Stream: t-compiler/wg-nll

Topic: weekly meeting October 2


pnkfelix (Oct 02 2018 at 19:24, on Zulip):

Just a quick heads up, @WG-compiler-nll weekly meting will being in about six minutes

lqd (Oct 02 2018 at 19:27, on Zulip):

hello everyone

pnkfelix (Oct 02 2018 at 19:31, on Zulip):

Oh, right, I suppose greeting people is nice. :smile:

pnkfelix (Oct 02 2018 at 19:31, on Zulip):

hi everyone!

pnkfelix (Oct 02 2018 at 19:31, on Zulip):

As usual we have our NLL Triage Paper

pnkfelix (Oct 02 2018 at 19:32, on Zulip):

lets start with the unassigned issues for rc2

pnkfelix (Oct 02 2018 at 19:33, on Zulip):

I'm going to tag #47184 as a metabug and tracking issue. Or at least, I think that's what its current status is ... ?

pnkfelix (Oct 02 2018 at 19:33, on Zulip):

aka "NLL should identify and respect the lifetime annotations that the user wrote" #47184

pnkfelix (Oct 02 2018 at 19:34, on Zulip):

no not a tracking issue; that's just for unstable features. Sorry.

memoryruins (Oct 02 2018 at 19:34, on Zulip):

makes sense ^^

memoryruins (Oct 02 2018 at 19:34, on Zulip):

nvm ^^

pnkfelix (Oct 02 2018 at 19:34, on Zulip):

basically, I just want some sort of tag that helps me remember that its ... okay ... that this is unassigned?

pnkfelix (Oct 02 2018 at 19:35, on Zulip):

or maybe ... I'll just assign it to myself, as a reminder to actually go through and follow up on the subissues

pnkfelix (Oct 02 2018 at 19:35, on Zulip):

next: "[nll] better error message when returning refs to upvars" #53040

pnkfelix (Oct 02 2018 at 19:35, on Zulip):

mm

pnkfelix (Oct 02 2018 at 19:36, on Zulip):

does anyone want to take point on this?

nikomatsakis (Oct 02 2018 at 19:36, on Zulip):

I was going to try and write down some notes on that one next

davidtwco (Oct 02 2018 at 19:36, on Zulip):

I can take a look.

pnkfelix (Oct 02 2018 at 19:36, on Zulip):

okay lets tentatively assign to @davidtwco , great

pnkfelix (Oct 02 2018 at 19:37, on Zulip):

next: "Lifetime inference failed on HRTB input" #54124

pnkfelix (Oct 02 2018 at 19:37, on Zulip):

I think I'll take this one

pnkfelix (Oct 02 2018 at 19:37, on Zulip):

unless someone else wants it

pnkfelix (Oct 02 2018 at 19:37, on Zulip):

(in which case: go ahead and steal it back from me)

pnkfelix (Oct 02 2018 at 19:38, on Zulip):

next: "NLL: unused_mut lint triggered by for _ in <unreachable>" #54586

nikomatsakis (Oct 02 2018 at 19:38, on Zulip):

the question on that one .. well, I'll leave a comment

pnkfelix (Oct 02 2018 at 19:38, on Zulip):

@blitzerr has been working on #54586, right?

pnkfelix (Oct 02 2018 at 19:38, on Zulip):

and there was also some input from @varkor

lqd (Oct 02 2018 at 19:38, on Zulip):

and/or varkor

pnkfelix (Oct 02 2018 at 19:39, on Zulip):

maybe I'll assign to @blitzerr for now

pnkfelix (Oct 02 2018 at 19:40, on Zulip):

that is: @blitzerr , if you're not comfortable being assigned to it, just let me know and/or remove yourself from the ticket

nikomatsakis (Oct 02 2018 at 19:40, on Zulip):

seems like we could assign to @varkor

pnkfelix (Oct 02 2018 at 19:40, on Zulip):

next: "NLL doesn't respect user types in closure signatures" #54692

nikomatsakis (Oct 02 2018 at 19:40, on Zulip):

I think they were prepared with a fix

pnkfelix (Oct 02 2018 at 19:40, on Zulip):

@nikomatsakis I was trying to understand the conversation that went by about that

nikomatsakis (Oct 02 2018 at 19:40, on Zulip):

I left some initial notes on #54692

pnkfelix (Oct 02 2018 at 19:40, on Zulip):

that @varkor had prepped something in passing?

nikomatsakis (Oct 02 2018 at 19:40, on Zulip):

yes

blitzerr (Oct 02 2018 at 19:41, on Zulip):

I am fine with it but varkor already has a fix he is tinkering with

nikomatsakis (Oct 02 2018 at 19:41, on Zulip):

though they did not confirm their desire to open a separate PR :)

nikomatsakis (Oct 02 2018 at 19:41, on Zulip):

but it seems independent from the original PR

pnkfelix (Oct 02 2018 at 19:41, on Zulip):

I could also assign them both, though sometimes that leads to confusion about whom takes point

nikomatsakis (Oct 02 2018 at 19:42, on Zulip):

not that big a deal I imagine

pnkfelix (Oct 02 2018 at 19:43, on Zulip):

last one: "NLL doesn't respect user types in closure signatures" #54692

nikomatsakis (Oct 02 2018 at 19:44, on Zulip):

I also left some notes on that one

pnkfelix (Oct 02 2018 at 19:45, on Zulip):

any volunteers?

nikomatsakis (Oct 02 2018 at 19:45, on Zulip):

I think I will take https://github.com/rust-lang/rust/issues/54574 away from @mikhail-m1 =)

nikomatsakis (Oct 02 2018 at 19:45, on Zulip):

based on prior conversation

pnkfelix (Oct 02 2018 at 19:45, on Zulip):

@nikomatsakis sounds good

nikomatsakis (Oct 02 2018 at 19:45, on Zulip):

and because it's a !@#!$! annoying one

blitzerr (Oct 02 2018 at 19:45, on Zulip):

I can take that as well

pnkfelix (Oct 02 2018 at 19:46, on Zulip):

@blitzerr #54692? Great

blitzerr (Oct 02 2018 at 19:46, on Zulip):

Ya

blitzerr (Oct 02 2018 at 19:46, on Zulip):

I would appreciate some mentoring notes though :grinning:

pnkfelix (Oct 02 2018 at 19:46, on Zulip):

okay, all issues for RC2 are assigned now

nikomatsakis (Oct 02 2018 at 19:47, on Zulip):

I would appreciate some mentoring notes though :grinning:

I did leave some notes but they are not as detailed; it's gonna be a mildly tricky bug as it will require coordinates changes to various parts of the compiler

pnkfelix (Oct 02 2018 at 19:47, on Zulip):

there's only one nominated issue, and its really targetted at the Lang Design Team: "borrow-checker allows partial reinit of struct that has been moved away, but no use of it." #21232

nikomatsakis (Oct 02 2018 at 19:47, on Zulip):

not super hard though, just touches a lot of things

pnkfelix (Oct 02 2018 at 19:51, on Zulip):

okay

pnkfelix (Oct 02 2018 at 19:52, on Zulip):

so, is there any other big business for this meeting...

pnkfelix (Oct 02 2018 at 19:52, on Zulip):

I will just make a quick note that earlier today @nikomatsakis and I decided to mark #54208 as NLL-deferred

Keith Yeung (Oct 02 2018 at 19:52, on Zulip):

all those new issues linked with #47184 are accounted for, yeah?

pnkfelix (Oct 02 2018 at 19:53, on Zulip):

all those new issues linked with #47184 are accounted for, yeah?

I hope so. might be worth quickly double checking

pnkfelix (Oct 02 2018 at 19:54, on Zulip):

I don't see an actual issue for "Associated constants in patterns" ...

nikomatsakis (Oct 02 2018 at 19:54, on Zulip):

I don' think we filed issues for everything

pnkfelix (Oct 02 2018 at 19:54, on Zulip):

just associated constants in expressions (#54571)

pnkfelix (Oct 02 2018 at 19:54, on Zulip):

there's issues for almost everything, it seems...

pnkfelix (Oct 02 2018 at 19:56, on Zulip):

@nikomatsakis was there anything else you wanted to cover? More performance stuff? Polonius? Rah go team?

nikomatsakis (Oct 02 2018 at 19:57, on Zulip):

Rah rah! not really

nikomatsakis (Oct 02 2018 at 19:57, on Zulip):

I'm just getting nervous about meeting the Oct 25 deadline :)

nikomatsakis (Oct 02 2018 at 19:57, on Zulip):

but I think we're doing good

nikomatsakis (Oct 02 2018 at 19:58, on Zulip):

we've got a lot of open issues though

pnkfelix (Oct 02 2018 at 19:58, on Zulip):

three weeks, I think we'll be okay

pnkfelix (Oct 02 2018 at 19:58, on Zulip):

but yes, lots to do

nikomatsakis (Oct 02 2018 at 19:58, on Zulip):

many of which are .. in progress .. so that's a good thing

pnkfelix (Oct 02 2018 at 19:59, on Zulip):

Okay well in that spriit

pnkfelix (Oct 02 2018 at 19:59, on Zulip):

I'm going to get back to hacking!

pnkfelix (Oct 02 2018 at 19:59, on Zulip):

Bye all, thanks for attending!

nikomatsakis (Oct 02 2018 at 20:00, on Zulip):

Ciao!

lqd (Oct 02 2018 at 20:00, on Zulip):

(I've opened the cfg compression PR for polonius so we can discuss it someday (as it's almost certainly wrong anyway); in the meantime I will have a little time in the coming week if there's an issue I can help with)

memoryruins (Oct 02 2018 at 20:06, on Zulip):

ooooh those are really solid improvements (modulo possible caveats)

lqd (Oct 02 2018 at 20:08, on Zulip):

let's make sure it does what it should before talking about those heh (as I said it's probably wrong :) (it would also be interesting to try with leapfrog)

lqd (Oct 03 2018 at 08:16, on Zulip):

(for those interested, leapfrog is around -30%)

Jake Goulding (Oct 03 2018 at 16:06, on Zulip):

FWIW, I find performance percentages completely useless. Does positive % mean faster, or does negative? Can something be 100% faster (the answer should be yes, but not how many people use the term)

Jake Goulding (Oct 03 2018 at 16:07, on Zulip):

I find it much more clear to say "before X it takes Y timeunits, afterwards it takes Z timeunits"

Jake Goulding (Oct 03 2018 at 16:13, on Zulip):

And the word faster makes it worse

Jake Goulding (Oct 03 2018 at 16:14, on Zulip):

For example, a test takes 100s before a commit and 80s after. The previous speed was 0.01 tps and the new speed is 0.0125 tps. The ratio of new speed/old speed is 125%. But then people will say that it's 125% faster, or will they say its 25% faster?

nikomatsakis (Oct 03 2018 at 16:34, on Zulip):

I'm a fan of ratios; new/old. Easy peezy. :)

Jake Goulding (Oct 03 2018 at 17:56, on Zulip):

I often find myself stating "it is XX% <units> compared to the previous code" or similar in a commit message

Jake Goulding (Oct 03 2018 at 17:56, on Zulip):

By fooing the bar, the code now takes 45% of the previous time

lqd (Oct 03 2018 at 18:01, on Zulip):

ok let’s say that by treefrogging the leapjoins, the code now takes 70% of the previous :frog: and making me uneasy at the same time :p

lqd (Oct 03 2018 at 18:11, on Zulip):

(for peace of mind, I’ll try to make use of Matthew’s awesome work tomorrow and see what I broke)

Last update: Nov 21 2019 at 14:10UTC