[binary132]

One thing that the latest generation of languages has taught me (or that I have learned in the process?) is that “languages” aren’t “really a real thing” in and of themselves as much as they are merely composable APIs over compilers / interpreters (which are basically just dynamic compilers, or compilers are interpreters?), and compilers are an insane, fragmented dumpster fire with no or barely usable APIs. Only LLVM, GCC, TinyCC, and Terra admit this reality, neither LLVM nor GCC’s APIs are really user-oriented, and Terra isn’t really a realistic option for most people. I also personally don’t really feel that LLVM has accomplished (or can accomplish) its potential for a variety of reasons — most LLVM projects are forks or patches, although Zig has done well at this — and GCC has limitations as well although I’m very pleased with gccjit in Emacs. I’m very curious to see what the future holds especially if we can get an optionally-verifiable yet usable compiler API.