Hare is a boring programming language November 27, 2022 by Drew DeVault
Hare is a boring programming language. It has relatively few features, exhaustively detailed in a short 67-page language specification. There are no macros, generics, or metaprogramming features. You just write the code and it does what you said to. Hare aims to never surprise you.
Many new systems programming languages are exploring big ideas and novel tools. Zig has comptime, Rust has procedural macros, and so on — but Hare lacks any grand ideas. Our biggest idea is ad-hoc tagged unions, which is not particularly interesting or innovative.
Hare’s boringness is a feature, not a bug. Our goal is not to make programming exciting again, but to make it easy to write simple, obvious programs which are optimized for reliability and longevity. An exciting programming language cannot meet that goal as effectively as Hare does. We have instead sought out the smallest and simplest language design which accommodates these goals. Because we have relatively few features, Hare programs tend to converge upon the single obvious way of solving their problems with the tools at their disposal.
Modules in the standard library, for example, are generally not surprising. If you ask yourself how a given problem might be solved in Hare, an obvious solution will probably present itself to you, and the relevant standard library module will implement that solution. It’s not clever or interesting — it just works, so that you can use it and move on to the next task.
Most Hare programs just need to be written once, and then they can be relied upon to do their job for a long time. We don’t often find ourselves having to revisit the Hare code we write. This differs from a common notion in free software where the activity of a project is used as a proxy for its quality — instead, the ideal Hare project completes its goals and stops, fixing bugs and adding minor improvements in the long tail.
It’s in the service of these goals that we plan to freeze the language once it reaches 1.0 and cease development of new language features. Following 1.0, the only specification changes will be clarifications and minor corrections. We won’t have the perfect language, and we’ll have to live with our oversights, but that’s okay: we’ll let the next language improve on our ideas. What we want is a language that we can rely on for as long as possible, and that requires a deep commitment to stability.
A Hare program written on the day that 1.0 releases will still compile in 50 years on contemporary Hare compilers. In fact, a Hare compiler written on the same release date will compile new Hare programs 50 years from now. That’s something which hasn’t really been done before, and I’m looking forward to seeing it usher in a renewed commitment to stability and long-term reliability in the software ecosystem.
Hare is boring, and that excites me.