@WG-prioritization issue #71359 has been requested for prioritization.
This might be expected fallout, but it's good to check. Can users set the recursion limit to get around this?
Having a backtrace for the error would be helpful
(Running with RUST_BACKTRACE does not change the output.)
If we run with
-Ztreat-err-as-bug, we should get a backtrace where that error is being generated.
@Wesley Wiser I tried to run:
RUST_BACKTRACE=1 cargo +nightly rustc -- -Ztreat-err-as-bug
But it didn't ICE, so I have no backtrace
Any reason why?
Huh, I've never seen
treat-err-as-bug not work...
this is a bit surprising to me too this is not showing any pointer to what's going on
If it's related to #32498 as the author mentions, I suspect the error is coming from the codegen backend which might not respect
I guess it may be nice to see what happens if we revert #70896
just in case and maybe ping @cuviper
this is the last unprioritized one :)
I think there may be a standing policy for this kind of issue
@Wesley Wiser what do you mean?
I feel like I've seen this kind of issue before
And I feel like this maybe just be acceptable breakage
Sort of in the same vein as the macro recursion limit stuff
I see, maybe it's good to nominate it?
Maybe @pnkfelix has an opinion?
TLDR: a recently nightly caused a crate to start breaking during codegen with type length limit exceeded errors
maybe P-medium but nominate for discussion...
@pnkfelix do you want to nominate it to just discuss this particular issue or what @Wesley Wiser have said, that we need a policy for this kind of issues?
Issue #71359's prioritization request has been removed.
I think it's probably worth having a small discussion in the meeting about this
Santiago Pastorino said:
pnkfelix do you want to nominate it to just discuss this particular issue or what Wesley Wiser have said, that we need a policy for this kind of issues?
Maybe both. At first my intent with my nomination suggestion was for more of a "announcement, we prioritized this as P-medium but it may be of general interest/concern. But now I am also thinking that @Wesley Wiser is making a good point that we should discuss what the policy is for cases like this. We don't think we want instances of this problem to just gradually accumulate...
From what I've seen in the Crater report, 6 crates broke because of this
There's nothing particularly worrying IMHO