Stream: t-lang/meta

Topic: lang team procedure meeting


view this post on Zulip nikomatsakis (Oct 19 2020 at 20:35):

So @Josh Triplett regarding scheduling a design meeting to discuss lang team procedure -- I think a good idea is for us to write up a specific proposal first

view this post on Zulip nikomatsakis (Oct 19 2020 at 20:36):

in particular I wanted to maybe find a time for you, me, @pnkfelix, @Ashley Mannix to chat about consolidating the various procedures

view this post on Zulip nikomatsakis (Oct 19 2020 at 20:37):

or maybe it's better to just start by sketching things out

view this post on Zulip nikomatsakis (Oct 19 2020 at 20:37):

I'm reminded, @simulacrum, of the various rustbot changes we were contemplating to support MCPs, as well

view this post on Zulip simulacrum (Oct 19 2020 at 20:37):

I think they're all mostly waiting on specs (or at least first drafts)

view this post on Zulip nikomatsakis (Oct 19 2020 at 20:38):

One thing I was thinking @Josh Triplett after our conversation was that delineating what sorts of decisions want "seconding" vs "full FCP" makes sense, and I was wondering if it is ultimately not correlated with controversy per se (though that may be a factor) but also irreversibility. In particular, I think that part of the problem with e.g. the rename from src/libstd to src/library or whatever that made it controversial was that while we can technically reverse it, we can't really

view this post on Zulip nikomatsakis (Oct 19 2020 at 20:39):

seconding just doesn't give as much "weight" as full check off

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:22):

Just saw this thread, after seeing the scheduling thread.

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:23):

I do agree that the fundamental factor for "second and no objections" vs "consensus and no objections" is reversibility.

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:23):

Closing an issue is trivially reversed, and the only issue is that people may have lost interest.

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:23):

Stabilizing something isn't reversible.

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:24):

Postpone is reversible.

view this post on Zulip Josh Triplett (Oct 19 2020 at 21:25):

"Chartering a project group" is socially painful to reverse, but not irreparable. It's effectively "we're receptive to this, so we think you wouldn't be wasting your time to put a proposal together".

view this post on Zulip simulacrum (Oct 19 2020 at 21:29):

I don't think we'd want to -- in general -- reverse direction on a project group. We might, though, want to say that given the exploration done of the problem/solution space, we do not think there is a pressing need anymore and/or solutions are too hard/expensive to design.

view this post on Zulip nikomatsakis (Oct 19 2020 at 21:34):

This I think is the question -- how reversible is it. I still kind of want agreeing to a project group to represent a sort of commitment, though I wonder if the commitment should come a bit later. i.e., give people a bit of time to explore before we make a harder choice.

view this post on Zulip nikomatsakis (Oct 19 2020 at 21:34):

I am reminded of stage0 / stage1 in TC39

view this post on Zulip nikomatsakis (Oct 19 2020 at 21:34):

I think the bar for stage0 might be quite low, and having some interesting lang team member who is willing to second may suffice

view this post on Zulip Josh Triplett (Oct 19 2020 at 22:01):

That's fair.

view this post on Zulip Josh Triplett (Oct 19 2020 at 22:02):

I was always imagining the initial approval as "yeah, we'd like to see that".

view this post on Zulip Josh Triplett (Oct 19 2020 at 22:02):

So a two-phase approval where the first phase is just a second, perhaps?

view this post on Zulip nikomatsakis (Oct 19 2020 at 22:09):

I will say this

view this post on Zulip nikomatsakis (Oct 19 2020 at 22:09):

I think the experience of "I have an idea!" where the immediate next step is "I create a charter than then sits in FCP limbo for a month" is not great

view this post on Zulip nikomatsakis (Oct 19 2020 at 22:10):

I'd rather have it be "any lang team member can second and charter a group that is in an experimental state; we create a stream and people can start chatting", but I do worry about somebody running off and pursuing an idea others are really not keen on, and this setting up for some kind of problem

view this post on Zulip nikomatsakis (Oct 19 2020 at 22:10):

maybe there's some kind of in-between, not sure

view this post on Zulip simulacrum (Oct 19 2020 at 23:17):

I think I'd be surprised with that being the case with current lang team at least :)

view this post on Zulip simulacrum (Oct 19 2020 at 23:18):

i.e., at least from my perspective it feels like there is normally agreement on "whether" in some sense, though of course not necessarily "how"

view this post on Zulip simulacrum (Oct 19 2020 at 23:19):

And in any case, there'd be roughly 12ish runaway projects at once presuming we limit to 2-3 per member or so

view this post on Zulip nikomatsakis (Oct 20 2020 at 19:34):

Yeah, I do think getting agreement on "whether" is important


Last updated: Jan 26 2022 at 08:02 UTC