Good afternoon @WG-compiler-nll — sync meeting in a few minutes
hey everyone; sorry, i'm moving a little slow today
we've got some recent injections that are probably due to the recent
E as T support
did we open an issue on that?
there's at least one
but e.g. https://github.com/rust-lang/rust/issues/55241 is ... another?
maybe what I really meant is
should we add those to the milestone
@RalfJ pointed out a possible fix over in #t-compiler
so the ICE's aren't mitigated
in this topic — but perhaps I misunderstood
oh I hadn't seen @RalfJ's note
/me was literally lying down five minutes ago
I can take a look at fixing that this afternoon perhaps
my suspicion is that the ICE's dont block the release
but they should be targetted for backports to beta if they don't get in before the rollover
(has the rollover from nightly to beta already happened?)
that's tomorrow, right?
FWIW meeting is kinda late, beta/stable promotion needs to happen Wednesday at the absolute latest
that's what @simulacrum said earlier
I don't 100% know what means
should we remove that from the milestone, then, and move to the "release" milestone?
"meeting is kinda late" ?
anyway, we're kind of in the weeds a bit here :)
this meeting is "kinda late"?
oh, no, that was referreing to something else
the "rust release" mtg I think
so @pnkfelix and I had been talking about using this meeting to talk "post NLL shipping" plans
but I will say that reviewing the list of work makes me realize there's still plenty of little things to handle ;)
but probably it's good to talk about that anyway
/me hopes https://github.com/rust-lang/rust/pull/55229 will land dang it
right. but I'll say up front that my incination was "we're pretty close to being 'done' with NLL as an initial product"
yep, I agree
and so I wasn't sure whether we needed this standing meeting as it currently stands
(who's here today, btw?)
looks like @Matthew Jasper and @davidtwco
(from paper doc)
(and from the waves above)
I left some notes in the paper doc (e.g., here)
just kind of "brain dumping"
it might be worth looking at the A-nll bugs though; it seems to me that when it comes to diagnostics we are basically at the "good enough" category, though I still think it'd be great to have a renewed push around simplifying our vocabulary etc, but I might like to get a bit of experience first
yeah I still need to do the .stderr review
notably we still have 19 bugs tagged to the final release in 6 weeks
a bunch of those are NLL-fixed-by-NLL though
the NLL-deferred might be more worrisome
8 are diagnostics
7 are soundness
basically all user type annot
and 3 NLL-deferred issues :)
anyway I think all that this means is that there is (sadly) still some work to do on user-type annotations post RC2
this brings up that question I raised of whether we expect to backport those things
plus I imagine there'll be more ICEs
@nikomatsakis @pnkfelix FWIW nightly/beta promotion is early next week, that was specifically referencing anything that needs to be in current beta (i.e., Thursday's stable)
@simulacrum I see
Which was I think a non-NLL thing that I was referencing
Is there an NLL thing that I should be aware of?
not related to current stable
we are trying to close out all the RC2 blockers
okay, sounds good
really right now we've just got one
which is being tested by bors right now
there are some ICEs that have been cropping up
so those might be "nice to haves"
once we have a fix
wait nightly/beta promotion is next week ... based on, whatever state nightly is in at that time next week?
well, it was supposed to be today
as in, the October 25th deadline is ... very soft indeed?
I've not had a chance to pickup a new issue yet - if there's anything that has a priority that you think I could handle I can dump some time into it over the next few days.
Yes, because it's RC2
The next milestone is not debatable/modifiable
@davidtwco maybe one of these ICEs
i see, so the rollover date for RC2 is slipping a few days then?
just trying to make sure I understand
Yes, because I think there's NLL work and a few other things that want to be in that beta so we'll just wait instead of backporting
All three of the
I-ICE issues are assigned to you or @pnkfelix - are there others that aren't tagged like that?
I don't know of one, we could maybe unassign
(I think it makes the release slightly smoother too, in some sense -- less things that need to happen this week)
Just to make sure NLL has no work that needs to land in current stable, right?
I believe that is the case, yes (that there is no NLL work that needs to land in the current stable)
Okay -- if not, please ping me so I can be aware of it before the release :)
I don't think we even have anything that needs to lands in the current beta, either ... right?
Hmm @davidtwco these ICEs don't all look the same; some of the ones in #55241 may be fixed by my PR
we basically were deciding about whether to try to force some stuff in, or plan on backports, or just let certain subtle things ship in a buggy state initially
@pnkfelix well I had #55229 as a blocker, since it's ready and fixes various ICEs and unsoundnesses
but if push came to shove...
anyway, it should hopefully land very soon
er, sorry, wrong link (fixed)
right, but that isn't targeting current beta? Just nightly?
I guess I may not be talking clearly about "current beta" vs "next beta' ?
sorry I misunderstood you I guess
I was just reacting to @simulacrum 's Q "Just to make sure NLL has no work that needs to land in current stable, right?"
right, I think your answer is correct
anyway it certainly sounds like #55229 will land in time
Holy crow its been 29minutes and we haven't even mentioned what I thought was going to be the main topic of the meeting
If you want to pick something suitable and assign me it then I'll take a look at that.
I'll also pick up random diagnostic issues throughout the week if I have time.
@pnkfelix which is the
&'staticissue we were talking about again...?
/me puts it on the milestone
so, @davidtwco, I think we should investigate that one https://github.com/rust-lang/rust/issues/55288 :)
might require a bit of context
but we can chat about it over in another topic
I just assigned you tho
the topic that @pnkfelix and I had intended to cover was sort of the "post shipping plans"... you can see I left some notes about possible directions we can go ...
@pnkfelix and I were chatting this morning and in particular debating about whether to push more on polonius or wait until we've "cleaned up" the existing codebase more
since right now it is kind of a mess — for example, we have two borrow checkers and two region checkers :)
(3 if you count polonius in its current state...)
Double the safety
we sort of surprised ourselves by finding that it seemed like it made sense to push on polonius — basically because it shouldn't impede the other stuff, hopefully, and that stuff kind of has to bake before we can do much simplification
I think that over the next N weeks though a lot depends on how much work we wind up having to do getting NLL formally shipped
Anyway, I guess we'll talk about it more next time. I hope to put in some time bringing polonius back in cache this week, though I've said that before. I'd like to at least start documenting poloinius in rustc-guide and decide whether that case that was a discrepancy between opt and naive was indeed a bug
it occurs to me btw @pnkfelix that this is another work item — documenting
e.g., rustc-guide, but also rust reference etc
I hope to get to that before too long
for the reference at least
to what specifically? ah
well I guess we can talk about it more next week
besides that ICE, I think the top priority should be the user-type-annot bugs
not sure how many of those are really left
There are open PRs for the two major cases
yeah we've got some PR's to land
The rest appear to be unfortunate edge cases
and normalization related bugs
not sure if that's "edge case" or not
yeah I want to dig into that one more...
sorry I couldn’t participate today, travelling to Mexico
gonna read the log tomorrow