Stream: t-compiler

Topic: weekly meeting 2019-09-19 #54818


pnkfelix (Sep 19 2019 at 12:12, on Zulip):

Hi @T-compiler/meeting ; the triage meeting will be starting in 1 hours 48 minutes

pnkfelix (Sep 19 2019 at 12:13, on Zulip):

I will be doing pre-triage in a parallel topic

pnkfelix (Sep 19 2019 at 12:13, on Zulip):

this week we are scheduled to have a check-in from RLS 2.0 and Self-Profile

pnkfelix (Sep 19 2019 at 12:13, on Zulip):

@mw or @Wesley Wiser : can one of you do the summary for WG-self-profile?

pnkfelix (Sep 19 2019 at 12:14, on Zulip):

@matklad : can you do the summary for WG-rls-2.0?

pnkfelix (Sep 19 2019 at 12:14, on Zulip):

(to be clear, I'm just trying to give you all "plenty" of notice ahead of time)

matklad (Sep 19 2019 at 12:15, on Zulip):

@pnkfelix sure! What kind of summary would be most interesting? last week / last month / next month?

pnkfelix (Sep 19 2019 at 12:16, on Zulip):

I assume that everything that's happened since the last check-in with WG-rls-2.0 is up for grabs.

pnkfelix (Sep 19 2019 at 12:16, on Zulip):

/me looks to see when that was

pnkfelix (Sep 19 2019 at 12:17, on Zulip):

last sync was ... in May (?)

pnkfelix (Sep 19 2019 at 12:18, on Zulip):

oh well the minutes from the last N triage meetings seems to be missing from this site...

pnkfelix (Sep 19 2019 at 12:20, on Zulip):

/me wonders why that is; the minutes in repo seems up to date

matklad (Sep 19 2019 at 12:20, on Zulip):

yeah, it's been a while since the last check-in, which definitely happen before mozilla all-hands and vacations. Guess I'll give an n-months update then

pnkfelix (Sep 19 2019 at 12:21, on Zulip):

@Wesley Wiser any idea why the site doesn't seem to include the triage meeting notes since 2019-06-20 (despite their presence in repo source) ?

Santiago Pastorino (Sep 19 2019 at 12:23, on Zulip):

@pnkfelix if I'm not wrong that list is hard coded, I noticed that and opened an issue a while back

Santiago Pastorino (Sep 19 2019 at 12:23, on Zulip):

let me check

Santiago Pastorino (Sep 19 2019 at 12:24, on Zulip):

https://github.com/rust-lang/compiler-team/issues/153

Santiago Pastorino (Sep 19 2019 at 12:25, on Zulip):

was @Wesley Wiser who opened the issue actually :)

pnkfelix (Sep 19 2019 at 12:25, on Zulip):

thanks to both of you, okay

mw (Sep 19 2019 at 12:29, on Zulip):

@pnkfelix Yes, I can give an update for WG-self-profile

nikomatsakis (Sep 19 2019 at 12:48, on Zulip):

pnkfelix if I'm not wrong that list is hard coded, I noticed that and opened an issue a while back

oh dear we should... fix that :)

Santiago Pastorino (Sep 19 2019 at 12:54, on Zulip):

pnkfelix if I'm not wrong that list is hard coded, I noticed that and opened an issue a while back

oh dear we should... fix that :)

yes, I can try to do something simple I guess

pnkfelix (Sep 19 2019 at 13:25, on Zulip):

:construction_worker: request for help: "Better error message when attempting to instantiate tuple structs with private fields" #58017 is a bug that has been waiting for someone to fix it since January (potentially even earlier, as it hasn't even been bisected beyond knowing that its a regression compared to 1.16); there's lots of ways anyone can help here.

pnkfelix (Sep 19 2019 at 13:26, on Zulip):

(and also, if no one takes it, I might downgrade it to P-medium :wolf: )

simulacrum (Sep 19 2019 at 13:26, on Zulip):

It does not seem like a P-high bug to me :)

pnkfelix (Sep 19 2019 at 14:00, on Zulip):

okay

pnkfelix (Sep 19 2019 at 14:00, on Zulip):

Hi @T-compiler/meeting

pnkfelix (Sep 19 2019 at 14:01, on Zulip):

there are five minutes for general announcements.

nikomatsakis (Sep 19 2019 at 14:02, on Zulip):

Reminder for @T-compiler/meeting: There is a design meeting tomorrow at this time. The subject will be the the **target tier policy** meeting that @Josh Triplett proposed. You might want to take a look at the gist that they prepared which contains more details.

pnkfelix (Sep 19 2019 at 14:02, on Zulip):

:construction_worker: here is a link to my earlier request for help

nikomatsakis (Sep 19 2019 at 14:03, on Zulip):

Also, the week after that is the next planning meeting -- we've got a few meeting proposals, so maybe take a look. Maybe open one of your own. Don't feel intimidated -- even just a few sentences is good enough, I'd say!

pnkfelix (Sep 19 2019 at 14:03, on Zulip):

more generally,; there are currently eight unassigned P-high issues

pnkfelix (Sep 19 2019 at 14:04, on Zulip):

Maybe open one of your own.

(indeed, part of my plan for this meeting is to liberally turn nominated issues into meeting proposals, if we cannot resolve them in the time alloted here)

pnkfelix (Sep 19 2019 at 14:05, on Zulip):

but i'd welcome other people doing that in my stead. :)

pnkfelix (Sep 19 2019 at 14:06, on Zulip):

okay so that's all the announcements for now. (If anyone has other announcements, please private message me and I'll allocate time at the end for them.)

pnkfelix (Sep 19 2019 at 14:07, on Zulip):

so, first things first: beta-nominations

pnkfelix (Sep 19 2019 at 14:07, on Zulip):

we have two

pnkfelix (Sep 19 2019 at 14:07, on Zulip):

beta nom 1/2: "fn ptr is structural match" #64431

centril (Sep 19 2019 at 14:08, on Zulip):

I say backport; well tested, narrow code changes.

pnkfelix (Sep 19 2019 at 14:08, on Zulip):

yeah I think this is uncontroversial

pnkfelix (Sep 19 2019 at 14:09, on Zulip):

(devil's advocate: the use-case being addressed is pretty niche, and does have a work around.)

pnkfelix (Sep 19 2019 at 14:09, on Zulip):

but okay, lets just approve it.

pnkfelix (Sep 19 2019 at 14:10, on Zulip):

beta nom 2/2: "fix nounwind attribute logic" #63909

centril (Sep 19 2019 at 14:11, on Zulip):

move on, not getting backported.

pnkfelix (Sep 19 2019 at 14:11, on Zulip):

I haven't had a chance to review the dialogue here

pnkfelix (Sep 19 2019 at 14:12, on Zulip):

at time when niko posted "rfcbot merge", @nikomatsakis said "like to go ahead and merge and beta-nominate this"

pnkfelix (Sep 19 2019 at 14:13, on Zulip):

however, I will admit

pnkfelix (Sep 19 2019 at 14:13, on Zulip):

...

pnkfelix (Sep 19 2019 at 14:13, on Zulip):

/me re-reads

mw (Sep 19 2019 at 14:13, on Zulip):

shouldn't this at least land first?

centril (Sep 19 2019 at 14:13, on Zulip):

@pnkfelix I would suggest not spending too much meeting time on this; it's clearly not getting backported

pnkfelix (Sep 19 2019 at 14:13, on Zulip):

I don't see how that follows

pnkfelix (Sep 19 2019 at 14:13, on Zulip):

#62603 landed

pnkfelix (Sep 19 2019 at 14:14, on Zulip):

and was backported

pnkfelix (Sep 19 2019 at 14:14, on Zulip):

/me still re-reading

centril (Sep 19 2019 at 14:14, on Zulip):

I'm opposed to #63909 landing on nightly ~> not landing on beta

mw (Sep 19 2019 at 14:14, on Zulip):

#63909 is a PR that has seen no testing in the wild and we are days before beta promotion

pnkfelix (Sep 19 2019 at 14:15, on Zulip):

right, I am basically inclined to agree with that logic

pnkfelix (Sep 19 2019 at 14:15, on Zulip):

that we need to have the PR get testing in the wild

pnkfelix (Sep 19 2019 at 14:15, on Zulip):

but I also want to understand

pnkfelix (Sep 19 2019 at 14:15, on Zulip):

what is the ramification of landing and backporting #62603

pnkfelix (Sep 19 2019 at 14:15, on Zulip):

without landing #63909

pnkfelix (Sep 19 2019 at 14:16, on Zulip):

i.e. what state is the beta in now, in terms of what behavior one gets here?

centril (Sep 19 2019 at 14:16, on Zulip):

My understanding is that this has been the status quo for years.

pnkfelix (Sep 19 2019 at 14:16, on Zulip):

oh right oh right

pnkfelix (Sep 19 2019 at 14:16, on Zulip):

god what a mess

pnkfelix (Sep 19 2019 at 14:17, on Zulip):

okay so landing #62603 and not landing #63909 is in fact maintaining status quo

pnkfelix (Sep 19 2019 at 14:17, on Zulip):

is my understanding

pnkfelix (Sep 19 2019 at 14:17, on Zulip):

it may be non-ideal from various other viewpoints

centril (Sep 19 2019 at 14:17, on Zulip):

(I'm gonna remove beta-nominated until such time as this lands to not revisit this each time.)

nikomatsakis (Sep 19 2019 at 14:17, on Zulip):

My current position:

pnkfelix (Sep 19 2019 at 14:17, on Zulip):

but our status quo is pretty much in a perpetual state of non-idealness.

mw (Sep 19 2019 at 14:18, on Zulip):

but our status quo is pretty much in a perpetual state of non-idealness.

well, it's software, right? :P

centril (Sep 19 2019 at 14:19, on Zulip):

@mw hey, at least it's not hardware ;)

pnkfelix (Sep 19 2019 at 14:20, on Zulip):

okay so that's all the beta nominations

pnkfelix (Sep 19 2019 at 14:20, on Zulip):

there are no stable nominations

pnkfelix (Sep 19 2019 at 14:21, on Zulip):

there is one PR marked "waiting on team": "Turn HIR indexing into a query" #59064

pnkfelix (Sep 19 2019 at 14:21, on Zulip):

but I have allocated compiler-team#175 for this

nikomatsakis (Sep 19 2019 at 14:21, on Zulip):

I believe you scheduled a design meeting proposal around this topic

pnkfelix (Sep 19 2019 at 14:22, on Zulip):

so we need not discuss it here)

pnkfelix (Sep 19 2019 at 14:22, on Zulip):

so that just leaves ... five nominated issues

pnkfelix (Sep 19 2019 at 14:22, on Zulip):

nom 1/5: "Duplicate error code usage lint removed" #64426

pnkfelix (Sep 19 2019 at 14:22, on Zulip):

so I don't know when this happened

pnkfelix (Sep 19 2019 at 14:23, on Zulip):

but apparently we lost those checks for ensuring you didn't have two different diagnostics using the same error code

pnkfelix (Sep 19 2019 at 14:23, on Zulip):

and this issue asks the question: Do we need those lints anyway?

pnkfelix (Sep 19 2019 at 14:23, on Zulip):

personally, I don't think it was that terrible to deal with

centril (Sep 19 2019 at 14:23, on Zulip):

Recently, @simulacrum removed the hardcoded register macros and made them into ordinary macros...?

pnkfelix (Sep 19 2019 at 14:24, on Zulip):

(and I say this as someone who had to deal with a ton of diagnostics "duplication" between AST-borrowck and NLL)

pnkfelix (Sep 19 2019 at 14:24, on Zulip):

and I am scared of the potential for error codes to silently collide due to a bors merge

centril (Sep 19 2019 at 14:25, on Zulip):

I personally don't think the lints add lotsa values, but if we have them, let's do it via tidy and not via compiler hacks

nikomatsakis (Sep 19 2019 at 14:25, on Zulip):

I was not a fan of the old setup, but I agree with @pnkfelix that duplicates will arise if we don't have some mechanism for checking.

nikomatsakis (Sep 19 2019 at 14:25, on Zulip):

I don't care much whether it's in tidy or elsewhere

simulacrum (Sep 19 2019 at 14:25, on Zulip):

My opinion is that we probably need or want the definition duplication error but don't need use site checking

nikomatsakis (Sep 19 2019 at 14:26, on Zulip):

What was use-site checking?

pnkfelix (Sep 19 2019 at 14:26, on Zulip):

ah interesting

nikomatsakis (Sep 19 2019 at 14:26, on Zulip):

I found it annoying that there could be exactly one use

nikomatsakis (Sep 19 2019 at 14:26, on Zulip):

I often wanted more than one

nikomatsakis (Sep 19 2019 at 14:26, on Zulip):

(this led to things like span_err("{}", the_message))

simulacrum (Sep 19 2019 at 14:26, on Zulip):

I wasn't able to find the time to investigate what we used to do for use site checking but it seems like no one knows why that's useful

pnkfelix (Sep 19 2019 at 14:26, on Zulip):

okay, yes

mw (Sep 19 2019 at 14:26, on Zulip):

yes, tidy or something should check that no error code is re-used for different things, but not much else is needed

nikomatsakis (Sep 19 2019 at 14:26, on Zulip):

(anyway, another alternative would be to move from "numeric" error codes to more descriptive ones, like ERR_ILLEGAL_BORROW or something)

centril (Sep 19 2019 at 14:26, on Zulip):

@nikomatsakis agreed; I've often wanted to duplicate sometimes when the error was very similar

pnkfelix (Sep 19 2019 at 14:26, on Zulip):

the main motivation for use-site checking that I could imagine would be to try to prevent divergence in the messaging

pnkfelix (Sep 19 2019 at 14:27, on Zulip):

but 1. it doesn't prevent people working around it (as in, injecting divergence deliberately), and 2 ...

centril (Sep 19 2019 at 14:27, on Zulip):

more context: https://github.com/rust-lang/rust/pull/64139 was the PR that did the change; pretty nice cleanups!

pnkfelix (Sep 19 2019 at 14:27, on Zulip):

2. I'm willing to let those things crop up and deal with it as its discovered.

simulacrum (Sep 19 2019 at 14:28, on Zulip):

Okay, so action item is to make sure we check for no duplicate definitions but otherwise we will not readd functionality of use site checking

nikomatsakis (Sep 19 2019 at 14:28, on Zulip):

(anyway, another alternative would be to move from "numeric" error codes to more descriptive ones, like ERR_ILLEGAL_BORROW or something)

(this has the advantage of making merge conflicts less problematic)

simulacrum (Sep 19 2019 at 14:28, on Zulip):

I think someone is already doing this (but I can be the assignee to make sure it happens)

nikomatsakis (Sep 19 2019 at 14:28, on Zulip):

that sounds @simulacrum like the easiest short-term thing; I'd be in favor of reconsidering the whole execution of codes, but I also don't consider it very urgent

centril (Sep 19 2019 at 14:28, on Zulip):

@simulacrum are we doing this in tidy?

eddyb (Sep 19 2019 at 14:29, on Zulip):

cc @Manish Goregaokar last time this came up elsewhere, I brought up the potential integration with translation, or more structured diagnostics, too

eddyb (Sep 19 2019 at 14:29, on Zulip):

(and there might already be people working on those things)

pnkfelix (Sep 19 2019 at 14:29, on Zulip):

lets not let this blow up into a second system syndrome

pnkfelix (Sep 19 2019 at 14:29, on Zulip):

(or in our case, 22nd system ...)

eddyb (Sep 19 2019 at 14:29, on Zulip):

catch 22 system (as is appropriate for bootstrapped compilers)

centril (Sep 19 2019 at 14:29, on Zulip):

(the old system was ungreat for compiling crates separately... cc the IDE/library-ification discussions)

pnkfelix (Sep 19 2019 at 14:30, on Zulip):

okay so is there anyone who objects to @simulacrum 's plan?

pnkfelix (Sep 19 2019 at 14:30, on Zulip):

alright lets move along

pnkfelix (Sep 19 2019 at 14:30, on Zulip):

I'm going to assume that #63909 is nominated for the lang team

pnkfelix (Sep 19 2019 at 14:31, on Zulip):

nom 3/5: "Make the generator-locals debuginfo test resilient to re-ordering." #63742

nikomatsakis (Sep 19 2019 at 14:31, on Zulip):

this should probably be tagged A-async-await

mw (Sep 19 2019 at 14:31, on Zulip):

I think the nomination can be cleared

nikomatsakis (Sep 19 2019 at 14:31, on Zulip):

and let us triage it

nikomatsakis (Sep 19 2019 at 14:31, on Zulip):

(unless we've resolved it to our satisfaction)

mw (Sep 19 2019 at 14:31, on Zulip):

we talked about it last week

pnkfelix (Sep 19 2019 at 14:31, on Zulip):

okay yeah

pnkfelix (Sep 19 2019 at 14:32, on Zulip):

sorry, I really shoud be checking that as part of pre-triage each week

pnkfelix (Sep 19 2019 at 14:32, on Zulip):

/me makes a note

pnkfelix (Sep 19 2019 at 14:32, on Zulip):

nom 4/5: "Codegen forgets array size when used with iterators" #63552

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

ah I bet this was nominated for prioritization

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

does anyone want to take it on?

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

(its only P-medium)

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

((but might be fun))

nikomatsakis (Sep 19 2019 at 14:33, on Zulip):

good example for LLVM ICE-breakers team that I've not quite gotten around to setting up

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

in any case, clearing nomination.

nikomatsakis (Sep 19 2019 at 14:33, on Zulip):

I'd say de-nominate and move on :)

centril (Sep 19 2019 at 14:33, on Zulip):

maybe @scottmcm is interested in ^--- also

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

nom 5/5: "Stable rustc always panics on arm/musl" #60297

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

someone please take this bug

pnkfelix (Sep 19 2019 at 14:33, on Zulip):

:wink:

pnkfelix (Sep 19 2019 at 14:34, on Zulip):

I downgraded it to P-medium because the platform is tier 2, IIUC

pnkfelix (Sep 19 2019 at 14:34, on Zulip):

and oh

pnkfelix (Sep 19 2019 at 14:34, on Zulip):

@nikomatsakis , you said some people tweeted back to you that this may be working now?

nikomatsakis (Sep 19 2019 at 14:35, on Zulip):

yeah, they did, not sure what's going on

pnkfelix (Sep 19 2019 at 14:35, on Zulip):

maybe I'll just close it and let people reopen it if its indeed still broken.

pnkfelix (Sep 19 2019 at 14:35, on Zulip):

okay

mw (Sep 19 2019 at 14:35, on Zulip):

Someone said they heard on twitter that it's working now. Closing!

pnkfelix (Sep 19 2019 at 14:35, on Zulip):

you think that's a joke

pnkfelix (Sep 19 2019 at 14:36, on Zulip):

to me, its a solution to a long standing problem.

pnkfelix (Sep 19 2019 at 14:36, on Zulip):

okay

pnkfelix (Sep 19 2019 at 14:36, on Zulip):

Awesome we actually discussed all the nominated issues ....?!?

nikomatsakis (Sep 19 2019 at 14:36, on Zulip):

https://twitter.com/raggi/status/1172161378024476682

pnkfelix (Sep 19 2019 at 14:36, on Zulip):

and I didn't transcribe any of them to design meeting proposals ???!

nikomatsakis (Sep 19 2019 at 14:36, on Zulip):

I guess maybe a different scenario :)

pnkfelix (Sep 19 2019 at 14:36, on Zulip):

well this is awesome

pnkfelix (Sep 19 2019 at 14:37, on Zulip):

@mw you still here? Want to provide a checkin from WG-self-profile ?

mw (Sep 19 2019 at 14:37, on Zulip):

yup

mw (Sep 19 2019 at 14:37, on Zulip):

WG-self-profile actually had its first meeting in months today.

mw (Sep 19 2019 at 14:38, on Zulip):

There are 3 main things the WG is looking into at the moment:

mw (Sep 19 2019 at 14:38, on Zulip):
mw (Sep 19 2019 at 14:38, on Zulip):

^^ perf.rlo will collect profiling data and make comparison view available that let's you compare time spent for each query

mw (Sep 19 2019 at 14:38, on Zulip):

that's item one

mw (Sep 19 2019 at 14:38, on Zulip):

item two: Improving the ergonomics of self-profiling

mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):

item three: Use self-profiling to implement a feature that allows to attribute compile time to specific parts of the source code

mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:39, on Zulip):
mw (Sep 19 2019 at 14:40, on Zulip):
mw (Sep 19 2019 at 14:40, on Zulip):

questions?

Wesley Wiser (Sep 19 2019 at 14:41, on Zulip):

We're also soliciting feedback from people on irlo who have thoughts about the existing tooling. So if you have thoughts, please let us know!

https://internals.rust-lang.org/t/compiler-profiling-survey/10969

pnkfelix (Sep 19 2019 at 14:41, on Zulip):

@mw you shared some links with me back when I started using the tool last week

nikomatsakis (Sep 19 2019 at 14:41, on Zulip):

very exciting!

pnkfelix (Sep 19 2019 at 14:42, on Zulip):

they were helpful in terms of getting myself started

mw (Sep 19 2019 at 14:42, on Zulip):

@pnkfelix this one probably: https://github.com/rust-lang/measureme/blob/master/summarize/Readme.md

nikomatsakis (Sep 19 2019 at 14:42, on Zulip):

Have we thought about adding in any links to the rustc manual (not the "learn how to hack on rustc" guide, but the "learn how to use rustc") to cover this feature? (I think we should at least add some links to the measureme docs)

nikomatsakis (Sep 19 2019 at 14:42, on Zulip):

(minor thing obviously)

mw (Sep 19 2019 at 14:42, on Zulip):

I think we are not there yet

nikomatsakis (Sep 19 2019 at 14:42, on Zulip):

I think that's a mistake :)

Wesley Wiser (Sep 19 2019 at 14:42, on Zulip):

There is at least a mention on Profiling the Compiler: https://rust-lang.github.io/rustc-guide/profiling.html

nikomatsakis (Sep 19 2019 at 14:43, on Zulip):

well, not a mistake, I just mean: I wind up pointing at the measureme docs pretty regularly.. it'd be nice if they were easier to find

nikomatsakis (Sep 19 2019 at 14:43, on Zulip):

anyway not imp't, sorry to derail

pnkfelix (Sep 19 2019 at 14:43, on Zulip):

@mw at the very least, the measureme/Readme.md should provide pointer to measureme/summarize/Readme.md

pnkfelix (Sep 19 2019 at 14:43, on Zulip):

;)

pnkfelix (Sep 19 2019 at 14:43, on Zulip):

because the latter is way better than the former

mw (Sep 19 2019 at 14:43, on Zulip):

yeah, from my point of view this is an alpha level feature still

pnkfelix (Sep 19 2019 at 14:43, on Zulip):

okay anyway

nikomatsakis (Sep 19 2019 at 14:43, on Zulip):

I think that's a mistake :)

what I meant by this was: I think users can understand that the feature is in alpha. Maybe I'll stop complaining and just open a PR..

pnkfelix (Sep 19 2019 at 14:44, on Zulip):

this is great stuff

pnkfelix (Sep 19 2019 at 14:44, on Zulip):

though of course, only of use to rustc devs

pnkfelix (Sep 19 2019 at 14:44, on Zulip):

so lets hear about great stuff for everyone else

mw (Sep 19 2019 at 14:44, on Zulip):

I think that's a mistake :)

what I meant by this was: I think users can understand that the feature is in alpha. Maybe I'll stop complaining and just open a PR..

OK, I can understand that :)

pnkfelix (Sep 19 2019 at 14:44, on Zulip):

(that's my segue to @matklad from WG-rls-2.0)

matklad (Sep 19 2019 at 14:44, on Zulip):

:)

matklad (Sep 19 2019 at 14:45, on Zulip):

So, WG-rls-2.0 havent' reported to t-compiler for a long while, so I'll try to summarize large things that happedn

matklad (Sep 19 2019 at 14:47, on Zulip):
mw (Sep 19 2019 at 14:47, on Zulip):

nice!

matklad (Sep 19 2019 at 14:49, on Zulip):
pnkfelix (Sep 19 2019 at 14:49, on Zulip):

what does "pretty incremental" mean here?

matklad (Sep 19 2019 at 14:49, on Zulip):

Note that the code in rust-analyzer for this is not in a very good shape though: if someone wants to drive name resolution and macro exapnsion in rust-analyzer, feel free to talk to me :)

matklad (Sep 19 2019 at 14:50, on Zulip):

pretty incremental == incremental enough to make completion fast

matklad (Sep 19 2019 at 14:51, on Zulip):
matklad (Sep 19 2019 at 14:52, on Zulip):

(btw, we have a meeting about rls/rust-analyzer strategy later today, if you are interested, feel free to join, there's a neightouring zulip thread for it)

pnkfelix (Sep 19 2019 at 14:52, on Zulip):

where "bridge" here means: Provide utility that effectively matches the diagnostic feedback provided by RLS 1.0, but actually use none of the RLS 1.0 code?

matklad (Sep 19 2019 at 14:53, on Zulip):

@pnkfelix something like this: we guaratee that error highgliting is precise (but not fast), and that completion is fast (but incomplete). As opposed to having only one of these two things

matklad (Sep 19 2019 at 14:54, on Zulip):

And, with RLS moving to out of process architecture, I actually hope to share the code, but I won't talk about it in this meeting :D

matklad (Sep 19 2019 at 14:55, on Zulip):
pnkfelix (Sep 19 2019 at 14:55, on Zulip):

(btw, we have a meeting about rls/rust-analyzer strategy later today, if you are interested, feel free to join, there's a neightouring zulip thread for it)

link to zulip thread

pnkfelix (Sep 19 2019 at 14:56, on Zulip):

/me is forced to admit that he did use play with rust-analyzer earlier in the year but has not kept up with it

matklad (Sep 19 2019 at 14:56, on Zulip):
centril (Sep 19 2019 at 14:57, on Zulip):

@pnkfelix do it! rust-analyzer is GREAT for hacking on rustc!

matklad (Sep 19 2019 at 14:58, on Zulip):

Yeah, it looks like folks find rust-analyzer actually useful: https://www.reddit.com/r/rust/comments/d59c41/comment/f0kmc9p

matklad (Sep 19 2019 at 15:00, on Zulip):

And yeah, that's obvious, but call for participation: t-compiler experience is sorely needed in rust-analyzer :)

pnkfelix (Sep 19 2019 at 15:01, on Zulip):

okay well this definitely sounds exciting

pnkfelix (Sep 19 2019 at 15:01, on Zulip):

and hopefully motivating for everyone to go out and try to use rust-analyzer

pnkfelix (Sep 19 2019 at 15:01, on Zulip):

(I was scared to even propose the Q of whether it could deal with the rustc code base, so I'm glad @centril beat me to it... assuming their comment wasn't in jest...)

pnkfelix (Sep 19 2019 at 15:02, on Zulip):

So, yeah. This is all awesome

pnkfelix (Sep 19 2019 at 15:02, on Zulip):

And let me just say: It feels really good to have a meeting where we actually went through all the nominated issues

pnkfelix (Sep 19 2019 at 15:03, on Zulip):

next week, maybe we'll shoot for going through all the P-high issues. All ... thirty-f... never mind.

centril (Sep 19 2019 at 15:03, on Zulip):

@pnkfelix not even a little bit in jest; switched from RLS to RA on my 7 year old Windows laptop; the code itself lives on my linux build server to which I connect via a sshfs share -- it works pretty smoothly, the types are mostly there, and it's snappy

pnkfelix (Sep 19 2019 at 15:03, on Zulip):

So, great meeting, thanks for coming everyone in @T-compiler/meeting

centril (Sep 19 2019 at 15:03, on Zulip):

Definitely improved my rustc hackery productivity

Last update: Nov 20 2019 at 02:10UTC