Operating system development

Post Reply New Topic

I want to develop some kind of operating system for myself one day, I don't know how insane that is but I want to eventually code all the tools myself and have complete control. I understand not a lot of people do this. I am able to learn a lot of things like assembly code and CPU features. Does anybody else have experience as a hobby developer? What is it like?

Linux is good place to start, modify a number of things, play, don't stop, play

Starting from scratch is possible, but it leaves open a lot of problems. Most importantly, if you don't have others involved as well, you are unlikely to get anything that is highly useful.

Putting your effort into an existing open source operating system would be much more useful. There is not only Linux, but there are also some like NetBSD, FreeBSD, and OpenBSD. One that I'm tempted to play with a bit is OpenVMS which was derived from the old VAX/VMS operating system.

I'd say that being on the spectrum makes it more likely that you can do a lot as a single developer, based on my personal experience with large projects.

I've been gardening for over twenty years and now have hundreds of flowering shrubs in my yard!
I saw a big black bumblebee today as it was unseasonably warm in Connecticut.

The great thing about doing something as a hobby is that it doesn't have to be practical or "make sense."

Here are some links for you:

osdev.org
^^ Seems like a lot of information ^^

linux from scratch
^^ you can learn a lot about how a Linux based system works here ^^
(You could also try running Gentoo for a little while if you haven't already.)

redox-os
^^ this started as a project by a single developer, it really has a lot of potential ^^

plan 9 from bell labs
^^ there is a LOT of information in the docs section of this site ^^

As was already mentioned, Linux and the *BSD's . Their manpages, their device driver and kernel module programming guides.

I have the unpopular opinion that you should not start a new Operating System. Look how long it took for Linux to even get the 3 or 4% share it has of the desktop market today: 30+ years. Do you think your project has 30+ years of momentum available to it? If not, maybe make a smaller contribution to some other pre-existing project which is on the verge of making a difference in your world. Can you help that project cross the tipping point to making a difference?

If you don't have 30+ years to devote to this idea, but still insist on a new OS in 2025, then maybe aim lower: for example, start with constrained hardware - say, an RP2350, and ask yourself if you can create an homage to some tiny, defunct OS with a simple GUI (like NextStep OS; bringing back WindowMaker, or maybe PalmOS or GeOS) on it.

Another opinion: C is a great OS-level language; no need to learn assembler. I coded in both of these back in my uni days. Liked C, hated assembler.

If you want something more modern than C, I hear Zig is a good "succesor" to C. Is still very fast, allows low-level control.

I'm not that advanced in computer programming.

But one thing I do know is that with enough time and drive, people can do big undertakings such as this.

I have many projects in the same vein, as I have a lot of time, the drive slowly faded. But I think I learned some valuable skills from each of those projects.

Good luck if you're still considering making an OS from scratch!

It would totally make my day if Linus Torvalds got diagnosed with Autism, joined this forum, then gave his opinion here.

""In the old days, you would chastise people for reinventing the wheel. Now we beg, ‘Oh, please, please reinvent the wheel." - Alan Kay (Computer Scientist, Pioneer of Object-Oriented Programming)

All comments here to this post hold valid hints and responses, for me it' more about picking a 'reasonably direction' and just go, most of the time it is not the direction you go in wich will define where you end up but how you aproach the subject matter.
3 directions I see today for instance would be:
-Machine-Oriented Programming
-Web-Oriented Programming
-Network-Oriented Programming

best regards,
Kada

*1* Machine-Oriented Programming
Focus: Low-level programming, systems, embedded, and performance-critical code.

-MIT OpenCourseWare – Computer Systems
Deep dive into C, assembly, and operating systems.
-CS50 by Harvard (edX)
Covers C, memory, and systems thinking—great for beginners.
-The Odin Project – Foundations
Includes a section on command line, Git, and basic C concepts.
-Learn C the Hard Way
A practical, project-based approach to C programming.

*2* Web-Oriented Programming
Focus: Frontend, backend, full-stack development.

-freeCodeCamp
Offers full certifications in responsive web design, JavaScript, APIs, and more.
-The Odin Project
Full-stack curriculum with HTML, CSS, JavaScript, Node.js, and React.
-Frontend Masters
Advanced courses on modern frontend frameworks and tooling.
-Codecademy Web Development Path
Interactive and beginner-friendly.

*3* Network-Oriented Programming
Focus: Networking protocols, socket programming, cybersecurity.

-Stanford CS144: Computer Networking
A rigorous course on TCP/IP, HTTP, and building a network stack.
-Beej’s Guide to Network Programming
A classic, beginner-friendly guide to socket programming in C.
-Coursera – Computer Communications by University of Colorado
Covers the OSI model, protocols, and real-world networking.
-Cisco Networking Academy
Industry-standard training in networking and security.

^ Nice recommendations.

BTW: a priceless post on Mastodon, exactly along the lines of the OP:
https://fox.nexus/@theresnotime/114619566141646769