https://news.ycombinator.com/item?id=26271117 Redbean (2011) is a single file distributable web server. Here's a video of me giving a talk about it https://youtu.be/1ZTRb-2DZGs
https://news.ycombinator.com/item?id=24256883 αcτµαlly pδrταblε εxεcµταblε (2020) lets you build c/c++/fortran/etc. code so that it runs on seven operating systems
https://news.ycombinator.com/item?id=13769727 Operation Rosehub (2017) where I helped save a lot of people from an Apache security bug similar to Log4j issue today
I like doing these projects because they all fit together so well into a consistent story. For example, a lot of these past projects were what helped to make SectorLISP better. Thanks to APE you now have a 20kb quickly deployable and embeddable LISP interpreter. Thanks to Redbean we're going to have an online pastebin service that'll let you publish SectorLISP gists in a few days, for example: https://lisp.pub/1
Thank you Hacker News community for being so supportive and encouraging.
> αcτµαlly pδrταblε εxεcµταblε
You mean ακτυαλλγ πορταβλε εχεκυταβλε?
but for the question: for your projects, do you have like a stable buildplatform (like nixos, minimal debian stable or sth like that) for which you publish your build instructions for your project?
I like Make since it's able to build a repository with 17k .o files, 80 .a archives, and 661 .com executables from scratch in under a minute on one personal computer (if the kernel page cache is warm). I wrote a couple small helper commands to make the make config more manageable, like package.com, mkdeps.com, compile.com, ar.com, zipobj.com, and runit.com.
The reason why Make works for me, is because I think the root cause of needing things like Autoconf and Cmake is because most projects need to depend on seven different C libraries. I decided that, rather than focusing on writing a better build config, I'd rather use an unfancy build system and instead devote my energy towards having a single C library that runs on all seven of the platforms I'm targeting. I owe a lot of thanks to projects like musl, dlmalloc, dtoa, llvm, etc. from whom I borrowed source code. That enabled me to abstract portability at the libc level, rather than punting to #ifdefs and configs.