This will be useful for papers / thesis.
Use ScalaMeter to create the infrastructure for micro benchmarks of compiler transformations. This will also give us some simple statistics and reporting for free. Benchmarks can then be run on a bunch of Scala code like scala-repos. The tricky part is making sure the transformations are lenient enough to not fail on random input.
After this we can define a few variants of the core transformations like ANF and DSCF to compare: using the Transversers API, recursive methods, stateful transformers from scalac, quasiquotes vs the AST API, etc.