It still seems odd to me that there are many different NaN values in floating point arithmetic. Those bits could have improved precision!
miniblog.
Related Posts
Applying an SMT solver to find missed optimisation opportunities in LLVM's arithmetic tooling: https://blog.regehr.org/archives/1714
Prettier 1.13 has had a ton of polish, but this feature really caught my eye: it explicitly adds parentheses to arithmetic operations, to avoid confusion regarding precedence!
Wuffs: a programming language that prevents null pointer defence, buffer overflow, and arithmetic overflow, all at compile time! Designed for file format parsing.