[yashaggarwal85d]

Fixes #164999

When a const arg is just a path like Self::N and the gca features are on, the error now suggests wrapping it in gca! instead of telling you to make a type const item:

help: consider wrapping `Self::N` in the `gca!` macro
  |
8 -     fn foo() -> [u8; Self::N];
8 +     fn foo() -> [u8; gca!(Self::N)];

It only fires when both gca_min_const_items and gca_const_items are enabled (with just the min feature the wrapped form still errors), and it skips const { .. } blocks, non-path expressions, macro spans, and inherent assoc consts, since gca! doesn't help there yet. Those keep the old help.

[rustbot]

HIR ty lowering was modified

cc @fmease

[rustbot]

r? @JonathanBrouwer

rustbot has assigned @JonathanBrouwer.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 77 candidates
• Random selection from 20 candidates

[yashaggarwal85d]

FYI, used LLM for writing tests.

[JonathanBrouwer]

r? @khyperia
I'm not sure how useful this suggestion is given that the gca! syntax is probably temporary

[estebank]

@JonathanBrouwer that is true, but if there will be any syntax, a suggestion will be needed and the filtering logic will likely be the same, just with different output. I find that spending a little bit of effort to improde diagnostics for nightly features helps set the tone for their maintainers to continue improving them and makes the stabilization rollout smoother because there needs to be less of a mad scramble to improve diagnostics right before or right after.

[JonathanBrouwer]

That makes sense, thanks for sharing!
I'll still defer to Ashley as she can better judge of when exactly we should suggest gca!