bump MSRV to 1.83

[carljm]

According to our new MSRV policy (see #16370 ), bump our MSRV to 1.83 (N - 2), and autofix some new clippy lints.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

Formatter (stable)

✅ ecosystem check detected no format changes.

Formatter (preview)

✅ ecosystem check detected no format changes.

[MichaReiser]

This is technically a breaking change for anyone building Ruff from source because they now need a newer Rust version to do so.

I have a slight preference to keep the current MSRV unless there's a lot of boilerplate involved.

The other thing we have to consider when bumping the MSRV is whether downstream packaging tools have updated to this Rust version (e.g. homebrew). This should be fine, considering that uv already uses 1.83

[carljm]

Ok I'll try the boilerplate tomorrow. Just annoying writing uglier code that shouldn't need to be that way to accommodate older rust versions :/ doesn't seem like it should be hard for anyone to use 1.83 instead of 1.80, and I'm sure we'll bump it up sometime...

[MichaReiser]

I'd say it depends on how much more boilerplate it requires. If it's something you can do in 15min, let's keep the MSRV as is. If not, bump it (although a lower MSRV might also be important for other projects using salsa)

[carljm]

If it's something you can do in 15min

I wasn't able to find something that compiles within 15min, but maybe you know how to do it.

The problem is at https://github.com/salsa-rs/salsa/blob/1ff42613cbaa699567009895a6d00b87bf7772e6/src/cycle.rs#L97 -- hashset iterators only starting implementing Default in Rust 1.83.

In the owning IntoIterator impl above I can just replace .unwrap_or_default() with .unwrap_or_else(|| FxHashSet::default().into_iter()). But in the borrowed version I can't figure out what I can put in the .unwrap_or_else() without getting errors about borrowing from a local, or returning the wrong type. Even if I remove the Copied and let it be just an iterator over &DatabaseKeyIndex, I still get errors about borrowing from a local if I try to use FxHashSet::default().iter().

[MichaReiser]

Yeah, I think you'd have to write your own iterator. Which is definitely less nice but it's not a show-stopper:

impl std::iter::IntoIterator for CycleHeads {
    type Item = DatabaseKeyIndex;
    type IntoIter = std::collections::hash_set::IntoIter<Self::Item>;

    fn into_iter(self) -> Self::IntoIter {
        self.0
            .map(|heads| *heads)
            .unwrap_or_else(|| FxHashSet::default())
            .into_iter()
    }
}

pub struct CycleHeadsIter<'a>(Option<std::collections::hash_set::Iter<'a, DatabaseKeyIndex>>);

impl<'a> Iterator for CycleHeadsIter<'a> {
    type Item = DatabaseKeyIndex;

    fn next(&mut self) -> Option<Self::Item> {
        self.0.as_mut()?.next().copied()
    }

    fn last(self) -> Option<Self::Item> {
        self.0?.last().copied()
    }
}

impl FusedIterator for CycleHeadsIter<'_> {}

impl<'a> std::iter::IntoIterator for &'a CycleHeads {
    type Item = DatabaseKeyIndex;
    type IntoIter = CycleHeadsIter<'a>;

    fn into_iter(self) -> Self::IntoIter {
        CycleHeadsIter(self.0.as_ref().map(|heads| heads.iter()))
    }
}

[carljm]

Thanks! Added this to the Salsa PR to maintain 1.80 compat, will close this for now.