Stream: t-compiler/wg-rls-2.0

Topic: metrics


matklad (Jul 24 2020 at 13:54, on Zulip):

I finally got to setting up metrics

Laurențiu Nicola (Jul 24 2020 at 14:00, on Zulip):

Metrics?

matklad (Jul 24 2020 at 14:01, on Zulip):

Compile time, time to typecheck regex, number of type errors

matklad (Jul 24 2020 at 14:01, on Zulip):

so that we can track perf over time

Laurențiu Nicola (Jul 24 2020 at 14:01, on Zulip):

Like https://edwin0cheng.github.io/github-action-usage-test/?

matklad (Jul 24 2020 at 14:02, on Zulip):

yup, like that

matklad (Jul 24 2020 at 14:03, on Zulip):

wait, does it already track time?

matklad (Jul 24 2020 at 14:10, on Zulip):

Hm, I think it still makes sense to do this in a somewhat more first-class way...

Florian Diebold (Jul 24 2020 at 14:24, on Zulip):

yeah, I think that page was always more intended as a test anyway, as the URL says ;)

Laurențiu Nicola (Jul 30 2020 at 18:09, on Zulip):

image.png

So right is newer, right? Or maybe not..

matklad (Jul 30 2020 at 18:10, on Zulip):

I think right should be newer, yes

matklad (Jul 30 2020 at 18:10, on Zulip):

Also, does anyone know if it's possible to enable perf_event_open syscall on github actions?

matklad (Jul 30 2020 at 18:11, on Zulip):

I've added instruction tracking today only to discover that it just doesn't work on CI :(

Laurențiu Nicola (Jul 30 2020 at 18:12, on Zulip):

So replacing rand reduced the memory usage by 16 MB? :confused:

matklad (Jul 30 2020 at 18:13, on Zulip):

doesn't seem unreasonable -- it's a big crate

matklad (Jul 30 2020 at 18:13, on Zulip):

@Laurențiu Nicola if you patch code at the gh-pages to show commit message on hover, you'll know for sure ;-)

Laurențiu Nicola (Jul 30 2020 at 18:14, on Zulip):

Does it return EPERM?

matklad (Jul 30 2020 at 18:14, on Zulip):

i think so

matklad (Jul 30 2020 at 18:15, on Zulip):

(something about permissions -- there's error in the stderr on CI)

Laurențiu Nicola (Jul 30 2020 at 18:17, on Zulip):

I see there are some problematic attributes like exclude_kernel, but perf-event enables them (which is better)

Laurențiu Nicola (Jul 30 2020 at 18:19, on Zulip):

I can't find the analysis-stats run on CI

Laurențiu Nicola (Jul 30 2020 at 18:19, on Zulip):

Only for nightly

matklad (Jul 30 2020 at 18:21, on Zulip):

https://github.com/rust-analyzer/rust-analyzer/runs/928737985?check_suite_focus=true#step:4:372

Laurențiu Nicola (Jul 31 2020 at 07:39, on Zulip):

PR is up

Last update: Sep 27 2020 at 14:15UTC