OSDev.org

Hi,

Kevin wrote:Actually, drivers aren't what I'm worried about. If I have some basic support for keyboard, mouse, storage, network and sound (which I do on one laptop at least), that's all that I really need from the hardware before I could call it somewhat usable. Having an actually useful userspace is the hard part, especially if you don't want to be fully POSIX. vim/bash/gcc/make is nice, but actually useful is a broswer, mail client, etc.

I agree.

For device drivers, the most important thing is the interfaces and functionality that drivers depend on. Beyond that you can get by with some generic drivers (AHCI, EHCI, "raw framebuffer", HD audio, etc) and hope that eventually one day volunteers write drivers for the OS.

For "user-space without porting stuff"; I'm expecting to spend a decade on developer tools (IDE, compiler, etc) before anyone is able to write a simple "Hello world" application.


Cheers,

Brendan

It's my focus to make Sortix something that can actually be used, something one actually wants to use as a development platform.

Compared to some hobby operating systems out there with advanced GUI, it is possible to interactively install my OS, there is a lot of documentation, and such installations come with the full toolchain and are self hosting. That goes a long way towards becoming a primary OS.

My recent work has been in porting editors (emacs, vim, nano, ed), networking (wget, libressl, openssl), even a port of qemu and networked package management. I have also done a small GUI with simple terminals and games. I can do with less GUI features for now, but I want a solid system underneath. It's really cool to be able to ssh and scp out. Native git pushed actually work. I just need the ability to build all ports nativity and I'll be truly self hosting. If I can get a https working, I could even host.

One can do a lot with a little. USB, sound, real graphics drivers, SMP, etc aren't that super important but just nice. I do want those in the future.

But still, porting major GUI applications like Firefox and VLC, it is really its own harder category. Once I implement a back end for GTK and other toolkits, they make start becoming doable.

So how hard it is to become your own primary OS really depends on your needs. Browsers are probably the hardest part of it since they build on everything else. Still one may be able to do with terminal browsers or netsurf, but it is not exactly lovely for the web as it is.

And then there is the category of OSes that are probably technically self-hosting, but actually doing that is kind of impractical. I did compile a kernel on it, I did install GRUB2 from it and I even wrote some driver code for real hardware natively on it because it was easier than rebooting all the time, but there is no way I would use tyndur for regular development. There is no real build system, so it would be a lot of manual work, the editor sucks for real work and 80x25 text mode isn't optimal either, and of course ported programs are usually almost impossible to build without a POSIX shell.

So while I'm positive that in theory I could build and install at least the core system natively, I would never actually try to do this at this point.