Sunder is a Sufficiently Stable Programming Language

This week I tagged what I consider to be the first stable-ish release of Sunder, my C-like systems programming language and compiler for x86-64 Linux. The Sunder project does not use semantic versioning, and I am not going to make any future compatibility promises, but I can say that as someone who has spent hundreds of hours reading, writing, and thinking about Sunder code, this is the first release of the project where everything feels like it is in the right place. The Sunder issue tracker has no outstanding bugs (at the time of writing). The Sunder standard library contains enough tools to write non-trivial software while leaving plenty of room for future development. And the Sunder compiler seems to be relatively jank-free¹.

It is difficult to convey just how incredible of a journey it has been working on Sunder over this past year and a half. Although I have been in the programming language space for a number of years, Sunder is the first PL project of mine that has felt like a true success. Sunder is not just a good programming language, it is my favorite programming language. It is the language that I wish C and C++ could be. It is a language that is an absolute joy to develop with, and a language that I look forward to using for years to come.

With this stable-ish language milestone achieved, the focus of Sunder will shift away from language and standard library development to compiler cleanup and portability concerns. Currently, Sunder can compile programs for any architecture and operating system as long as it is x86-64 Linux. Focusing on a single architecture and operating system up to this point was definitely the right choice, but it would be nice to see Sunder code running on some of my other devices. Additionally, targeting different architectures and operating systems will help uncover potential language-level bugs and weak standard library abstractions. It would also be nice to explore more sophisticated compiler backends such as QBE or LLVM. The current backend strategy of "just output and assemble un-optimized NASM code lmao" was the right choice for the language up to this point, but it might be a good time to investigate potential alternatives. Sunder is also in desparate need of some documentation. Without any sort of reference guide the only real way to onboard as a Sunder user is to read through the example programs and the standard library sources which is... less than ideal.

On a more personal note, I am going to take a bit of a break from significant work on the Sunder project now that this stable-ish release is finally out. Side-project burnout is a serious concern of mine, and it is kind of a miracle that I was able to work as intensely as I did for a year and a half without suffering severe fatigue. With a project of this scope there is always a problem to solve, always code that could use refactoring, always an idea for a new feature, always something. I feel like my brain has been unable to stop thinking about this project since development began, and I need to spend some time away from the project repository if for no other reason than to shut my brain up for a while. Now is probably a good time to do some development with Sunder rather than on Sunder. Project Euler problems, toy applications, and blog posts featuring example code written in Sunder are all great ways that I can enjoy using the language in a low-stress environment without hyper-fixating on compiler internals.

This release is a big deal for me. My work on Sunder is probably the most effort I have put into a personal project (programming or otherwise) in my life, and it is pretty surreal seeing everything come together so nicely. I would be remiss if I didn't take the opportunity to acknowledge my amazing partner who has put up with way more compiler ramblings than any person should ever have to endure. I would also like to thank my wonderful dog who has been an invaluable companion and source of emotional support over the duration of this project.

If you would like to check out Sunder then you can find the source code and installation instructions on GitHub with a mirror on SourceHut. The language is developed and tested on Debian Stable, but one should be able to get the Sunder toolchain running on any modern x86-64 Linux system with access to a C99 compiler, ld, and either nasm or yasm². Cheers, and happy Sundering!

Footnotes

1. Every compiler (and really every large software project) is going to have some level of jank. There will always be abstractions that could be cleaner, error messages that could be clearer, and architectural changes that would allow for simpler code. It is almost impossible to get everything right on the first try, and it is perfectly acceptable to take on additional unnecessary complexity (i.e. jank) within a project if you believe the benefits of that complexity outweigh the drawbacks within the context of the project as a whole. As a compiler author I know that there is plenty of jank to be fixed within the compiler, but as a user I am generally not exposed to the internal compiler jank. This shielding of jank from the user is what I mean when I say the compiler seems to be relatively jank-free.

2. If not then please send me an email or file an issue on GitHub, it would be much appriciated! 😊