Symbolic Modeling and Transformation of Java Code
summary
Project Babylon represents Java programs as an in-memory code model: a tree of operation, body, and block nodes in SSA form with block parameters, explicit conversions, and reflective ops (field refs, method refs and invoke ops) so normal Java code maps directly to symbolic models. The implementation places ~99% of code in an incubating module, integrates with Onyx and a heterogeneous accelerator toolkit (Panama-backed), and ships a Gary Frost viewer that visualizes models and SSA graphs.
AI-written summary · about
chapters (5)
- 0:00Introduction and status
- 2:36Project goals and metadata
- 3:48Code model structure
- 7:00Modeling Java constructs
- 25:02Tooling and conclusion
related talks
53mProject Babylon - Code Reflection
43mReflecting on HAT: A Project Babylon Case Study
42mCode Reflection
35mJava Source to Byte Code Transformation - The Hidden Story!
40mCode Reflection in Action - Translating Java to SPIR-V
45mMaking Java more dynamic
42mEternal Sunshine of the Spotless Macro
44mJava for AI