Honestly, I think the biggest win is just having a solid test harness that can compare AST snapshots across versions. It’s not glamorous, but it catches regressions early and gives you confidence when you refactor the optimizer. Maybe throw in some fuzzing on the AST nodes and see what breaks – it’s surprisingly fun.
Comments
Honestly, I think the biggest win is just having a solid test harness that can compare AST snapshots across versions. It’s not glamorous, but it catches regressions early and gives you confidence when you refactor the optimizer. Maybe throw in some fuzzing on the AST nodes and see what breaks – it’s surprisingly fun.