Docs regression on crate teloxide-0.10.1

[notriddle]

Behavior, describe in zulip thread

https://docs.rs/teloxide/0.10.1/teloxide/struct.Bot.html has an inherent impl block, but it is not shown. https://docs.rs/teloxide/0.10.0/teloxide/struct.Bot.html (which was built with an earlier version of rustdoc) shows it.

Regression information

searched nightlies: from nightly-2022-08-03 to nightly-2022-08-04
regressed nightly: nightly-2022-08-04
searched commit range: 4493a0f...1b57946
regressed commit: 7308c22

bisected with cargo-bisect-rustc v0.6.4

Host triple: x86_64-unknown-linux-gnu
Reproduce with:

cargo bisect-rustc --prompt --script=/teloxide/run.sh --start=2022-08-03 --end=2022-08-04 --test-dir=/teloxide/ 

run.sh:

#!/bin/sh
cargo doc

[Mark-Simulacrum]

@notriddle it sounds like the regression here was introduced in #99738, which was backported to 1.63 beta (releasing in <1 week). Is that right?

If so, I assume that we should review + backport #100207 into 1.63 and 1.64, correct?

[Mark-Simulacrum]

On the other hand, it looks like #99738 was a stable/stable regression, so it may make sense to back out the backport of that PR instead -- it makes me a little nervous that we already encountered one regression introduced by it (i.e., this issue); we don't have a lot of time for things to bake and get reviewed.

[inquisitivecrystal]

On the other hand, it looks like #99738 was a stable/stable regression, so it may make sense to back out the backport of that PR instead -- it makes me a little nervous that we already encountered one regression introduced by it (i.e., this issue); we don't have a lot of time for things to bake and get reviewed.

I think the main question here is whether we'd rather have docs fail entirely, or have them build but be incorrect. Personally, I really don't like having either of those on stable, but I'm not sure what to do about it.