Consider suggesting using `HashMap<String, String>` instead of `HashSet<String, String>` when `PartialEq` derive macro is used

[zlepper]

Code

use std::collections::HashSet;

#[derive(PartialEq)]
pub struct MyStruct {
    pub parameters: HashSet<String, String>,
}

Current output

Compiling playground v0.0.1 (/playground)
error[E0369]: binary operation `==` cannot be applied to type `HashSet<String, String>`
   --> src/lib.rs:5:5
    |
  3 | #[derive(PartialEq)]
    |          --------- in this derive macro expansion
  4 | pub struct MyStruct {
  5 |     pub parameters: HashSet<String, String>,
    |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    |
note: the foreign item type `String` doesn't implement `BuildHasher`
   --> /playground/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/alloc/src/string.rs:360:1
    |
360 | pub struct String {
    | ^^^^^^^^^^^^^^^^^ not implement `BuildHasher`

For more information about this error, try `rustc --explain E0369`.
error: could not compile `playground` (lib) due to 1 previous error

Desired output

Compiling playground v0.0.1 (/playground)
error[E0369]: binary operation `==` cannot be applied to type `HashSet<String, String>`
   --> src/lib.rs:5:5
    |
  3 | #[derive(PartialEq)]
    |          --------- in this derive macro expansion
  4 | pub struct MyStruct {
  5 |     pub parameters: HashSet<String, String>,
    |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    |
note: the foreign item type `String` doesn't implement `BuildHasher`
   --> /playground/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/alloc/src/string.rs:360:1
    |
360 | pub struct String {
    | ^^^^^^^^^^^^^^^^^ not implement `BuildHasher`

Suggestion: Did you perhaps intend to use `HashMap<String, String>` instead?

For more information about this error, try `rustc --explain E0369`.
error: could not compile `playground` (lib) due to 1 previous error

Rationale and extra context

I have now messed this up several times. The first time it keep me about an hour to notice that I had used the wrong hash collection. The third time this morning it thankfully only took me 5 minutes.

The current compiler error is absolutely correct, and i'm the fool who can't properly articulate what I thought i was doing, and what I was actually doing.

When I expect to have typed HashMap, and i'm like 99.99% sure that strings as supported as keys in HashMaps, then it absolutely confuses by brain enough that I don't catch the difference between HashSet<String, T> and HashMap<String, T>, considering there are only 3 letters difference in the middle.

A way of catching this issue and emitting diagnostics could be look for HashSet declarations that specifies both T and S, where T doesn't implement BuildHasher.

Other cases

Rust Version

rustc 1.86.0 (05f9846f8 2025-03-31)
binary: rustc
commit-hash: 05f9846f893b09a1be1fc8560e33fc3c815cfecb
commit-date: 2025-03-31
host: x86_64-unknown-linux-gnu
release: 1.86.0
LLVM version: 19.1.7

Anything else?

Playground link: https://play.rust-lang.org/?version=stable&mode=debug&edition=2024&gist=4aa004ed6a442e7b902c1d30d5800aa0 (1.90 as of the making of this issue)

[hanna-kruppe]

More generally, any diagnostic that involves “second type parameter to HashSet doesn’t implement BuildHasher” might benefit from pointing out the potential HashSet/HashMap typo (at least if the type does implement Hash+Eq).

[Qelxiros]

@rustbot claim