Stream: t-compiler

Topic: forge vs compiler-team vs rustc-dev-guide


Santiago Pastorino (Apr 28 2020 at 20:45, on Zulip):

during a @WG-rustc-dev-guide meeting we were discussing the difference between what should go in forge, compiler-team and rustc-dev-guide

Santiago Pastorino (Apr 28 2020 at 20:45, on Zulip):

in some sense it seems like rustc-dev-guide should be about understanding how the compiler works

Santiago Pastorino (Apr 28 2020 at 20:45, on Zulip):

in forge we have some procedures

Santiago Pastorino (Apr 28 2020 at 20:46, on Zulip):

but it's not 100% clear what should go on each place

Santiago Pastorino (Apr 28 2020 at 20:46, on Zulip):

for instance:

Santiago Pastorino (Apr 28 2020 at 20:47, on Zulip):

we have a procedures section in compiler-team https://rust-lang.github.io/compiler-team/procedures/crates/

Santiago Pastorino (Apr 28 2020 at 20:47, on Zulip):

but some other procedures were moved to forge

Santiago Pastorino (Apr 28 2020 at 20:48, on Zulip):

and we have things like https://rustc-dev-guide.rust-lang.org/ice-breaker/about.html what's an ice-breaker group and how works, that seems a bit like a procedure to me and also things like https://rustc-dev-guide.rust-lang.org/stabilization_guide.html

Santiago Pastorino (Apr 28 2020 at 20:50, on Zulip):

there are a lot of things that are debatable and we may try to figure out but I guess a high level idea of what should go on each would be great :)

Santiago Pastorino (Apr 28 2020 at 20:50, on Zulip):

cc @nikomatsakis @pnkfelix

nagisa (Apr 28 2020 at 21:19, on Zulip):

to me forge was always about having a higher-level view around our processes. Examples:

  1. When is the next release?
  2. What targets do we support?

and similar. I could see compiler-team page being a subsection in forge instead.

Santiago Pastorino (Apr 28 2020 at 21:24, on Zulip):

@nagisa in that sense, where do you think https://rustc-dev-guide.rust-lang.org/ice-breaker/about.html, https://rustc-dev-guide.rust-lang.org/stabilization_guide.html and https://rust-lang.github.io/compiler-team/procedures/crates/ should go? :)

nikomatsakis (Apr 28 2020 at 22:05, on Zulip):

So, I put the ice-breaker stuff in rustc-dev-guide because it was connected to "contribution"

nikomatsakis (Apr 28 2020 at 22:05, on Zulip):

i.e., it's a way to get involved

nikomatsakis (Apr 28 2020 at 22:06, on Zulip):

and not so much a "procedure"

nikomatsakis (Apr 28 2020 at 22:06, on Zulip):

though I agree it could potentially live in both places, and one could imagine forge having a "forwarding link" to rustc-dev-guide

nikomatsakis (Apr 28 2020 at 22:06, on Zulip):

Stabilizing features is particularly complicated because it overlaps to some extent with other teams, like lang team etc

nikomatsakis (Apr 28 2020 at 22:07, on Zulip):

but I guess that by the logic of "kinds of contributions" it fits

nikomatsakis (Apr 28 2020 at 22:07, on Zulip):

I hvae to say that Part 1 of the rustc-dev-guide kind of looks like a hodge podge right now as I skim ot iver

nikomatsakis (Apr 28 2020 at 22:07, on Zulip):

https://rust-lang.github.io/compiler-team/procedures/crates/

I think this clearly belongs in forge, I think that the compiler-team website should be deprecated in favor of forge for this sort of thing

nikomatsakis (Apr 28 2020 at 22:08, on Zulip):

I think the role of the compiler-team website is probably just limited to tracking minutes, proposals, and working groups and things

nikomatsakis (Apr 28 2020 at 22:08, on Zulip):

I'm not 100% convinced though that the "ICE-breaker content" belongs in rustc-dev-guide, and could be persuaded that it should move to forge, along with other "contributor like" material

nikomatsakis (Apr 28 2020 at 22:09, on Zulip):

in that case, rustc-dev-guide would be really focused more on documenting rustc and not so much the procedure around contribution, issue triage, etc

nikomatsakis (Apr 28 2020 at 22:09, on Zulip):

I think I would probably also move out then the stabilization PRs, as well as like "walking you through your first contribution"

nikomatsakis (Apr 28 2020 at 22:09, on Zulip):

and have a link over to forge for that sort of thing

Santiago Pastorino (Apr 29 2020 at 11:17, on Zulip):

@nikomatsakis this is exactly what I was thinking :)

Santiago Pastorino (Apr 29 2020 at 11:17, on Zulip):

make rustc-dev-guide more focused on documenting rustc

Santiago Pastorino (Apr 29 2020 at 11:18, on Zulip):

it should totally have a contributing section but that section could be short, with links to ice breakers page and things like that that live in forge

Santiago Pastorino (Apr 29 2020 at 11:19, on Zulip):

so we can make rustc-dev-guide more readable and more focused on understanding how the compiler works

mark-i-m (Apr 30 2020 at 02:34, on Zulip):

nikomatsakis said:

I hvae to say that Part 1 of the rustc-dev-guide kind of looks like a hodge podge right now as I skim ot iver

Yes, this sprint the wg wants to fix this...

mark-i-m (Apr 30 2020 at 02:35, on Zulip):

Personally, I lean towards having the guide be a 1-stop shop for all the things a contributor could need, but others in the WG would like it to be more book-like.

nikomatsakis (Apr 30 2020 at 14:22, on Zulip):

I think that having Part 1 have some pointers to forge, appendix, or other content can be a middle ground

nikomatsakis (Apr 30 2020 at 14:22, on Zulip):

it's what I would prefer, I think

nikomatsakis (Apr 30 2020 at 14:22, on Zulip):

I was saying to @Santiago Pastorino that it seems like a good "rule of thumb" might be -- the rustc-guide is all about what you can do after you clone the git repo. It's not about things that require interacting with github issues =)

mark-i-m (Apr 30 2020 at 21:03, on Zulip):

I guess my question is why? What goal does it accomplish? Making the rustc-dev-guide readable as book is cool, but does it serve any real need? Personally, when I was starting out in contributing, I was a bit bewildered by the fact that information is spread out in soooo many places. We use 4-5 different platforms for discussion, the forge, compiler-team, rustc-dev-guide, tracking issues, random hackmds and dropbox papers... It just seems like it would be helpful to consolidate information somewhere.

nikomatsakis (May 02 2020 at 10:28, on Zulip):

I think my concern is that the more stuff we add into rustc-dev-guide, the more it itself is bewildering.

nikomatsakis (May 02 2020 at 10:28, on Zulip):

But I feel like having it be a link to forge

nikomatsakis (May 02 2020 at 10:28, on Zulip):

Serves two purposes

nikomatsakis (May 02 2020 at 10:28, on Zulip):

First off, it lets you know forge exists :)

nikomatsakis (May 02 2020 at 10:29, on Zulip):

And it helps us to keep the rustc-dev-guide ToC readable and focused

nikomatsakis (May 02 2020 at 10:29, on Zulip):

I do think it's important that it has appendices and sections

nikomatsakis (May 02 2020 at 10:29, on Zulip):

I guess the alternative is to move all compiler team material into rustc-dev-guide --

nikomatsakis (May 02 2020 at 10:29, on Zulip):

but I think that team procedures and things will overlap with other teams

nikomatsakis (May 02 2020 at 10:29, on Zulip):

e.g., I'd like to have centralized docs for "so you want to make a change ..." that guides folks on whether they need to write an RFC etc

nikomatsakis (May 02 2020 at 10:30, on Zulip):

and I think we can centralize all the team docs for that on forge

nikomatsakis (May 02 2020 at 10:30, on Zulip):

I don't think it would make sense to put all that into rustc-dev-guide

Santiago Pastorino (May 04 2020 at 15:56, on Zulip):

agree with @nikomatsakis here, I think consolidating is good, forge for all the processes of all the teams, rustc-dev-guide for all that is how the compiler works

Santiago Pastorino (May 04 2020 at 15:56, on Zulip):

anyway, I agree that we can link from rustc-dev-guide to forge in a lot of places

mark-i-m (May 04 2020 at 16:38, on Zulip):

I guess I would still rather have everything in one place, but I seem to be in the minority. I'm willing to give it try and maybe evaluate again afterwards

Santiago Pastorino (May 04 2020 at 17:00, on Zulip):

@mark-i-m yeah, we can try things and revert if they don't look great

Last update: May 29 2020 at 16:15UTC