Stream: rustdoc

Topic: backport procedure


view this post on Zulip Noah Lev (Feb 12 2021 at 02:42):

It seems like it would be a good idea for us to come up with a beta (and stable) backport procedure. Right now it's ad hoc — e.g., https://github.com/rust-lang/rust/pull/81831#issuecomment-777930430 — so it would be nice to have a standard procedure.

view this post on Zulip Noah Lev (Feb 12 2021 at 02:42):

It might be nice if we could use a "mini FCP" — one that perhaps didn't require as many approvals and was over a shorter time period.

view this post on Zulip Noah Lev (Feb 12 2021 at 02:43):

But that would require changes to rfcbot.

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:43):

if we could get rfcbot to remove the "at most 2 votes outstanding" requirement I think normal FCPs would work pretty well

view this post on Zulip Noah Lev (Feb 12 2021 at 02:44):

However, we could emulate the mini FCP by pinging @rust-lang/rustdoc and then having people add their names in an edit to the comment (or just posting new comments).

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:44):

that's basically what we do now, isn't it?

view this post on Zulip Noah Lev (Feb 12 2021 at 02:44):

Joshua Nelson said:

if we could get rfcbot to remove the "at most 2 votes outstanding" requirement I think normal FCPs would work pretty well

That and the 10 day time period. That seems too long for backports.

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:44):

oh yeah I wouldn't want any waiting period at all

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:44):

the worst that happens is some people don't see it :shrug: I've never had anyone object to a backport

view this post on Zulip Noah Lev (Feb 12 2021 at 02:44):

Joshua Nelson said:

that's basically what we do now, isn't it?

Yes, but not the editing the comment part. And also we could just stick to the ad-hoc approach but make it offical.

view this post on Zulip Noah Lev (Feb 12 2021 at 02:45):

So maybe the procedure could be "ping the team and wait for at least 2 other people to approve"? Maybe more people for stable backports...?

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:46):

I think currently T-compiler is actually in charge of backports

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:46):

we should probably talk to them

view this post on Zulip Noah Lev (Feb 12 2021 at 02:47):

Really? My understanding is that each team is responsible for accepting or rejecting backports.

view this post on Zulip Noah Lev (Feb 12 2021 at 02:47):

And then T-release (or someone else, but that's what I thought) does the actual backport once the team has accepted.

view this post on Zulip Noah Lev (Feb 12 2021 at 02:48):

https://forge.rust-lang.org/release/beta-backporting.html#backporting-in-rust-langrust

view this post on Zulip Joshua Nelson (Feb 12 2021 at 02:48):

well yes but @WG-prioritization definitely puts T-rustdoc issues on the weekly meeting agenda

view this post on Zulip Léo Lanteri Thauvin (Feb 12 2021 at 05:59):

I think currently T-compiler is actually in charge of backports

This is also what I reemember. The rustdoc team asked T-compiler to handle the rustdoc backports

view this post on Zulip apiraino (Feb 12 2021 at 09:48):

Léo Lanteri Thauvin said:

I think currently T-compiler is actually in charge of backports

This is also what I remember. The rustdoc team asked T-compiler to handle the rustdoc backports

correct, and T-rustdoc backports are approved (I would say) easily if a rustdoc team member r'ed the backport. Last example I can think of was #81288, meeting discussion

view this post on Zulip apiraino (Feb 12 2021 at 09:49):

just to understand the context, @Camelid the discussion here is about #81831 beta nomination that didnt make it into the agenda so it will slip to the next beta cycle?

view this post on Zulip Noah Lev (Feb 13 2021 at 01:43):

apiraino said:

just to understand the context, Camelid the discussion here is about #81831 beta nomination that didnt make it into the agenda so it will slip to the next beta cycle?

I'm not sure what you mean. The issue is that it landed a few days after beta was branched and so the current status is:


Last updated: Oct 21 2021 at 19:46 UTC