Fix division by zero

[raviqqe]

No description provided.

[codspeed-hq]

Merging this PR will not alter performance

✅ 41 untouched benchmarks
⏩ 10 skipped benchmarks¹

---

_{Comparing bug/div-to-nan (0bcd11a) with main (5cc30e0)}

Footnotes

1. 10 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[Copilot]

Pull request overview

This PR updates Float62’s division behavior to avoid undefined/panicking integer operations when dividing by zero, and aligns integer-division-by-zero results with IEEE-754 float semantics (±∞ for nonzero / 0, NaN for 0 / 0).

Changes:

• Prevent modulo-by-zero by guarding the integer fast-path with y != 0 before x % y.
• For integer division by zero, fall back to float division (x as f64 / y as f64) to produce ±∞/NaN instead of a hardcoded NaN.
• Update division-by-zero tests to assert ±∞ and NaN outcomes.

Comments suppressed due to low confidence (1)

src/f62.rs:305

• Float62::from_integer only reliably encodes 63-bit signed integers (see box_integer docs). In the integer fast-path, x / y can produce a quotient outside that range (e.g. x == -(1<<62) and y == -1 yields 1<<62). Boxing that value wraps and decodes to the wrong integer. Consider range-checking the quotient and falling back to float when it is not representable as an integer payload (and add a regression test for this boundary case).

        if y != 0 && x % y == 0 {
            Self::from_integer(x / y)
        } else {
            Self::from_float(x as f64 / y as f64)
        }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.