Oh, today is a meeting day.
We have 2 beta-nominated issues we need to vote on
We also have a few nominated issues to talk about.
I think we do that.
For the first nominated issue:
Avoid ICE when referencing desugared local binding in borrow error #63051
Actual code change seems tiny + safe.
Yeah, and something that is very nice to have backported too, although probably not critical
any complaints since this has been merged?
no complaints to my knowledge at least
Preemptively marking as accepted
Don't access a static just for its size and alignment #62982 is the next one
(I'm a bit thunderstruck that there was no test for this before this PR)
Not as simple of a PR, but fixes a stable-to-beta regression
Does @oli want to weigh in?
I’m not opposed to backporting this just because this is a regression fix
I’m in favour just because of that, actually
Seems fine to backport but I don't know the code really
Fix has been in for 11 days -- not a very long time
don’t want to have a broken beta become stable
Trains move next week.
So there is no time for this to bake further.
let's do it then
11 days is not nothing
and there's still some time
Yeah I agree; I'm in favor of backport but concurrently unhappy about things ;)
Onto I-nominated stuff Recompiling causes panic #63349
This looks like codegen unit collection thing probably
but the only real conclusion I can draw is that this needs investigation.
btw, cc @Pietro Albini ; you now have 2 new beta backports to land ^^
I'll do it sometime in the next few days
I can try to reproduce the problem
yeah, let's call it p-high for now
Next one is ICE when constructing monomorphic vtable in a static in dependent crate. #63226
btw... could we change the ICE message to ask folks to minimize the problem before filing an issue (if they can)?
well an issue on ICE is desirable regardless of whether they have a minimal reproducer or not.
Sure it is; I'm just saying that a minimization is even better
and it might not occur to people to minimize otherwise
I think for #63226 I can only suggest @oli to sync with @mw
and discuss options and whatnot
yes, I might be able to help there
Does this have t-lang implications?
I don't know
Assigned both of you and marked P-high
Compiler panic on a macro invocation combined with invalid pub declaration #63223 already has @Vadim Petrochenkov assigned.
Doesn’t appear like there is anything to discuss here.
ICE in macro: doc meta with expr on an item, string concat, stringify!(...) #55414 appears to be non-regression ICE
As usual for ICEs P-high.
Seems like @Vadim Petrochenkov's wheelhouse
or maybe one of the compiler contributors?
Pinged @Vadim Petrochenkov, leaving unassigned for now.
ICE when using anonymization in impl Trait with HRTB #54895
The remaining 3 issues in general appear to share the same theme
need investigation, a long-standing ICE
Actually Trait bounds are not checked on type aliases until they are used #51626 is different
Trait bounds are not checked on type aliases until they are used #51626 looks like T-lang to me actually
not sure if we want to enforce anything on aliases.
Long standing well known bug; type aliases do not WF-check the RHS
we tried to fix it in the 2018 edition
This wants lazy normalization afaik
I’ll leave this nominated then, because @nikomatsakis is way better equipped to make the call on this
For the other two, P-high, leaving unassigned.
That’s all for I-nominated issues.
There are no stable-to-beta regressions to look at
There are a number of stable-to-nightly ones.
Link errors on MSVC Nightly since around 2019-07-13 #63361 appears to be the only new one
Alex seems to have commented on what the next step should be
I generally agree with them
this is not the first time
target-cpu=native is not being very well behaved.
Assign to yourself + alex?
Sure, although I do not anticipate myself looking at it.
There are two PRs that are waiting on us
Make SyncDroplessArena allocations thread-local in the fast path #61873
Seems like this is just waiting for aturon or somebody else to review it.
Reverted to waiting-on-review.
(It would be really good if those large unsafe blocks had comments.)
It would be good to know why this is needed
feel free to type your reviews on the PR itself.
Turn HIR indexing into a query #59064 appears to need more than 10 minutes of a random meeting with low participation.
so leaving it alone.
With that all waiting-on-team are covered as well.
I think it is a good spot to wrap this meeting up, having covered all the most important bases.
(I hear that there should be checkin calendar somewhere, but I still have no idea where)
Thanks everyone for attending!
thanks a lot for driving, @nagisa!