Stream: t-compiler

Topic: beta nominations


Pietro Albini (Oct 02 2018 at 10:24, on Zulip):

Hey! There are a lot of beta nominations for T-compiler, could you check them out?

pnkfelix (Oct 02 2018 at 11:27, on Zulip):

@Pietro Albini We typically go through the beta-nominations at our weekly meeting. Last week's got a bit bogged down, so we didn't properly go through the list at that time, and then clearly the list has also grown since then...

pnkfelix (Oct 02 2018 at 11:28, on Zulip):

so the question becomes: Do we try to apply some sort of asynchronous and/or ad-hoc system for approving backports in the meantime, or do we wait until the meeting coming this Thursday

pnkfelix (Oct 02 2018 at 11:29, on Zulip):

I know that @nikomatsakis often will approve things for beta-backporting that are "obviously fine" based on his personal evaluation. Its possible that I (or he, or another member of @T-compiler ) should go through that process today or tomorrow, just to reduce the size of the list going into the meeting on thursday...

oli (Oct 02 2018 at 11:33, on Zulip):

I tried that earlier. I wasn't very sure about my judgements so I left almost all of them alone. I can do another round and be more "aggressive"

pnkfelix (Oct 02 2018 at 11:35, on Zulip):

I personally would like to try to ensure that members of the compiler team at least get a chance to object

pnkfelix (Oct 02 2018 at 11:36, on Zulip):

So what i'm currently thinking is this: Its two days until the meeting anyway. We could have a policy that any member of the compiler team can write a comment stating that they think a beta-nominated PR should be backported

pnkfelix (Oct 02 2018 at 11:37, on Zulip):

and then once we get to the meeting, that is the synchronization point

pnkfelix (Oct 02 2018 at 11:37, on Zulip):

But instead of slowing waiting on each issue to see if anyone objects

pnkfelix (Oct 02 2018 at 11:37, on Zulip):

we instead quickly enumerate the ones that have already been conditionally approved

pnkfelix (Oct 02 2018 at 11:38, on Zulip):

at least, that sounds quicker to me than what I had been doing...

pnkfelix (Oct 02 2018 at 11:38, on Zulip):

One might argue that the act of applying a beta-nomination tag should already count as the upvote for a backport

pnkfelix (Oct 02 2018 at 11:39, on Zulip):

(and that I have simply been too slow in going through them.)

oli (Oct 02 2018 at 11:48, on Zulip):

ups, well. I did those that I felt confident about (in necessity, backportability, code and the fact that the nomination wasn't mine)

pnkfelix (Oct 02 2018 at 11:58, on Zulip):

Its fine, I'm more just trying to figure out what sort of long term precedent to set.

pnkfelix (Oct 02 2018 at 11:59, on Zulip):

if a person is confident that a given PR is fine for backport, then it does seem wasteful to force it to wait until the team meeting.

pnkfelix (Oct 02 2018 at 12:00, on Zulip):

@Oli I just want to confirm: when you removed the beta-nominated tag from PR's like #54650, is there a corresponding merge to beta for that backport?

pnkfelix (Oct 02 2018 at 12:01, on Zulip):

@Oli i.e. I believe our protocol is that we leave PR's tagged as both beta-nominated and beta-accepted when they are in the state that the given PR has been both "approved for backport" and that particular backport hasn't happened yet; we remove the beta-nominated tag after the backport happens.

pnkfelix (Oct 02 2018 at 12:03, on Zulip):

... given that the last commit to beta was 2 days ago, I am inferring that @Oli was not aware of the aforementioned protocol ...

pnkfelix (Oct 02 2018 at 12:06, on Zulip):

(okay, I think I fixed all the cases of the problem in the previous comment now...)

Pietro Albini (Oct 02 2018 at 12:09, on Zulip):

@pnkfelix actually, beta-nominated should be removed when the backport is r+ed

Pietro Albini (Oct 02 2018 at 12:10, on Zulip):

I'll do a backport PR now

pnkfelix (Oct 02 2018 at 12:10, on Zulip):

okay, either way, there were not yet backport PR's in this case, right?

Pietro Albini (Oct 02 2018 at 12:11, on Zulip):

nope, the only backport PR is #54748 by @Oli, for #54693 which is not yet approved

oli (Oct 02 2018 at 12:21, on Zulip):

Yea, I wasn't aware of that protocol. Thanks for the update

Pietro Albini (Oct 02 2018 at 12:21, on Zulip):

did a rollup PR for the backports :)

oli (Oct 02 2018 at 12:22, on Zulip):

Ah zulip was eating notifications. Fixed now

Last update: Nov 22 2019 at 04:35UTC