Stream: t-compiler/wg-nll

Topic: weekly meeting 2019.01.16


pnkfelix (Jan 16 2019 at 20:00, on Zulip):

am I right that this week's meeting is starting 30minutes earlier than normal? Or did I misread that in the notes from last week?

Matthew Jasper (Jan 16 2019 at 20:01, on Zulip):

It's supposed to be 1h long. I'm not sure if the start time or the end time was moved.

pnkfelix (Jan 16 2019 at 20:02, on Zulip):

ah. unfortunate that that was left unspecified

Santiago Pastorino (Jan 16 2019 at 20:26, on Zulip):

hi @pnkfelix

Santiago Pastorino (Jan 16 2019 at 20:26, on Zulip):

welcome back

Santiago Pastorino (Jan 16 2019 at 20:27, on Zulip):

I couldn’t fill the paper doc about what I’ve done, will do later

Santiago Pastorino (Jan 16 2019 at 20:28, on Zulip):

I’m on vacations 🏖

nikomatsakis (Jan 16 2019 at 20:30, on Zulip):

um shoot

nikomatsakis (Jan 16 2019 at 20:30, on Zulip):

I can't really attend for 1hr today

nikomatsakis (Jan 16 2019 at 20:30, on Zulip):

:)

nikomatsakis (Jan 16 2019 at 20:30, on Zulip):

I just realized

pnkfelix (Jan 16 2019 at 20:30, on Zulip):

should we not do steering mtg today then?

pnkfelix (Jan 16 2019 at 20:30, on Zulip):

I haven't done any triage

nikomatsakis (Jan 16 2019 at 20:31, on Zulip):

well let's se

nikomatsakis (Jan 16 2019 at 20:31, on Zulip):

I could imagine trying to discuss e.g. @Matthew Jasper's borowing PR

nikomatsakis (Jan 16 2019 at 20:32, on Zulip):

or something like that

nikomatsakis (Jan 16 2019 at 20:32, on Zulip):

else we could do some triage and do that another time -- either next week or schedule a meeting for it

nikomatsakis (Jan 16 2019 at 20:32, on Zulip):

(or just async)

pnkfelix (Jan 16 2019 at 20:32, on Zulip):

what is @Matthew Jasper 's borrowing PR ?

Matthew Jasper (Jan 16 2019 at 20:33, on Zulip):

#57609 ?

nikomatsakis (Jan 16 2019 at 20:33, on Zulip):

I was referring to https://github.com/rust-lang/rust/pull/57609

nikomatsakis (Jan 16 2019 at 20:33, on Zulip):

yes

nikomatsakis (Jan 16 2019 at 20:34, on Zulip):

sorry I had some stuff come up today

nikomatsakis (Jan 16 2019 at 20:34, on Zulip):

trying to get my thoughts together

nikomatsakis (Jan 16 2019 at 20:34, on Zulip):

so before we said

- how to organize now that NLL has shipped and it’s the new year?
- what are our big goals? (e.g., finishing the migration, polonius, plus other related features)

nikomatsakis (Jan 16 2019 at 20:35, on Zulip):

but yeah maybe that wants 1hr to really get into?

davidtwco (Jan 16 2019 at 20:36, on Zulip):

:wave:

pnkfelix (Jan 16 2019 at 20:37, on Zulip):

@Matthew Jasper did that crater run just fail spuriously 100% of the time? What happened there?

Matthew Jasper (Jan 16 2019 at 20:37, on Zulip):

#53114 and #48225 are nominated. Re: the second point I'm trying to get to the point where we can enable migrate mode on the 2015 edition, and have some things that I want to discuss at some point soon for that.

pnkfelix (Jan 16 2019 at 20:37, on Zulip):

(my Q was regarding #57609 )

nikomatsakis (Jan 16 2019 at 20:38, on Zulip):

oh, I see @Matthew Jasper's writeup comment now for #57609

nikomatsakis (Jan 16 2019 at 20:39, on Zulip):

my sense is maybe that we should not dig into #57609 in detail here but rather schedule a dedicated slot

nikomatsakis (Jan 16 2019 at 20:39, on Zulip):

and we might as well do a bit of triage

nikomatsakis (Jan 16 2019 at 20:39, on Zulip):

just because of time :)

Matthew Jasper (Jan 16 2019 at 20:39, on Zulip):

no: see the full report: https://crater-reports.s3.amazonaws.com/pr-57609/full.html

nikomatsakis (Jan 16 2019 at 20:39, on Zulip):

I'd be up to try and do that on friday though if that works for people, or else next week during this meeting

nikomatsakis (Jan 16 2019 at 20:39, on Zulip):

I do think we need to spend some dedicated time on the 2PB questions and I feel like it's been hard for me to carve time, so I want to schedule it

pnkfelix (Jan 16 2019 at 20:40, on Zulip):

Oh okay thanks @Matthew Jasper

lqd (Jan 16 2019 at 20:40, on Zulip):

re: #48225 we saw this getopts behaviour in the NLL crater runs, my understanding is it seemed to be "fixed by NLL", however one of Felix's examples in there seemed subtly different that maybe it deserves discussion/investigation

nikomatsakis (Jan 16 2019 at 20:40, on Zulip):

should we start there then?

nikomatsakis (Jan 16 2019 at 20:40, on Zulip):

I do remember there being a getopts "fixed" bug

lqd (Jan 16 2019 at 20:41, on Zulip):

yeah

nikomatsakis (Jan 16 2019 at 20:41, on Zulip):

From Felix's comment:

ah I think the reduced test case compiles because it reduces the intervening "read" of state to match state { State::A => false } and it looks like MIR borrowck manages to ... treat that as a no-op?

I think that is legit for NLL to do -- there is no "test" happening

nikomatsakis (Jan 16 2019 at 20:41, on Zulip):

similarly I expect match () { () => ... } to be "nothing"

nikomatsakis (Jan 16 2019 at 20:42, on Zulip):

and yeah the example from my comment looks .. familiar, I think that was some "fixed by NLL" case

pnkfelix (Jan 16 2019 at 20:42, on Zulip):

I don't remember if I ended up looking at the non-reduced original getopts test input

lqd (Jan 16 2019 at 20:42, on Zulip):

how do y'all feel about the 2nd example here ?

Matthew Jasper (Jan 16 2019 at 20:43, on Zulip):

It looks fine.

pnkfelix (Jan 16 2019 at 20:43, on Zulip):

oh I guess we did look at non-reduced getopts and decided that this should never have compiled (even under AST-borrowck)?

nikomatsakis (Jan 16 2019 at 20:44, on Zulip):

I basically think we should close this bug, yeah. everything seems fine

lqd (Jan 16 2019 at 20:44, on Zulip):

it came up in the run here for example

lqd (Jan 16 2019 at 20:45, on Zulip):

(from my 4th run results PR which wasn't merged I guess — yeah let's close :)

nikomatsakis (Jan 16 2019 at 20:46, on Zulip):

Closed

nikomatsakis (Jan 16 2019 at 20:46, on Zulip):

What about https://github.com/rust-lang/rust/issues/53114 ?

nikomatsakis (Jan 16 2019 at 20:47, on Zulip):

Comment here from Felix

nikomatsakis (Jan 16 2019 at 20:47, on Zulip):

Seems like this particular issue can be closed, though https://github.com/rust-lang/rust/issues/54003 remains open (and was spun off from that)

lqd (Jan 16 2019 at 20:48, on Zulip):

@Matthew Jasper did you want a test for #53114 here ?

nikomatsakis (Jan 16 2019 at 20:48, on Zulip):

i.e., if this issue is specific to match foo { _ => ... }, then I think it's "ok" and we have some tests for it, though I should look more closely at those tests

pnkfelix (Jan 16 2019 at 20:48, on Zulip):

yeah there's still the issue of needing tests for e.g. let _ = ...;

Matthew Jasper (Jan 16 2019 at 20:48, on Zulip):

Or changed to E-needstest

nikomatsakis (Jan 16 2019 at 20:49, on Zulip):

yeah there's still the issue of needing tests for e.g. let _ = ...;

oh, I see, #54003 is really specific to the unsafety check

nikomatsakis (Jan 16 2019 at 20:49, on Zulip):

ok I guess it's E-needstest

nikomatsakis (Jan 16 2019 at 20:49, on Zulip):

maybe somebody wants to knock that out this week? :) good easy starting point :)

nikomatsakis (Jan 16 2019 at 20:49, on Zulip):

or we can tweet it perhaps and try to mentor

lqd (Jan 16 2019 at 20:50, on Zulip):

mentoring could be nice, otherwise I can take it

nikomatsakis (Jan 16 2019 at 20:52, on Zulip):

I think part of my motivation for nominating this particular issue was trying to close out the NLL-complete label

pnkfelix (Jan 16 2019 at 20:53, on Zulip):

maybe we can use a different label here

pnkfelix (Jan 16 2019 at 20:53, on Zulip):

well

nikomatsakis (Jan 16 2019 at 20:54, on Zulip):

let's just close the dang issue :)

nikomatsakis (Jan 16 2019 at 20:54, on Zulip):

maybe we want to use last few minutes to talk over when to talk over the 2PB stuff?

nikomatsakis (Jan 16 2019 at 20:54, on Zulip):

do others agree some kind of "focused time" on this would be good?

pnkfelix (Jan 16 2019 at 20:54, on Zulip):

(if you close it I'll just reopen it and self-assign again. If I notice. :wink:)

nikomatsakis (Jan 16 2019 at 20:54, on Zulip):

heh, I meant "by adding the test"

pnkfelix (Jan 16 2019 at 20:54, on Zulip):

oh yeah

lqd (Jan 16 2019 at 20:54, on Zulip):

focused time for 2PB apart from the other "steering topics" ?

nikomatsakis (Jan 16 2019 at 20:55, on Zulip):

right, specifically on the "future proofing" question and how to achieve it

nikomatsakis (Jan 16 2019 at 20:55, on Zulip):

maybe the conversation can be conducted async

nikomatsakis (Jan 16 2019 at 20:55, on Zulip):

I guess one way to do it is that I can just add some time in my calendar to force myself to do it :P

nikomatsakis (Jan 16 2019 at 20:56, on Zulip):

specifically, I think we need to:

Matthew Jasper (Jan 16 2019 at 20:56, on Zulip):

Pretty much.

nikomatsakis (Jan 16 2019 at 20:56, on Zulip):

I think an auxiliary item is that it'd be great to start writing up why we desugar matches the way we do as part of a design document etc

pnkfelix (Jan 16 2019 at 20:56, on Zulip):

is there already a T-compiler steering mtg this friday?

nikomatsakis (Jan 16 2019 at 20:56, on Zulip):

there is one scheduled

pnkfelix (Jan 16 2019 at 20:56, on Zulip):

k

nikomatsakis (Jan 16 2019 at 20:57, on Zulip):

well, I have to review the PR gain and peruse @Matthew Jasper's comment, maybe we can just discuss it in the relevant zulip topic for now

Matthew Jasper (Jan 16 2019 at 20:58, on Zulip):

sounds good.

nikomatsakis (Jan 16 2019 at 20:59, on Zulip):

anything else ? I have to run sadly

lqd (Jan 16 2019 at 21:00, on Zulip):

do we know when we want to do the 1hr NLL steering meeting ?

Matthew Jasper (Jan 16 2019 at 21:00, on Zulip):

I'm trying to get to the point where we can enable migrate mode on the 2015 edition, and have some things that I want to discuss at some point soon for that.

Should I create a topic for this, so that people have some time to see what I'm thinking here.

nikomatsakis (Jan 16 2019 at 21:01, on Zulip):

that sounds good, I missed that comment

Matthew Jasper (Jan 16 2019 at 21:01, on Zulip):

and we can discuss next week or whenever

lqd (Jan 16 2019 at 21:02, on Zulip):

in case we're done, have a nice afternoon/evening everyone :wave:

Last update: Nov 22 2019 at 00:15UTC