Method set_optimize

src/compilation_pipeline.rs:621–623

Enable IR-level optimization passes (constant folding, dead-code elimination). Off by default so IR/assembly goldens stay stable; when on, the passes run on the lowered IR before backend lowering.

(&mut self, opts: OptOptions)

    /// Enable the conservative assembly peephole pass. Off by default, because it
    /// rewrites the token stream and the `.s` text rendered from it.
    pub fn set_peephole(&mut self, enabled: bool) {
        self.peephole = enabled;
    }

    /// Keep hot scalars in callee-saved registers (s2-s11) instead of stack slots.
    /// On by default; off gives the pure stack-slot lowering for shape comparisons.