Stream: t-compiler/wg-nll

Topic: weekly meeting September 4


pnkfelix (Sep 04 2018 at 19:23, on Zulip):

hi @WG-compiler-nll ; meeting will be starting here in about 7 minutes

nikomatsakis (Sep 04 2018 at 19:30, on Zulip):

:wave:

pnkfelix (Sep 04 2018 at 19:32, on Zulip):

(sorry all, I'm a little distracted because my son won't sleep)

pnkfelix (Sep 04 2018 at 19:32, on Zulip):

(but I'll try to start driving now)

pnkfelix (Sep 04 2018 at 19:33, on Zulip):

so as usual we have our NLL triage Paper

pnkfelix (Sep 04 2018 at 19:34, on Zulip):

I'd like to suggest that we try to get through all the nominated NLL issues today

nikomatsakis (Sep 04 2018 at 19:34, on Zulip):

sounds reasonable; I was going to suggest something similar

pnkfelix (Sep 04 2018 at 19:34, on Zulip):

(and clear the I-nominated tag so that we'll stop repeatedly looking at them)

pnkfelix (Sep 04 2018 at 19:34, on Zulip):

;)

nikomatsakis (Sep 04 2018 at 19:34, on Zulip):

I didn't have a chance yet today to look for untriaged stuff

pnkfelix (Sep 04 2018 at 19:35, on Zulip):

the link for the NLL nominated issues

nikomatsakis (Sep 04 2018 at 19:35, on Zulip):

/me still working through backlog

pnkfelix (Sep 04 2018 at 19:35, on Zulip):

I'll start with #53764

pnkfelix (Sep 04 2018 at 19:35, on Zulip):

I nominated this because I wanted to discuss what priority it should get

pnkfelix (Sep 04 2018 at 19:36, on Zulip):

(its the thing I've been working on trying to resolve)

pnkfelix (Sep 04 2018 at 19:36, on Zulip):

(by porting as many run-pass tests as possible to ui/ tests)

pnkfelix (Sep 04 2018 at 19:37, on Zulip):

any objections to me marking this P-high and putting it on ... well I don't know if Edition RC is realistic at this point, but how about RC 2?

nikomatsakis (Sep 04 2018 at 19:37, on Zulip):

I think RC2 seems reasonable

nikomatsakis (Sep 04 2018 at 19:37, on Zulip):

I feel like we should do it before we ship

nikomatsakis (Sep 04 2018 at 19:37, on Zulip):

I don't have an opinion about P-high or not

nikomatsakis (Sep 04 2018 at 19:38, on Zulip):

how much time until Edition RC ?

pnkfelix (Sep 04 2018 at 19:38, on Zulip):

Thats something I should know, but do not

Matthew Jasper (Sep 04 2018 at 19:38, on Zulip):

Milestone says it's the 13th

nikomatsakis (Sep 04 2018 at 19:39, on Zulip):

ok so ~1.5 weeks

pnkfelix (Sep 04 2018 at 19:39, on Zulip):

if you had asked me last week I would have said "oh yeah it will be done by then."

nikomatsakis (Sep 04 2018 at 19:39, on Zulip):

one thing I would hope we can do (but we'll have to move fast) is to make sure we've set our Edition RC goals

nikomatsakis (Sep 04 2018 at 19:39, on Zulip):

I think we're doing pretty ok...

pnkfelix (Sep 04 2018 at 19:40, on Zulip):

"set our edition goals" as in, formally state them?

nikomatsakis (Sep 04 2018 at 19:40, on Zulip):

ideally we'd have triaged those before this mtg :)

nikomatsakis (Sep 04 2018 at 19:40, on Zulip):

sorry, I think I meant "triage the list of Edition RC issues and bump and/or ensure everything is on path to completion"

pnkfelix (Sep 04 2018 at 19:40, on Zulip):

ah okay. Yes that's something I wish I had spent some time on today

nikomatsakis (Sep 04 2018 at 19:40, on Zulip):

anyway, I think we should put that run-pass issue for RC2, as suggested

nikomatsakis (Sep 04 2018 at 19:40, on Zulip):

ah well we can try to do that tomorrow

nikomatsakis (Sep 04 2018 at 19:41, on Zulip):

I think we're in decent shape

pnkfelix (Sep 04 2018 at 19:41, on Zulip):

the meeting time is limited so I don't think its the best time to attempt (re)triage of the list of Edition RC issues

nikomatsakis (Sep 04 2018 at 19:41, on Zulip):

agreed

pnkfelix (Sep 04 2018 at 19:41, on Zulip):

anyway, next nominated issue: #53569

nikomatsakis (Sep 04 2018 at 19:41, on Zulip):

so I believe I nominated this because I wanted to discuss the best fix

nikomatsakis (Sep 04 2018 at 19:42, on Zulip):

I think that @Matthew Jasper made a topic for it even

nikomatsakis (Sep 04 2018 at 19:42, on Zulip):

but I don't know that we followed up :)

Matthew Jasper (Sep 04 2018 at 19:42, on Zulip):

Yes, there is a topic with a suggestion in it.

nikomatsakis (Sep 04 2018 at 19:42, on Zulip):

this topic here

pnkfelix (Sep 04 2018 at 19:42, on Zulip):

yeah, https://rust-lang.zulipchat.com/#narrow/stream/122657-wg-nll/subject/.2353569.20even.20more.20dangly.20paths

nikomatsakis (Sep 04 2018 at 19:43, on Zulip):

I think we can prob un-nominate for now

nikomatsakis (Sep 04 2018 at 19:43, on Zulip):

I just added a to-do item to look at @Matthew Jasper 's suggestion (to my personal todo list)

pnkfelix (Sep 04 2018 at 19:43, on Zulip):

should it block the Edition RC?

pnkfelix (Sep 04 2018 at 19:43, on Zulip):

(I know we just said we wouldn't talk about that)

nikomatsakis (Sep 04 2018 at 19:43, on Zulip):

heh :)

pnkfelix (Sep 04 2018 at 19:43, on Zulip):

(but ... I figured if its already in mental cache)

nikomatsakis (Sep 04 2018 at 19:43, on Zulip):

I think my sense is it is a "nice to have"

Santiago Pastorino (Sep 04 2018 at 19:44, on Zulip):

I there anything I can help with that it's urgent?, gonna be back to rustc tomorrow and the thing I have on my plate is a refactor, which I'd say is not urgent

pnkfelix (Sep 04 2018 at 19:44, on Zulip):

Its one of those cases where I'd say the migration path gives us breathing room

nikomatsakis (Sep 04 2018 at 19:44, on Zulip):

we should definitely figure out a plan and start working on it now but don't have to freak out ..

pnkfelix (Sep 04 2018 at 19:44, on Zulip):

i.e. it will only warn in short term on edition=2018

nikomatsakis (Sep 04 2018 at 19:44, on Zulip):

I there anything I can help with that it's urgent?, gonna be back to rustc tomorrow and the thing I have on my plate is a refactor, which I'd say is not urgent

unclear as of yet, I guess

pnkfelix (Sep 04 2018 at 19:45, on Zulip):

okay lets try to settle the Milestone Q for that one tomorrow, I guess

pnkfelix (Sep 04 2018 at 19:45, on Zulip):

next #53528

Basile Desloges (Sep 04 2018 at 19:46, on Zulip):

FYI I tried to start working on this today.

pnkfelix (Sep 04 2018 at 19:46, on Zulip):

hmm

Basile Desloges (Sep 04 2018 at 19:46, on Zulip):

It seems linked to #53450

nikomatsakis (Sep 04 2018 at 19:47, on Zulip):

this is a weird sort of issue

nikomatsakis (Sep 04 2018 at 19:47, on Zulip):

not obviously an "NLL bug" exactly

pnkfelix (Sep 04 2018 at 19:47, on Zulip):

which in turn may be a variance issue, as niko noted

Matthew Jasper (Sep 04 2018 at 19:47, on Zulip):

I think this is a case of dropck respecting opaqueness and us not giving the "value later used here" for Drops of temporaries.

Basile Desloges (Sep 04 2018 at 19:47, on Zulip):

And I also encountered #52706 while trying to have a minimal case

nikomatsakis (Sep 04 2018 at 19:48, on Zulip):

I think this is a case of dropck respecting opaqueness and us not giving the "value later used here" for Drops of temporaries.

yeah so it might not be variance, it might be the "outlives" code I guess

pnkfelix (Sep 04 2018 at 19:48, on Zulip):

I'll assign it to myself to at least investigate further tomorrow

nikomatsakis (Sep 04 2018 at 19:48, on Zulip):

seems like we need to figure out precisely what the problem is, yes

pnkfelix (Sep 04 2018 at 19:49, on Zulip):

next: #53121

nikomatsakis (Sep 04 2018 at 19:49, on Zulip):

so this is another one that I nominated because the right fix is not entirely obvious

nikomatsakis (Sep 04 2018 at 19:49, on Zulip):

though I've had .. some thoughts since then, I think I noted down some of them

pnkfelix (Sep 04 2018 at 19:50, on Zulip):

your comments don't elaborate where the dependence on crate-type=lib comes from

nikomatsakis (Sep 04 2018 at 19:50, on Zulip):

we can probably un-nominate for now and I can try to write out a plan and maybe we can find someone to take it on? I'm not sure how easy/hard it's gonna be though, kind of a pain. The problem here is that if you have something like <T as Foo<'0>>::Bar: '1, where '0 and '1 are inference variables, you don't know whether you will need a constraint that '0: '1 or not.

nikomatsakis (Sep 04 2018 at 19:50, on Zulip):

oh, I think that's a red herring

pnkfelix (Sep 04 2018 at 19:51, on Zulip):

If you have any insight there, I'd appreciate it

pnkfelix (Sep 04 2018 at 19:51, on Zulip):

sure, but it definitely seems ... goofy ...

nikomatsakis (Sep 04 2018 at 19:51, on Zulip):

yeah so.. it is :)

nikomatsakis (Sep 04 2018 at 19:51, on Zulip):

what happens is that

nikomatsakis (Sep 04 2018 at 19:51, on Zulip):

iirc

pnkfelix (Sep 04 2018 at 19:51, on Zulip):

actually

pnkfelix (Sep 04 2018 at 19:51, on Zulip):

lets not take time on it now

nikomatsakis (Sep 04 2018 at 19:51, on Zulip):

yeah — basically we wind up requesting the optimized MIR of something we wouldn't otherwise — I forget the details

nikomatsakis (Sep 04 2018 at 19:51, on Zulip):

a bit weird

pnkfelix (Sep 04 2018 at 19:52, on Zulip):

next #52934

pnkfelix (Sep 04 2018 at 19:52, on Zulip):

this doesn't need to be nominated anymore, I think.

nikomatsakis (Sep 04 2018 at 19:52, on Zulip):

oh, you did a lot of bug minimzing

pnkfelix (Sep 04 2018 at 19:53, on Zulip):

though I wouldn't mind someone sanity check the reasoning I presented in #53948

pnkfelix (Sep 04 2018 at 19:53, on Zulip):

(I mean, that would essentially amount to a review. :wink:)

nikomatsakis (Sep 04 2018 at 19:53, on Zulip):

I'll put that on my to do list too

pnkfelix (Sep 04 2018 at 19:53, on Zulip):

likewise the nomination of #52768 can be cleared

pnkfelix (Sep 04 2018 at 19:54, on Zulip):

next: #52706

pnkfelix (Sep 04 2018 at 19:54, on Zulip):

if this manifests without NLL, it might not be our issue to resolve

pnkfelix (Sep 04 2018 at 19:55, on Zulip):

I'll assign to self to invesigate, and leave nominated for discussion at T-compiler mtg

nikomatsakis (Sep 04 2018 at 19:55, on Zulip):

if this manifests without NLL, it might not be our issue to resolve

ah, yes, I may have missed that

pnkfelix (Sep 04 2018 at 19:55, on Zulip):

clearing nomination tag on #50467

pnkfelix (Sep 04 2018 at 19:56, on Zulip):

oh wait, maybe this ICE wasn't fixed?

pnkfelix (Sep 04 2018 at 19:57, on Zulip):

well I'll try to figure out what the original ICE by backtracking, if I have spare cycles.

lqd (Sep 04 2018 at 19:57, on Zulip):

the nomination for this seems to be from july

pnkfelix (Sep 04 2018 at 19:57, on Zulip):

but I'm going to leave the issue closed and the nomination tag cleared; there's no reason to make people spend time on it

pnkfelix (Sep 04 2018 at 19:58, on Zulip):

right, I was going to blindly clear the tag becasue the issue was closed

pnkfelix (Sep 04 2018 at 19:58, on Zulip):

(and the nomination was so old)

nikomatsakis (Sep 04 2018 at 19:58, on Zulip):

seems fine

pnkfelix (Sep 04 2018 at 19:58, on Zulip):

but the recent comment from simonvandel indicates that they are ... well, I wouldn't say that they are skeptical. Just pointing out that the actual bug was never identified.)

pnkfelix (Sep 04 2018 at 19:59, on Zulip):

okay next: #47349

pnkfelix (Sep 04 2018 at 19:59, on Zulip):

okay @nikomatsakis can you summarize the nominaton and status for #47349?

nikomatsakis (Sep 04 2018 at 20:00, on Zulip):

ok well

nikomatsakis (Sep 04 2018 at 20:00, on Zulip):

this is a somewhat common source of regressions in real crates

nikomatsakis (Sep 04 2018 at 20:00, on Zulip):

the old AST checker was unsound here

nikomatsakis (Sep 04 2018 at 20:00, on Zulip):

however, the code (in most cases) is actually fine

nikomatsakis (Sep 04 2018 at 20:01, on Zulip):

it's for scenarios like vec[i] += vec[j]

nikomatsakis (Sep 04 2018 at 20:01, on Zulip):

probably the only way to accept the code would be an aggressive expansion of 2PB, which we decided was out of scope

nikomatsakis (Sep 04 2018 at 20:01, on Zulip):

so the question was can we use better diagnostics?

nikomatsakis (Sep 04 2018 at 20:01, on Zulip):

@Keith Yeung and I dug in a bit

nikomatsakis (Sep 04 2018 at 20:02, on Zulip):

we found that not so easy to do

nikomatsakis (Sep 04 2018 at 20:02, on Zulip):

basically, this is a kind of "code motion" question — can you pull the vec[j] up so it occurs before vec[i] is executed?

nikomatsakis (Sep 04 2018 at 20:02, on Zulip):

of course, maybe we don't have to be very precise about it

pnkfelix (Sep 04 2018 at 20:02, on Zulip):

the milder diagnostic you suggested in your comment, ""note: consider moving this expression so that it occurs before the mutable borrow begins", just seems like it would be very generally applicable

nikomatsakis (Sep 04 2018 at 20:02, on Zulip):

yes, that too

pnkfelix (Sep 04 2018 at 20:02, on Zulip):

is this something we have to put into the note diagnostics?

pnkfelix (Sep 04 2018 at 20:03, on Zulip):

wha does the HELP for this class of error messages say?

nikomatsakis (Sep 04 2018 at 20:03, on Zulip):

so we have to sort of pick between:

1. do nothing
2. maybe offer helpful tips..kind of all the time?
3. maybe some other heuristic?
4. put in a bunch of effort to give a more targeted heuristic?

pnkfelix (Sep 04 2018 at 20:03, on Zulip):

in other words, I imgagine that people used to working with borrowck errors are used to making such transformations?

nikomatsakis (Sep 04 2018 at 20:03, on Zulip):

wha does the HELP for this class of error messages say?

I ... think nothing?

pnkfelix (Sep 04 2018 at 20:03, on Zulip):

and so I'm musing that its begineers who need the guidance

nikomatsakis (Sep 04 2018 at 20:03, on Zulip):

you might think so

pnkfelix (Sep 04 2018 at 20:03, on Zulip):

heh.

pnkfelix (Sep 04 2018 at 20:04, on Zulip):

"you might think so" is like "may you live in interesting times"

nikomatsakis (Sep 04 2018 at 20:04, on Zulip):

I feel like we might want to target index impls specifically somehow

nikomatsakis (Sep 04 2018 at 20:04, on Zulip):

I guess that's a way to narrow it down

nikomatsakis (Sep 04 2018 at 20:04, on Zulip):

because it's specifically around things like foo[i] += where the problems arise..

pnkfelix (Sep 04 2018 at 20:04, on Zulip):

I personally am unconvinced that this is worth trying to force into the RC milestone.

nikomatsakis (Sep 04 2018 at 20:04, on Zulip):

that desugars to AddAssign(&mut foo[i], ...) essentially

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

yeah, I think I too am leaning towards "give up for now"

pnkfelix (Sep 04 2018 at 20:05, on Zulip):

I'd call it "nice to have", but I just ... am not seeing the motivation as being as strong as other cases in the RC milestone

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

definitely not for RC

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

it seems like the sort of polish -- right, a nice to have. I guess that sums it up.

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

ok, let's defer then

pnkfelix (Sep 04 2018 at 20:05, on Zulip):

okay. the comment history shows that its milestone assignment bounced around a little

pnkfelix (Sep 04 2018 at 20:05, on Zulip):

okay. clearing nomination

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

in my ideal world, we'd figure out if the transfomration is almost ceratinly fine and then offer an automated suggestion that introduces a let =)

nikomatsakis (Sep 04 2018 at 20:05, on Zulip):

but that is a ton of work to do properly

nikomatsakis (Sep 04 2018 at 20:06, on Zulip):

but dang it'd be cool

nikomatsakis (Sep 04 2018 at 20:06, on Zulip):

(particularly if we had fix-it tooltips in VSCode)

pnkfelix (Sep 04 2018 at 20:06, on Zulip):

last but definitely not least, #47184

pnkfelix (Sep 04 2018 at 20:07, on Zulip):

it looks like things are going well here

nikomatsakis (Sep 04 2018 at 20:07, on Zulip):

yeah so with my PR we are kind of down to two things, I think...

nikomatsakis (Sep 04 2018 at 20:07, on Zulip):

let (x, y): T; <-- no initializer

nikomatsakis (Sep 04 2018 at 20:07, on Zulip):

and some cases of constants, I suspect

nikomatsakis (Sep 04 2018 at 20:08, on Zulip):

I was hoping today to look into that a bit more but didn't get to it

pnkfelix (Sep 04 2018 at 20:08, on Zulip):

FYI your PR is still failing travis

nikomatsakis (Sep 04 2018 at 20:08, on Zulip):

bother

pnkfelix (Sep 04 2018 at 20:08, on Zulip):

not that this should block my review. :)

pnkfelix (Sep 04 2018 at 20:08, on Zulip):

okay. The main Q is whether this needs to stay nominated

nikomatsakis (Sep 04 2018 at 20:08, on Zulip):

oh yeah I had started to rebase and fix that

pnkfelix (Sep 04 2018 at 20:08, on Zulip):

the milestone assignment seems right

nikomatsakis (Sep 04 2018 at 20:08, on Zulip):

I think probably nomination not necessary

pnkfelix (Sep 04 2018 at 20:09, on Zulip):

okay

pnkfelix (Sep 04 2018 at 20:09, on Zulip):

(its old, we probably talked about it last week and then forgot to clear)

pnkfelix (Sep 04 2018 at 20:09, on Zulip):

fantastic

pnkfelix (Sep 04 2018 at 20:09, on Zulip):

so that's all the nominated issues

pnkfelix (Sep 04 2018 at 20:10, on Zulip):

I unfortunately have to go afk because my toddler is still refusing to sleep (he's been playing beside me all this time, occasionally in my lap)

pnkfelix (Sep 04 2018 at 20:10, on Zulip):

so now I have to go be his jail guard.

nikomatsakis (Sep 04 2018 at 20:10, on Zulip):

:)

pnkfelix (Sep 04 2018 at 20:10, on Zulip):

bye all

lqd (Sep 04 2018 at 20:10, on Zulip):

ciao felix, and toddler :)

pnkfelix (Sep 04 2018 at 20:10, on Zulip):

(thanks for attending! and please do chat in my absence, I'll review the log sometime after he's asleep!)

lqd (Sep 04 2018 at 20:12, on Zulip):

is there anything urgent we need to do this week ?

simulacrum (Sep 04 2018 at 20:13, on Zulip):

NLL needs to be stabilized if it is to make it for RC

lqd (Sep 04 2018 at 20:16, on Zulip):

stabilized as in, not requiring nightly, features, the edition preview flags, and so on, or stable as bugfixing ?

simulacrum (Sep 04 2018 at 20:17, on Zulip):

Not requiring nightly / feature flags

simulacrum (Sep 04 2018 at 20:18, on Zulip):

We essentially need to stabilize everything in rust_2018_preview , which currently includes NLL

lqd (Sep 04 2018 at 20:19, on Zulip):

sounds like that would basicaly be similar to what was done for the crater run, modulo I'm assuming mir borrowck should be stabilized in migrate mode

simulacrum (Sep 04 2018 at 20:19, on Zulip):

sure, yeah -- we'll just want to land that PR or a similar one

lqd (Sep 04 2018 at 20:19, on Zulip):

alright I can do that

lqd (Sep 04 2018 at 20:20, on Zulip):

do we have an issue for this maybe ?

simulacrum (Sep 04 2018 at 20:21, on Zulip):

We sort of do, let me find it

simulacrum (Sep 04 2018 at 20:22, on Zulip):

https://github.com/rust-lang/rust/issues/43234#issuecomment-410263039

simulacrum (Sep 04 2018 at 20:23, on Zulip):

@nikomatsakis Could we resolve some of the concerns there perhaps? NLL is a critical part of the edition feature set and we're announcing RC1 next week hopefully

lqd (Sep 04 2018 at 20:23, on Zulip):

stabilization seems to depend on #47184 which we talked about earlier

simulacrum (Sep 04 2018 at 20:25, on Zulip):

yeah that looks like one of if not the critical blocker

simulacrum (Sep 04 2018 at 20:26, on Zulip):

hopefully we have some progress on resolving that

lqd (Sep 04 2018 at 20:27, on Zulip):

yeah niko has a PR for parts of it

lqd (Sep 04 2018 at 20:28, on Zulip):

as for the "perf comparison vs incremental" concern, could we/would we need to track it on the perf NLL dashboard ?

lqd (Sep 04 2018 at 20:28, on Zulip):

or would this just be a one-off like when we compared to modern crate versions before ?

simulacrum (Sep 04 2018 at 20:29, on Zulip):

Depending on whether we're stabilizing NLL in both 2015 and 2018, I think

simulacrum (Sep 04 2018 at 20:29, on Zulip):

If it's just 2018 then comparison becomes harder but if both then it'll show up on normal compare runs

lqd (Sep 04 2018 at 20:29, on Zulip):

I think the plan is 2018 for now, see how the migrate mode goes, then in some time flip it on for 2015

simulacrum (Sep 04 2018 at 20:29, on Zulip):

hm, well, we can for sure flip it on for both in a PR and then view compare.html output

simulacrum (Sep 04 2018 at 20:30, on Zulip):

that'll give us the comparison data we need

lqd (Sep 04 2018 at 20:30, on Zulip):

could we use the try build we did for the crater run ? or would that be too old maybe (and perf is still being worked on as we speak, for example the ucd case)

memoryruins (Sep 04 2018 at 20:31, on Zulip):

before running the next crater, #53745 should be merged

lqd (Sep 04 2018 at 20:31, on Zulip):

(the try build may be a bit old https://github.com/rust-lang/rust/pull/53426#issuecomment-413576495)

simulacrum (Sep 04 2018 at 20:31, on Zulip):

I think we can probably reuse the same PR -- just rerun try on it

lqd (Sep 04 2018 at 20:31, on Zulip):

yeah no crater for a while

lqd (Sep 04 2018 at 20:32, on Zulip):

the ICEs are fixed but IIRC niko wanted a bit more of the #47184 work to land before doing another crater run

simulacrum (Sep 04 2018 at 20:32, on Zulip):

If you could rebase the PR and repush that'd be great, then we can queue perf for it

lqd (Sep 04 2018 at 20:33, on Zulip):

sure

lqd (Sep 04 2018 at 20:33, on Zulip):

probably tomorrow if that's ok :)

simulacrum (Sep 04 2018 at 20:33, on Zulip):

yeah, shouldn't be a problem

lqd (Sep 04 2018 at 20:34, on Zulip):

did anyone want to talk about something else in particular ?

lqd (Sep 04 2018 at 20:36, on Zulip):

@simulacrum do you happen to know if mw wanted to see perf in migrate mode ? that would still run AST borrowck IIRC

lqd (Sep 04 2018 at 20:36, on Zulip):

I guess they wouldn't

lqd (Sep 04 2018 at 20:38, on Zulip):

and if not, good afternoon/evening everyone :) :wave:

memoryruins (Sep 04 2018 at 20:44, on Zulip):

you too ^^

Keith Yeung (Sep 04 2018 at 20:51, on Zulip):

is there anything else that we need to work on?

Keith Yeung (Sep 04 2018 at 20:55, on Zulip):

hmm... i'm guessing the edition RC2 issues would be a start

simulacrum (Sep 04 2018 at 22:04, on Zulip):

I think probably whatever we're stabilizing is what we should be testing

simulacrum (Sep 04 2018 at 22:04, on Zulip):

Basically we don't want beta to be a serious performance regression

nikomatsakis (Sep 05 2018 at 12:55, on Zulip):

@simulacrum sorry, missed these comments yesterday. I forgot that we planned to do the "double beta" trick.

nikomatsakis (Sep 05 2018 at 12:55, on Zulip):

I think the plan is 2018 for now, see how the migrate mode goes, then in some time flip it on for 2015

hmm, this is an interesting question

Last update: Nov 21 2019 at 13:05UTC