regression: overflow evaluating requirement

[Mark-Simulacrum]

• https://crater-reports.s3.amazonaws.com/beta-1.71-4/beta-2023-06-25/reg/html-0.5.2/log.txt

[INFO] [stdout] error[E0275]: overflow evaluating the requirement `html_sys::scripting::noscript::NoScript: Unpin`
[INFO] [stdout]     |
[INFO] [stdout]     = help: consider increasing the recursion limit by adding a `#![recursion_limit = "256"]` attribute to your crate (`test`)
[INFO] [stdout] note: required because it appears within the type `NoScript`
[INFO] [stdout]    --> /opt/rustwide/workdir/src/generated/noscript.rs:8:16
[INFO] [stdout]     |
[INFO] [stdout] 8   |     pub struct NoScript {
[INFO] [stdout]     |                ^^^^^^^^

[hameerabbasi]

This seems to already have been fixed on the main branch and also on cca7ee5 (which is what I get if I run cargo-bisect-rustc with the HEAD of the beta branch).

[Mark-Simulacrum]

Hm, that rollup doesn't have anything obvious in it - can we bisect on main (master) or check the individual PRs in the rollup? Sounds like there may be a missing backport?

[hameerabbasi]

I'm sorry, I might have miscommunicated. Running cargo-bisect-rustc --start 1.70.0 and cargo-bisect-rustc --start 1.70.0 --end cca7ee58110726983951a19d5fb7316d9243925d both say that the issue isn't present on both endpoints.

Sounds like there may be a missing backport?

That's exactly my understanding as well. However, I don't see an obvious way to find out which PR fixed/caused the issue. My first intuition was to check the beta branch, but cargo-bisect-rustc only seems to work on master.

[compiler-errors]

@hameerabbasi make sure you're building all tests too. this continues to fail on master.

[hameerabbasi]

Ah -- my mistake. I'll do cargo test, I was doing cargo build before.

[apiraino]

this now landed on stable. Following the suggestion of rustc to increase the recursion limit fixes the tests. This can be a legit solution, right? maybe worth reporting to the maintainer of the html repository?

[apiraino]

WG-prioritization assigning priority (Zulip discussion).

@rustbot label -I-prioritize +P-medium

[Skepfyr]

I suspect that this is a real recursion limit hit because the html crate is a bit weird in being an automatically generated reflection of the HTML spec. However, if anyone's interested the regresison is somewhere in nightly-2023-05-24.
cargo-bisect-rustc found 8 bors merge commits in the specified range
commit[0] 2023-05-22: Auto merge of #111848 - Dylan-DPC:rollup-7jqydzg, r=Dylan-DPC
commit[1] 2023-05-23: Auto merge of #110204 - compiler-errors:new-solver-hir-typeck-hacks, r=lcnr
commit[2] 2023-05-23: Auto merge of #110519 - loongarch-rs:ci, r=Mark-Simulacrum
commit[3] 2023-05-23: Auto merge of #111807 - erikdesjardins:noalias, r=oli-obk
commit[4] 2023-05-23: Auto merge of #111869 - Dylan-DPC:rollup-9pydw08, r=Dylan-DPC
commit[5] 2023-05-23: Auto merge of #107294 - JamieCunliffe:neon-fp, r=Amanieu
commit[6] 2023-05-23: Auto merge of #111882 - matthiaskrgr:rollup-1xyv5mq, r=matthiaskrgr
commit[7] 2023-05-23: Auto merge of #111867 - RalfJung:miri, r=RalfJung

[compiler-errors]

Oh, for the record, that's due to #111807 it seems. New is_unpin check in codegen...