Path Sensitive Type Inference for REST APIs

May 8, 2023
11:00am EST
Cummings 402
Speaker: Mark Aldrich - Quals Talk
Host: Jeffrey Foster

Abstract

Quals talk:

REST APIs provide a simple interface for client/server communication using HTTP requests and JSON. As single endpoints often express multiple type signatures depending on their inputs, generating accurate documentation has become a key issue for both clients and maintainers. We propose path sensitive type inference to statically and precisely infer the dynamic behavior of API endpoints. Drawing inspiration from symbolic execution, we reason through different paths throughout the program, but encode them as types. Running a program through a path sensitive type inference engine then yields a rich type suitable for API documentation.