Authorization is a fundamental problem in modern distributed systems, and the ''policies-as-code'' paradigm has emerged as a promising solution to decouple access control logic from application code. However, most policy languages lack the ability to handle stateful policies directly. This limitation forces developers to manage policy-related state within the application code, reintroducing the very coupling that policies as code aims to eliminate and opening the door to security vulnerabilities. To address this gap, we introduce Strobilus, a language designed to express effects over policy-specific data. Strobilus is built to seamlessly complement and integrate with Amazon's Cedar, allowing developers to write stateful policies without modifying Cedar's core syntax or evaluation engine. Strobilus is distinguished by its formal semantics, a strong typing system, and a guarantee of termination, which facilitates rigorous analysis and verification of policies. We have developed a prototype implementation in Rust, which demonstrates that Strobilus may lead to significant performance improvements over external methods for policy data management. This approach aims to fully realizes the promise of ''policies-as-code'' by providing a comprehensive, safe, and verifiable solution for both stateless and stateful authorization policies.

Layout analysis is a task of uttermost importance in ancient handwritten document analysis and represents a fundamental step toward the simplification of subsequent tasks such as optical character recognition and automatic transcription. However, many of the approaches adopted to solve this problem rely on a fully supervised learning paradigm. While these systems achieve very good performance on this task, the drawback is that pixel-precise text labeling of the entire training set is a very time-consuming process, which makes this type of information rarely available in a real-world scenario. In the present paper, we address this problem by proposing an efficient few-shot learning framework that achieves performances comparable to current state-of-the-art fully supervised methods on the publicly available DIVA-HisDB dataset.