Stream: t-compiler/wg-nll

Topic: weekly meeting June 26


nikomatsakis (Jun 26 2018 at 19:35, on Zulip):

:wave:

qmx (Jun 26 2018 at 19:36, on Zulip):

I've been swamped with work, and besides that been working on some non-nll things (edition-related stuff)

qmx (Jun 26 2018 at 19:36, on Zulip):

so just lurking here

nikomatsakis (Jun 26 2018 at 19:37, on Zulip):

ok so

nikomatsakis (Jun 26 2018 at 19:37, on Zulip):

a couple of things

nikomatsakis (Jun 26 2018 at 19:37, on Zulip):

first of all, you may have noticed that @pnkfelix hasn't been around — for personal reasons, he's got to be away, not sure when he'll be back

nikomatsakis (Jun 26 2018 at 19:38, on Zulip):

hopefully soon

nikomatsakis (Jun 26 2018 at 19:38, on Zulip):

secondly, I spent a long time this morning doing profiles and investigations

nikomatsakis (Jun 26 2018 at 19:38, on Zulip):

if you go to the "next steps" section of the triage document

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

you'll see that I wrote out a bunch of ways we can improve performance

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

mostly small, but afaict there is no single "home run" available to us at this point

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

at the moment the most promising thing seems to be making a series of small improvements

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

but I think there are still plenty of opportunities

Santiago Pastorino (Jun 26 2018 at 19:39, on Zulip):

cool

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

and maybe once we're done, we'll see some bigger things we can do

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

or we'll just be done :)

Santiago Pastorino (Jun 26 2018 at 19:39, on Zulip):

are you going to open issues about all this stuff?

nikomatsakis (Jun 26 2018 at 19:39, on Zulip):

already done

nikomatsakis (Jun 26 2018 at 19:40, on Zulip):

they are linke from the document

Santiago Pastorino (Jun 26 2018 at 19:40, on Zulip):

ahh I see the document now

nikomatsakis (Jun 26 2018 at 19:40, on Zulip):

or available with the NLL-performant label

nikomatsakis (Jun 26 2018 at 19:40, on Zulip):

you're assigned to one of them @Santiago Pastorino :)

Santiago Pastorino (Jun 26 2018 at 19:40, on Zulip):

yep :)

Santiago Pastorino (Jun 26 2018 at 19:41, on Zulip):

are you going to keep profiling stuff?

nikomatsakis (Jun 26 2018 at 19:41, on Zulip):

I would like to dig into at least one other test case

nikomatsakis (Jun 26 2018 at 19:41, on Zulip):

maybe a few more

nikomatsakis (Jun 26 2018 at 19:41, on Zulip):

this is all from clap-rs

nikomatsakis (Jun 26 2018 at 19:41, on Zulip):

another thing is that I want to start making a big push to grow the WG

nikomatsakis (Jun 26 2018 at 19:42, on Zulip):

my sense is that -- for various reasons -- we've had a kind of attrition over time

nikomatsakis (Jun 26 2018 at 19:42, on Zulip):

such that the set of active hacking people is starting to get low :)

Santiago Pastorino (Jun 26 2018 at 19:42, on Zulip):

:(

nikomatsakis (Jun 26 2018 at 19:42, on Zulip):

if you have any suggestions, I'm all ears, my first step was just this tweet :)

https://twitter.com/nikomatsakis/status/1011681895221022720

nikomatsakis (Jun 26 2018 at 19:42, on Zulip):

not just if any people are here because of it (if so, Hi!)

Santiago Pastorino (Jun 26 2018 at 19:42, on Zulip):

I would like to dig into at least one other test case

I've asked because I'd like to see you going over stuff, may be helpful

nikomatsakis (Jun 26 2018 at 19:43, on Zulip):

yeah would be happy to do it "live" with a few others

nikomatsakis (Jun 26 2018 at 19:43, on Zulip):

I was actually contemplating that just for fun

davidtwco (Jun 26 2018 at 19:43, on Zulip):

I'd like to start picking up some more tasks, not taken on a lot in the past few weeks, I've reckon I've got the bandwidth for more.

nikomatsakis (Jun 26 2018 at 19:43, on Zulip):

ok, sounds great, main thing is: I would say don't sign up for stuff unless you are actively hacking on it

nikomatsakis (Jun 26 2018 at 19:44, on Zulip):

to keep the pool of things open for otherse

nikomatsakis (Jun 26 2018 at 19:44, on Zulip):

I'm sure we can create more tasks if you feel like you are running low

nikomatsakis (Jun 26 2018 at 19:44, on Zulip):

one challenge here is that I've been finding it hard to keep up with writing mentoring instructions, but I decided to just do a less good job ;)

lqd (Jun 26 2018 at 19:44, on Zulip):

could be nice if the official rust-lang account / Steve retweeted it for bigger reach

nikomatsakis (Jun 26 2018 at 19:44, on Zulip):

that is, give a few pointers and see if people can jump in and come to the channel

nikomatsakis (Jun 26 2018 at 19:45, on Zulip):

yeah I guess steve didn't see it, seems fine, the follow-ups are prob more imp't

nikomatsakis (Jun 26 2018 at 19:45, on Zulip):

anyway, enough of that

nikomatsakis (Jun 26 2018 at 19:45, on Zulip):

I don't know if it makes sense to go over the performance stuff in more detail -- I'm guessing probably not? — but know that there are a number of issues and I think they're all worth doing

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

(and we can discuss individual things in more detail on distinct topics)

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

I didn't really get time to go over diagnostics yet

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

but there was one other thing I wanted to talk about for sure, which was the idea of doing a crater run

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

### Crater run

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

/me expriemnts with section headings

nikomatsakis (Jun 26 2018 at 19:46, on Zulip):

I think we ought to try to bootstrap rustc and do a crater run wih NLL enabled

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

and then go over the results and look for burning bugs

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

as well as "total diagnostic fails" :)

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

I am actually not sure if we can do a crater run without bootstrap working

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

and I not sure if bootstrap works (maybe it does?)

lqd (Jun 26 2018 at 19:47, on Zulip):

how would one test NLL w/ bootstrapping ?

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

but it seems like something I should know the answer to

nikomatsakis (Jun 26 2018 at 19:47, on Zulip):

well, we would want to make -Zborrowck=mir the default

lqd (Jun 26 2018 at 19:48, on Zulip):

I thought x.py check would work but I guess it's just using stage0 (ie, beta ?) — but since there was a release it might also change this

nikomatsakis (Jun 26 2018 at 19:48, on Zulip):

ah I remember duh

nikomatsakis (Jun 26 2018 at 19:48, on Zulip):

what I did before was

nikomatsakis (Jun 26 2018 at 19:48, on Zulip):

we have a variable... I think it's called RUSTFLAGS_NOT_STAGE0 or something

nikomatsakis (Jun 26 2018 at 19:48, on Zulip):

and I did RUSTFLAGS_NOT_STAGE0='-Zborrowck=mir' ./x.py build

nikomatsakis (Jun 26 2018 at 19:48, on Zulip):

cc @simulacrum are you around by any chance?

lqd (Jun 26 2018 at 19:48, on Zulip):

oh right

nikomatsakis (Jun 26 2018 at 19:49, on Zulip):

@simulacrum I'm wondering if you know whether crater can be run with custom flags

nikomatsakis (Jun 26 2018 at 19:49, on Zulip):

ideally we would decouple bootstrap from crater run

lqd (Jun 26 2018 at 19:49, on Zulip):

yeah

nikomatsakis (Jun 26 2018 at 19:49, on Zulip):

we could potentially do that by checking the RUST_BOOTSTRAP env variable or something

nikomatsakis (Jun 26 2018 at 19:49, on Zulip):

actually ideally bootstrap would work :)

lqd (Jun 26 2018 at 19:50, on Zulip):

I started testing bootstrap last week but as you can imagine, it happens I was mostly testing the old beta's NLL :3

nikomatsakis (Jun 26 2018 at 19:50, on Zulip):

I guess my question is:

Does somebody want to take charge of trying to push bootstrap through and reporting back?

lqd (Jun 26 2018 at 19:50, on Zulip):

sure I can try

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

<3

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

we should file an issue to track I guess

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

the strategy I was trying to do before was

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

to find places where bootstrap failed

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

minimize the errors

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

and then find workarounds :)

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

i.e., file an issue with a minimzed example for the future

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

we'll see how necessary that is

nikomatsakis (Jun 26 2018 at 19:51, on Zulip):

oh shoot

nikomatsakis (Jun 26 2018 at 19:52, on Zulip):

I am remembering now what the blocker was before

nikomatsakis (Jun 26 2018 at 19:52, on Zulip):

I had kind of forgotten about that

nikomatsakis (Jun 26 2018 at 19:52, on Zulip):

well, we can discuss off line :)

nikomatsakis (Jun 26 2018 at 19:52, on Zulip):

we should find out whether we can do the crater run w/ custom flags; I'll go ask in #rust-infra on IRC

lqd (Jun 26 2018 at 19:53, on Zulip):

(the env var to use for testing bootstrap is probably RUSTFLAGS_STAGE_NOT_0)

nikomatsakis (Jun 26 2018 at 19:54, on Zulip):

@lqd I filed https://github.com/rust-lang/rust/issues/51823 and assigned you =)

nikomatsakis (Jun 26 2018 at 19:55, on Zulip):

I'll find out about crater run possibilities

lqd (Jun 26 2018 at 19:55, on Zulip):

nice thank you

nikomatsakis (Jun 26 2018 at 19:55, on Zulip):

(already asked in #rust-infra)

nikomatsakis (Jun 26 2018 at 19:55, on Zulip):

ok so I guess the question is "what else" -- I know that @Santiago Pastorino is looking into that one issue, and @David Wood is looking into the region output

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

ah I see some updates from @Matthew Jasper too :)

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

I guess an open question that I'd like your thoughts on

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

is how to prioritize "completeness" and "soundness" bugs -- that is, things like ICEs, or failing to compile some code we think we should, or failing to report errors

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

I'm having a hard time deciding :)

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

I think maybe the first thing is to try and get a bit more of a survey though

nikomatsakis (Jun 26 2018 at 19:56, on Zulip):

(e.g., via crater)

nikomatsakis (Jun 26 2018 at 19:57, on Zulip):

I'd really like to get to the point where we can enable NLL warnings...performance still seems like a blocker there

nikomatsakis (Jun 26 2018 at 19:57, on Zulip):

@Matthew Jasper I reassigned your PR from pnkfelix to me, seeing as they are out (as I wrote earlier)

lqd (Jun 26 2018 at 19:58, on Zulip):

and the one ICE we talked about earlier might to be blocking some nightly users as well (even though I guess disabling NLL might have been an option)

nikomatsakis (Jun 26 2018 at 19:58, on Zulip):

I should probaly do that for the rest of their PRs

nikomatsakis (Jun 26 2018 at 19:59, on Zulip):

yeah, I guess my sense is that we should do a "triage look" over ICEs and things

lqd (Jun 26 2018 at 19:59, on Zulip):

will eddy review your huge perf PR niko since felix is unavailable ?

nikomatsakis (Jun 26 2018 at 19:59, on Zulip):

looking for ones that are either (a) blocking or affecting a lot of people or (b) will require some nontrivial refactorings to fix

nikomatsakis (Jun 26 2018 at 19:59, on Zulip):

will eddy review your huge perf PR niko since felix is unavailable ?

hopefully :stuck_out_tongue:

nikomatsakis (Jun 26 2018 at 19:59, on Zulip):

I should ping them

lqd (Jun 26 2018 at 19:59, on Zulip):

:)

nikomatsakis (Jun 26 2018 at 19:59, on Zulip):

(done)

nikomatsakis (Jun 26 2018 at 20:00, on Zulip):

OK so — here are two organizational tasks I plan to do:

nikomatsakis (Jun 26 2018 at 20:00, on Zulip):

what I may do then is announce when I'll do them them and just open up a appear.in channel for anyone who wants to hang out ;)

nikomatsakis (Jun 26 2018 at 20:00, on Zulip):

I think this sort of thing would be slow to do over zulip

qmx (Jun 26 2018 at 20:01, on Zulip):

niko, if you schedule it in advance more people can join

qmx (Jun 26 2018 at 20:01, on Zulip):

:)

nikomatsakis (Jun 26 2018 at 20:01, on Zulip):

it prob won't be today

nikomatsakis (Jun 26 2018 at 20:01, on Zulip):

I was planning on picking some times tomorrow

Jon Gjengset (Jun 26 2018 at 20:01, on Zulip):

even though I guess disabling NLL might have been an option

just a minor point on this: in my case, disabling NLL now that we've had it for a while would be quite a pain :)

nikomatsakis (Jun 26 2018 at 20:02, on Zulip):

@Jon Gjengset which is your case?

nikomatsakis (Jun 26 2018 at 20:02, on Zulip):

is that the ICE that @lqd was referring to?

lqd (Jun 26 2018 at 20:02, on Zulip):

yeah

Jon Gjengset (Jun 26 2018 at 20:03, on Zulip):

yup, https://github.com/rust-lang/rust/issues/51348#issuecomment-400398622

Jon Gjengset (Jun 26 2018 at 20:03, on Zulip):

(I have to run to a meeting, but will keep this open in case there are questions for me that I can answer later)

nikomatsakis (Jun 26 2018 at 20:04, on Zulip):

ok, I'll try to poke that that briefly

nikomatsakis (Jun 26 2018 at 20:04, on Zulip):

I'll open a dedicated Zulip topic on it and ping you from there

nikomatsakis (Jun 26 2018 at 20:04, on Zulip):

I guess we're done here with this meeting?

nikomatsakis (Jun 26 2018 at 20:04, on Zulip):

anybody have anything else on their minds?

lqd (Jun 26 2018 at 20:04, on Zulip):

what to do with the perf issues ?

nikomatsakis (Jun 26 2018 at 20:05, on Zulip):

I'd say if you see one you'd like to tackle, leave a comment and/or assign yourself, and feel free to ping me here

lqd (Jun 26 2018 at 20:05, on Zulip):

say, do we wait a bit, in case more people want to pick them up ?

nikomatsakis (Jun 26 2018 at 20:05, on Zulip):

I would not wait, if you have time and energy to hack :)

nikomatsakis (Jun 26 2018 at 20:05, on Zulip):

I'm just saying don't like go grab 2 or 3 ;)

lqd (Jun 26 2018 at 20:05, on Zulip):

:) understood

nikomatsakis (Jun 26 2018 at 20:05, on Zulip):

that said, I think there may be a "connection" between 2 of them that I was pinging you about @lqd

nikomatsakis (Jun 26 2018 at 20:06, on Zulip):

in particular #51819 and #51813 seem .. related?

nikomatsakis (Jun 26 2018 at 20:06, on Zulip):

well, maybe the thing to do would be to focus just on liveness first

lqd (Jun 26 2018 at 20:06, on Zulip):

yeah I saw you were saying it would make sense to do both at the same time

nikomatsakis (Jun 26 2018 at 20:06, on Zulip):

and then split out the do_dataflow part, which is conceptually related but -- in terms of code -- entirely independent

nikomatsakis (Jun 26 2018 at 20:06, on Zulip):

that is, I should prob split #51813 into two bugs

nikomatsakis (Jun 26 2018 at 20:07, on Zulip):

but I think what I will do

nikomatsakis (Jun 26 2018 at 20:07, on Zulip):

is to focus #51813 on do_dataflow and focus #51819 on liveness -- both its memory alloc problem + the laziness bit

lqd (Jun 26 2018 at 20:07, on Zulip):

do you think bootstrap (or crater, for someone else say @Jake Goulding wink wink) more "important" our new perf issues, seems like no ?

tmandry (Jun 26 2018 at 20:08, on Zulip):

@nikomatsakis is the nll label the right place to start for anyone new looking to making a contribution?

lqd (Jun 26 2018 at 20:09, on Zulip):

here the new ones are WG-compiler-nll + NLL-performant (the performance ones)

nikomatsakis (Jun 26 2018 at 20:10, on Zulip):

@Tyler Mandry yeah so one thing I want to do is to simplify the NLL tracking issue

nikomatsakis (Jun 26 2018 at 20:10, on Zulip):

but the NLL-performant and NLL-diagnostics labels are probably the two "main sources" of problems, though really all the NLL-foo labels are relevant

nikomatsakis (Jun 26 2018 at 20:10, on Zulip):

if you see one that interests you :)

nikomatsakis (Jun 26 2018 at 20:11, on Zulip):

do you think bootstrap (or crater, for someone else say @Jake Goulding wink wink) more "important" our new perf issues, seems like no ?

@lqd good question -- I think perf is prob more imp't. Bootstrapping in partcular is more of a "background activity", I think

nikomatsakis (Jun 26 2018 at 20:11, on Zulip):

the main work for crater run will be going through the results anyway

lqd (Jun 26 2018 at 20:11, on Zulip):

don't we have traits + NLL intersections where @Tyler Mandry could shine ?

nikomatsakis (Jun 26 2018 at 20:11, on Zulip):

ah so there are a few things I didn't get arond to filing that I just remembered

nikomatsakis (Jun 26 2018 at 20:12, on Zulip):

I want to verify, but I think that one of my PRs may have broken some of the rustc-perf test cases (notably futures)

nikomatsakis (Jun 26 2018 at 20:12, on Zulip):

it was causing ICEs but @pnkfelix r+'d anyway and I didn't notice (though it's prob not a bad thing that it landed)

nikomatsakis (Jun 26 2018 at 20:12, on Zulip):

in any case I think the fix is to introduce a new "query"

nikomatsakis (Jun 26 2018 at 20:12, on Zulip):

(in the sense of chalk queries, @Tyler Mandry)

nikomatsakis (Jun 26 2018 at 20:13, on Zulip):

I'll try to file an issue about that

nikomatsakis (Jun 26 2018 at 20:13, on Zulip):

there is also the bug on speeding up canonicalization

nikomatsakis (Jun 26 2018 at 20:14, on Zulip):

this one: https://github.com/rust-lang/rust/issues/48417

nikomatsakis (Jun 26 2018 at 20:14, on Zulip):

I just tagged it with NLL-performant

nikomatsakis (Jun 26 2018 at 20:14, on Zulip):

I'm not sure what technicalguy's story is, they didn't answer my pings over on discord

tmandry (Jun 26 2018 at 20:15, on Zulip):

okay sounds good, I'll take a closer look at that one soon. also ping me if you create the other issue.

nikomatsakis (Jun 26 2018 at 20:16, on Zulip):

ok, let me go see if there are still problems with ICEs

nikomatsakis (Jun 26 2018 at 20:16, on Zulip):

I also want to investigate @Jon Gjengset's problem

nikomatsakis (Jun 26 2018 at 20:16, on Zulip):

ok, end of meeting? :)

lqd (Jun 26 2018 at 20:16, on Zulip):

:thumbs_up: :)

nikomatsakis (Jun 26 2018 at 20:17, on Zulip):

thanks all! <3

lqd (Jun 26 2018 at 20:17, on Zulip):

cheers everyone

nikomatsakis (Jun 26 2018 at 21:08, on Zulip):

okay sounds good, I'll take a closer look at that one soon. also ping me if you create the other issue.

@Tyler Mandry so I can no longer reproduce those failures, but I feel like the underlying problem is not fixed, I think we're just missing the asserts now =) gonna try to add them back...

nikomatsakis (Jun 26 2018 at 21:09, on Zulip):

hm, maybe not

nikomatsakis (Jun 26 2018 at 21:10, on Zulip):

well I still think there's a problem and will try to write it up, but I guess the urgency is low until I find a way to have something fail ;)

Eh2406 (Jun 27 2018 at 15:00, on Zulip):

I just joined zulip to ping you, @nikomatsakis. Am I pinging correctly?

Eh2406 (Jun 27 2018 at 15:02, on Zulip):

Specifically, I may have a spare cycal to take a stab at 51821.

nikomatsakis (Jun 27 2018 at 15:16, on Zulip):

you are :)

nikomatsakis (Jun 27 2018 at 15:16, on Zulip):

great!

nikomatsakis (Jun 27 2018 at 15:16, on Zulip):

did the instructions make sense to you?

Eh2406 (Jun 27 2018 at 15:20, on Zulip):

I think 50-70%. I am working on getting an jetbrains/rls to work on the relevant files so I can fill in some of the blanks, or adeast get to where I can articulate what I don't understand.

nikomatsakis (Jun 27 2018 at 15:21, on Zulip):

ok, sounds good

Eh2406 (Jun 27 2018 at 15:23, on Zulip):

From a clean checkout I seem to be getting "failed to read C:\rust_things\rust\src\tools\clippy\Cargo.toml" thoughts? Do I need to start a build :-(

nikomatsakis (Jun 27 2018 at 15:31, on Zulip):

probably

nikomatsakis (Jun 27 2018 at 15:31, on Zulip):

and/or various git commands to update the submodules

nikomatsakis (Jun 27 2018 at 15:31, on Zulip):

I would do ./x.py check

Jake Goulding (Jun 27 2018 at 15:38, on Zulip):

maybe git submodule update

Santiago Pastorino (Jun 27 2018 at 15:51, on Zulip):

I'd suggest git submodule update --init --recursive if you're having submodule issues

Jon Gjengset (Jun 27 2018 at 15:54, on Zulip):

@nikomatsakis you mentioned you'd open a dedicated topic for https://github.com/rust-lang/rust/issues/51348. I'm new to Zulip — where would I find that topic?

Eh2406 (Jun 27 2018 at 16:08, on Zulip):

So while that runs I see 2 questions, that I may be able to answer for myself with a working editor, but you may just know:
1, how does the vec constraints end up calling add_region?
2, is ConstraintIndex use anywhere? The places IndexVec are added to from 51821 don't record it.

davidtwco (Jun 27 2018 at 16:11, on Zulip):

Just an FYI @Eh2406, you should create a new topic in the #wg-nll stream to keep any questions related to your issue in one place - it's easier to find later on and for catching up.

Eh2406 (Jun 27 2018 at 16:20, on Zulip):

@David Wood never used zulip before how do I do that, sorry.

davidtwco (Jun 27 2018 at 16:21, on Zulip):

@Eh2406 No worries, if you don't have a message to reply to selected (if you do, there's a little cross to close that above the text box on the right) then there's a "new topic" button. Then you just type a name and a message.

nikomatsakis (Jun 27 2018 at 16:54, on Zulip):

@Jon Gjengset I don't think I ever did, I think I said I would

nikomatsakis (Jun 27 2018 at 16:54, on Zulip):

but I didn't get to it :)

Jon Gjengset (Jun 27 2018 at 17:18, on Zulip):

Haha, I see. No worries! I see you managed to build an even simpler reproducing example!

nikomatsakis (Jun 27 2018 at 17:19, on Zulip):

yeah so I see what the problem is, but not the ideal fix

nikomatsakis (Jun 27 2018 at 17:19, on Zulip):

it intersects with recent work by @pnkfelix that I have to review

nikomatsakis (Jun 27 2018 at 17:20, on Zulip):

however, if you need a workaround, it seems clear you could have one

Jon Gjengset (Jun 27 2018 at 17:20, on Zulip):

Oh, yeah, we could easily re-write the code to not have that happen

Jon Gjengset (Jun 27 2018 at 17:20, on Zulip):

Now that we know what the issue is

Last update: Nov 21 2019 at 13:10UTC