Stream: t-compiler

Topic: weekly meeting 2019-06-06 #54818


pnkfelix (Jun 06 2019 at 10:42, on Zulip):

Hi @T-compiler/meeting ; the triage meeting will be starting in about 3 hours and 15 minutes

pnkfelix (Jun 06 2019 at 10:43, on Zulip):

As usual I will be doing pre-triage in a parallel topic

pnkfelix (Jun 06 2019 at 10:51, on Zulip):

We didn't have a meeting last week. Skimming over the WG check-in schedule, it seems lke we haven't heard from wg-parallel-rustc, wg-mir-opt, nor wg-pipelining in a while.

pnkfelix (Jun 06 2019 at 10:52, on Zulip):

I know @mw is on leave. @Zoxc , are you around? Is there anything to report about parallel-rustc?

pnkfelix (Jun 06 2019 at 10:52, on Zulip):

And for WG-mir-opt: @oli , are you around today?

pnkfelix (Jun 06 2019 at 10:53, on Zulip):

(I'm hesitant to JIT-schedule an update from WG-pipelining because the two leads of that WG do not typically come the rustc triage. Better to give them more advance notice, I think.)

oli (Jun 06 2019 at 10:54, on Zulip):

I'm around, but maybe not at the beginning of the mtg

oli (Jun 06 2019 at 10:54, on Zulip):

can I have a slot in the end?

pnkfelix (Jun 06 2019 at 10:54, on Zulip):

@oli sure, the WG checkin's rarely start until the half-hour mark, at the earliest, right?

pnkfelix (Jun 06 2019 at 10:54, on Zulip):

and since we missed a meeting last week, I can imagine triage taking up the full 30 minutes easily

oli (Jun 06 2019 at 10:54, on Zulip):

ah cool, I'll do the mir-opt check-in then

Zoxc (Jun 06 2019 at 11:59, on Zulip):

@pnkfelix I'm half awake, but there's nothing interesting to report

eddyb (Jun 06 2019 at 12:57, on Zulip):

oh right I should make sure Zulip is working in case https://github.com/rust-lang/rust/issues/61539 gets discussed

pnkfelix (Jun 06 2019 at 14:06, on Zulip):

Hi @T-compiler/meeting , time to start, better late than never, right?

pnkfelix (Jun 06 2019 at 14:06, on Zulip):

So, are there any announcements?

pnkfelix (Jun 06 2019 at 14:08, on Zulip):

I'll start off by just noting that I didn't finish the pre-triage, but I did at least get through all the nominated issues that lacked priority labels...

simulacrum (Jun 06 2019 at 14:08, on Zulip):

As a potentially important note to y'all, cfg(stage0) is no longer a thing, use cfg(bootstrap) instead

pnkfelix (Jun 06 2019 at 14:08, on Zulip):

/me briefly thought (presumably incorrectly) that meant stage0 was no longer a thing

nikomatsakis (Jun 06 2019 at 14:08, on Zulip):

hey all :wave: sorry i'm late, my daughter had a school event that ran over

simulacrum (Jun 06 2019 at 14:09, on Zulip):

Yeah, no, stage0 is still a thing just renamed the cfg flag

pnkfelix (Jun 06 2019 at 14:09, on Zulip):

Also, just note we have 14 open P-high bugs with no assignee

eddyb (Jun 06 2019 at 14:10, on Zulip):

one of the ideas where cfg(bootstrap) came up in is not setting it when stage0 is a recent nightly or similar (as opposed to starting the build at stage1) - this could simplify builds for some people

eddyb (Jun 06 2019 at 14:10, on Zulip):

(maybe @simulacrum or I should've opened an issue about it)

pnkfelix (Jun 06 2019 at 14:10, on Zulip):

and that's not counting the 14 beta-regressions with no assignee that I didn't manage to pre-triage

simulacrum (Jun 06 2019 at 14:11, on Zulip):

@eddyb probably not the time to discuss it but that's one of the reasons for the rename :)

pnkfelix (Jun 06 2019 at 14:11, on Zulip):

so arguably that's 28 distinct issues that have no owners

eddyb (Jun 06 2019 at 14:11, on Zulip):

bringing it up in case someone gets interested it in and it's less forgotten

eddyb (Jun 06 2019 at 14:11, on Zulip):

also, @Zoxc merged the removal of the 'gcx/'tcx split (except everything still has the lifetimes. I'm toying around with getting rid of some of that, hopefully I won't step on anyone's toes :P)

main thing is: there's going to be some friction in the future along those lines until the transition is finished - maybe we should discuss on Friday

pnkfelix (Jun 06 2019 at 14:12, on Zulip):

@eddyb you mean you expect there to be a high rate of non-mergable PR collisions, right?

eddyb (Jun 06 2019 at 14:13, on Zulip):

and bitrot, yeah

pnkfelix (Jun 06 2019 at 14:13, on Zulip):

right. But the "removal" has been merged?

nikomatsakis (Jun 06 2019 at 14:13, on Zulip):

and that's not counting the 14 beta-regressions with no assignee that I didn't manage to pre-triage

are these individual crates or have they been "merged" in some sense?

nikomatsakis (Jun 06 2019 at 14:13, on Zulip):

I guess we need to go through those one by one

eddyb (Jun 06 2019 at 14:14, on Zulip):

yeah, we no longer throw away types that contain inference variables after every e.g. function body typeck

pnkfelix (Jun 06 2019 at 14:14, on Zulip):

are these individual crates or have they been "merged" in some sense?

I think there's been some amount of merging

simulacrum (Jun 06 2019 at 14:14, on Zulip):

they've been merged as much as is feasible without serious investigation, I expect few duplicates

pnkfelix (Jun 06 2019 at 14:14, on Zulip):

yeah, we no longer throw away types that contain inference variables after every e.g. function body typeck

So we're paying the memory cost but not yet getting the Development Experience benefit, right?

eddyb (Jun 06 2019 at 14:15, on Zulip):

right, memory usage will not go up from now on, but the code will gradually get cleaner

pnkfelix (Jun 06 2019 at 14:15, on Zulip):

okay

eddyb (Jun 06 2019 at 14:15, on Zulip):

(assuming we can't do it all at once because oh-god-so-much-code-to-clean-up-I-wish-I-never-did-this-in-the-first-place)

pnkfelix (Jun 06 2019 at 14:15, on Zulip):

@eddyb do you think it would be worthwhile to try to have a concentrated focus on the cleanup?

nikomatsakis (Jun 06 2019 at 14:15, on Zulip):

/me notes that down for future reference; this means I can really simplify the chalk integration :)

eddyb (Jun 06 2019 at 14:16, on Zulip):

@pnkfelix let's land a first step from @Zoxc or I and then we can discuss it further

pnkfelix (Jun 06 2019 at 14:16, on Zulip):

I.e. if its mechanical ... then maybe it can be distributed. But that's assuming the work can be parallelized, which is unlikely, right?

nikomatsakis (Jun 06 2019 at 14:16, on Zulip):

My opinion is that we should be looking at these regressions / P-high bugs at the moment, not 'gcx/'tcx :)

eddyb (Jun 06 2019 at 14:16, on Zulip):

could be a week, could be a month, not sure yet

nikomatsakis (Jun 06 2019 at 14:16, on Zulip):

(though I do think that's a good thing to discuss in general)

eddyb (Jun 06 2019 at 14:16, on Zulip):

yes sorry I just wanted to mention it in case anyone missed it

eddyb (Jun 06 2019 at 14:16, on Zulip):

and then move on :)

pnkfelix (Jun 06 2019 at 14:16, on Zulip):

My opinion is that we should be looking at these regressions / P-high bugs at the moment, not 'gcx/'tcx :)

it was mostly the collisions I was worried about

pnkfelix (Jun 06 2019 at 14:17, on Zulip):

those P-high bugs won't get resolved if PR's keep bouncing off of people removing liftetmes.

pnkfelix (Jun 06 2019 at 14:17, on Zulip):

but that's more a question of coordinating the timing of the removal itself

pnkfelix (Jun 06 2019 at 14:17, on Zulip):

Which we do not need to discuss now.

pnkfelix (Jun 06 2019 at 14:17, on Zulip):

So yeah

eddyb (Jun 06 2019 at 14:17, on Zulip):

I have some ideas for that, we can discuss later if you want

pnkfelix (Jun 06 2019 at 14:18, on Zulip):

there are 29 open P-high T-compiler bugs, 15 of them are assigned

nikomatsakis (Jun 06 2019 at 14:18, on Zulip):

JFYI: I'm going through the regressions and I'll take a few notes

pnkfelix (Jun 06 2019 at 14:18, on Zulip):

and as mentioned, 14 are unassigned

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

it would be good to try to get owners for each of these

pnkfelix (Jun 06 2019 at 14:20, on Zulip):

I'm going to pick out a few that caught my eye

pnkfelix (Jun 06 2019 at 14:20, on Zulip):

first, "Creating a recursive type with infinite size leads to internal compiler error" #61323

pnkfelix (Jun 06 2019 at 14:20, on Zulip):

this is something going astray in the dependency graph code

pnkfelix (Jun 06 2019 at 14:21, on Zulip):

does anyone own that besides @mw ?

pnkfelix (Jun 06 2019 at 14:21, on Zulip):

(@mw is on parental leave, IIUC)

nikomatsakis (Jun 06 2019 at 14:21, on Zulip):

I can own it

pnkfelix (Jun 06 2019 at 14:21, on Zulip):

you have a lot of stuff on your plate

pnkfelix (Jun 06 2019 at 14:21, on Zulip):

is there anyone else you can think of?

pnkfelix (Jun 06 2019 at 14:22, on Zulip):

(in general I've been wondering if we might need a WG-incr-comp, to address some things that have arisen)

simulacrum (Jun 06 2019 at 14:22, on Zulip):

@Zoxc perhaps?

pnkfelix (Jun 06 2019 at 14:22, on Zulip):

(but this not the time to discuss forming a WG, i think. or maybe there already is one, I cannot recall)

nikomatsakis (Jun 06 2019 at 14:22, on Zulip):

I don't think we have one

pnkfelix (Jun 06 2019 at 14:22, on Zulip):

@Zoxc are you well versed in the dependency-graph code? Would you be able to look at bugs like #61323 and #57373 ?

nikomatsakis (Jun 06 2019 at 14:23, on Zulip):

you have a lot of stuff on your plate

I do, I wonder though if somebody would be game to help me :)

nikomatsakis (Jun 06 2019 at 14:23, on Zulip):

i.e., I'd be happy to sit for an hour with somebody and try to debug it

pnkfelix (Jun 06 2019 at 14:23, on Zulip):

okay. Myabe I will assign to niko

nikomatsakis (Jun 06 2019 at 14:23, on Zulip):

(hint hint)

davidtwco (Jun 06 2019 at 14:23, on Zulip):

I've got a bunch of things I'm already behind on, but I'm always up for adding to that.

pnkfelix (Jun 06 2019 at 14:23, on Zulip):

with the intention that niko should focus on trying to delegate it to someone else

pnkfelix (Jun 06 2019 at 14:23, on Zulip):

okay?

nikomatsakis (Jun 06 2019 at 14:23, on Zulip):

lol -- depends how urgent they are I guess :)

simulacrum (Jun 06 2019 at 14:23, on Zulip):

I can probably try and help with the dep graph stuff, yeah

simulacrum (Jun 06 2019 at 14:24, on Zulip):

(given Niko's help :)

nikomatsakis (Jun 06 2019 at 14:24, on Zulip):

let's open a topic to discuss

nikomatsakis (Jun 06 2019 at 14:24, on Zulip):

I see no reason it has to be just 1 person either :)

pnkfelix (Jun 06 2019 at 14:25, on Zulip):

okay, here is the another bug I noticed (I don't think we'll have time to look at many, at this rate): "ICE when trying to match on non-PartialEq slice." #61188

pnkfelix (Jun 06 2019 at 14:26, on Zulip):

@eddyb made a recent comment there following up on a question posed

eddyb (Jun 06 2019 at 14:26, on Zulip):

wait there was a backtrace, I haven't read it

pnkfelix (Jun 06 2019 at 14:26, on Zulip):

it might actually be a decent mentorship bug

eddyb (Jun 06 2019 at 14:27, on Zulip):

oh it is in mir_borrowck

eddyb (Jun 06 2019 at 14:27, on Zulip):

weird that it calls into codegen_fulfill_obligation

eddyb (Jun 06 2019 at 14:27, on Zulip):

(something something we should've renamed that to not say "codegen" :P)

pnkfelix (Jun 06 2019 at 14:27, on Zulip):

is that not an effect of lazily evaluating things via queries?2

eddyb (Jun 06 2019 at 14:28, on Zulip):

ehm, sorry, mir_build called from mir_borrowck

pnkfelix (Jun 06 2019 at 14:28, on Zulip):

anyway

eddyb (Jun 06 2019 at 14:28, on Zulip):

can we get a backtrace with debuginfo? this one is almost useless save for the queries

pnkfelix (Jun 06 2019 at 14:28, on Zulip):

at this rate I guess I will follow niko's example and assign to self with intent to mentor

pnkfelix (Jun 06 2019 at 14:28, on Zulip):

unless someone else wants to do that

eddyb (Jun 06 2019 at 14:28, on Zulip):

which I should've read first, it says "#1 [mir_built] processing main"

eddyb (Jun 06 2019 at 14:29, on Zulip):

so my suspicion is now that the MIR building code uses Instance::resolve for something that rustc_typeck didn't check

pnkfelix (Jun 06 2019 at 14:29, on Zulip):

right.

pnkfelix (Jun 06 2019 at 14:30, on Zulip):

@eddyb keep in mind that the goal right now is not to actually resolve the bug, unless its trivial, but rather to identify someone who can own it

eddyb (Jun 06 2019 at 14:31, on Zulip):

sorry, yes. @davidtwco and @varkor seem good candidates for this area

eddyb (Jun 06 2019 at 14:31, on Zulip):

or @Matthew Jasper who reported it

pnkfelix (Jun 06 2019 at 14:31, on Zulip):

Unfortunately we do not have time to go through all the P-high bugs in this manner

nikomatsakis (Jun 06 2019 at 14:31, on Zulip):

OK, I finished going through the beta regressions. I took summaries in this hackmd doc -- the good news is that many are expected or seem similar. But there are a few question marks. I'll try to organize a bit better.

nikomatsakis (Jun 06 2019 at 14:32, on Zulip):

(If @QuietMisdreavus is around, perhaps they can confirm whether there have been any rustdoc fixes that would cause us to detect more tests than we did before)

pnkfelix (Jun 06 2019 at 14:32, on Zulip):

in particular, I want to make sure we get to the backport nominations before we run out of time

eddyb (Jun 06 2019 at 14:32, on Zulip):

also cc @Guillaume

pnkfelix (Jun 06 2019 at 14:32, on Zulip):

so lets look at those now

pnkfelix (Jun 06 2019 at 14:33, on Zulip):

first, beta-nominations for T-compiler

pnkfelix (Jun 06 2019 at 14:33, on Zulip):

"Fix regression 61475" #61500

simulacrum (Jun 06 2019 at 14:33, on Zulip):

(If QuietMisdreavus is around, perhaps they can confirm whether there have been any rustdoc fixes that would cause us to detect more tests than we did before)

fwiw I think I checked all of those and it's breaking on tests that _ran_ successfully on stable

QuietMisdreavus (Jun 06 2019 at 14:34, on Zulip):

@nikomatsakis there was a pulldown-cmark update that had a bug

pnkfelix (Jun 06 2019 at 14:34, on Zulip):

for reference, the bug that fixes is "regression: expected expression, found &&" #61475

QuietMisdreavus (Jun 06 2019 at 14:34, on Zulip):

i approved a backport that updated it again to fix that bug

nikomatsakis (Jun 06 2019 at 14:34, on Zulip):

@QuietMisdreavus yes that was listed as possibly fixing some of the regressions

nikomatsakis (Jun 06 2019 at 14:34, on Zulip):

we should test though...

eddyb (Jun 06 2019 at 14:34, on Zulip):

@nikomatsakis for #61476 I think we could grep the history for changes that mention Borrow in the diff - not sure I remember how to but I can do it if you want

nikomatsakis (Jun 06 2019 at 14:35, on Zulip):

fwiw I think I checked all of those and it's breaking on tests that _ran_ successfully on stable

oh? it didn't look that way to me. but I'll double check.

pnkfelix (Jun 06 2019 at 14:35, on Zulip):

...

pnkfelix (Jun 06 2019 at 14:35, on Zulip):

Okay so I'll let this conversation run its course and then restart the nominations.

nikomatsakis (Jun 06 2019 at 14:35, on Zulip):

sorry :)

pnkfelix (Jun 06 2019 at 14:36, on Zulip):

no its okay, you posted that hackmd, I probably should have let there be some time to discuss it

pnkfelix (Jun 06 2019 at 14:36, on Zulip):

but I also got a side-channel request asking that we not miss the backport nominations

pnkfelix (Jun 06 2019 at 14:36, on Zulip):

so I was rushing it

pnkfelix (Jun 06 2019 at 14:37, on Zulip):

is there more to discuss then regarding the beta regressions? I know everything looks a bit out of control.

nikomatsakis (Jun 06 2019 at 14:37, on Zulip):

@simulacrum am I correct that the "start" links are stable? (e.g., in #61562)?

nikomatsakis (Jun 06 2019 at 14:37, on Zulip):

is there more to discuss then regarding the beta regressions? I know everything looks a bit out of control.

let's do the beta nominations

centril (Jun 06 2019 at 14:38, on Zulip):

Please make sure to discuss the stable nominations

nikomatsakis (Jun 06 2019 at 14:38, on Zulip):

we can do the regressions a bit after if needed

centril (Jun 06 2019 at 14:38, on Zulip):

They are arguably most important

pnkfelix (Jun 06 2019 at 14:38, on Zulip):

okay so link to first beta nom

pnkfelix (Jun 06 2019 at 14:39, on Zulip):

it seems pretty safe to backport to me, and the votes so far seem to agree.

pnkfelix (Jun 06 2019 at 14:39, on Zulip):

beta backport? : "Turn turbo :fish: :ice_cream: into an error" #61189

pnkfelix (Jun 06 2019 at 14:40, on Zulip):

(this has also been nominated for a stable backport; but lets cover beta on its own first.)

pnkfelix (Jun 06 2019 at 14:41, on Zulip):

it seems sensible to beta backport that

pnkfelix (Jun 06 2019 at 14:42, on Zulip):

okay lets now ask the follow-up Q

pnkfelix (Jun 06 2019 at 14:42, on Zulip):

stable backport? : "Turn turbo :fish: :ice_cream: into an error" #61189

pnkfelix (Jun 06 2019 at 14:43, on Zulip):

yeah okay I guess I don't have a problem with a stable backport either

pnkfelix (Jun 06 2019 at 14:44, on Zulip):

but what was #61085 then?

pnkfelix (Jun 06 2019 at 14:44, on Zulip):

oh oh I see, that didn't land

Esteban Küber (Jun 06 2019 at 14:44, on Zulip):

Stable predates the defid-> res change right?

nikomatsakis (Jun 06 2019 at 14:44, on Zulip):

It seems ok to stable backport. I'm not sure how strongly motivated it is if there aren't other things.

simulacrum (Jun 06 2019 at 14:44, on Zulip):

fwiw once y'all approve then release team will decide whether to actually release

pnkfelix (Jun 06 2019 at 14:44, on Zulip):

@nikomatsakis yeah the impression I have is that we're just supposed to make individual decisions about what to put into stable branch, and then release team decides whether its actually worth putting out a point release?

simulacrum (Jun 06 2019 at 14:45, on Zulip):

at this point I'm not expecting one but I haven't investigated fall out of all the stable backport reasons too thoroughly yet

pnkfelix (Jun 06 2019 at 14:45, on Zulip):

or is the decision about whether it even goes into the stable branch itself also the purview of the release team?

pnkfelix (Jun 06 2019 at 14:46, on Zulip):

anyway I digressed

pnkfelix (Jun 06 2019 at 14:46, on Zulip):

last beta nomination: "Fix overflowing literal lint in loops" #61098

pnkfelix (Jun 06 2019 at 14:47, on Zulip):

okay, all have been beta accepted, and #61189 was also stable accepted.

nagisa (Jun 06 2019 at 14:47, on Zulip):

/me shows up

pnkfelix (Jun 06 2019 at 14:47, on Zulip):

lets go through the other stable nominations now, and then we'll do WG-checkin

pnkfelix (Jun 06 2019 at 14:48, on Zulip):

stable backport? : "Fix more escaping ReScopes" #60765

pnkfelix (Jun 06 2019 at 14:49, on Zulip):

(I infer this is already in beta due to how the beta-cut was timed, right?)

nagisa (Jun 06 2019 at 14:49, on Zulip):

Is this a soundness issue fix?

pnkfelix (Jun 06 2019 at 14:49, on Zulip):

I'm not so sure we need to backport this one

nagisa (Jun 06 2019 at 14:49, on Zulip):

the original issue does not have a tag to that effect

oli (Jun 06 2019 at 14:50, on Zulip):

no, just an ICE fix

centril (Jun 06 2019 at 14:50, on Zulip):

(but an ICE fix that unbreaks stable working code?)

nikomatsakis (Jun 06 2019 at 14:51, on Zulip):

Is this a soundness issue fix?

it might be -- that old code was flat out wrong

pnkfelix (Jun 06 2019 at 14:51, on Zulip):

what do people think? THe original bug was a stable-to-stable regression, yielding an ICE on working code

nikomatsakis (Jun 06 2019 at 14:51, on Zulip):

I was very happy when I rebased over this PR with the work i'm doing on improving region inference and opaque types :)

pnkfelix (Jun 06 2019 at 14:52, on Zulip):

the risk itself seems pretty low, I guess

nagisa (Jun 06 2019 at 14:52, on Zulip):

I marked :shrug: because this is "just" an ICE fix, rather than a soundness one. I would vote :back: if there was a known stable point release coming but I don’t think we have plans for one quite yet?

centril (Jun 06 2019 at 14:53, on Zulip):

I would vote :back: if there was a known stable point release coming but I don’t think we have plans for one quite yet?

@nagisa you should focus on the individual fix; not make decisions based on the chances of a point release.

pnkfelix (Jun 06 2019 at 14:53, on Zulip):

I would vote :back: if there was a known stable point release coming but I don’t think we have plans for one quite yet?

I think, see above, that the intent is that we should just try to put stable channel into an idealized state

pnkfelix (Jun 06 2019 at 14:53, on Zulip):

and release team decides whether to release

simulacrum (Jun 06 2019 at 14:53, on Zulip):

When making stable backport decisions I would assume that there _is_ a point release because we put everything approved in one if it does happen

centril (Jun 06 2019 at 14:53, on Zulip):

The release team will decide if the total amount of stable backports merit a point release

simulacrum (Jun 06 2019 at 14:53, on Zulip):

(and if not, then nothing)

pnkfelix (Jun 06 2019 at 14:54, on Zulip):

(right, okay, see also my above question about whether the stable channel itself is maintained by T-compiler or by T-release )

pnkfelix (Jun 06 2019 at 14:54, on Zulip):

okay well I guess we'll go ahead and approve this for stable backport

pnkfelix (Jun 06 2019 at 14:55, on Zulip):

last one

pnkfelix (Jun 06 2019 at 14:55, on Zulip):

stable backport? : "Constrain all regions in the concrete type for an opaque type" #60449

simulacrum (Jun 06 2019 at 14:55, on Zulip):

stable branch is maintained by T-release with T-compiler approving new commits onto it (but not generally actually landing them)

pnkfelix (Jun 06 2019 at 14:57, on Zulip):

I nominated this, but that doesn't mean I was endorsing it. :)

pnkfelix (Jun 06 2019 at 14:57, on Zulip):

in particular, I'm not sure whether there's any stable-to-stable regression this fixes.

pnkfelix (Jun 06 2019 at 14:57, on Zulip):

oh no, #57464 was one

pnkfelix (Jun 06 2019 at 14:57, on Zulip):

so never mind me

pnkfelix (Jun 06 2019 at 14:58, on Zulip):

okay lets stable backport then

centril (Jun 06 2019 at 14:58, on Zulip):

(beta branch is also maintained by T-release but we don't require full team approval to land beta backports and just do them in rollups when various teams approve backports)

pnkfelix (Jun 06 2019 at 14:58, on Zulip):

Okay taht's all the backports

pnkfelix (Jun 06 2019 at 14:58, on Zulip):

Hey @oli you want to do a WG-mir-opt check in within ... 2 minutes ?

centril (Jun 06 2019 at 14:58, on Zulip):

(master branch is also sorta maintained by t-release in terms of managing the bors queue)

oli (Jun 06 2019 at 14:58, on Zulip):
oli (Jun 06 2019 at 14:58, on Zulip):

so we've been busy

oli (Jun 06 2019 at 14:59, on Zulip):

some perf tests have 30% improvements

oli (Jun 06 2019 at 14:59, on Zulip):

per PR

eddyb (Jun 06 2019 at 14:59, on Zulip):

@oli can you cc me on Place PRs? I might not review but I still am curious what's being done

oli (Jun 06 2019 at 14:59, on Zulip):

ok

centril (Jun 06 2019 at 14:59, on Zulip):

@oli is that 30% I-slow or I-compiletime?

oli (Jun 06 2019 at 14:59, on Zulip):

compile-time

nikomatsakis (Jun 06 2019 at 14:59, on Zulip):

(compile time, I presume, since we don't measure the latter -- unfortunately!)

centril (Jun 06 2019 at 14:59, on Zulip):

yeah that makes more sense =) 30% I-slow improvements would be ludicrous...

eddyb (Jun 06 2019 at 15:00, on Zulip):

(something something lolbench)

oli (Jun 06 2019 at 15:00, on Zulip):

comments on https://github.com/rust-lang/rust/pull/61532 would be great since we have a 4% regression that we have no idea where it comes from

oli (Jun 06 2019 at 15:00, on Zulip):

that's it from wg-mir-opt

eddyb (Jun 06 2019 at 15:01, on Zulip):

@oli you know, I should be in that WG, and I should be focusing on some of that stuff too - let's talk elsewhere after the meeting ;)

pnkfelix (Jun 06 2019 at 15:01, on Zulip):

thanks @oli

pnkfelix (Jun 06 2019 at 15:01, on Zulip):

okay then, thanks everyone in @T-compiler/meeting for attending

centril (Jun 06 2019 at 15:07, on Zulip):

@oli wg-grammar wants @eddyb all too ourselves =P Our precious...!

eddyb (Jun 06 2019 at 15:11, on Zulip):

@centril secure some funding :P

pnkfelix (Jun 06 2019 at 15:14, on Zulip):

Argh we totally skipped the I-nominated list

pnkfelix (Jun 06 2019 at 15:15, on Zulip):

luckily I think everything there I cared about can wait until next week

QuietMisdreavus (Jun 06 2019 at 15:22, on Zulip):

i did a quick check of the rustdoc ICEs by trying to run the same thing some of the affected crates on stable/beta/nightly, and the crates worked (or at least didn't ICE) on stable, ICE'd on beta, and worked (or at least didn't ICE) on nightly

simulacrum (Jun 06 2019 at 15:31, on Zulip):

okay, great, then once we have a new beta with the pulldown-cmark backport then we can probably close those once confirmed fixed

Last update: Nov 16 2019 at 02:40UTC