Stream: wg-traits

Topic: weekly meeting 2018-11-12


nikomatsakis (Nov 12 2018 at 18:59, on Zulip):

Hello all :wave:

nikomatsakis (Nov 12 2018 at 19:00, on Zulip):

Honestly I've spent most of today slowly catching up and I wasnt' around last week, curious to hear from other folks though. I hope to get on top of reviews today. =)

nikomatsakis (Nov 12 2018 at 19:00, on Zulip):

Dropbox document: link

nikomatsakis (Nov 12 2018 at 19:06, on Zulip):

so @scalexm you have a huge pending PR :)

nikomatsakis (Nov 12 2018 at 19:06, on Zulip):

which I can try to review today hopefully

nikomatsakis (Nov 12 2018 at 19:06, on Zulip):

what are the "next steps" in your view from what we have now?

nikomatsakis (Nov 12 2018 at 19:06, on Zulip):

when it comes to chalk integration

scalexm (Nov 12 2018 at 19:06, on Zulip):

not that huge, there is a bit of code reordering

scalexm (Nov 12 2018 at 19:07, on Zulip):

next step is to implement missing unification routines

scalexm (Nov 12 2018 at 19:07, on Zulip):

which I'm working on currently

scalexm (Nov 12 2018 at 19:08, on Zulip):

(i'll probably split this work in two PRs)

nikomatsakis (Nov 12 2018 at 19:08, on Zulip):

next step is to implement missing unification routines

do you mean the various chalk callback methods?

scalexm (Nov 12 2018 at 19:08, on Zulip):

yeah

nikomatsakis (Nov 12 2018 at 19:09, on Zulip):

ok, there is an open issue on that I think

nikomatsakis (Nov 12 2018 at 19:10, on Zulip):

ok well maybe we should take it easy this week and I can focus on reviewing then :)

nikomatsakis (Nov 12 2018 at 19:11, on Zulip):

and we'll try to circle back next week and get more active? :)

tmandry (Nov 12 2018 at 19:12, on Zulip):

I was pretty busy last week also. I'll at least wrap up the SLG solver parts of the guide

tmandry (Nov 12 2018 at 19:13, on Zulip):

a first crack, that is

scalexm (Nov 12 2018 at 19:13, on Zulip):

@nikomatsakis I think with myunification branch (WIP right now) we're basically seeing the end of the implementation of chalk Context trait and al

nikomatsakis (Nov 12 2018 at 19:13, on Zulip):

great!

scalexm (Nov 12 2018 at 19:19, on Zulip):

@nikomatsakis also did you see this issue I opened in chalk: https://github.com/rust-lang-nursery/chalk/issues/189

scalexm (Nov 12 2018 at 19:20, on Zulip):

of course it would be easy to hack around this in rustc (just write a "by-hand" coercion), but that's unfortunate

nikomatsakis (Nov 12 2018 at 19:20, on Zulip):

I saw it, but I usually solve this sort of thing by having an explicit upcast

nikomatsakis (Nov 12 2018 at 19:20, on Zulip):

in fact I think there are some existing methods of this kind already

nikomatsakis (Nov 12 2018 at 19:21, on Zulip):

e.g.,

    fn sink_answer_subset(
        &self,
        value: &C::CanonicalConstrainedSubst,
    ) -> I::CanonicalConstrainedSubst;

    fn lift_delayed_literal(
        &self,
        value: DelayedLiteral<I>,
    ) -> DelayedLiteral<C>;
nikomatsakis (Nov 12 2018 at 19:21, on Zulip):

we don't appear to have one for ExClause

scalexm (Nov 12 2018 at 19:21, on Zulip):

mmh I guess lift is for going from 'tcx to 'gcx, no?

nikomatsakis (Nov 12 2018 at 19:21, on Zulip):

yes

scalexm (Nov 12 2018 at 19:21, on Zulip):

but anyway we can write that by hand indeed

nikomatsakis (Nov 12 2018 at 19:21, on Zulip):

I don't think there's another option really

nikomatsakis (Nov 12 2018 at 19:22, on Zulip):

unless we change ExClause to not be parameterized by C

nikomatsakis (Nov 12 2018 at 19:22, on Zulip):

but rather by all the things it projects out from C

scalexm (Nov 12 2018 at 19:22, on Zulip):

well we could make ExClause etc fully parametric, i.e. not using associated items :) but that would be a bit verbose

nikomatsakis (Nov 12 2018 at 19:22, on Zulip):

right

nikomatsakis (Nov 12 2018 at 19:22, on Zulip):

no perfect sol'n

nikomatsakis (Nov 12 2018 at 19:22, on Zulip):

I think of the two I'd find an upcast a touch easier?

scalexm (Nov 12 2018 at 19:22, on Zulip):

ok so let's go for the explicit upcast

Alexander Regueiro (Nov 12 2018 at 19:28, on Zulip):

Hi

Alexander Regueiro (Nov 12 2018 at 19:28, on Zulip):

Good to hear of Chalk being so close to integration. :-)

Alexander Regueiro (Nov 12 2018 at 20:24, on Zulip):

@nikomatsakis @scalexm weird... mir emitting seems broken.

Alexander Regueiro (Nov 12 2018 at 22:25, on Zulip):

@scalexm What's the status of https://github.com/rust-lang/rust/issues/44265 right now, do you know

Alexander Regueiro (Nov 12 2018 at 22:25, on Zulip):

?

Gankro (Nov 16 2018 at 17:10, on Zulip):

is the performance impact of chalk being tracked anywhere?

Last update: Nov 12 2019 at 16:55UTC