Taming Type Annotations in Gradual Typing
summary
Paper claims static analysis can find incorrect annotations that later cause runtime cast errors by combining type inference with variational types and a lifted compatibility relation, producing concrete annotation conflicts (e.g., suggesting change from int to string). Implements the approach as Pyhound using exploratory typing and constraint solving, handles dynamic features and conditionals, and evaluates by seeding cast errors into Python programs: the tool locates the injected errors and suggests fixes ahead of time.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:30Dynamic Typing Basics def append_underscore(s)
- 1:33Getting Gradual
- 2:29Gradual Runtime Errors
- 4:03Issues With Wrong Annotations and Runtime Errors Three issues
- 5:36Goals
- 6:26How to Discover and Fix Errors
- 7:44Initial Approach
- 8:37Exploratory Typing
- 9:32Variational Inference Finish
- 10:11Solving Equations
- 12:02Pack Results Back Into Variation
- 13:00The Full Paper
- 14:22Contact Me
