Stream: t-compiler

Topic: steering meeting 2019.03.01 #58850


nikomatsakis (Feb 28 2019 at 20:39, on Zulip):

Hey @T-compiler/meeting -- I just posted a comment to internals requested feedback on agenda items for tomorrow's steering meeting. Sorry for leaving this a bit late. I'd love your feedback!

Here's the comment inline:

We have a steering meeting tomorrow. Here are is an array of possible topics that have come up lately:

There are also a few things that came up that might fit better as "announcements", although maybe there is enough to discuss:

Other suggestions, announcements, etc?

I mentioned "how to maintain the agenda" :)

nikomatsakis (Feb 28 2019 at 20:40, on Zulip):

Probably best to leave comments on internals tho

nikomatsakis (Feb 28 2019 at 20:40, on Zulip):

I feel like I had more thoughts but those are all that came to mind right now, anyway

mw (Mar 01 2019 at 10:18, on Zulip):

https://internals.rust-lang.org/t/compiler-steering-meeting/8588/45

davidtwco (Mar 01 2019 at 10:22, on Zulip):

That was discussed briefly at the #t-compiler/wg-meta meeting this week. I'll be submitting a WIP PR to the compiler-team repo shortly writing up some of the ideas and thoughts from when this was discussed at the All Hands. I expect I'll need to make some changes based on feedback from the compiler team.

davidtwco (Mar 01 2019 at 10:31, on Zulip):

draft policy

mw (Mar 01 2019 at 11:36, on Zulip):

thanks, @davidtwco! I added some comments to the PR.

nikomatsakis (Mar 01 2019 at 14:05, on Zulip):

I created a hackmd to collect Steering Meeting Agenda Ideas

nikomatsakis (Mar 01 2019 at 14:54, on Zulip):

@mw is your sense that said policies would be a good topic for this meeting?

nikomatsakis (Mar 01 2019 at 14:54, on Zulip):

I'm trying to firm up what to propose :)

nikomatsakis (Mar 01 2019 at 14:56, on Zulip):

Hello @T-compiler/meeting! Steering meeting in 5 minutes.

mw (Mar 01 2019 at 14:56, on Zulip):

@nikomatsakis we could discuss it, yes

mw (Mar 01 2019 at 14:57, on Zulip):

although I don't know if this would a be a topic that would need some preparation

mw (Mar 01 2019 at 14:57, on Zulip):

e.g. people trying to go through the process and taking note of the questions that arise

nikomatsakis (Mar 01 2019 at 14:58, on Zulip):

that is partly what I was wondering

mw (Mar 01 2019 at 14:59, on Zulip):

@WG-self-profile will probably run into this soon

nikomatsakis (Mar 01 2019 at 15:01, on Zulip):

OK, let's quickly discuss the agenda

nikomatsakis (Mar 01 2019 at 15:01, on Zulip):

Here is a list of my thoughts:

Santiago Pastorino (Mar 01 2019 at 15:01, on Zulip):

@nikomatsakis just started this https://hackmd.io/Izvor8KZRiqUgcUyd2CYTw# as a place holder

nikomatsakis (Mar 01 2019 at 15:01, on Zulip):

Ah, nice

Santiago Pastorino (Mar 01 2019 at 15:01, on Zulip):

couldn't add that to your agenda

Santiago Pastorino (Mar 01 2019 at 15:02, on Zulip):

I have some things written down, will fill when I can. Just wanted you to put a link in the agenda :)

matklad (Mar 01 2019 at 15:02, on Zulip):

/me is here, but will run away in 20 minutes

nikomatsakis (Mar 01 2019 at 15:02, on Zulip):

I think @mw is right that it'd be good to go over the out-of-tree crate policy a bit. We probably could use some prep, but I think we probably wouldn't get much of that anywaY :P

nikomatsakis (Mar 01 2019 at 15:02, on Zulip):

And in prtaicular there exist a fair number of crates that either exist today or will exist

nikomatsakis (Mar 01 2019 at 15:03, on Zulip):

So maybe it's good to review the proposed thoughts and add comments etc

pnkfelix (Mar 01 2019 at 15:03, on Zulip):

given that matklad will have to go soon

pnkfelix (Mar 01 2019 at 15:03, on Zulip):

should we consider waiting until after they leave to add/discuss add'l agenda items?

nikomatsakis (Mar 01 2019 at 15:03, on Zulip):

If we have time, I have some other thoughts (what I wrote as "triage-like")

pnkfelix (Mar 01 2019 at 15:04, on Zulip):

(I mean this in the sense of "lets maximize the amount of time matklad can spend on the RLS 2.0 item)

nikomatsakis (Mar 01 2019 at 15:04, on Zulip):

Yeah, so, unless anyone objects, I propose we kick it off, and indeed we can start out with a brief discussion of RLS 2.0 "call for invovement"

nikomatsakis (Mar 01 2019 at 15:05, on Zulip):

The summary being that at the end of the triage meeting yesterday, @matklad expressed a desire to grow the set of mentors etc. I definitely think that "RLS 2.0" will be more successful the more people get involved.

nikomatsakis (Mar 01 2019 at 15:06, on Zulip):

(Actually one interjection: I just thought I'd mention that the t-compiler/meeting alias is open to anyone who wants to get pings for these sorts of things, and said folks should just requested to be added)

matklad (Mar 01 2019 at 15:07, on Zulip):

Yeah, and there's the idea that sharing expertise between "RLS 2.0" and rustc would be very helpful to make the experiment more valuable.

nikomatsakis (Mar 01 2019 at 15:07, on Zulip):

I'm not really sure what else to say about this though, not sure @matklad if you had other things you wanted to say?

nikomatsakis (Mar 01 2019 at 15:07, on Zulip):

(cc @Igor Matuszewski, btw)

nikomatsakis (Mar 01 2019 at 15:07, on Zulip):

I guess one thing we should mention is that we've been doing some video sessions

nikomatsakis (Mar 01 2019 at 15:07, on Zulip):

recording walkthroughs of what rust-analyzer has today

nikomatsakis (Mar 01 2019 at 15:07, on Zulip):

We are planning a discussion next week about the type checker

nikomatsakis (Mar 01 2019 at 15:08, on Zulip):

There is a video up now of the syntax tree stuff that @matklad has done (which is quite cool)

nikomatsakis (Mar 01 2019 at 15:08, on Zulip):

I promise it will make you drool just a bit when compaerd with what it's like to manipulate HIR ;)

pnkfelix (Mar 01 2019 at 15:08, on Zulip):

Question: do we believe that the T-compiler members interested in helping with RLS 2.0 have already identified themselves and have been added to the WG-rls2.0 ?

matklad (Mar 01 2019 at 15:08, on Zulip):

So, basically, there's rust-analyzer repo, it has documentation (guides and a couple of videos), a set of issues with E-mentor, etc etc. If you feel like digging into requirenments of IDEs, it's a good chance, while the code base is small and fits one's head :)

Igor Matuszewski (Mar 01 2019 at 15:09, on Zulip):

I believe the E-mentor and these posted resources are key element moving forward if we want to expand the set of people involved/being able to mentor others

Igor Matuszewski (Mar 01 2019 at 15:10, on Zulip):

(Which is basically :+1: from me to what we're doing now)

davidtwco (Mar 01 2019 at 15:10, on Zulip):

I've mentioned interest before, but I'll try to actually carve some time out tonight to jump in and start working on something.

pnkfelix (Mar 01 2019 at 15:10, on Zulip):

or maybe let me put my Question another way: Are we seeking to increase involvement from people who are in this meeting now? Or are we seeking to increase involvement beyond that?

Igor Matuszewski (Mar 01 2019 at 15:10, on Zulip):

I'll start digging into an issue @matklad kindly pointed to this weekend, so I'll dig in slowly as well

mw (Mar 01 2019 at 15:11, on Zulip):

I won't have time in the medium term unfortunately. Pity, it looks really interesting

nikomatsakis (Mar 01 2019 at 15:11, on Zulip):

Are we seeking to increase involvement from people who are in this meeting now?

I think so, yes

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

I carved out some time two/three weeks ago, and am hoping to devote more time to it in the near future

Igor Matuszewski (Mar 01 2019 at 15:12, on Zulip):

I was doing some spring cleaning in the RLS these last weeks and should be able to devote more time now

nikomatsakis (Mar 01 2019 at 15:13, on Zulip):

(side note, @matklad, we should link to the RLS 2.0 video playlist from the RLS 2.0 readme)

nikomatsakis (Mar 01 2019 at 15:13, on Zulip):

OK, well, shall we move to next topic? There was one more very brief announcement.

matklad (Mar 01 2019 at 15:13, on Zulip):

:+1: , let's move

nikomatsakis (Mar 01 2019 at 15:14, on Zulip):

nikomatsakis just started this https://hackmd.io/Izvor8KZRiqUgcUyd2CYTw# as a place holder

@Santiago Pastorino is creating an Expert Map -- this HackMd URL is the holding spot -- that aims to document the parts of the compiler and who is a good person to ask about them

nikomatsakis (Mar 01 2019 at 15:14, on Zulip):

Please feel free to add things to it or add your name places

nikomatsakis (Mar 01 2019 at 15:14, on Zulip):

It's pretty minimal right now :)

nikomatsakis (Mar 01 2019 at 15:15, on Zulip):

The intention with this is to help people who are looking to ask questions

nikomatsakis (Mar 01 2019 at 15:15, on Zulip):

But also (I think) to help us get started on documenting the overall structure of the compiler and so forth, in its own way

nikomatsakis (Mar 01 2019 at 15:15, on Zulip):

At least that's something I'd like to see written down

nikomatsakis (Mar 01 2019 at 15:16, on Zulip):

Anyway, if you have thoughts or suggestions, please feel free to give them -- we should create a topic in #t-compiler/wg-meta I guess

nikomatsakis (Mar 01 2019 at 15:16, on Zulip):

(Anything else to say on that?)

mw (Mar 01 2019 at 15:16, on Zulip):

creating a separate topic for this is a good idea

nikomatsakis (Mar 01 2019 at 15:18, on Zulip):

Done

Santiago Pastorino (Mar 01 2019 at 15:18, on Zulip):

was going to do that :)

nikomatsakis (Mar 01 2019 at 15:18, on Zulip):

@Santiago Pastorino you can add a link to the hackmd in there, at least :)

nikomatsakis (Mar 01 2019 at 15:18, on Zulip):

OK, so on to the main topic?

nikomatsakis (Mar 01 2019 at 15:18, on Zulip):

@davidtwco are you here?

davidtwco (Mar 01 2019 at 15:19, on Zulip):

:thumbs_up:

nikomatsakis (Mar 01 2019 at 15:19, on Zulip):

So, the proposed main topic here is to discuss a bit more our policy on out-of-tree crates, particularly those that we author. Existing examples would be things like polonius and chalk, future examples might be "libnag" (diagnostics), self-profiling, etc

nikomatsakis (Mar 01 2019 at 15:20, on Zulip):

We discussed this at the Rust All Hands and @davidtwco did the work yesterday of trying to codify that discussion

pnkfelix (Mar 01 2019 at 15:20, on Zulip):

/me likes "libnag"

nikomatsakis (Mar 01 2019 at 15:20, on Zulip):

Credit goes to @oli =)

nikomatsakis (Mar 01 2019 at 15:21, on Zulip):

We thought we might briefly take a look at that summary during this meeting and discuss it... probably it makes sense to summarize the high points here in thread? Should we all take 5 minutes to read it? :)

nikomatsakis (Mar 01 2019 at 15:21, on Zulip):

Pull request link: https://github.com/rust-lang/compiler-team/pull/19

nikomatsakis (Mar 01 2019 at 15:22, on Zulip):

It's not very long. You can view in rendered form here

davidtwco (Mar 01 2019 at 15:23, on Zulip):

There's certainly some blank spots where things were mentioned in the meeting notes from the All Hands that I wasn't sure how to distil down into the document and @mw has raised some good questions on the PR. I'm happy to go away and find answers any questions we raise about what is included in the PR currently and make any additions we feel are warranted.

nikomatsakis (Mar 01 2019 at 15:23, on Zulip):

I think one key point for me

nikomatsakis (Mar 01 2019 at 15:23, on Zulip):

is that we should try to make the experience "uniform" as much as we can

nikomatsakis (Mar 01 2019 at 15:23, on Zulip):

I am primarily thinking of:

nikomatsakis (Mar 01 2019 at 15:24, on Zulip):
nikomatsakis (Mar 01 2019 at 15:24, on Zulip):

these are concrete pain points we've hit in the past

davidtwco (Mar 01 2019 at 15:24, on Zulip):

There were points in the All Hands meeting about what we could do from a tooling perspective to make out-of-tree crates as easy to work with but that didn't seem like it made sense to add that to the policy - since what we'd want out of tooling isn't part of the policy.

nikomatsakis (Mar 01 2019 at 15:25, on Zulip):

/me likes "libnag"

ps, @pnkfelix and I both actually had the same concern about this name -- that (in English, to my ears at least) it carries a subtle gender connotation that is actually not great. This brings up another point: how do we pick the names?

mw (Mar 01 2019 at 15:25, on Zulip):

I'd like a step-by-step guide that I can follow when creating a new crate

nikomatsakis (Mar 01 2019 at 15:25, on Zulip):

(I don't care to debate the name of that particular crate at this meeting)

nikomatsakis (Mar 01 2019 at 15:25, on Zulip):

@davidtwco I think "nice tooling" is jumping the gun

nikomatsakis (Mar 01 2019 at 15:25, on Zulip):

First step is to write down the steps to follow

nikomatsakis (Mar 01 2019 at 15:25, on Zulip):

(We can talk to @Pietro Albini perhaps as well)

mw (Mar 01 2019 at 15:26, on Zulip):

the guide for setting up a working group, for example, is very helpful

nikomatsakis (Mar 01 2019 at 15:26, on Zulip):

Once we know what the steps are -- and we've tried them out a bit to gain experience -- we can try to make a tool for it

nikomatsakis (Mar 01 2019 at 15:26, on Zulip):

Also, I think those steps should (where possible) include links to examples and things to copy-n-paste

nikomatsakis (Mar 01 2019 at 15:26, on Zulip):

Regarding naming, I really do think we should have some "process" -- it does't have to be heavy weight -- but the names of things in rust-lang are kind of part of our "public facing appearance"

nikomatsakis (Mar 01 2019 at 15:26, on Zulip):

And sometimes there are subtle issues

pnkfelix (Mar 01 2019 at 15:27, on Zulip):

there's a statement in What should these crates be named that I want to also bring up

pnkfelix (Mar 01 2019 at 15:27, on Zulip):

"this isn't appropriate as the crate was actually intended to be general-purpose."

pnkfelix (Mar 01 2019 at 15:27, on Zulip):

This is implying that if the code being factored out isn't very general purpose, then it shouldn't be factored out in the first place, right?

pnkfelix (Mar 01 2019 at 15:28, on Zulip):

I guess I'm wondering what the minimum bar for expected re-use is

nikomatsakis (Mar 01 2019 at 15:28, on Zulip):

Yeah, I don't really believe that

pnkfelix (Mar 01 2019 at 15:28, on Zulip):

to justify making the step

nikomatsakis (Mar 01 2019 at 15:28, on Zulip):

In fact, I'm a bit wary of trying to make the libraries too general purpose

nikomatsakis (Mar 01 2019 at 15:28, on Zulip):

because that carries a higher maintenance / breaking-change burden :)

pnkfelix (Mar 01 2019 at 15:28, on Zulip):

and also potential for second-system effect, too

mw (Mar 01 2019 at 15:29, on Zulip):

I guess this has to be decided on a case-by-case basis

mw (Mar 01 2019 at 15:29, on Zulip):

for small crates that provide a re-usable datastructure I wouldn't call it rustc_xxx

mw (Mar 01 2019 at 15:30, on Zulip):

a name like this almost precludes that the crate will be used outside of rustc

mw (Mar 01 2019 at 15:30, on Zulip):

at least I wouldn't be confident in using such a crate for fear of breakage and changing interfaces

nikomatsakis (Mar 01 2019 at 15:31, on Zulip):

As an aside, I think we should absolutely follow semver to the best of our ability

nikomatsakis (Mar 01 2019 at 15:31, on Zulip):

In case that was in doubt ;)

nikomatsakis (Mar 01 2019 at 15:31, on Zulip):

(But I have no problem with issues tons of major versions)

nikomatsakis (Mar 01 2019 at 15:32, on Zulip):

Do we want to have a review policy similar to rustc?

nikomatsakis (Mar 01 2019 at 15:32, on Zulip):

I think we .. probably do?

nikomatsakis (Mar 01 2019 at 15:32, on Zulip):

Another question is, can those crates have reviewers who are not reviewers of rustc? Seems like...yes?

nikomatsakis (Mar 01 2019 at 15:32, on Zulip):

I think we .. probably do?

by this I mean: we probably want to require reviews for major PRs

mw (Mar 01 2019 at 15:32, on Zulip):

huh, that's really tricky

nikomatsakis (Mar 01 2019 at 15:32, on Zulip):

You see why I'm asking :)

mw (Mar 01 2019 at 15:33, on Zulip):

I do :)

nikomatsakis (Mar 01 2019 at 15:33, on Zulip):

I guess I like the idea of crates having their own communities

nikomatsakis (Mar 01 2019 at 15:33, on Zulip):

OTOH maybe we should just say that if you're involved enough in some crate to review for it

nikomatsakis (Mar 01 2019 at 15:33, on Zulip):

we can give you r+ to rustc

nikomatsakis (Mar 01 2019 at 15:33, on Zulip):

I would do the same for people who've made major contributions in one area

nikomatsakis (Mar 01 2019 at 15:33, on Zulip):

(under the expectation that they will use it responsibly)

nikomatsakis (Mar 01 2019 at 15:34, on Zulip):

(though in the long run, having more libraries might offer us a way to have a more fine-grained notion of r+ that is appealing)

nikomatsakis (Mar 01 2019 at 15:34, on Zulip):

I guess I think we should not stress about this too hard but it's more a question to think about :)

mw (Mar 01 2019 at 15:36, on Zulip):

Should we try to walk through an example?

pnkfelix (Mar 01 2019 at 15:36, on Zulip):

long term we may want separate tests within rustc repo of its "third party" crate

pnkfelix (Mar 01 2019 at 15:36, on Zulip):

rather than trusting the third parties to remain compatible

mw (Mar 01 2019 at 15:37, on Zulip):

ok, so @WG-self-profile will probably create a crate for a mildly sophisticated on-disk string table

mw (Mar 01 2019 at 15:37, on Zulip):

that crate will be used by the compiler and an external tool doing some processing on data

mw (Mar 01 2019 at 15:37, on Zulip):

so it's a good candidate to split out

mw (Mar 01 2019 at 15:38, on Zulip):

how would we go about this?

nikomatsakis (Mar 01 2019 at 15:39, on Zulip):

It does seem like a great candidate

mw (Mar 01 2019 at 15:39, on Zulip):

do t-compiler members already have the GH rights to create the project?

nikomatsakis (Mar 01 2019 at 15:39, on Zulip):

I don't know, I can check

nikomatsakis (Mar 01 2019 at 15:40, on Zulip):

We immediately hit the issue of picking a name :)

pnkfelix (Mar 01 2019 at 15:41, on Zulip):

is it best to jump to a rust-lang repo, or start in a prototype hosted in user repo?

nikomatsakis (Mar 01 2019 at 15:41, on Zulip):

I say go to the rust-lang repo

nikomatsakis (Mar 01 2019 at 15:41, on Zulip):

We can always "archive" it or whatever

davidtwco (Mar 01 2019 at 15:41, on Zulip):

There are things that would need added (licenses, CoC, etc.), things that would need set up (bors, highfive, etc.) and then how it is added as a dependency to rustc. So the process we flesh out should include a list of all those things.

nikomatsakis (Mar 01 2019 at 15:42, on Zulip):

Yep, so presuming we settle on a name and add to the list, we create the repo and follow the steps to set it up

mw (Mar 01 2019 at 15:42, on Zulip):

also, do we have a guide on how to best add crates.io deps to rustc?

nikomatsakis (Mar 01 2019 at 15:42, on Zulip):

I don't think there's any discussion of that

Wesley Wiser (Mar 01 2019 at 15:42, on Zulip):

I assume we'd probably want anybody on the compiler team to be able to r+ stuff, but we'd probably only want "active participants" to be on the highfive rotation?

nikomatsakis (Mar 01 2019 at 15:43, on Zulip):

I do think that -- for a crate like this

nikomatsakis (Mar 01 2019 at 15:43, on Zulip):

it's a great chance to have active WG members get r+

nikomatsakis (Mar 01 2019 at 15:43, on Zulip):

i.e., for more collaboration

nikomatsakis (Mar 01 2019 at 15:43, on Zulip):

just coming back to that earlier question :)

mw (Mar 01 2019 at 15:44, on Zulip):

do we plan to automatically set up community infrastructure such as Zulip streams? (I'd say no)

nikomatsakis (Mar 01 2019 at 15:44, on Zulip):

I think that should be tied to the WG

mw (Mar 01 2019 at 15:44, on Zulip):

:+1:

nikomatsakis (Mar 01 2019 at 15:44, on Zulip):

I say go to the rust-lang repo

ps, do people disagree with this? :) (it's just my opinion, I'm open to other thoughts)

mw (Mar 01 2019 at 15:45, on Zulip):

I agree because it will make things like bors-ng and highfive integration easier

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

right, also team permissions

mw (Mar 01 2019 at 15:45, on Zulip):

yes

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

basically at the point where it would be in the compiler

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

it should be in rust-lang

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

I would also say that I've personally found it more convenient

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

to start there

nikomatsakis (Mar 01 2019 at 15:45, on Zulip):

whereas if it's in my repo

pnkfelix (Mar 01 2019 at 15:46, on Zulip):

presumably its not hard to migrate from a user-repo to the rust-lang one, if one does do initial private development there?

nikomatsakis (Mar 01 2019 at 15:46, on Zulip):

it's messier

nikomatsakis (Mar 01 2019 at 15:46, on Zulip):

presumably its not hard to migrate from a user-repo to the rust-lang one, if one does do initial private development there?

it is easy to transfer

mw (Mar 01 2019 at 15:46, on Zulip):

yes, so this is more like: there should be the option to start out immediately in rust-lang

nikomatsakis (Mar 01 2019 at 15:46, on Zulip):

because it overwrites the redirect =)

nikomatsakis (Mar 01 2019 at 15:47, on Zulip):

(you can email GH and they will fix it for you, i've done this e.g. for chalk, where I have a fork with a different name)

davidtwco (Mar 01 2019 at 15:47, on Zulip):

We should make sure that the responsible working group and that it is a compiler team repo is mentioned somewhere. Most of the other concerns about setting things up will need some collaboration with the infra team.

mw (Mar 01 2019 at 15:47, on Zulip):

do we have an example workflow written up somewhere on how to handle changes in external crates?

davidtwco (Mar 01 2019 at 15:48, on Zulip):

(my Zulip desktop client started swallowing messages there, so apologies if that showed up for you half a dozen times)

nikomatsakis (Mar 01 2019 at 15:48, on Zulip):

do we have an example workflow written up somewhere on how to handle changes in external crates?

this goes with your earlier question -- seems clear we need a rustc-guide section on this

nikomatsakis (Mar 01 2019 at 15:49, on Zulip):

I would also want to include (which we talked about earlier) guidance for how to make a local copy

nikomatsakis (Mar 01 2019 at 15:49, on Zulip):

for debugging

davidtwco (Mar 01 2019 at 15:49, on Zulip):

this goes with your earlier question -- seems clear we need a rustc-guide section on this

Where do you think the line is for what is part of the policy document and what is part of the rustc guide?

Wesley Wiser (Mar 01 2019 at 15:49, on Zulip):

It would be great if there was a standardized way of indicating what team(s) and/or WGs are responsible or a given repo. Some of the infrastructure setup like who is on highfive and has r+ could be automated at that point.

mw (Mar 01 2019 at 15:49, on Zulip):

I think the policy document should point to the rustc-guide for examples

nikomatsakis (Mar 01 2019 at 15:50, on Zulip):

Where do you think the line is for what is part of the policy document and what is part of the rustc guide?

Heh, I was wondering about that. I mean .. maybe that stuff should just go in the rustc-guide, but otherwise I'd say "things that affect the code go in rustc-guide", things that are more "meta" are "procedures", and sometimes a procedure might reference the rustc-guide (e.g., "see checklist in the rustc-guide")

davidtwco (Mar 01 2019 at 15:50, on Zulip):

Setting up a out of tree crate is in the policy document and working with one is in the guide?

pnkfelix (Mar 01 2019 at 15:50, on Zulip):

this goes with your earlier question -- seems clear we need a rustc-guide section on this

Where do you think the line is for what is part of the policy document and what is part of the rustc guide?

change to external crates is something that comes up naturally during development, and does not seem to require questions of policy, right?

pnkfelix (Mar 01 2019 at 15:50, on Zulip):

versus setting up new external crates is policy

nikomatsakis (Mar 01 2019 at 15:50, on Zulip):

It would be great if there was a standardized way of indicating what team(s) and/or WGs are responsible or a given repo.

To start I think we should have a template README that includes a link over to the working group etc

davidtwco (Mar 01 2019 at 15:50, on Zulip):

cc @Pietro Albini - there's probably a ton of interesting stuff for you in this thread.

nikomatsakis (Mar 01 2019 at 15:51, on Zulip):

But @Pietro Albini may have thoughts on that specific matter

nikomatsakis (Mar 01 2019 at 15:51, on Zulip):

heh, jinx

mw (Mar 01 2019 at 15:51, on Zulip):

a question we didn't answer yet: if someone wants to create a new crate, what do they do?

mw (Mar 01 2019 at 15:52, on Zulip):

where do they gather consent?

mw (Mar 01 2019 at 15:52, on Zulip):

I don't think the answer is: just create the project without asking :)

pnkfelix (Mar 01 2019 at 15:53, on Zulip):

Well

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

OK, so, where are we at.

pnkfelix (Mar 01 2019 at 15:53, on Zulip):

would it suffice as a T-compiler weekly meeting agenda item?

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

where do they gather consent?

so I created this "open a PR" step

pnkfelix (Mar 01 2019 at 15:53, on Zulip):

(creating a new crate, that is)

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

but it doesn't answer the core question

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

would it suffice as a T-compiler weekly meeting agenda item?

I think so

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

we could also use @rfcbot fcp poll

mw (Mar 01 2019 at 15:53, on Zulip):

would it suffice as a T-compiler weekly meeting agenda item?

I'd rather have something async here

nikomatsakis (Mar 01 2019 at 15:53, on Zulip):

but I think maybe it's overkill

pnkfelix (Mar 01 2019 at 15:54, on Zulip):

maybe both

pnkfelix (Mar 01 2019 at 15:54, on Zulip):

i.e. for people like me that cannot manage github notifications

pnkfelix (Mar 01 2019 at 15:54, on Zulip):

I'd prefer an announcement/inquiry the T-compiler setting

pnkfelix (Mar 01 2019 at 15:54, on Zulip):

but I think it makes sense to have async approval

nikomatsakis (Mar 01 2019 at 15:54, on Zulip):

For me personally, a ping in Zulip also works ;)

nikomatsakis (Mar 01 2019 at 15:55, on Zulip):

But yes to all the above

nikomatsakis (Mar 01 2019 at 15:55, on Zulip):

We should outline the goals of approval here:

nikomatsakis (Mar 01 2019 at 15:55, on Zulip):
nikomatsakis (Mar 01 2019 at 15:55, on Zulip):

I guess just some kind of sense for whether this is a "suitable" choice?

nikomatsakis (Mar 01 2019 at 15:56, on Zulip):

like, is it sufficiently disentangled from the compiler

nikomatsakis (Mar 01 2019 at 15:56, on Zulip):

I sort of feel like there's a general presumption that if somebody wants to, it's probably a good idea, and we're looking mostly for a few people to argee?

mw (Mar 01 2019 at 15:57, on Zulip):

Yeah, I'd assume people would get to an agreement within the resp. WG before

nikomatsakis (Mar 01 2019 at 15:57, on Zulip):

and the compiler meeting is also an option

nikomatsakis (Mar 01 2019 at 15:57, on Zulip):

(to draw attention to it)

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

wow, this idea of walking through an example was pretty good @mw ;)

mw (Mar 01 2019 at 15:58, on Zulip):

one thing about the compiler meeting: anything that is not a tagged GH issue is practically invisible

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

I was going to bring that up in a separate topic

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

But yeah

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

So, this would be a GH PR

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

but it'd be on the wrong repo

nikomatsakis (Mar 01 2019 at 15:58, on Zulip):

This is a general problem though

mw (Mar 01 2019 at 15:58, on Zulip):

about that: do we have a list of crates somewhere?

nikomatsakis (Mar 01 2019 at 15:59, on Zulip):

about that: do we have a list of crates somewhere?

oh, I sort of implied that we would add it to the compiler-team repository

nikomatsakis (Mar 01 2019 at 15:59, on Zulip):

(but maybe that's the wrong home)

nikomatsakis (Mar 01 2019 at 15:59, on Zulip):

i.e., we should be able to have P-high issues -- but on some library

this feels like something we won't fully solve today, but I guess it's a "to do" item -- adjust triage to allow us to track this sort of thing?

nikomatsakis (Mar 01 2019 at 16:00, on Zulip):

Regarding list of crates, I do think we should have one. I am agnostic about where, but i'd prefer not the main repo, because I'd prefer to not have to go through bors.

mw (Mar 01 2019 at 16:00, on Zulip):

huh, yeah, having multilpe GH projects makes searching for issues harder

nikomatsakis (Mar 01 2019 at 16:00, on Zulip):

This is plausibly something we can also take to GH

nikomatsakis (Mar 01 2019 at 16:00, on Zulip):

I wouldn't expect "quick" action

nikomatsakis (Mar 01 2019 at 16:00, on Zulip):

but it feels like a legit use case

nikomatsakis (Mar 01 2019 at 16:01, on Zulip):

It also further argues for using rust-lang

mw (Mar 01 2019 at 16:01, on Zulip):

I'm sure this could be automated with existing GH APIs though...

nikomatsakis (Mar 01 2019 at 16:01, on Zulip):

Right, I was just thinking that the usual fix here

nikomatsakis (Mar 01 2019 at 16:01, on Zulip):

is to create a "dashboard" that wraps GH

nikomatsakis (Mar 01 2019 at 16:01, on Zulip):

that's the kind of engineering project I'm trying not to suggest

Wesley Wiser (Mar 01 2019 at 16:01, on Zulip):

Isn't that sort of what octobox.io is?

nikomatsakis (Mar 01 2019 at 16:01, on Zulip):

Yeah, can it do this?

mw (Mar 01 2019 at 16:01, on Zulip):

I think the compiler-team repo might be a good place to have the list. because I don't want to have even more repos :)

Wesley Wiser (Mar 01 2019 at 16:01, on Zulip):

(I haven't used it)

nikomatsakis (Mar 01 2019 at 16:02, on Zulip):

I've never found that project very useful

davidtwco (Mar 01 2019 at 16:02, on Zulip):

We shouldn't have a repo to contain the list of repos? :stuck_out_tongue:

nikomatsakis (Mar 01 2019 at 16:02, on Zulip):

but dont' get me started about notifications. I have a lot of strong opinions :P

Wesley Wiser (Mar 01 2019 at 16:02, on Zulip):

From the screenshot it appears to have cross-repo aggregation with labels

nikomatsakis (Mar 01 2019 at 16:02, on Zulip):

i.e., it is basically defining "what is our jurisdiction"

nikomatsakis (Mar 01 2019 at 16:03, on Zulip):

Anyway, we're out of time for today, but it feels liike this was really helpful

nikomatsakis (Mar 01 2019 at 16:04, on Zulip):

I think compile-team should be it :) It's the root repo for "where to find out what's going on with the compiler team", seems like a plausible place for this.

also I just realized @davidtwco was joking :P

nikomatsakis (Mar 01 2019 at 16:04, on Zulip):

Ordinarily I write-up minutes from the meeting

nikomatsakis (Mar 01 2019 at 16:04, on Zulip):

So, I have a question

nikomatsakis (Mar 01 2019 at 16:04, on Zulip):

But I'm wondering if somebody else wants to take that on?

nikomatsakis (Mar 01 2019 at 16:04, on Zulip):

Ordinarily I write-up minutes from the meeting

not really minutes, more of a "collected summary" (example)

nikomatsakis (Mar 01 2019 at 16:05, on Zulip):

I'm also ok doing it here :) but it occured to me that this is the sort of job we ought to consider sharing :)

mw (Mar 01 2019 at 16:05, on Zulip):

I agree but don't want to do it :)

nikomatsakis (Mar 01 2019 at 16:05, on Zulip):

I mean this "laughter tears" emoji in both senses :P

davidtwco (Mar 01 2019 at 16:06, on Zulip):

I can write it up since I'll be going over this meeting for the draft policy changes anyway.

nikomatsakis (Mar 01 2019 at 16:07, on Zulip):

On, one last parting thing:

I've created a HackMd to collect steering meeting agenda ideas. In the future, when something comes up, please add it to the list. I'll be taking a look at it from time to time and trying to collate / collect.

nikomatsakis (Mar 01 2019 at 16:08, on Zulip):

I'm trying to get a bit more organized about this so that we can actually have a post with the agenda a few days before the meeting, instead of a few hours :P

pnkfelix (Mar 01 2019 at 16:08, on Zulip):

Can you maybe allocate a github issue and then link such things there?

pnkfelix (Mar 01 2019 at 16:08, on Zulip):

topic

pnkfelix (Mar 01 2019 at 16:08, on Zulip):

analogous to what we do for the triage meetings on Thursdays?

nikomatsakis (Mar 01 2019 at 16:08, on Zulip):

But I guess it's fine either wa

nikomatsakis (Mar 01 2019 at 16:08, on Zulip):

I'll do the isssue to make the link easier :)

nikomatsakis (Mar 01 2019 at 16:09, on Zulip):

and compiler-team repo can point at it

pnkfelix (Mar 01 2019 at 16:09, on Zulip):

i just like having links in topics

pnkfelix (Mar 01 2019 at 16:09, on Zulip):

since it is a piece of running context as the zulip-hosted dialogue moves along

pnkfelix (Mar 01 2019 at 16:11, on Zulip):

okay bye all gotta go!

nikomatsakis (Mar 01 2019 at 16:11, on Zulip):

Thanks to everyone who participated! These meetings never fail to lift my spirits. :heart:

nikomatsakis (Mar 01 2019 at 16:23, on Zulip):

OK, I created #58850 and retconn'd it into the zulip topic ;)

Pietro Albini (Mar 01 2019 at 18:00, on Zulip):

thanks for the ping! I'll split out a few threads to reply to various things

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

(Right now it's not an active problem we have, right?)

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

@mw concrete example are always good, did you have one in mind?

mw (Mar 01 2019 at 18:02, on Zulip):

do we announce somewhere that we want to create this crate?

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

Probably we want a list -- maybe in the repository? -- of compiler crates?

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

I meant the compile-team repository

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

nah

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

it's just a bit annoying because then you still don't have your own fork, and it's hard to set one up

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

Let's assume yes for now

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

I was thinking that a first step might be to open a PR to add to that list :)

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

i.e., how to build rustc with your own local changes applied

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

my personal branches are mixed with the "real thing"

mw (Mar 01 2019 at 18:02, on Zulip):

ah yes, that seems like a good place

mw (Mar 01 2019 at 18:02, on Zulip):

a poll seems fine

pnkfelix (Mar 01 2019 at 18:02, on Zulip):

namely so that we have a link in the title of the steering meeting

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

I also like gh issues as a poor man's wiki

pnkfelix (Mar 01 2019 at 18:02, on Zulip):

(or is the world deciding that this is an anti-pattern?)

davidtwco (Mar 01 2019 at 18:02, on Zulip):

There should be some mention of which working group and team are responsible for the crate.

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

@davidtwco thanks! I was just thinking that I'd probably wind up forwarding you the same list anyway

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

I was debating about that. I can certainly do it, though I think a page in the compiler-team repo might make more sense

nikomatsakis (Mar 01 2019 at 18:02, on Zulip):

seems good

mw (Mar 01 2019 at 18:03, on Zulip):

let's say we already have a non-controversial name for it

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

(Probably including a cargo publish to reserve the name?)

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

i.e., we should be able to have P-high issues -- but on some library

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

we need to adjust our links

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

I think compile-team should be it :) It's the root repo for "where to find out what's going on with the compiler team", seems like a plausible place for this.

mw (Mar 01 2019 at 18:03, on Zulip):

e.g. in order to avoid multiple versions

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

I would say: create a rfcbot fcp poll and, on Zulip, make a topic and ping some likely folks and try to get them to five feedback :)

mw (Mar 01 2019 at 18:03, on Zulip):

you mentioned a PR to that list

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

i.e., have a issue just to serve as info about the meeting?

nikomatsakis (Mar 01 2019 at 18:03, on Zulip):

because it lets me have a personal "fork"

mw (Mar 01 2019 at 18:03, on Zulip):

e.g. for performance testing

davidtwco (Mar 01 2019 at 18:03, on Zulip):

Anyway, we're out of time for today, but it feels liike this was really helpful

I've certainly got plenty of things I can update the draft policy with.

mw (Mar 01 2019 at 18:04, on Zulip):

do we require to have standalone tests?

mw (Mar 01 2019 at 18:04, on Zulip):

:D

davidtwco (Mar 01 2019 at 18:05, on Zulip):

We should make sure that the responsible working group and that it is a compiler team repo is mentioned somewhere. Most of the other concerns about setting things up will need some collaboration with the infra team.

pnkfelix (Mar 01 2019 at 18:05, on Zulip):

dad jokes are okay, as long as they not offensive

davidtwco (Mar 01 2019 at 18:06, on Zulip):

Something really strange is going on with this topic. The old topic name keeps appearing with a few messages in it and the timestamps are off.

davidtwco (Mar 01 2019 at 18:07, on Zulip):

Messages I sent hours ago are showing with timestamps for five minutes ago.

nikomatsakis (Mar 01 2019 at 18:08, on Zulip):

right

davidtwco (Mar 01 2019 at 18:09, on Zulip):

Is anyone else seeing this?

davidtwco (Mar 01 2019 at 18:09, on Zulip):

And there's the other topic showing up again..

davidtwco (Mar 01 2019 at 18:15, on Zulip):

It looks like random messages have just been pulled out of the actual chat and had their timestamp updated to the last 20 minutes.

nikomatsakis (Mar 01 2019 at 18:58, on Zulip):

maybe editing the topic wasn't so wise ;)

davidtwco (Mar 02 2019 at 13:29, on Zulip):

@nikomatsakis here's the PR with the minutes

davidtwco (Mar 02 2019 at 13:29, on Zulip):

Apparently summarizing isn't my strong suit.

davidtwco (Mar 02 2019 at 21:05, on Zulip):

We should make sure that the responsible working group and that it is a compiler team repo is mentioned somewhere. Most of the other concerns about setting things up will need some collaboration with the infra team.

davidtwco (Mar 02 2019 at 21:05, on Zulip):

Anyway, we're out of time for today, but it feels liike this was really helpful

I've certainly got plenty of things I can update the draft policy with.

davidtwco (Mar 02 2019 at 21:05, on Zulip):

There should be some mention of which working group and team are responsible for the crate.

pnkfelix (Mar 04 2019 at 09:59, on Zulip):

namely so that we have a link in the title of the steering meeting

pnkfelix (Mar 04 2019 at 09:59, on Zulip):

dad jokes are okay, as long as they not offensive

pnkfelix (Mar 04 2019 at 09:59, on Zulip):

(or is the world deciding that this is an anti-pattern?)

Pietro Albini (Mar 04 2019 at 10:00, on Zulip):

ugh the topics for this steering meeting are so confusing

pnkfelix (Mar 04 2019 at 10:51, on Zulip):

what the heck, did I accidentally cut-and-paste stuff?

pnkfelix (Mar 04 2019 at 10:52, on Zulip):

oh I see, something went wonky (as noted by @davidtwco ), possibly due to the rename of the topic name

pnkfelix (Mar 04 2019 at 10:52, on Zulip):

or maybe is just a simple zulip bug

pnkfelix (Mar 04 2019 at 10:53, on Zulip):

(which is scary)

davidtwco (Mar 04 2019 at 11:08, on Zulip):

I suspected it was due to them doing a rolling upgrade of 2.0 but it looks to have happened again this morning.

nikomatsakis (Mar 04 2019 at 21:06, on Zulip):

nikomatsakis here's the PR with the minutes

@davidtwco thanks! :heart: I'll take a look

nikomatsakis (Mar 05 2019 at 22:14, on Zulip):

hey @davidtwco -- your summary looks good to me, the only thing that I might say is that there might be value in trying to summarize it not so much as a "conversation" but rather as a kind of "coherent proposal"

nikomatsakis (Mar 05 2019 at 22:14, on Zulip):

OTOH, I'm happy to merge those notes, since I think we'd be creating something like that for WG-meta anyway

davidtwco (Mar 05 2019 at 22:16, on Zulip):

I'd lean towards merging as is, there's a coherent proposal for that discussion in the crates policy draft.

davidtwco (Mar 05 2019 at 22:17, on Zulip):

In any future summaries, I'll try to aim for a more coherent proposal.

nikomatsakis (Mar 05 2019 at 22:22, on Zulip):

Sounds good to me!

nikomatsakis (Mar 05 2019 at 22:22, on Zulip):

Also, thanks for doing that :heart:

nikomatsakis (Mar 05 2019 at 22:22, on Zulip):

@davidtwco merged -- you want to post to the internals thread?

davidtwco (Mar 05 2019 at 22:27, on Zulip):

posted, thanks!

Last update: Nov 22 2019 at 05:20UTC