in the beach :)

pnkfelix (Dec 19 2018 at 20:43, on Zulip):

(at this point I think we can open things up to general discussion topics? I dont think I have anything else I wanted to discuss)

nikomatsakis (Dec 19 2018 at 20:44, on Zulip):

I don't have a lot but

Matthew Jasper (Dec 19 2018 at 20:44, on Zulip):

Are we any closer to a decision on #56254?

pnkfelix (Dec 19 2018 at 20:44, on Zulip):

not yet

pnkfelix (Dec 19 2018 at 20:44, on Zulip):

I need to look at it

nikomatsakis (Dec 19 2018 at 20:44, on Zulip):

one thing people might be interested in is this thread: link

nikomatsakis (Dec 19 2018 at 20:44, on Zulip):

can we maybe discuss the "next steps"?

nikomatsakis (Dec 19 2018 at 20:45, on Zulip):

I feel like some form of "write-up" is needed and perhaps a sync discussion with Ralf and others present?

nikomatsakis (Dec 19 2018 at 20:45, on Zulip):

to try and dive in?

pnkfelix (Dec 19 2018 at 20:45, on Zulip):

my plan was to try to make docs for two-phase borrows in parallel with investigating this Q

pnkfelix (Dec 19 2018 at 20:45, on Zulip):

so I'd be happy to work on a write-up.

pnkfelix (Dec 19 2018 at 20:46, on Zulip):

though I may want multiple reviewers

pnkfelix (Dec 19 2018 at 20:46, on Zulip):

to sanity check it

nikomatsakis (Dec 19 2018 at 20:46, on Zulip):

sounds good, I was thinking I might try to leave a comment on the issue with my current 'mental state' just to have it documented somewhere before I forget

nikomatsakis (Dec 19 2018 at 20:47, on Zulip):

I suspect that some members of lang team (notably Josh Tripplet, but maybe others) would also like to be part of discussion

nikomatsakis (Dec 19 2018 at 20:47, on Zulip):

Josh in particular seemed in favor of continuing to accept the code in question

nikomatsakis (Dec 19 2018 at 20:49, on Zulip):

(ok, well, shall we adjourn for this week?)

pnkfelix (Dec 19 2018 at 20:49, on Zulip):

is there anything w.r.t. polonius to discuss?

pnkfelix (Dec 19 2018 at 20:49, on Zulip):

or better to leave all that to your forthcoming blog post?

nikomatsakis (Dec 19 2018 at 20:49, on Zulip):

well, I started working on a blog post (draft) covering some of the missing pieces,

nikomatsakis (Dec 19 2018 at 20:50, on Zulip):

it's more or less comlete but I didn't publish it yet because I wanted to revise my description,

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

but it doesn't seem super high priority.

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

I guess we still need a sort of "gameplan" for how to transition, but I feel like it would be unwise to start doing any major moves at this point, we're still letting NLL "settle in"

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

one thing we could discuss is whether/when to start enabling NLL more broadly but it seems clearly premature to me at this point

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

at least I would want to have NLL-sound issues closed,

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

and stuff like 2PB discussion settled

nikomatsakis (Dec 19 2018 at 20:51, on Zulip):

maybe it's worth making a kind of "criteria" though

lqd (Dec 19 2018 at 20:52, on Zulip):

(just a tiny datafrog PR, for discussion wrt stats gathering, on my part for polonius)

nikomatsakis (Dec 19 2018 at 20:52, on Zulip):

oh that reminds me I added highfive to polonius/datafrog

nikomatsakis (Dec 19 2018 at 20:52, on Zulip):

I think right now all PRs get assigned to me

nikomatsakis (Dec 19 2018 at 20:52, on Zulip):

but if other people want to add their names that is fine :)

nikomatsakis (Dec 19 2018 at 20:52, on Zulip):

but at least PRs shouldn't go unobserved for long periods of time

lqd (Dec 19 2018 at 20:53, on Zulip):

oh we can r? people now I assume (I don't think highfive picked this one up already)

nikomatsakis (Dec 19 2018 at 20:53, on Zulip):

you can do that too

nikomatsakis (Dec 19 2018 at 20:53, on Zulip):

if it's not working though let me know

nikomatsakis (Dec 19 2018 at 20:54, on Zulip):

but I only did that few days back

nikomatsakis (Dec 19 2018 at 20:54, on Zulip):

so perhaps there are some older PRs, I see there are 3 :)

lqd (Dec 19 2018 at 20:54, on Zulip):

will do ^^

pnkfelix (Dec 19 2018 at 20:54, on Zulip):

one thing we could discuss is whether/when to start enabling NLL more broadly but it seems clearly premature to me at this point

this could mean different things

lqd (Dec 19 2018 at 20:54, on Zulip):

(I also wanted to work on the followup to the older "specialization" PR I closed until treefrog was merged)

pnkfelix (Dec 19 2018 at 20:54, on Zulip):

it could mean "when do we enable NLL migrate mode in 2015 edition"

pnkfelix (Dec 19 2018 at 20:55, on Zulip):

but it could also mean "when do we move away from some of the hacks associated with migrate mode?"

nikomatsakis (Dec 19 2018 at 20:55, on Zulip):

true

nikomatsakis (Dec 19 2018 at 20:55, on Zulip):

I think I meant the former

pnkfelix (Dec 19 2018 at 20:55, on Zulip):

both Q's are important

pnkfelix (Dec 19 2018 at 20:55, on Zulip):

and I don't know if there may be an ordering to them

lqd (Dec 19 2018 at 20:55, on Zulip):

the polonius tweaks PR #56507 seems to fail late, doing documentation I think

pnkfelix (Dec 19 2018 at 20:55, on Zulip):

i.e. should we try to leverage the time with the 2018 edition

pnkfelix (Dec 19 2018 at 20:56, on Zulip):

to try to make migrate mode closer to what final NLL will be

Matthew Jasper (Dec 19 2018 at 20:56, on Zulip):

We have 9 issues still open marked NLL-sound https://github.com/rust-lang/rust/labels/NLL-sound (although some are being closed soon)

pnkfelix (Dec 19 2018 at 20:56, on Zulip):

before enabling migrate for 2015

nikomatsakis (Dec 19 2018 at 20:56, on Zulip):

the polonius tweaks PR #56507 seems to fail late, doing documentation I think

ugh

pnkfelix (Dec 19 2018 at 20:56, on Zulip):

(i wonder if we want an NLL-2pb label)

nikomatsakis (Dec 19 2018 at 20:56, on Zulip):

I think there are some unfiled issues, too

nikomatsakis (Dec 19 2018 at 20:56, on Zulip):

re NLL-sound

nikomatsakis (Dec 19 2018 at 20:57, on Zulip):

I just remembered I was meaning to file an issue anyway

nikomatsakis (Dec 19 2018 at 20:57, on Zulip):

let me do that before I forget again

lqd (Dec 19 2018 at 20:57, on Zulip):

(highfive seems to not answer to r? yet)

nikomatsakis (Dec 19 2018 at 20:58, on Zulip):

(highfive seems to not answer to r? yet)

yeah weird

nikomatsakis (Dec 19 2018 at 20:58, on Zulip):

We have 9 issues still open marked NLL-sound https://github.com/rust-lang/rust/labels/NLL-sound (although some are being closed soon)

in any case I feel like we should strive to close these before enabling NLL on 2015, probably?

nikomatsakis (Dec 19 2018 at 21:01, on Zulip):

filed the issue I was talking about: https://github.com/rust-lang/rust/issues/56993

nikomatsakis (Dec 19 2018 at 21:02, on Zulip):

maybe i'll self-assign to investigate

pnkfelix (Dec 19 2018 at 21:03, on Zulip):

You probably meant to write -Zborrowck=mir in that description.

pnkfelix (Dec 19 2018 at 21:04, on Zulip):

an issue like that is definitely an example of the second case I was describing ("when do we move away from migrate hacks:")

nikomatsakis (Dec 19 2018 at 21:04, on Zulip):

indeed

nikomatsakis (Dec 19 2018 at 21:04, on Zulip):

yeah it hadn't occurred to me that we might transition 2018 to "real NLL" (not migrate) first

nikomatsakis (Dec 19 2018 at 21:05, on Zulip):

I feel like we probably don't want to do that though

pnkfelix (Dec 19 2018 at 21:05, on Zulip):

really?

nikomatsakis (Dec 19 2018 at 21:05, on Zulip):

I had assumed we would want to end the migration period kind of "universally"

nikomatsakis (Dec 19 2018 at 21:05, on Zulip):

but now i'm wondering why I thought that

pnkfelix (Dec 19 2018 at 21:07, on Zulip):

well

pnkfelix (Dec 19 2018 at 21:07, on Zulip):

hmm

pnkfelix (Dec 19 2018 at 21:07, on Zulip):

I was more wondering if we could keep doing migration

pnkfelix (Dec 19 2018 at 21:08, on Zulip):

but make its semantics, in terms of what MIR-borrowck reports, actually match that of #![feature(nll)]

pnkfelix (Dec 19 2018 at 21:08, on Zulip):

because our current status is pretty confusing

pnkfelix (Dec 19 2018 at 21:08, on Zulip):

and spreading that confusion to the 2015 edition seems ... not great...

nikomatsakis (Dec 19 2018 at 21:11, on Zulip):

i.e., we would not emit the region errors?

nikomatsakis (Dec 19 2018 at 21:11, on Zulip):

we could perhaps due that by accumulating them in the typeck-tables or something

pnkfelix (Dec 19 2018 at 21:12, on Zulip):

anyway I'm signing off

pnkfelix (Dec 19 2018 at 21:12, on Zulip):

thanks for attending, everyone

Last update: Nov 21 2019 at 13:25UTC