Stream: t-compiler

Topic: weekly meeting 2019-07-11 #54818


pnkfelix (Jul 11 2019 at 12:37, on Zulip):

Hi @T-compiler/meeting ; the triage meeting will be starting in 1 hours 23 minutes

pnkfelix (Jul 11 2019 at 12:38, on Zulip):

I will be doing pre-triage in a parallel topic

pnkfelix (Jul 11 2019 at 12:39, on Zulip):

are either of @Alex Crichton or @nnethercote around? I was wondering if we might get a check-in from WG-pipelining today

pnkfelix (Jul 11 2019 at 14:01, on Zulip):

Hi @T-compiler/meeting , lets get this meeting started!

pnkfelix (Jul 11 2019 at 14:01, on Zulip):

first up, lets see if there are any announcements.

pnkfelix (Jul 11 2019 at 14:02, on Zulip):

I'll start: this was already discussed in last Friday's meeting, but just so everyone knows: I'm going on parental leave for 8 weeks, starting this coming monday

pnkfelix (Jul 11 2019 at 14:02, on Zulip):

@nikomatsakis and @nagisa will be running the meetings (or delegating thereof) in my absence.

Alex Crichton (Jul 11 2019 at 14:05, on Zulip):

I'm around now!

nikomatsakis (Jul 11 2019 at 14:06, on Zulip):

:wave:

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

@Alex Crichton great! would you be prepared to give a WG-pipelining check in in say ... 40 minutes?

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

@Alex Crichton if you're not available then and are in fact ready now, we could do the checkin first, before the triage stuff

Alex Crichton (Jul 11 2019 at 14:07, on Zulip):

sure! I can check in whenever

pnkfelix (Jul 11 2019 at 14:07, on Zulip):

but if you'd prefer time to prepare (by which I mean get jazzed up on some coffee), we can keep it at the end of the meeting

pnkfelix (Jul 11 2019 at 14:07, on Zulip):

okay then lets aim to hear from alex in about 38 minutes

pnkfelix (Jul 11 2019 at 14:08, on Zulip):

a follow-up announcement: @nikomatsakis do you want to mention your own PTO ?

pnkfelix (Jul 11 2019 at 14:08, on Zulip):

(again, I know we covered this on friday, but I don't want to assume everyone here was there)

nikomatsakis (Jul 11 2019 at 14:09, on Zulip):

Sure! As I do most years, I'm going to be on PTO for roughly the next 4 weeks. I expect to announce a more detailed schedule, since I will likely be spending some days checking in and keeping track of things.

nikomatsakis (Jul 11 2019 at 14:09, on Zulip):

(But probably less than most years, where I've often done like 50% time.)

nikomatsakis (Jul 11 2019 at 14:10, on Zulip):

I guess most likely I would be around for some parts of the week of July 22

nikomatsakis (Jul 11 2019 at 14:10, on Zulip):

And I think I would be happy to commit to running the meeting that week, but @nagisa let's talk in a separate topic I guess

nagisa (Jul 11 2019 at 14:10, on Zulip):

ACK

pnkfelix (Jul 11 2019 at 14:11, on Zulip):

okay, so lets get to the standing agenda stuff

pnkfelix (Jul 11 2019 at 14:11, on Zulip):

there are five P-high unassigned T-compiler issues

pnkfelix (Jul 11 2019 at 14:11, on Zulip):

two of them are regressions of some kind

pnkfelix (Jul 11 2019 at 14:12, on Zulip):

though the two regressions, #62546 and #61560, are P-high only in the sense that I/we want someone to go in and double check that things aren't actually horribly broken

pnkfelix (Jul 11 2019 at 14:12, on Zulip):

I would assign these to myself, but, see above announcement about going on leave.

pnkfelix (Jul 11 2019 at 14:13, on Zulip):

anyway, if no one self-assigns these, I'll probably hand out auto-assignments tomorrow as one of my parting blows

nikomatsakis (Jul 11 2019 at 14:14, on Zulip):

I think @Esteban Küber might be a good candidate to investigate #62546

nikomatsakis (Jul 11 2019 at 14:14, on Zulip):

Since they do a lot of work around parse errors and recovery

pnkfelix (Jul 11 2019 at 14:14, on Zulip):

right

pnkfelix (Jul 11 2019 at 14:14, on Zulip):

okay, we have two beta-nominations

pnkfelix (Jul 11 2019 at 14:14, on Zulip):

first beta-nom: "Raise the default recursion limit to 128" #62450

nagisa (Jul 11 2019 at 14:15, on Zulip):

For this one I believe the regression with rustdoc got to beta via trains, but I never really verified that.

nikomatsakis (Jul 11 2019 at 14:16, on Zulip):

Seems pretty harmless.

pnkfelix (Jul 11 2019 at 14:16, on Zulip):

Does anyone want to object to #62450 ? I can imagine arguments such as "slippery slope"

pnkfelix (Jul 11 2019 at 14:16, on Zulip):

but mostly I am in camp "harmless"

nagisa (Jul 11 2019 at 14:17, on Zulip):

There is an objection by IDE folks about compiler needing bounded runtime and whatnot

nikomatsakis (Jul 11 2019 at 14:17, on Zulip):

I think that was more of a theoretical point

pnkfelix (Jul 11 2019 at 14:17, on Zulip):

can one override this value via the interface that the IDE uses ?

nikomatsakis (Jul 11 2019 at 14:18, on Zulip):

i.e., IDEs will need limits to ensure respnsiveness

centril (Jul 11 2019 at 14:18, on Zulip):

I say let's revisit "slippery slope" if we come to this point again (i.e. raising to 256)

nikomatsakis (Jul 11 2019 at 14:18, on Zulip):

vs an argument against this particular change

pnkfelix (Jul 11 2019 at 14:18, on Zulip):

yeah, I took the IDE point being more in response to the comment "Ideally we wouldn't have such arbitrary limits"

pnkfelix (Jul 11 2019 at 14:18, on Zulip):

as "here is a reason why we need such limits"

pnkfelix (Jul 11 2019 at 14:18, on Zulip):

anyway, looks like we all agree we should accept, at least in this instance.

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

My understanding is that we have these limits due to undecidability?

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

@centril (in some cases I think its more that no one wants to bother putting a better check in ?)

pnkfelix (Jul 11 2019 at 14:20, on Zulip):

((but I could be wrong; I'm thinking in some cases of recursive data structures where I was/am pretty sure the matter is decidable but we rely on recursion limits))

pnkfelix (Jul 11 2019 at 14:20, on Zulip):

okay, anyway, next beta nom

eddyb (Jul 11 2019 at 14:20, on Zulip):

oh I kept thinking you meant "nom" (the crate) not "nomination"

pnkfelix (Jul 11 2019 at 14:20, on Zulip):

beta nom: "Fix ICEs when Self is used in type aliases" #62417

pnkfelix (Jul 11 2019 at 14:21, on Zulip):

oh I kept thinking you meant "nom" (the crate) not "nomination"

I could write "nom'" but I bet that would be even more confusing

centril (Jul 11 2019 at 14:21, on Zulip):

One byte at a time, omnomnom

nikomatsakis (Jul 11 2019 at 14:21, on Zulip):

beta nom: "Fix ICEs when Self is used in type aliases" #62417

this appears to be a reverted commit?

pnkfelix (Jul 11 2019 at 14:21, on Zulip):

what?

pnkfelix (Jul 11 2019 at 14:22, on Zulip):

/me looks

nikomatsakis (Jul 11 2019 at 14:22, on Zulip):

sorry, I was referring to @Alexander Regueiro's comments

nikomatsakis (Jul 11 2019 at 14:22, on Zulip):

i.e., the change was kind of reverting some commit that they felt was causing problems?

nikomatsakis (Jul 11 2019 at 14:22, on Zulip):

I'm just verifying

Alexander Regueiro (Jul 11 2019 at 14:22, on Zulip):

yeah, it was more or less a revert

Alexander Regueiro (Jul 11 2019 at 14:22, on Zulip):

just of a small hunk

nikomatsakis (Jul 11 2019 at 14:22, on Zulip):

got it

centril (Jul 11 2019 at 14:23, on Zulip):

Feels safe to backport

pnkfelix (Jul 11 2019 at 14:23, on Zulip):

okay looks good then, lets backport

Alexander Regueiro (Jul 11 2019 at 14:23, on Zulip):

@nikomatsakis BTW, think you'll get time to review that PR of mine before you go on hols? specific tips on the coercion code to modify w.r.t. trait upcasting would be great too, if that doesn't distract you too much.

pnkfelix (Jul 11 2019 at 14:24, on Zulip):

we have no stable backport nominations to review

nikomatsakis (Jul 11 2019 at 14:24, on Zulip):

@pnkfelix I believe I nominated one thing for beta backport

nikomatsakis (Jul 11 2019 at 14:25, on Zulip):

maybe I messed up :)

pnkfelix (Jul 11 2019 at 14:25, on Zulip):

there are two PR's marking S-waiting-on-team, but one of them is the ancient PR #59064 from @Zoxc , and the other is I-nominated, so we'll get to it in the nominated issues

nikomatsakis (Jul 11 2019 at 14:25, on Zulip):

https://github.com/rust-lang/rust/pull/61853

nikomatsakis (Jul 11 2019 at 14:25, on Zulip):

I think I neglected the T-compiler label

nikomatsakis (Jul 11 2019 at 14:25, on Zulip):

Emit warning when trying to use PGO in conjunction with unwinding on … #61853

nikomatsakis (Jul 11 2019 at 14:25, on Zulip):

I was asked if we could backport this by the FF team, since it is blocking their use of PGO

nikomatsakis (Jul 11 2019 at 14:26, on Zulip):

It is a pretty mild change (error -> warn)

pnkfelix (Jul 11 2019 at 14:26, on Zulip):

(hmm I must have forgotten to look at all beta-nominations without team labels. I thought I did)

pnkfelix (Jul 11 2019 at 14:26, on Zulip):

hmm

centril (Jul 11 2019 at 14:27, on Zulip):

Feels like a pretty semantic change on the other hand

pnkfelix (Jul 11 2019 at 14:27, on Zulip):

the potential danger seems high, no?

nikomatsakis (Jul 11 2019 at 14:27, on Zulip):

how so?

pnkfelix (Jul 11 2019 at 14:27, on Zulip):

"it sometimes does not crash and will probably generate a corrupted binary"

pnkfelix (Jul 11 2019 at 14:27, on Zulip):

these things sound bad to me

nikomatsakis (Jul 11 2019 at 14:27, on Zulip):

nothing that works today is affected, obviously.

nikomatsakis (Jul 11 2019 at 14:28, on Zulip):

I mean it sounds like you're arguing against the change itself

nikomatsakis (Jul 11 2019 at 14:28, on Zulip):

(vs the risk of backporting)

pnkfelix (Jul 11 2019 at 14:28, on Zulip):

That's probably true

centril (Jul 11 2019 at 14:28, on Zulip):

Yeah that's what I'm hearing

nikomatsakis (Jul 11 2019 at 14:28, on Zulip):

which is potentially fair, I just want to be clear :)

centril (Jul 11 2019 at 14:28, on Zulip):

and thinking myself also

centril (Jul 11 2019 at 14:28, on Zulip):

But if Felix is having doubts about its existence on master it seems bad to backport, no?

pnkfelix (Jul 11 2019 at 14:28, on Zulip):

so if we don't backport

pnkfelix (Jul 11 2019 at 14:28, on Zulip):

that gives us time to debate

pnkfelix (Jul 11 2019 at 14:29, on Zulip):

but I'm going on leave; I'm in no position to support my side of argument

nikomatsakis (Jul 11 2019 at 14:29, on Zulip):

I hear that but I think the risk is being overblown, personally. That is, I think that people who are building with PGO may encounter some problems perhaps, but in the meantime we block the feature from being used at all

pnkfelix (Jul 11 2019 at 14:29, on Zulip):

Is there a way we could give this to FF but not the world at large?

pnkfelix (Jul 11 2019 at 14:29, on Zulip):

maybe that's the wrong attitude

simulacrum (Jul 11 2019 at 14:29, on Zulip):

We could stick it in a lint group that is forbid by default, I guess?

simulacrum (Jul 11 2019 at 14:29, on Zulip):

(That might be hard)

nikomatsakis (Jul 11 2019 at 14:29, on Zulip):

Moreover, the people driving PGO is basically FF, so they're the ones who would be encountering the problems (and fixing them)

pnkfelix (Jul 11 2019 at 14:30, on Zulip):

I guess the NLL migrate borrowck currently emits (stern) warnings about things that might actually be weaponizable

pnkfelix (Jul 11 2019 at 14:30, on Zulip):

so okay, I'm not going to fight this.

centril (Jul 11 2019 at 14:31, on Zulip):

so > does not crash and will probably generate a corrupted binary.

does that mean it might lead to unsoundness?

pnkfelix (Jul 11 2019 at 14:31, on Zulip):

let me actually put up the link with title and emojis for people to click

nikomatsakis (Jul 11 2019 at 14:31, on Zulip):

I guess I think my take is that, if we had a lot of people around, it seems ok, but in the meantime I think it's ok to issue a warning. Messing around with things like PGO (to me) and linker behavior is fairly low level hackery and it is possible for things to go wrong.

pnkfelix (Jul 11 2019 at 14:31, on Zulip):

last beta nom: "Emit warning when trying to use PGO in conjunction with unwinding on [Windows]" #61853

centril (Jul 11 2019 at 14:32, on Zulip):

@nikomatsakis but how wrong are we talking?

centril (Jul 11 2019 at 14:32, on Zulip):

if it justs leads to a crash and not UB then fine

nikomatsakis (Jul 11 2019 at 14:33, on Zulip):

I don't know exactly

nikomatsakis (Jul 11 2019 at 14:33, on Zulip):

Maybe a good idea at minimum is to update the text on https://github.com/rust-lang/rust/issues/61002

pnkfelix (Jul 11 2019 at 14:33, on Zulip):

there is bug connected to a old versions of GNU ld we discussed during pre-triage that one might put in a similar boat, in terms of asking "what is our obligation when it comes to linker issues"

centril (Jul 11 2019 at 14:34, on Zulip):

So I'd like to be sure that this won't lead to unsoundness at least, which I would not be fine with

nikomatsakis (Jul 11 2019 at 14:34, on Zulip):

In the meantime, I expect mw will investigate this in more detail when they return from leave in ... 4 weeks or so

Wesley Wiser (Jul 11 2019 at 14:34, on Zulip):

Is it possible to just move the check later in the compilation process? EricRahm says the issue is that the error is being emitted too eagerly ie even when they're not asking rustc to actually do PGO https://github.com/rust-lang/rust/issues/61002#issuecomment-500075739

nikomatsakis (Jul 11 2019 at 14:34, on Zulip):

It may be

pnkfelix (Jul 11 2019 at 14:35, on Zulip):

but then again I assume EricRahm is fine with the current "solution" under discussino since they authored it?

pnkfelix (Jul 11 2019 at 14:36, on Zulip):

We can make a decision (one way or the other) right now. Or we can say "lets delay until next week", but @nikomatsakis and I will both be absent then. Is that okay?

centril (Jul 11 2019 at 14:37, on Zulip):

@Wesley Wiser so if you do "use PGO in conjunction with unwinding on Windows" and actually use PGO you'd get an error?

pnkfelix (Jul 11 2019 at 14:37, on Zulip):

it seems like we should probably investigate at least a tiny bit more before we approve for backport. Is someone here willing to take point on that investigation in the absence of myself and @nikomatsakis ?

centril (Jul 11 2019 at 14:37, on Zulip):

I feel a distinct lack of information to say I'm fine

Wesley Wiser (Jul 11 2019 at 14:37, on Zulip):

@centril Yeah.

centril (Jul 11 2019 at 14:37, on Zulip):

@Wesley Wiser oh; well in that case if there's an error later in compilation then it cannot be a problem in terms of soundness and whatnot?

pnkfelix (Jul 11 2019 at 14:38, on Zulip):

Is someone here willing to take point on that investigation in the absence of myself and @nikomatsakis ?

(if I don't see any takers on this question, then I feel pressured to make a decision, one way or the other, right now...)

nikomatsakis (Jul 11 2019 at 14:38, on Zulip):

If we can find an alternate fix, that's great. Maybe we can move the check and make it more precise. I'd like it if we identified who will investigate, though.

centril (Jul 11 2019 at 14:39, on Zulip):

@Wesley Wiser would you like to take it on perhaps?

Wesley Wiser (Jul 11 2019 at 14:39, on Zulip):

I can do some investigation and try to report back next week with my findings.

pnkfelix (Jul 11 2019 at 14:39, on Zulip):

(or maybe @varkor since they reviewed the original PR ... )

nikomatsakis (Jul 11 2019 at 14:39, on Zulip):

I'll ping Eric to try and get a contact who might have more info from FF

pnkfelix (Jul 11 2019 at 14:39, on Zulip):

okay, great, thanks @Wesley Wiser !

pnkfelix (Jul 11 2019 at 14:42, on Zulip):

okay, we have five nominated issues but only 4 minutes before I promised @Alex Crichton a chance to speak. :)

pnkfelix (Jul 11 2019 at 14:42, on Zulip):

nominated: "ICE: thread 'rustc' panicked at 'capacity overflow'" #62554

pnkfelix (Jul 11 2019 at 14:42, on Zulip):

I only nominated this so people are aware I tagged it as P-medium. :)

pnkfelix (Jul 11 2019 at 14:42, on Zulip):

if you think it should be P-high, speak up.

pnkfelix (Jul 11 2019 at 14:42, on Zulip):

(preferably on issue)

pnkfelix (Jul 11 2019 at 14:43, on Zulip):

next: "1.30 -> 1.31 dylib late-binding regression with GNU binutils 2.28 or older." #61539

pnkfelix (Jul 11 2019 at 14:43, on Zulip):

this is the bug I mentioned up above with things with GNU linker versions

pnkfelix (Jul 11 2019 at 14:43, on Zulip):

I nominated it to raise awareness of this stray thought

pnkfelix (Jul 11 2019 at 14:44, on Zulip):

namely, the idea of using dynamic linker version detect " just to drive the emission of a diagnostic warning. Where said diagnostic tells people to take one of the aforementioned steps to work around the bug."

pnkfelix (Jul 11 2019 at 14:44, on Zulip):

I wanted to take peoples temperature about that idea.

pnkfelix (Jul 11 2019 at 14:44, on Zulip):

But again, please feel free to respond on ticket, we don't have to do it live here.

nagisa (Jul 11 2019 at 14:45, on Zulip):

Well, the warning message based on just linker version would lead to false positives more often than not, but I don’t know of any other solution that’s not reenabling plt by default.

pnkfelix (Jul 11 2019 at 14:45, on Zulip):

next, I nominated "Creating a recursive type with infinite size leads to internal compiler error" #61323 in hopes of finding someone other than niko to look into it

pnkfelix (Jul 11 2019 at 14:46, on Zulip):

@nagisa is there particular pattern of codegen we could associate the diagnostic with, to try to lower the false positive rate?

pnkfelix (Jul 11 2019 at 14:46, on Zulip):

Or is it basically going to happen all the time?

nagisa (Jul 11 2019 at 14:47, on Zulip):

Maybe? Ultimately to trigger that particular bug a dynamic library needs to be loaded at runtime (via e.g. dlopen)

pnkfelix (Jul 11 2019 at 14:47, on Zulip):

mmm

pnkfelix (Jul 11 2019 at 14:47, on Zulip):

okay

nagisa (Jul 11 2019 at 14:48, on Zulip):

Most relevant case where this almost always happen are compiler plugins

pnkfelix (Jul 11 2019 at 14:48, on Zulip):

two other nominations; "Add a "diagnostic item" scheme for lints referring to libstd items" #60966 (is, according to @eddyb , waiting on @oli , who is on vacation for another week). and "Stable rustc always panics on arm/musl" #60297 is still looking for some love

eddyb (Jul 11 2019 at 14:48, on Zulip):

@nagisa oh yeah I even forgot how rare this is

pnkfelix (Jul 11 2019 at 14:49, on Zulip):

(does anyone here have access to arm/musl? Hey @Alex Crichton , do you? want to do some rustc hacking for us?)

Alex Crichton (Jul 11 2019 at 14:49, on Zulip):

Heh I do not outside of qemu emulation

pnkfelix (Jul 11 2019 at 14:49, on Zulip):

that seems as good a segue as any to a checkin from WG-pipelining, take it away @Alex Crichton

nagisa (Jul 11 2019 at 14:49, on Zulip):

(10 mins left)

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

ok! so the general tracking issue is still at https://github.com/rust-lang/rust/issues/60988

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

and everything is set enough that I've been proposing stabilizing the pieces in rustc so we can stabilize the support in Cargo

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

and get it all out to everyone by default

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

that requires two sub-issues to be stabilized

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

one is https://github.com/rust-lang/rust/issues/60987#issuecomment-509310492, where a --json flag is proposed to replace --json-rendered and it takes a list of arguments of what to print

Alex Crichton (Jul 11 2019 at 14:50, on Zulip):

(and various configurations and whatnot)

Alex Crichton (Jul 11 2019 at 14:51, on Zulip):

The next is https://github.com/rust-lang/rust/issues/60419#issuecomment-502228173 which is stabilizing the fact that JSON notifications come out for artifacts when you compile rustc and turn on json messages

eddyb (Jul 11 2019 at 14:51, on Zulip):

would --json imply JSON output without any other flags?

nikomatsakis (Jul 11 2019 at 14:51, on Zulip):

it's worth pointing out that there is a pending FCP proposal and @Zoxc, @Esteban Küber, @nagisa, @oli, @Vadim Petrochenkov, and @varkor still pending

eddyb (Jul 11 2019 at 14:52, on Zulip):

(because that would probably be neat)

Alex Crichton (Jul 11 2019 at 14:52, on Zulip):

in that latter issue in the discussion it was brought up that a -C flag isn't really necessary and we can just infer from json output

Alex Crichton (Jul 11 2019 at 14:52, on Zulip):

@eddyb perhaps eventually, but seems like a separable question

Alex Crichton (Jul 11 2019 at 14:52, on Zulip):

but yes the main thing now is ticky boxes and I don't mind doing the legwork actually stabilizing support

Alex Crichton (Jul 11 2019 at 14:52, on Zulip):

long-term there's more possibilities for pipelining still, but for now this is just focused on what we have working today becoming stable

Alex Crichton (Jul 11 2019 at 14:53, on Zulip):

and that's all I've got for pipelining!

nikomatsakis (Jul 11 2019 at 14:53, on Zulip):

I guess there are two pending proposals

eddyb (Jul 11 2019 at 14:53, on Zulip):

IMO one thing I'd be fine with is just increasing the amount of detail in the JSON output, and expect any consumers to adapt

eddyb (Jul 11 2019 at 14:53, on Zulip):

as an alternative to increasing amounts of configurability

nikomatsakis (Jul 11 2019 at 14:54, on Zulip):

The second one is waiting on @Zoxc, @nagisa, @Vadim Petrochenkov, and @pnkfelix

nikomatsakis (Jul 11 2019 at 14:54, on Zulip):

Seems like it'd be good to get something here out the door...

pnkfelix (Jul 11 2019 at 14:55, on Zulip):

-Z flags need to go through rfcbot ?

pnkfelix (Jul 11 2019 at 14:55, on Zulip):

oh, the PR issue title is misleading!

Alex Crichton (Jul 11 2019 at 14:56, on Zulip):

yeah sorry if anything is unclear let me know and I can try to clear it up!

Alex Crichton (Jul 11 2019 at 14:56, on Zulip):

these were stabilization proposals on the tracking issue rather than creation of a new issue proposing stabilization

nikomatsakis (Jul 11 2019 at 14:58, on Zulip):

I have a separate question -- there is a rustc guide that documents how to use rustc -- not to be confused with the one that teaches you how to hack on rustc, we should fix that...

nikomatsakis (Jul 11 2019 at 14:59, on Zulip):

does anybody know how complete that is or whether it's maintained?

nikomatsakis (Jul 11 2019 at 14:59, on Zulip):

seems like a useful artifact

nikomatsakis (Jul 11 2019 at 14:59, on Zulip):

(I was thinking that both PGO + pipelining would be nice things to document in there)

pnkfelix (Jul 11 2019 at 15:00, on Zulip):

@Alex Crichton it seems like @eddyb and @Zoxc both care to some degree about whether the flag for #60419 goes under -C or [ ... somewhere else ...]

centril (Jul 11 2019 at 15:00, on Zulip):

(https://doc.rust-lang.org/rustc/lints/levels.html should probably be moved partially to the reference)

pnkfelix (Jul 11 2019 at 15:00, on Zulip):

should we try to resolve that question right here, while people are present?

eddyb (Jul 11 2019 at 15:00, on Zulip):

I'd argue it should be called rustc CLIor even an outright manpage :P

eddyb (Jul 11 2019 at 15:00, on Zulip):

rustc guide seems confusing

nikomatsakis (Jul 11 2019 at 15:00, on Zulip):

I'm :+1: on resolving questions now if we can

eddyb (Jul 11 2019 at 15:00, on Zulip):

@pnkfelix I prefer no flag, just add it to the JSON output by default

pnkfelix (Jul 11 2019 at 15:01, on Zulip):

pnkfelix I prefer no flag, just add it to the JSON output by default

yeah I found @Alex Crichton 's response to this suggestion confusing

Alex Crichton (Jul 11 2019 at 15:01, on Zulip):

sorry to clarify, I'm fine with whatever, if someone feels strongly I'm happy to go that route

Alex Crichton (Jul 11 2019 at 15:01, on Zulip):

I have no preference myself other than to get this stabilized

pnkfelix (Jul 11 2019 at 15:01, on Zulip):

why wouldn't we do what @eddyb suggests?

Alex Crichton (Jul 11 2019 at 15:01, on Zulip):

It has a higher risk of being a breaking change

eddyb (Jul 11 2019 at 15:01, on Zulip):

(also relevant for other JSON additions)

pnkfelix (Jul 11 2019 at 15:01, on Zulip):

just to make --json independently testable?

Alex Crichton (Jul 11 2019 at 15:02, on Zulip):

if consumers of json output expect a particular structure

pnkfelix (Jul 11 2019 at 15:02, on Zulip):

but we don'

pnkfelix (Jul 11 2019 at 15:02, on Zulip):

don't even hav a --json flag right now, right?

eddyb (Jul 11 2019 at 15:02, on Zulip):

shouldn't they handle failure gracefully? I thought we've been adding stuff to it over the past couple years

pnkfelix (Jul 11 2019 at 15:02, on Zulip):

/me looks

Alex Crichton (Jul 11 2019 at 15:02, on Zulip):

I'm just saying "higher risk", I'm not saying "something will break"

eddyb (Jul 11 2019 at 15:03, on Zulip):

/me is probably overoptimistic tbh

Alex Crichton (Jul 11 2019 at 15:03, on Zulip):

we do not have --json now, no

nikomatsakis (Jul 11 2019 at 15:03, on Zulip):

I do think we should feel free to add to the json

nikomatsakis (Jul 11 2019 at 15:03, on Zulip):

However

nikomatsakis (Jul 11 2019 at 15:04, on Zulip):

I also think we should avoid bikeshedding this to death =)

pnkfelix (Jul 11 2019 at 15:04, on Zulip):

oh maybe I misunderstood @eddyb 's suggestion. I took it to mean that the change would be implied by the --json flag

nikomatsakis (Jul 11 2019 at 15:04, on Zulip):

I also can't remember how much we've extended the json before :) at least sometimes, I think

pnkfelix (Jul 11 2019 at 15:05, on Zulip):

Okay well I'm going to officially say :shrug: and check off my box, as I need to go

pnkfelix (Jul 11 2019 at 15:05, on Zulip):

thanks to everyone in @T-compiler/meeting for attending!

pnkfelix (Jul 11 2019 at 15:07, on Zulip):

(i do think the --cargo=foo=bar options sounds like a good way to introduce functionality like this in the future)

nikomatsakis (Jul 11 2019 at 15:09, on Zulip):

I think my take on the issue is that I like @eddyb's suggestion but I would be quite happy to stabilize what's there and then introduce a "grand unified mechanism" separately.

nikomatsakis (Jul 11 2019 at 15:10, on Zulip):

And yes I sort of like having a "mid-way point" that lets us create explicitly temporary interfaces that can get "stabilized" separately

nnethercote (Jul 11 2019 at 21:24, on Zulip):

Unfortunatey I will never be around for the meeting due to timezones :/

Last update: Nov 16 2019 at 01:05UTC