Stream: t-compiler

Topic: weekly meeting 2020-05-07 #54818


Santiago Pastorino (May 06 2020 at 18:54, on Zulip):

Hi @T-compiler/meeting; the triage meeting will be starting in ~ 19 hours

Santiago Pastorino (May 06 2020 at 18:55, on Zulip):

The @WG-prioritization will be doing pre-triage in #t-compiler/wg-prioritization > pre-meeting triage 2020-05-07 #54818

Santiago Pastorino (May 06 2020 at 18:55, on Zulip):

During pre-triage we will be preparing the meeting agenda

Santiago Pastorino (May 06 2020 at 18:56, on Zulip):

We will have checkins from @WG-rls2.0 and @WG-self-profile

Santiago Pastorino (May 06 2020 at 18:57, on Zulip):

@matkladdo you have something you want to share about @WG-rls2.0?

Santiago Pastorino (May 06 2020 at 18:57, on Zulip):

@Wesley Wiser do you have something you want to share about @WG-self-profile?

Wesley Wiser (May 06 2020 at 18:58, on Zulip):

Sure!

matklad (May 07 2020 at 11:58, on Zulip):

will do!

Santiago Pastorino (May 07 2020 at 13:28, on Zulip):

Hi @T-compiler/meeting, triage meeting will be starting in ~ 30 minutes

Santiago Pastorino (May 07 2020 at 13:28, on Zulip):

Check out the meeting agenda

pnkfelix (May 07 2020 at 14:01, on Zulip):

Hi @T-compiler/meeting! Add a :wave: emoji to show you're here :)

pnkfelix (May 07 2020 at 14:01, on Zulip):

we will start off with 5 minutes for ...

Announcements

pnkfelix (May 07 2020 at 14:02, on Zulip):
pnkfelix (May 07 2020 at 14:02, on Zulip):
pnkfelix (May 07 2020 at 14:02, on Zulip):
pnkfelix (May 07 2020 at 14:03, on Zulip):

(is there anyone here who doesn't know what I'm referencing by the "State of the Compiler Team form" ? Feel free to speak up, or to PM if you prefer)

Santiago Pastorino (May 07 2020 at 14:03, on Zulip):

On friday we have our Compiler Team Planning meeting.

Have the need to fix this because I redacted it yesterday and may look confusing, to be clear the meeting is tomorrow friday 8th

nikomatsakis (May 07 2020 at 14:05, on Zulip):

there is 1 new major change proposal, I guess that means

pnkfelix (May 07 2020 at 14:06, on Zulip):

yeah

nikomatsakis (May 07 2020 at 14:06, on Zulip):

however

nikomatsakis (May 07 2020 at 14:06, on Zulip):

I think that is incomplete

pnkfelix (May 07 2020 at 14:06, on Zulip):

I'm now wondering if we shuold post every MCP that has not been seconded

nikomatsakis (May 07 2020 at 14:06, on Zulip):

for example, there are two in FCP

nikomatsakis (May 07 2020 at 14:06, on Zulip):

actually 4

pnkfelix (May 07 2020 at 14:06, on Zulip):

until they are either declined or seconded?

nikomatsakis (May 07 2020 at 14:06, on Zulip):

or maybe some of those already completed and we need to move

nikomatsakis (May 07 2020 at 14:06, on Zulip):

but we should definitely be announcing them

nikomatsakis (May 07 2020 at 14:06, on Zulip):
nikomatsakis (May 07 2020 at 14:07, on Zulip):
nikomatsakis (May 07 2020 at 14:07, on Zulip):

Move src/test/run-fail tests to UI compiler-team#274 -- I think this should be considered "accepted"

nikomatsakis (May 07 2020 at 14:07, on Zulip):

oh, that's it, compiler-team#272 is a dummy

nikomatsakis (May 07 2020 at 14:07, on Zulip):

I am reminded I wanted to update and merge the RFC about these..

nikomatsakis (May 07 2020 at 14:07, on Zulip):

but definitely check out those first two :)

Santiago Pastorino (May 07 2020 at 14:07, on Zulip):

these two were announced already but maybe we should define how to announce this, going to create a topic to discuss this

pnkfelix (May 07 2020 at 14:07, on Zulip):

okay. @Santiago Pastorino and I will work on revising the agenda template

nikomatsakis (May 07 2020 at 14:08, on Zulip):

sorry, maybe I was just confused :) but can't hurt to announce again

pnkfelix (May 07 2020 at 14:08, on Zulip):

yeah I don't think it hurts to remind people

pnkfelix (May 07 2020 at 14:09, on Zulip):

(meta-thought: I was just about to say "it doesn't hurt to repeatedly mention it, since the intention is not to open up immediate discussion." but then I realized: If the intention isn't to open up discussion, then why mention it at the synchronous meeting?)

nikomatsakis (May 07 2020 at 14:09, on Zulip):

people are looking

pnkfelix (May 07 2020 at 14:09, on Zulip):

I'll have to think more about that later.

pnkfelix (May 07 2020 at 14:10, on Zulip):

anyway that's all the announcements

pnkfelix (May 07 2020 at 14:10, on Zulip):

oh, well there's one more semi-announcement

pnkfelix (May 07 2020 at 14:11, on Zulip):

I did a bunch of unilateral beta-acceptances over the last week:

pnkfelix (May 07 2020 at 14:11, on Zulip):

Beta-acceptances from past week

pnkfelix (May 07 2020 at 14:11, on Zulip):

leaving us with just a single beta-nomination to synchronously discuss

pnkfelix (May 07 2020 at 14:11, on Zulip):

Beta-nominations

T-compiler beta noms

pnkfelix (May 07 2020 at 14:12, on Zulip):

often I'd just unilaterally accept a version update

pnkfelix (May 07 2020 at 14:12, on Zulip):

but I wasn't clear on how risky this one was

pnkfelix (May 07 2020 at 14:13, on Zulip):

and so I wanted to take people's temperature on whether it was worth backporting

pnkfelix (May 07 2020 at 14:13, on Zulip):

so far the votes seem pretty clear....

nikomatsakis (May 07 2020 at 14:13, on Zulip):

@Alex Crichton writes

Given bugs like alexcrichton/cc-rs#493 I think this may also be a good candidate for backporting since it should be pretty low-risk and may be good to accelerate getting this out to users.

pnkfelix (May 07 2020 at 14:14, on Zulip):

yeah. I just had a hard time processing the diff that @simulacrum pointed at

pnkfelix (May 07 2020 at 14:15, on Zulip):

anyway looks like everyones cool with a backport

nikomatsakis (May 07 2020 at 14:15, on Zulip):

(true, I did too :)

pnkfelix (May 07 2020 at 14:15, on Zulip):

so beta-accepted, sounds great

nikomatsakis (May 07 2020 at 14:15, on Zulip):

but I think it's ok

pnkfelix (May 07 2020 at 14:16, on Zulip):

okay. there weren't any stable-noms nor any PR's marked S-waiting-on-team

pnkfelix (May 07 2020 at 14:16, on Zulip):

Issues of Note

Short Summary

There are 1 less P-critical issues and 2 less P-high issues in comparison with last week.

pnkfelix (May 07 2020 at 14:17, on Zulip):

P-critical

Santiago Pastorino (May 07 2020 at 14:17, on Zulip):

P-high issues are consistently going down!

pnkfelix (May 07 2020 at 14:17, on Zulip):

So I wanted to pause here and try to get some feedback

pnkfelix (May 07 2020 at 14:18, on Zulip):

@WG-prioritization noted that the agenda this week was going to be light, and so it would make sense to try to discuss some things that we often don't have time for

pnkfelix (May 07 2020 at 14:18, on Zulip):

and the basic debate, if you can call it that, was

pnkfelix (May 07 2020 at 14:18, on Zulip):

which is more important for us to discuss (slash trying to find volunteers to address)

pnkfelix (May 07 2020 at 14:18, on Zulip):

between:

pnkfelix (May 07 2020 at 14:19, on Zulip):
  1. unassigned P-medium stable-to-beta regressions, or
pnkfelix (May 07 2020 at 14:19, on Zulip):
  1. unassigned P-high issues ?
pnkfelix (May 07 2020 at 14:19, on Zulip):

here's an emoji vote

pnkfelix (May 07 2020 at 14:19, on Zulip):

(hmm I thought there were numbers available)

pnkfelix (May 07 2020 at 14:20, on Zulip):

I'll just use thumbs instead

pnkfelix (May 07 2020 at 14:20, on Zulip):

For the purposes of this meeting, I decided I'd rather try to solicit volunteers to take on unassigned P-high issues

nikomatsakis (May 07 2020 at 14:20, on Zulip):

I'm not sure what makes a stable-to-beta regression "p-medium", I guess

pnkfelix (May 07 2020 at 14:21, on Zulip):

something that we're willing to accept as breakage, I think

nikomatsakis (May 07 2020 at 14:21, on Zulip):

(I voted but I'd be curious if there's a "capsule summary" of the case for P-medium stable-to-beta regressions)

nikomatsakis (May 07 2020 at 14:21, on Zulip):

i.e., it seems sort of like "we would've called it p-high if it was more important" :)

pnkfelix (May 07 2020 at 14:21, on Zulip):

take a look: https://github.com/rust-lang/rust/issues?q=is%3Aopen+label%3Aregression-from-stable-to-beta+label%3AP-medium

Santiago Pastorino (May 07 2020 at 14:21, on Zulip):

so a personal opinion, when I see that something is a regression I kind of bump up a bit my rating

Santiago Pastorino (May 07 2020 at 14:22, on Zulip):

with that reasoning P-high things would be preferred

pnkfelix (May 07 2020 at 14:22, on Zulip):

although maybe I'd like to see this one fixed: "forbid overwritten by later allow on the same "scope level"" #70819

Santiago Pastorino (May 07 2020 at 14:23, on Zulip):

Santiago Pastorino said:

with that reasoning P-high things would be preferred

because if it's a regression and P-medium it's probably not really important, but I have no idea if all P-medium regressions are rated using that logic

pnkfelix (May 07 2020 at 14:23, on Zulip):

I guess I should have listed each individually to allow some discussion

pnkfelix (May 07 2020 at 14:23, on Zulip):

but anyway

pnkfelix (May 07 2020 at 14:23, on Zulip):

the whole point was that I didn't think we should jump into them in this meeting

pnkfelix (May 07 2020 at 14:23, on Zulip):

and listing them individually would defeat that. :)

pnkfelix (May 07 2020 at 14:24, on Zulip):

Anyway lets look at the unassigned P-high issues

pnkfelix (May 07 2020 at 14:24, on Zulip):

that I picked

pnkfelix (May 07 2020 at 14:24, on Zulip):

(I chose the ones that are non-ICE's that are not marked E-needs-mcve)

pnkfelix (May 07 2020 at 14:25, on Zulip):

or wait

pnkfelix (May 07 2020 at 14:25, on Zulip):

all I did was nominate them

pnkfelix (May 07 2020 at 14:25, on Zulip):

so we will just go through the nominated issues now

pnkfelix (May 07 2020 at 14:25, on Zulip):

and you'll see which ones are the one's I threw on the list. :)

pnkfelix (May 07 2020 at 14:25, on Zulip):

Nominated Issues

T-compiler I-nominated

pnkfelix (May 07 2020 at 14:25, on Zulip):
pnkfelix (May 07 2020 at 14:26, on Zulip):

oh I guess we can drop the nomination here?

Santiago Pastorino (May 07 2020 at 14:26, on Zulip):

yes

pnkfelix (May 07 2020 at 14:26, on Zulip):
pnkfelix (May 07 2020 at 14:27, on Zulip):

so this is an example of an unassigned P-high issue

nikomatsakis (May 07 2020 at 14:28, on Zulip):

regressed in 1.30.0, so a long time back, causes a segfault

pnkfelix (May 07 2020 at 14:29, on Zulip):

I'm confused though

pnkfelix (May 07 2020 at 14:29, on Zulip):

@eddyb 's notes from mid February

pnkfelix (May 07 2020 at 14:29, on Zulip):

indicate that this was stack overflowing from the beginning?

pnkfelix (May 07 2020 at 14:30, on Zulip):

(that the cases that were compiling were false-negatives because godbolt was compiling in library mode)

eddyb (May 07 2020 at 14:30, on Zulip):

yeah that causes non-pub functions to not be codegen'd

eddyb (May 07 2020 at 14:30, on Zulip):

so this never actually regressed

nikomatsakis (May 07 2020 at 14:30, on Zulip):

oh, I see

pnkfelix (May 07 2020 at 14:31, on Zulip):

so my first question: Is this really P-high ?

pnkfelix (May 07 2020 at 14:31, on Zulip):

I'm tempted to reclassify as P-medium and let it go

nikomatsakis (May 07 2020 at 14:31, on Zulip):

I think "no"

pnkfelix (May 07 2020 at 14:31, on Zulip):

okay cool. lets move on then

pnkfelix (May 07 2020 at 14:32, on Zulip):
pnkfelix (May 07 2020 at 14:34, on Zulip):

a user says they have a RISC-V repro

pnkfelix (May 07 2020 at 14:34, on Zulip):

which makes this interesting because it implies the bug may be target agnostic

nikomatsakis (May 07 2020 at 14:34, on Zulip):

it seems like this bug is in lld?

pnkfelix (May 07 2020 at 14:35, on Zulip):

that is indeed a question: Is there an upstream bug filed against lld?

pnkfelix (May 07 2020 at 14:35, on Zulip):

oh the user says their repro works on x86 too

pnkfelix (May 07 2020 at 14:36, on Zulip):

okay you know what, I'll take a look at it then

nikomatsakis (May 07 2020 at 14:36, on Zulip):

(that said, this comment by @Daniel Egger might suggest the problem is in rust-lld..? I can't quite tell.)

pnkfelix (May 07 2020 at 14:36, on Zulip):

see if I can find or file an upstream LLD bug report

pnkfelix (May 07 2020 at 14:36, on Zulip):

nikomatsakis said:

(that said, this comment by Daniel Egger might suggest the problem is in rust-lld..? I can't quite tell.)

yeah, that may be part of the problem. Which ... would put it back in our court, right?

pnkfelix (May 07 2020 at 14:36, on Zulip):

anyway I will self-assign.

nikomatsakis (May 07 2020 at 14:36, on Zulip):

I don't really know who maintains rust-lld :)

simulacrum (May 07 2020 at 14:37, on Zulip):

rust-lld is just "lld in the llvm submodule"

eddyb (May 07 2020 at 14:38, on Zulip):

lld built from our llvm-project fork, right?

pnkfelix (May 07 2020 at 14:38, on Zulip):

I'll follow-up. Self-assigned #65391

pnkfelix (May 07 2020 at 14:38, on Zulip):
eddyb (May 07 2020 at 14:39, on Zulip):

this sounds like something I should've followed up on :(

simulacrum (May 07 2020 at 14:39, on Zulip):

yeah lld from our fork

simulacrum (May 07 2020 at 14:39, on Zulip):

well "fork"

eddyb (May 07 2020 at 14:39, on Zulip):

shallow fork for backports and Rust-specific patches :P

pnkfelix (May 07 2020 at 14:39, on Zulip):

@eddyb well it looked like @Esteban Küber was going to look into it too, right?

eddyb (May 07 2020 at 14:39, on Zulip):

right but I think they might've been waiting for something from me

pnkfelix (May 07 2020 at 14:39, on Zulip):

(and I also said I might look at it... big collective "oh maybe I will..." from each of us)

pnkfelix (May 07 2020 at 14:40, on Zulip):

anyway @eddyb would you like to take point here for the short term?

pnkfelix (May 07 2020 at 14:41, on Zulip):

if you do not have time then I can

eddyb (May 07 2020 at 14:41, on Zulip):

I guess I'll try

pnkfelix (May 07 2020 at 14:41, on Zulip):

but I figured you might have more immediate context in your ... well, maybe not mental L1 cache, but perhaps at least your mental L3 cache

eddyb (May 07 2020 at 14:41, on Zulip):

try corrupted tape backups

Esteban Küber (May 07 2020 at 14:41, on Zulip):

I looked into it, but couldn't make it not regress the output

pnkfelix (May 07 2020 at 14:42, on Zulip):

okay assigning to @eddyb

Esteban Küber (May 07 2020 at 14:42, on Zulip):

Should be easy for someone with more familiarity

Esteban Küber (May 07 2020 at 14:42, on Zulip):

We have an indirection call there that can be direct recursion

eddyb (May 07 2020 at 14:43, on Zulip):

(supposedly #56655 might've helped but I don't remember why I dropped that PR)

pnkfelix (May 07 2020 at 14:43, on Zulip):

last nomination

pnkfelix (May 07 2020 at 14:43, on Zulip):
pnkfelix (May 07 2020 at 14:44, on Zulip):

hmm this was nominated for lang team meeting

Santiago Pastorino (May 07 2020 at 14:44, on Zulip):

I was going to say that, I've skipped it because of that :)

pnkfelix (May 07 2020 at 14:44, on Zulip):

but it was on our agenda?

pnkfelix (May 07 2020 at 14:44, on Zulip):

did I do that?

pnkfelix (May 07 2020 at 14:44, on Zulip):

anyway okay maybe I messed that up last night

Santiago Pastorino (May 07 2020 at 14:45, on Zulip):

I thought you re-added because you wanted to discuss it due to lack of things :)

pnkfelix (May 07 2020 at 14:45, on Zulip):

great great great, done with T-compiler nominations

pnkfelix (May 07 2020 at 14:45, on Zulip):

so, WG checkin's!

pnkfelix (May 07 2020 at 14:45, on Zulip):

lets hear from @matklad , right?

matklad (May 07 2020 at 14:46, on Zulip):

:thumbs_up:

matklad (May 07 2020 at 14:46, on Zulip):

Let me paste the report...

matklad (May 07 2020 at 14:46, on Zulip):

We've opened an RFC for making rust-analyzer the official LSP server impl:

https://github.com/rust-lang/rfcs/pull/2912

:tada:

Right now, the primary focus is on closing the immediate blockers surfaced in the RFC (protocol conformance primarily).

Internally, the biggest change is probably switch to recursive chalk solver.
It, and many other improvements everywhere now make rust-analyzer infer types correctly in most cases.
Feels like a real IDE now (auto-imports are also a big part here) :-)
Biggest missing pieces are type inference around impl Trait, supertraits of dyn Trait and .await.

We also added initial support for procedural macros, and can expand derives (though the feature is disabled by default). This works on quasi-stable -- we copy-paste proc-macro server crate into rust-analyzer, and use the current ABI to load proc macros. Macros are executed in separate process, to avoid bringing down the whole IDE if the macro aborts.

matklad (May 07 2020 at 14:46, on Zulip):

Oh, the tracking issue for tracking RFC blockers is https://github.com/rust-analyzer/rust-analyzer/issues/4224

nikomatsakis (May 07 2020 at 14:47, on Zulip):

some of those blockers missing pieces (dyn Trait supertraits...) have pending PRs related to them in chalk, I know

matklad (May 07 2020 at 14:48, on Zulip):

To be clear, dyn Trait and nother type-inference failures are not really blockers, imo, type inference works ok-ish even without htme

matklad (May 07 2020 at 14:48, on Zulip):

rahter, they are the low-hanging fruits for improving experience for thouse who already use rust-analzer

matklad (May 07 2020 at 14:48, on Zulip):

blockers are for those, who can't use ra for whatever reason

pnkfelix (May 07 2020 at 14:49, on Zulip):

but were able to use previous RLS, right?

matklad (May 07 2020 at 14:49, on Zulip):

yup

matklad (May 07 2020 at 14:49, on Zulip):

Though, there's a bunch of people who can't use rls, but use rust-analyzer as well I think

pnkfelix (May 07 2020 at 14:50, on Zulip):

sure

matklad (May 07 2020 at 14:50, on Zulip):

In general, in terms of "user-visible functionality", I am pretty happy about current state. I am much less happy about "two compilers" situation

pnkfelix (May 07 2020 at 14:50, on Zulip):

just wanted to make clear that we wouldn't block switching to RA just because, I dunno, someone is mad that it doesn't work with their fork of vim on their Raspberry PI

matklad (May 07 2020 at 14:51, on Zulip):

(but we probably should switch to the other WG? Seems like I am going on a tangent :)

pnkfelix (May 07 2020 at 14:51, on Zulip):

yeah okay

pnkfelix (May 07 2020 at 14:51, on Zulip):

@Wesley Wiser posted their update for @WG-self-profile in the agenda, so I'll cut and paste it here.

Wesley Wiser (May 07 2020 at 14:51, on Zulip):

My update is short so if there's more to cover with ra, we can circle back.

pnkfelix (May 07 2020 at 14:51, on Zulip):

@WG-self-profile checkin by @Wesley Wiser:

Things have settled down quite a lot since the last check-in.
We achieved our MVP goal and since then, lots of people have been using the tools either directly or via the perf.rlo integration.

I think the only major news is that we have documentation everwhere that has been requested:

* GitHub readme files for each individual tool.
* rustc -Zhelp has some short help text describing the options.
* The Unstable Book has more complete documentation about the -Z flags.
* Rustc Dev Guide has a mention.

I also wrote a rather extensive post on the Inside-Rust blog about the tooling including a complete step-by-step tutorial.

While there additional polish that could be done, there are no remaining "must-have" features that I'm aware of.

pnkfelix (May 07 2020 at 14:51, on Zulip):

(oh, they were here!)

Wesley Wiser (May 07 2020 at 14:52, on Zulip):

No worries! Thanks for posting.

nikomatsakis (May 07 2020 at 14:52, on Zulip):

I have a question @Wesley Wiser

Wesley Wiser (May 07 2020 at 14:52, on Zulip):

Sure

nikomatsakis (May 07 2020 at 14:52, on Zulip):

something I hear fairly consistently is about the desire to figure out which parts of one's code are responsible for compilation time

nikomatsakis (May 07 2020 at 14:52, on Zulip):

I remember there being some amount of work in that direction

nikomatsakis (May 07 2020 at 14:53, on Zulip):

(and .. I think .. some requests for help (from me...eek) about accounting for time in trait solving)

pnkfelix (May 07 2020 at 14:53, on Zulip):

@nikomatsakis you mean which parts of one's source code being fed into rustc, right?

nikomatsakis (May 07 2020 at 14:53, on Zulip):

but that wasn't part of MVP, right?

nikomatsakis (May 07 2020 at 14:53, on Zulip):

@pnkfelix I don't know if I understand the distinction

pnkfelix (May 07 2020 at 14:53, on Zulip):

vs code of rustc itself

nikomatsakis (May 07 2020 at 14:53, on Zulip):

I'm mostly just wondering if there's some status update, and the status might be "yeah that's kind of stalled out" (and that seems fine)

eddyb (May 07 2020 at 14:54, on Zulip):

@nikomatsakis poke me later about trait solving stuff but that's definitely something I'm supposed to be working on

nikomatsakis (May 07 2020 at 14:54, on Zulip):

@pnkfelix ok right, i.e., if I were to edit my crate, what changes might help reduce my compilation time

Wesley Wiser (May 07 2020 at 14:54, on Zulip):

So we have support for recording query keys which means we can blame queries back to a DefId. I think the open question is how to represent blame because of the effects of query caching. If I have

fn foo() {
  bar::<u8>();
}

fn foo2() {
  bar::<u8>();
}

Do we blame (for example) well formd-ness checking of barr::<u8>() on foo or foo2?

Wesley Wiser (May 07 2020 at 14:55, on Zulip):

That might be a bad specific example but I think it gets the point across.

nikomatsakis (May 07 2020 at 14:55, on Zulip):

yeah, that's a tricky problem

eddyb (May 07 2020 at 14:55, on Zulip):

recording query keys can help today but it's hard to see them without e.g. using the Chrome profiler view

Wesley Wiser (May 07 2020 at 14:55, on Zulip):

We do have an issue for this request but no-one has come up with a good solution to that.

eddyb (May 07 2020 at 14:55, on Zulip):

and even then there's no aggregation by proximity in the source code

eddyb (May 07 2020 at 14:55, on Zulip):

you just have to get lucky with spotting an anomaly

Wesley Wiser (May 07 2020 at 14:55, on Zulip):

Yeah, that's a great point

pnkfelix (May 07 2020 at 14:56, on Zulip):

is there a way to at least add multiple defid's in, in response to the 2nd, 3rd, etc queries?

pnkfelix (May 07 2020 at 14:56, on Zulip):

i.e. be able to represent the reuse?

pnkfelix (May 07 2020 at 14:56, on Zulip):

(and "spread the blame", so to speak?)

nikomatsakis (May 07 2020 at 14:56, on Zulip):

I have thoughts but, among other things, I would note that this problem doesn't affect a "flat view", so much

nikomatsakis (May 07 2020 at 14:56, on Zulip):

i.e., just looking at self-time?

Wesley Wiser (May 07 2020 at 14:56, on Zulip):

I believe there's support for tracking query cache hits as well but last I recall it slowed down the compiler significantly

nikomatsakis (May 07 2020 at 14:56, on Zulip):

(but maybe that is just not esp. useful in practice)

Wesley Wiser (May 07 2020 at 14:56, on Zulip):

We issue a lot of queries

eddyb (May 07 2020 at 14:57, on Zulip):

OTOH that (chrome profile) file is JSON and I have done some analysis using jq on it

nikomatsakis (May 07 2020 at 14:57, on Zulip):

I don't mean a flat view of queries

nikomatsakis (May 07 2020 at 14:57, on Zulip):

I mean that if you just showed (e.g.) the def-ids

Wesley Wiser (May 07 2020 at 14:57, on Zulip):

Yeah

nikomatsakis (May 07 2020 at 14:57, on Zulip):

(like, how much time was spent in any query tagged with a particular def-id)

Wesley Wiser (May 07 2020 at 14:57, on Zulip):

Perhaps it's worth trying the naive thing and seeing if people actually find that useful.

nikomatsakis (May 07 2020 at 14:57, on Zulip):

with some amount being kind of "unknown" (because it's not ascribed to a single fn)

eddyb (May 07 2020 at 14:57, on Zulip):

yeah I can do stuff like that by hand now with jq

nikomatsakis (May 07 2020 at 14:58, on Zulip):

Wesley Wiser said:

Perhaps it's worth trying the naive thing and seeing if people actually find that useful.

I think it would be worth it, if it's not too hard

pnkfelix (May 07 2020 at 14:58, on Zulip):

I'm sorry, what is jq @eddyb ?

eddyb (May 07 2020 at 14:58, on Zulip):

JSON query tool

Wesley Wiser (May 07 2020 at 14:58, on Zulip):

Probably not! Certainly worth a few hours :slight_smile:

pnkfelix (May 07 2020 at 14:58, on Zulip):

/me googles

eddyb (May 07 2020 at 14:58, on Zulip):

it's basically awk/perl one-liners but for the JSON age :P

nikomatsakis (May 07 2020 at 14:59, on Zulip):

jq is nifty

eddyb (May 07 2020 at 14:59, on Zulip):

the other place I've used it in recently is https://www.reddit.com/r/rust/comments/g9mw57/oneliner_to_correctly_list_all_uses_ofunsafe_in/

eddyb (May 07 2020 at 14:59, on Zulip):

to filter Cargo output by lint

pnkfelix (May 07 2020 at 14:59, on Zulip):

/me reading jq home page, falls into the RIIR trap

pnkfelix (May 07 2020 at 15:01, on Zulip):

hup!

pnkfelix (May 07 2020 at 15:01, on Zulip):

that's an hour, folks!

pnkfelix (May 07 2020 at 15:01, on Zulip):

Thank you to everyone in @T-compiler/meeting for attending. And again, please do fill out those "State of the Compiler Team" surveys; they should not take long, its only like four or five questions.

pnkfelix (May 07 2020 at 15:01, on Zulip):

Stay safe, stay healthy!

Santiago Pastorino (May 07 2020 at 15:03, on Zulip):

eddyb said:

the other place I've used it in recently is https://www.reddit.com/r/rust/comments/g9mw57/oneliner_to_correctly_list_all_uses_ofunsafe_in/

-Funsafe-code flag is awesome, best name ever :slight_smile:

eddyb (May 07 2020 at 15:03, on Zulip):

it's just short for "forbid" :P

Santiago Pastorino (May 07 2020 at 15:03, on Zulip):

:P

pnkfelix (May 07 2020 at 15:04, on Zulip):

@eddyb c'mon take credit for even the accidental puns

Santiago Pastorino (May 07 2020 at 15:04, on Zulip):

gonna use this fun and safe code flag ... :slight_smile:

eddyb (May 07 2020 at 15:05, on Zulip):

@pnkfelix don't make me spend two hours in git blame to find whoever added these flags back in the day,,,

pnkfelix (May 07 2020 at 15:05, on Zulip):

some people spend ages coming up with things like Libiberty to achieve such goals

eddyb (May 07 2020 at 15:10, on Zulip):

don't remind me I have to ping those people again because they've been ignoring my rust demangler patch for probably months now

eddyb (May 07 2020 at 15:10, on Zulip):

I wonder if the person who owns the demanglers part of libiberty is ok

Santiago Pastorino (May 07 2020 at 15:14, on Zulip):

I needed to do this https://twitter.com/spastorino/status/1258414788822130689 sorry :joy:

oli (May 07 2020 at 15:21, on Zulip):

:D I saw on twitter before I saw this message

Last update: Nov 25 2020 at 02:15UTC