I just read Logic-flow analysis of higher-order programs by Matt Might.
Great paper, and had a lot of the ideas that were spinning around in my mind. Matt Might has done a ton of work in so many areas of control flow analysis that it boggles my mind.
The related work section at the end includes many references that I'd like to read to understand more in this area. However, I feel like egglog supersedes this work, with a lot of the same ideas but in a better framework for extending and experimenting with the ideas.
Additionally I read the Entangled abstract domains for higher order programs paper.
The last paragraph maybe introduces some further research directions to look. Though again, I think I'd rather look at egglog and understand more about z3 for now. However, it does claim to be a direct consequence from failings of logic flow analysis, so if I were to try to implement one of them, I should probably do this one.
Weekend Plans:
- Work on full CFA for delimited continuations, or implement the double CPS transform for the interpreter in Racket, Dart or Koka
- And either
- Work on Koka framework for functional reactive programming
- Work on Koka libuv support
- Work on Koka parser
- Work on Koka vector library
- Think about Demand CFA for Algebraic Effects
Papers to read next week:
- One from defunctionalization related to analysis
- Control flow analysis for delimited continuations
- One Chapter from Compiling with Continuations
- One paper from the egglog related work section
- One paper from the logic flow analysis related work section