How do humans learn complex rules from small available data? Rule et al introduce metaprogram learning as a model that fits human behaviour in complex learning situations. Unlike previous approaches, ...

I've written about MOTHER's philosophy and its mechanics. This post is about what I realized after I had some time to step back and reflect on the system once it was in place. MOTHER is a metaprogram...

Posted by Jonathan Brossard on Aug 01 Dear list, I hope this email finds you well. I'd like to bring attention to a soundness vulnerability in the Lean 4 theorem prover kernel that allows a malicious metaprogram to cause the kernel to accept invalid proofs, including proofs of false statements such as False and 0 = 1, with no axioms and full kernel checking enabled. --[ Summary The Lean 4 kernel fails to validate that projection expressions in nested inductive type...

When is it appropriate to metaprogram? Today's episode looks at one situation in which it may be a good choice.

This article demonstrates how modern C++ lets you perform computations at compile time using templates, constexpr, and related features. It covers classic template-based metaprogramming (e.g., factorial and Fibonacci calculations), then shows how constexpr functions, loops, classes, and other modern constructs simplify and optimize compile-time calculations. You’ll learn best practices, common pitfalls, and how to balance readability, performance, and debugging complexity when leveraging compile-time computations in your code.