Reproduce “Classifying Capabilities (Extended Version)” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Cao Nguyen Pham, Oliver Bračevac, Yichen Xu, Yaoyu Zhao, Martin Odersky

Researchers at EPFL extended Scala 3's capture checking with capability classifiers, enabling types to filter computational effects based on their semantic kind rather than individual identity. This enhancement allows for precise type signatures using inclusion (.only[C]) and exclusion (.except[C]) operators, addressing prior limitations in typing standard library constructs and increasing capture-checked standard library coverage from 66.7% to 81.7%.

Key insights

  • A tree-structured, user-extensible classifier hierarchy with a decidable kind algebra ensures modularity and extensibility in an open programming ecosystem without compromising soundness.
  • Utilizing a checked big-step semantics for the formal system enables rigorous metatheoretical guarantees, including type safety, effect safety, and precise prediction of used and captured capabilities.
  • Fine-grained classification of capabilities by their semantic kind, rather than just identity, provides expressive power for specifying complex effect policies and improving API precision.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.