Latte: Lightweight Aliasing Tracking for Java

Explore a 33-minute conference talk on Latte, a lightweight aliasing tracking system for Java. Delve into the innovative approach proposed by Conrad Zimmerman, Catarina Gamboa, Alcides Fonseca, and Jonathan Aldrich to minimize annotations and simplify invariants for reasoning about aliasing in object-oriented languages with mutation. Discover how Latte requires annotations only for parameters and fields while inferring them for local variables, and learn about its relaxed uniqueness concept that allows precise aliasing among local variables. Understand the benefits of this design, including support for destructive reads without altering language or run-time semantics. Examine practical applications of Latte in tracking uniqueness and aliasing in local sequential settings, such as modeling a stack, and gain insights into how this system compares to existing aliasing and uniqueness tracking systems in terms of expressiveness and developer effort.

[HATRA23] Latte: Lightweight Aliasing Tracking for Java