regression: abs_diff conflict leading to compilation failures

[Mark-Simulacrum]

The abs_diff API was added to integer types in #88780, which is leading to compilation failures in:

• https://crater-reports.s3.amazonaws.com/beta-1.57-1/beta-2021-10-22/reg/genevo-0.5.0/log.txt
• https://crater-reports.s3.amazonaws.com/beta-1.57-1/beta-2021-10-22/gh/rodya-mirov.cadmium-ostrich/log.txt

Since it's just a couple cases I suspect we can just accept this regression, but wanted to surface it to T-libs-api. (Maybe a rename or something is in order).

[nbdd0121]

Similar things have caused breakage a few times. I'll investigate.

@rustbot claim

[nbdd0121]

Okay, found the issue. The issue is that we try "by_value" before "autoref", and then within each we try "stable" before "unstable", so "unstable by_value" takes priority over "stable autoref". The fix should simply be try all stable candidates first and then try unstable ones. Working on a fix now.

[m-ou-se]

Discussed in the meeting. This is expected and acceptable. Improvements to name resolution to avoid issues like this in the future would be nice though!