Stream: wg-traits

Topic: meeting 2018-10-22


nikomatsakis (Oct 22 2018 at 18:00, on Zulip):

cc @WG-compiler-traits — sync time

Dropbox paper link.

nikomatsakis (Oct 22 2018 at 18:01, on Zulip):

I'll give a few minutes for folks to show up :)

nikomatsakis (Oct 22 2018 at 18:04, on Zulip):

OK, let's see...

nikomatsakis (Oct 22 2018 at 18:04, on Zulip):

@scalexm quick question about the bound-ty stuff, are you pursuing bringing all of that together under the general "late-bound" plan?

scalexm (Oct 22 2018 at 18:05, on Zulip):

@nikomatsakis yes, my plan is to get rid of ReCanonical as well, and adapt the shift_etc routines to take into account bound tys

nikomatsakis (Oct 22 2018 at 18:05, on Zulip):

great

nikomatsakis (Oct 22 2018 at 18:05, on Zulip):

I was just rebasing the universe PR over those changes and I was confused for a sec because ReCanonical persisted

nikomatsakis (Oct 22 2018 at 18:05, on Zulip):

ok, so, this week is the RC2 release

nikomatsakis (Oct 22 2018 at 18:06, on Zulip):

so that is kind of consuming me but I am hopeful to be able to start devoting more dedicated time after that is done

nikomatsakis (Oct 22 2018 at 18:06, on Zulip):

I don't see much activity on the chalk-integration issues since last week

nikomatsakis (Oct 22 2018 at 18:06, on Zulip):

(meta-question: is GH back up and operational yet?)

Sunjay Varma (Oct 22 2018 at 18:07, on Zulip):

(meta-answer: https://status.github.com )

nikomatsakis (Oct 22 2018 at 18:08, on Zulip):

@scalexm one thing we should perhaps discuss-- but probably not in depth right now. I've been looking today at https://github.com/rust-lang/rust/issues/55222, which has to do with trying to enforce some subset of bounds for type aliases (e.g., type Foo = Bar...)

scalexm (Oct 22 2018 at 18:08, on Zulip):

btw @nikomatsakis I opened a chalk PR which updates the assoc type rule, as we discussed https://github.com/rust-lang-nursery/chalk/pull/184

nikomatsakis (Oct 22 2018 at 18:08, on Zulip):

...for ... um ... "historical reasons", we do not do that presently, but it is interesting to discuss what strategy we want to be shooting for when it comes to implied bounds etc

nikomatsakis (Oct 22 2018 at 18:09, on Zulip):

ok, great, that reminds me that I was hoping to get some focus on further work within chalk, probably starting with trying to refactor to be more demand driven

nikomatsakis (Oct 22 2018 at 18:09, on Zulip):

(so that we can lower some traits but not all)

scalexm (Oct 22 2018 at 18:10, on Zulip):

also I'm trying to think about how to represent trait objects in chalk, the main thing I have in mind is https://github.com/rust-lang/rust/issues/44454 (which I'm not sure I have a solution for yet :P)

nikomatsakis (Oct 22 2018 at 18:10, on Zulip):

ah, yes, I keep forgetting that we need to think about that, too

scalexm (Oct 22 2018 at 18:10, on Zulip):

I think that's the main blocker

nikomatsakis (Oct 22 2018 at 18:10, on Zulip):

do you think they should be integrated as a kind of builtin?

scalexm (Oct 22 2018 at 18:11, on Zulip):

because the current object safety rules are easily translated into logical rules I think

scalexm (Oct 22 2018 at 18:11, on Zulip):

well yes I was thinking of having them as some builtin types

scalexm (Oct 22 2018 at 18:12, on Zulip):

in chalk this would just amount to have another variant in enum TypeName so that trait objects behave as special ApplicationTy

nikomatsakis (Oct 22 2018 at 18:12, on Zulip):

we probably want to keep in mind genearlizing to dyn (Foo + Bar)

scalexm (Oct 22 2018 at 18:12, on Zulip):

yeah right

nikomatsakis (Oct 22 2018 at 18:13, on Zulip):

we should probably make a list of "things chalk needs to model but doesn't" and tag said issues with some special label

nikomatsakis (Oct 22 2018 at 18:13, on Zulip):

beyond dyn traits, there is specialization, but are there other things I am overlooking? :)

scalexm (Oct 22 2018 at 18:14, on Zulip):

I don't think so, presuming that we treat other associated items like fn and constants roughly like associated types

nikomatsakis (Oct 22 2018 at 18:14, on Zulip):

we kind of have to work out resolving methods to a specific impl, though I had hoped to handle that as associated types, exactly

nikomatsakis (Oct 22 2018 at 18:14, on Zulip):

I guess it's a bit of an "open question" how much special treatment constants need. Seems like not much, that's kind of pushed to unification

nikomatsakis (Oct 22 2018 at 18:15, on Zulip):

@Sunjay Varma what's your schedule looking like? :)

nikomatsakis (Oct 22 2018 at 18:16, on Zulip):

not sure if you are just lurking or actively looking for ways to spend your time :)

nikomatsakis (Oct 22 2018 at 18:16, on Zulip):

(either of which are fine, of course...)

scalexm (Oct 22 2018 at 18:16, on Zulip):

about the remaining chalk integration issues, I see https://github.com/rust-lang/rust/issues/55096 which is fairly easy, so it might be a good thing to find someone to do that

scalexm (Oct 22 2018 at 18:16, on Zulip):

https://github.com/rust-lang/rust/issues/55097 does not seem too hard either

nikomatsakis (Oct 22 2018 at 18:17, on Zulip):

yeah

nikomatsakis (Oct 22 2018 at 18:17, on Zulip):

I tried to identify those

nikomatsakis (Oct 22 2018 at 18:17, on Zulip):

maybe a good work item is to put better instructions and tag with E-mentor

nikomatsakis (Oct 22 2018 at 18:17, on Zulip):

seems doable

scalexm (Oct 22 2018 at 18:18, on Zulip):

and once we have bound types + universes + https://github.com/rust-lang/rust/issues/55098 I guess we are basically done?

scalexm (Oct 22 2018 at 18:19, on Zulip):

ah I forgot https://github.com/rust-lang/rust/issues/54935 of course

nikomatsakis (Oct 22 2018 at 18:19, on Zulip):

one thing at a time :)

nikomatsakis (Oct 22 2018 at 18:19, on Zulip):

so, some goals for the week, let's see... I think adding mentoring to those bugs + opening universe PRs will do it for me :)

nikomatsakis (Oct 22 2018 at 18:20, on Zulip):

I guess you'll keep pushing @scalexm on the bound-ty stuff?

scalexm (Oct 22 2018 at 18:20, on Zulip):

@nikomatsakis yes, hopefully I'll have something working in the next few days

nikomatsakis (Oct 22 2018 at 18:20, on Zulip):

we should try to make https://github.com/rust-lang/rust/labels/chalk-integration be a "milestone label" of sorts

nikomatsakis (Oct 22 2018 at 18:20, on Zulip):

so we can "drill it down"

nikomatsakis (Oct 22 2018 at 18:20, on Zulip):

I think that was my intention, not sure if there is anything extra int here

nikomatsakis (Oct 22 2018 at 18:20, on Zulip):

yes, hopefully I'll have something working in the next few days

sounds great

Sunjay Varma (Oct 22 2018 at 18:38, on Zulip):

not sure if you are just lurking or actively looking for ways to spend your time :)

A bit of both. I am interviewing for jobs for the next week or so. Things are busy, but I am relatively available for small tasks. I would like to keep contributing in some way and am basically just looking for a way to do that. :)

Sunjay Varma (Oct 22 2018 at 18:38, on Zulip):

@nikomatsakis ^

nikomatsakis (Oct 22 2018 at 18:44, on Zulip):

ok :)

nikomatsakis (Oct 22 2018 at 18:44, on Zulip):

hmm

Josh Huber (Oct 22 2018 at 19:05, on Zulip):

I am mostly lurking for the next week or two, following along and trying to come up to speed. So, thank you to everyone who's written rustc documentation, it's super appreciated. (probably more like necessary :) Still planning on poking my head in if/when there seems like something that's a good fit for me.

Alexander Regueiro (Oct 22 2018 at 19:32, on Zulip):

sorry I couldn't be there, was out tonight...

nikomatsakis (Oct 22 2018 at 19:33, on Zulip):

@uberjay maybe take a look at one of the issues we talked about; I'll try to write some more detailed comments on one of them

Alexander Regueiro (Oct 22 2018 at 19:33, on Zulip):

or is it still going on? :-)

nikomatsakis (Oct 22 2018 at 19:33, on Zulip):

@Alexander Regueiro :wave: I'm taking a look at your PR... of course GH outage is kind of a pain

Alexander Regueiro (Oct 22 2018 at 19:33, on Zulip):

Oh hey.

Alexander Regueiro (Oct 22 2018 at 19:33, on Zulip):

Great.

Alexander Regueiro (Oct 22 2018 at 19:34, on Zulip):

Yeah, it's quite annoying. I was grappling with it last night, almost 24 hours ago!

Alexander Regueiro (Oct 22 2018 at 19:59, on Zulip):

@nikomatsakis making sense to you?

Alexander Regueiro (Oct 22 2018 at 20:28, on Zulip):

@nikomatsakis you there?

Last update: Nov 12 2019 at 16:45UTC