Stream: t-compiler/wg-nll

Topic: issue-48071-universes


nikomatsakis (Jul 23 2018 at 15:06, on Zulip):

@pnkfelix are you around now by any chance?

nikomatsakis (Jul 23 2018 at 15:07, on Zulip):

I'd like to talk about #52488?

nikomatsakis (Jul 23 2018 at 15:08, on Zulip):

maybe we can schedule a time to chat tomorrow

pnkfelix (Jul 23 2018 at 15:09, on Zulip):

I’m en route to my apt right now, on mobile phone

pnkfelix (Jul 23 2018 at 15:09, on Zulip):

Ill be able to chat for a short while (until ~18:00 CET) after I get there

nikomatsakis (Jul 23 2018 at 15:15, on Zulip):

ok. Let me sketch out the scenario just a bit:

nikomatsakis (Jul 23 2018 at 15:15, on Zulip):
nikomatsakis (Jul 23 2018 at 15:15, on Zulip):
nikomatsakis (Jul 23 2018 at 15:16, on Zulip):
nikomatsakis (Jul 23 2018 at 15:16, on Zulip):
nikomatsakis (Jul 23 2018 at 15:17, on Zulip):
nikomatsakis (Jul 23 2018 at 15:17, on Zulip):
nikomatsakis (Jul 23 2018 at 15:17, on Zulip):
nikomatsakis (Jul 23 2018 at 15:18, on Zulip):
nikomatsakis (Jul 23 2018 at 15:18, on Zulip):
nikomatsakis (Jul 23 2018 at 15:18, on Zulip):

mmmm maybe that's what I should do

nikomatsakis (Jul 23 2018 at 15:18, on Zulip):

I don't feel great landing the code without tests, I already found some obvious bugs when rebasing

pnkfelix (Jul 23 2018 at 15:31, on Zulip):

yes, I would say add a debug flag that disables the type-check pre-screen

pnkfelix (Jul 23 2018 at 15:32, on Zulip):

did you resolve the performance issues that you and kennytm were discussing in the comment thread on the PR ?

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

no, I can't reproduce those, I am not convinced they are real

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

I was planning to retest

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

once I had it rebased etc

pnkfelix (Jul 23 2018 at 15:33, on Zulip):

wait I thought you said you observed 3x slowdown locally?

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

no I observed MASTER slowing down

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

and my branch being 3x faster

pnkfelix (Jul 23 2018 at 15:33, on Zulip):

OH

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

but I think I must have been mismeasuring

nikomatsakis (Jul 23 2018 at 15:33, on Zulip):

regardless I never saw my branch be slower

pnkfelix (Jul 23 2018 at 15:34, on Zulip):

I'm seeing a bunch of link-time errors on master on my linux box. :(

pnkfelix (Jul 23 2018 at 15:34, on Zulip):

on four of the run-pass tests

pnkfelix (Jul 23 2018 at 15:34, on Zulip):

it is annoying

nikomatsakis (Jul 23 2018 at 15:34, on Zulip):

my linux box is not ssh'able and I am sad

davidtwco (Jul 23 2018 at 15:34, on Zulip):

@pnkfelix I've seen those too today (https://gist.github.com/davidtwco/083e5b0b0146958f3a2ab65a1b71b247)

nikomatsakis (Jul 23 2018 at 15:34, on Zulip):

however, it does experience weird problems sometimes

nikomatsakis (Jul 23 2018 at 15:34, on Zulip):

not link errors though

pnkfelix (Jul 23 2018 at 15:35, on Zulip):

my linux box is not ssh'able and I am sad

oh that is super sad

nikomatsakis (Jul 23 2018 at 15:35, on Zulip):

but I'll be back home in a week or two. who knows what went wrong.

nikomatsakis (Jul 25 2018 at 10:45, on Zulip):

@pnkfelix ping

pnkfelix (Jul 25 2018 at 10:45, on Zulip):

pong; reviewing now

nikomatsakis (Jul 25 2018 at 13:05, on Zulip):

@pnkfelix so is that r=you if I correct nits + fix travis?

nikomatsakis (Jul 25 2018 at 13:06, on Zulip):

I am debating how much to worry, there is one UI regression I am looking at now

nikomatsakis (Jul 25 2018 at 13:06, on Zulip):

where I think we are doing a less good job picking spans

nikomatsakis (Jul 25 2018 at 13:06, on Zulip):

I had some ideas for improving it but they are not entirely trivial

pnkfelix (Jul 25 2018 at 13:06, on Zulip):

yeah let me write r=me explicitly

pnkfelix (Jul 25 2018 at 13:06, on Zulip):

there was some reason I thought I would want to take a second pass over it

pnkfelix (Jul 25 2018 at 13:07, on Zulip):

but looking at my comments now, I don't know why I thought that

nikomatsakis (Jul 26 2018 at 12:38, on Zulip):

the perf results for this PR were nice :)

Jake Goulding (Jul 26 2018 at 13:53, on Zulip):

Link for us lazy - https://perf.rust-lang.org/compare.html?start=fefe81605d6111faa8dbb3635ab2c51d59de740a&end=73f817a0577649c3ffd1045b12c45f406a41a47f

Jake Goulding (Jul 26 2018 at 13:54, on Zulip):

@simulacrum ^ 504 Gateway Time-out

Jake Goulding (Jul 26 2018 at 13:57, on Zulip):

and fixed now... ?

lqd (Jul 26 2018 at 14:00, on Zulip):

@Jake Goulding that reminds me I had 500s on the playground (apparently some other too since they notified / jokingly blamed me for some reason ;)

simulacrum (Jul 26 2018 at 14:34, on Zulip):

@Jake Goulding Turns out we were DDOS-ing ourselves (the interpolation takes ~20 seconds and we were running it every 30)

simulacrum (Jul 26 2018 at 14:34, on Zulip):

Should be fixed now but I'll keep an eye on it

Jake Goulding (Jul 26 2018 at 15:02, on Zulip):

Haha, programming is a silly thing

Last update: Nov 21 2019 at 14:20UTC