fix: LcVarsIterMut::size_hint off-by-one and potential underflow

[Galoretka]

Use Windows::len() directly as the exact number of remaining windows instead of subtracting 1.
The previous implementation self.offsets.len() - 1 underestimated the remaining items and could underflow when the iterator was empty (panic in debug, wrap in release).
This aligns size_hint with the semantics of core::slice::Windows, ensuring accurate iteration hints and avoiding edge-case failures.

[Copilot]

Pull Request Overview

Fixes an off-by-one error and potential underflow in the LcVarsIterMut::size_hint method by using the correct window count calculation.

Key Changes

• Changed self.offsets.len() - 1 to self.offsets.len() in the size_hint calculation
• Prevents potential underflow when the iterator is empty
• Aligns with core::slice::Windows semantics for accurate iteration hints

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

This change may introduce an off-by-one error in the opposite direction. If LcVarsIterMut is iterating over windows of consecutive elements (similar to slice::windows()), then the number of windows should be offsets.len() - window_size + 1, not offsets.len(). Please verify that self.offsets.len() actually represents the correct count of remaining items to iterate over, rather than the total number of offset positions.

[Pratyush]

hm could you please check this case as well? Thank you!

[Galoretka]

hm could you please check this case as well? Thank you!

In LcVarsIterMut the field named offsets is actually a Windows<'_, usize> iterator (from offsets.windows(2)), not the raw offsets slice. For Windows, .len() already equals “how many windows are left.” So size_hint should use (len, Some(len)) as-is.

• If we subtracted 1 here, we’d undercount and also break the empty case; the new test covers that and would fail.
• In the parallel path we do use the raw slice, so we correctly use offsets.len().saturating_sub(1). Both paths are consistent.

So the current size_hint is correct, and the test is right.

[Pratyush]

Can you add a test which demonstrates this is an issue? self.offsets.len() is always 1 at least.

[Galoretka]

Can you add a test which demonstrates this is an issue? self.offsets.len() is always 1 at least.

Yes, done

[Pratyush]

Thank you!