Racket 9.0 brings parallel threads, black-box wrapper, AArch64 natipkg packages, and more

Racket 9.0 has been released as the latest version of this functional and contractual programming language. With this major update, Racket introduces support for shared-memory threads capable of true parallel execution on multicore hardware. This differs from earlier Racket versions, where threads could only run concurrently, not in parallel.

Alongside the thread overhaul, developers gain access to a new black-box wrapper that prevents the optimizing compiler from eliminating certain computations. This feature helps ensure that benchmarks represent true performance characteristics. Additionally, the addition of the decompile-linklet function allows users to map compiled linklets back to s-expressions, providing deeper introspection when working with Racket internals.

Building on platform support, this release brings “natipkg” packages for AArch64, facilitating package-building and testing on Arm-based systems. When running on BC Racket, the processor-count function now always reports the system’s parallel processor count, providing more accurate resource information. Check Syntax tracking has been improved, allowing it to identify more deeply nested identifiers in syntax objects. The math library has also been enhanced with new support for Weibull distributions. In addition, this release features several repairs and updates to the documentation.