I'm legit thinking about creating a new desktop operating system.
Rough plan: 1. Find some cheap target hardware, likely ARM-based - something like the Raspberry Pi 2. Take an existing kernel (Linux) to simplify the task. Write a basic graphics system on top from scratch. Port a browser engine to it. Make it an ugly hacked-together monolothic mess with all the windowing, UI and browser shit hard-coded and hacked together to get it working and focus on getting shit working and get it working fast. Polish the shit out of it (I have some design ideas). 4. Make it a useful prototype and hope it picks up some traction. 5. Long-term goal is to use this prototype as a basis to build a proper fully-fledged standalone desktop OS but that will do nothing to impact short-term decisions
Does this seem feasible? I have a decent amount of systems programming background.
I have the computing experience of 25 years ago in mind, when UIs used to be simple and work well despite being incredibly low-specced by modern stadards.
My big question: Are low-powered computers fast enough to have a decent browsing experience on 2018's websites if you hack together and optimise shit? Cheap phones can do it, right, and Android isn't that well optimised. Imagine a browser on a shit smartphone except outputting 1080p on a monitor.
Your idea is shit and you're not even half as smart as you think you are.
Gavin Clark
>4. Make it a useful prototype and hope it picks up some traction.
it won't.
what's your USP?
you can already run raspian with a 1080p monitor. possibly more.
plus, hacker toy computers aren't competing with consumer microcomputers (chromebit / intel compute stick).
you think you can hold a candle to windows and chromeos?
Kayden Jones
just download rio or something
Logan Jones
I have some design ideas that have been bouncing around my head for many years about OSes should be like if they weren't so stuck in the past (whether unix or 90s guis). I'm willing to try to give it a go and build it so see if anyone is interested. ChromeOS is a good idea but way too bland and conservative in its approach to excite anyone
You're right that I should do more investigation on target hardware. Would need to be ARM as I don't want to be dealing with x86/x64 asm.
Nicholas Scott
I don't know what you mean by "design" but if it's just visual, you might as well just make a skin/shell instead of an os.
Robert Cruz
What would be the point of that? The world doesn't need another shitty OS, linux is already a great OS. Make a desktop enviroment if your unhappy with what linux currently offers.
Austin Brooks
Thinking about it, graphical shell is the term I've been looking for what I want to make. It's rarely used anymore.
But I deliberately don't want compatibility with other unix/Linux stuff.
Elijah Rogers
>But I deliberately don't want compatibility with other unix/Linux stuff.
you're not steve jobs
Ryder Taylor
First we make the logo.
Nathan White
You mean the guy who literally created a Unix system?
I don't like Unix/Linux stuff. I don't think it's very well designed, I think it's existence hurts the creation of better things.
Hunter Robinson
>But I deliberately don't want compatibility with other unix/Linux stuff. Can you explain why your idea for a new OS design paradigm cannot be built on top of UNIX/Linux? Is there some sort of fundamental incompatibility?
If not - If you really wanna buckle down and do it, you might eventually end up creating some sort of name for yourself... similar to Terry
Ryder Richardson
>Can you explain why Because he is relatively new to software creation, just discovered Linux recently, and doesn't realize what you can do with it. We all went through that phase.
Leo Reed
But what is step 3
Wyatt Fisher
>You mean the guy who literally created a Unix system?
no I just don't think you have the resources and the marketing department to be able to afford to build something that is virtually incompatible with the rest of current technology.
It could be, but if you're trying to make an interface that works in a fundsmentally different way. having people use their existing stuff alongside it fucks it up so you'd be better off not having it.
Think of it like the jump from Windows Mobile to Windows Phone - it changed too much to make compatibility sensible.
I'm literally suggesting building an OS based around a web browser. It will work with stuff on the web, won't work with existing native apps. There's like a zillion options if you just want a Linux shell of some description.
Adrian Morris
Are you still here OP? Make an an IRC channel on Rizon.
Dylan Long
>I'm literally suggesting building an OS based around a web browser. It will work with stuff on the web, won't work with existing native apps. There's like a zillion options if you just want a Linux shell of some description.
nigger
just take tinycore linux, install the modules you need, and build a shell on that, and call that "your" "os"
maybe build "your own browser" with a Qt webview.
there is no sense in writing all the hardware drivers. it's infeasible to maintain all your software to keep up with the internet standards.
Nathaniel Hill
I'm in the process of making something similar. I'm not interested in sharing my spaghetti code right now, but maybe in a few months I'll host it somewhere if I decide on proper licensing.
So allow me to offer my two cents here. Your first mistake is trying to target soymale hardware like a Raspberry Pi. ARM is very much a vendor locked platform. Until we get RISC-V or some good Chink MIPS-64 stuff again, I would recommend that you not invest too heavily in RISC. Instead, just go with x86-64 as your target. There may be a lot of ARM hardware out there, but most of it is far too locked down. x86 is the only realistic option at the moment, so suck it up and deal with it.
Your second mistake is wanting to base the kernel on Linux. Just don't. Linux is a giant ball of bullshit that has been slapped together over decades. For the love of all that is holy, do not fork that big ball of monolithic shit again. You should start working on a microkernel based OS. The monolithic kernel runs one single program that manages a bunch of stuff. With a microkernel, you break down the functions of the kernel into smaller processes. They act as tiny kernels, and the benefits are numerous. If the microkernel handling drivers were to shit the bed, it won't crash everything else. You can have this trigger another tiny kernel to come along and restart it. So you'd have to have nearly every kernel fail to crash the system, or break the hardware. You can also separate the kernels for security. So some of these could be offloaded into userspace as opposed to running in kernelspace. This is good for untrusted software like non-free driver binaries you want to run. I'm taking the hybrid kernel approach where there's a larger kernel that does a lot of neat stuff at boot time and manages the microkernels that will do other stuff.
And as for the UI, you should probably just fork and improve the Trinity desktop environment. Trinity is a KDE3 fork that works great on toasters.
Don't support all the retardedly complicated stuff (unicode, etc.)
Do an old school OS with your own networking protocols. Then develop a community of hobbyists running those networks.
Like good old BBSes.
Chase Gray
>I have a decent amount of systems programming background. lmao im calling bullshit, you sound clueless
Jaxon Nguyen
>Do an old school OS with your own networking protocols. Then develop a community of hobbyists running those networks. there is nobody on Sup Forums intelligent enough to formulate their own network protocols that could ever be usable
Joseph Gutierrez
This is the best template github.com/Harvey-OS/harvey Its file system is unbeatable, easily expanded to a networking aware file system.
show me a network protocol that you have used and integrated into an application that anyone other than you uses
here's your (you)
Tyler Morgan
>Take an existing kernel (Linux) You mean to make a desktop environment then.
Adam Johnson
have written***
Elijah Cook
>2. Take an existing kernel (Linux) to simplify the task. Write a basic graphics system on top from scratch. Port a browser engine to it. what /are/ you doing? it just sounds like you're competing with wayland/mir, not any OS
also, taking an existing kernel to "simplify" the task is an overstatement, like taking a car with no seats as a starting point for your new car
Isaiah Mitchell
you have ~literally~ re-conceived of ChomeOS whooptie doo
Juan Lee
not gonna reveal my identity for a short ego fap on Sup Forums.
I'm just gonna say that it's a special tcp based protocol that gets around the firewall issue through recursive holepunching to enable unsolicited connections.
Mason Roberts
>taking a car with no seats as a starting point for your new car This is exactly what OP is doing. We can end the thread now.
>tcp based protocol that gets around the firewall issue through recursive holepunching to enable unsolicited connections. so it's a protocol based on an already existing protocol? you're not making the sale user
William Powell
All wrong. You make a gun for bullets. Linus had made Linux/Freax for GNU programs. So, you should make 1. Compatible with Windows NT or Linux drivers 2. Compatible with Windows binaries (implement WinAPI) or Linux binaries 3. Open source under GPL. Other licenses are cancer and shouldn't be used for kernel. Bonus points for super fast x86-64 emulation on ARM64.
It is doable, ReactOS had successfully made a clone single-thread Windows NT 5.2 .
Carter Diaz
shouldn't sell things to retards anyways.
I know you like talking like a big boy but maybe you should understand some fundamentals first. College might be a good start.
Isaac Bennett
user, just stop posting, you seem really upset
Bentley Wilson
>that gets around the firewall issue what the fuck are you even talking about?
Holy shit you're retarded. You have a decent amount of experience in deluding yourself into believing you have any skill whatsoever. Please, do this, and remember my words here: You can't. You just can't. You wouldn't know properly where to even start. I am 100% certain that even if you took the JavaScript approach and just cobbled a monstrosity together with already working modules you'd still get absolutely nowhere.
Remember my words. Please, remember them when you give up. I want you to understand how completely incompetent you are. The only way to do this is to fail miserably and reflect a little. You need this because you sorely lack perspective. So fail, and understand: Building a functional OS with even 1990s level features is a gargantuan task. It is utterly beyond you.
Joshua Martinez
Operating systems are not rocket science. There's plenty of guides. Terry could do it. Fucking ponyfags could do it. The only reason all those small OSes aren't used is lack of drivers and software.
Daniel Hill
>also, taking an existing kernel to "simplify" the task is an overstatement, like taking a car with no seats as a starting point for your new car
That's literally how they build a lot of new cars. Strip them down to its core then rebuild then piece by piece. Lot easier than starting from scratch. Hyundai is especially known to do this starting with Kia cars. A car is more than its engine and gearbox
Parker Edwards
>Strip them down to its core are you suggesting everything but the seats is the 'core' of the car?
Gavin Carter
No, but that's a shit analogy that implies a kernel is what makes an OS and everything above is "seats" Android, Linux desktop, routers, TiVo, PS4 etc are all different OSes by any reasonable definition.
Cameron Harris
i can run the same application binaries on android, desktop linux, and routers (don't know about tivo, and ps4 isn't linux-based)
Eli Russell
i wrote a very basic kernel (no networking, 80x25 vga console) and minimal os in freepascal. it's a lot of fun and very rewarding but i wouldn't expect it to take off or anything.
be prepared to feel murderous rage towards all hardware companies. acpi, uefi, firmware, and microcode are all trash, the actual benefit of this garbage is nil. it's complexity for the sake of it because mediocre developers can't create anything new, so they instead butcher somebody else's work and stick a "2.0" on it. a case can be made for firmware and microcode but acpi and efi are indefensible vats of pus that oozed out of the decaying mindcunts of the world's worst engineers.
but other than a new found hatred for intel and microsoft it's a great project. have fun and learn as much as you can as you go.
Owen Peterson
Don't listen to the naysayers around here, they're mostly a bunch of holier-than-thou autists that know less than what they project.
That being said, it definitely /is/ an incredibly large undertaking that's a lot harder than you probably realize right now, but don't let that deter you. Instead, I suggest checking out what actual alternative operating system concepts have been researched already to get a good idea of what's been tried and what options are out there. Then maybe you can join your development efforts with one of those instead of starting totally from scratch. And if not, then at least you'll have a better idea of what exists out there, because there are plenty of existing interesting non-unix OSes to learn from: helenos.org haiku-os.org/ redox-os.org/ kolibrios.org/en/ menuetos.net/ icarosdesktop.org/ morphos.de/ sourceforge.net/projects/syllable/ riscosopen.org/ plan9.io/plan9/ capros.org reactos.org
>I want to make a new OS, but instead I am going to just make a GNU/Linux distro like a fucking nigger faggot would KYS
Angel Morales
he's talking about building an entirely different operating system on top of the Linux kernel, not making a distribution of the GNU operating system
Bentley Martin
Too late to the party, Google is already on it. Moreover linux is a shitty monstrous blob, if you're going to do it, pick a capability-based microkernel at least, why ignore 30 years of research? Doesn't have to be seL4, Google took a toy microkernel with ideas from late 90s to start from. If you want to actually contribute to something meaningful, work on camkes.
Liam Nguyen
Why not just use RISC-OS Open, which is already perfectly suited to ARM-based processors?
Mason Hill
ILL MAKE THE LOGO
Wyatt Stewart
Did either of you read his post? The guy is an idiot. He strung just enough buzzwords together in his head to think he has a clue.
Kevin Diaz
if OP is an idiot, let him be an idiot by trying something stupid and failing. You're a bigger idiot for autistically replying to such idiots with butthurt drivel making a big deal out of some naive kid's hobby project. No one asked you for your half-assed life advice.
>I'm going to write a new OS >Using linux You're writing a shitty distro. Just contribute to the existing software projects and make them less shit instead of reinventing the wheel
Henry Lee
All you are doing is making another Linux / UNIX distro.
Make your own kernal, fag.
Ryan Adams
Ask Terry Davis for advice, he's really good at making operating systems
Ryan Barnes
>1. Find some cheap target hardware, likely ARM-based - something like the Raspberry Pi That's a decent idea. There's a lot of baremetal tutorials for the pi, to get you quickly to the point of not being entirely clueless. >2. Take an existing kernel (Linux) to simplify the task. This is where you fuck up. Go bare metal, and start from scratch. Linux is insanely complicated and a bad starting point to learn anything.
Jordan Morales
Just make a new distro if you have to
Adrian Russell
take templeos and pimp it
Nathaniel Baker
u dont have the man years
Angel Russell
t. OP
Easton Jenkins
Why? LG just open sourced WebOS and it runs on the Pi.
Jason Perry
But Chrome OS is a good idea (web-browser-oriented OS) with the least inspiring possible implementation imaginable. It does literally nothing new or interesting with the idea.
Windows 98SE had more interesting ideas about how to merge a browser with an OS than Chrome OS does.
