Creating your own propreitary Linux OS: Where to start?

Not really answering your question but ...

For developing games: the OS isn't really a limiting factor - it's all about tools.
For running video games: you're not on a very good base with linux - there's enough troubles trying to keep driver and hardware support working in mainstream releases without adding another variant, plus linux is a vanishingly small market share of gamers.
For running other programs: linux is already just a giant developer toolkit, what would you do differently?

Honestly, I'd not worry about being sued - your target market would be so small and there'd be no money to sue for. Modern game engines are already close enough to a separate OS anyway but they're not trying to rewrite linux/windows - rather just focusing their platform on relevant tools inside an existing OS.

 

As long as you abide to the restrictions and responsibilities of gplv2 and other licenses you would have no problem.

 

You're describing a game engine. They already exist as free for hobbyists (eg unity) and in some cases as completely open source for a tiny cost to use for hobbyists (e.g. unreal engine). Anyone doing more than a hobby probably needs some of the support that comes with a small license fee and anyone bigger than that is going to be rewriting big chunks of the engine to customise performance. Modern game engines are hugely complex pieces of software - the very fact they're not tied to a specific operating system and are seperate programs is a good thing. Really, the OS aspect is a negligible part of the work in creating a modern game engine.

Some of the questions you ask are so massive in scope as to be hard to answer in a forum. This is one of those.

Step 1 would probably be don't tie any of your tools to any specific OS.

 

Wow flux I dont want to dampen your ambition but have you thought through the enormity of your task.
sort of reminds me of this
http://forum.beyond3d.com/showthread.php?t=62550

Edit: sorry that was a bit harsh

 

Do game developers really think "I wish my operating system could run the game engine and have no other function"
How do you plan to achieve this
start with just the kernel and add all the services and daemons until its capable of running a game engine
or start with a full o/s and rip out everything unnecessary ?

 

There are people here who could probably write a full competitive game engine on their own before they died if they worked on it full time. If you are that person you probably already work in the industry and certainly don't need to ask these questions.

As an absolute minimum requirement you'd have to know at least one low level language and one scripting language inside out. You'd have to be comfortable writing multithreaded performant algorithms and deeply understand all modern rendering techniques, shadowing, lighting, shaders, BRDF systems, etc. Assuming those basic requirements (which aren't trivial to gain) you'd need a massive amount of free time to actually write it all.

Realistically you don't get these skills without working in the industry and if you work in the industry you don't have the free time to do it properly... By the time you finished you'd probably be the only person who could use it and it'd be obsolete.

If I was starting out and wanted to do something useful I'd try writing a basic game either from scratch or inside an existing engine. By game I mean something simple but with loading screens, multiple different textured models, level loading, lighting and shadows, collision detection and some basic AI/pathing of enemies. eg: not just a rendering demo that drops into a sandbox world with everything already loaded.

If you get past that point - identify the bits you struggled with or thought were unnecessarily difficult to do and that's probably where the world would benefit from a better tool or a good tutorial. Unfortunately, until you really understand the problem it's very hard to write a tool that will solve the problem. You need to experience the problems and why they matter yourself. (imo of course).

 

I dont think he has any plans to create an engine, just use someone else's

 

If I'm reading it correctly, he's considering integrating a "license free engine of similar quality and performance to unity(and supporting tools) and unreal(and supporting tools) that is built into the os?"

If he's not considering writing one then he's just saying can I run an existing engine on an existing OS... Tieing an existing engine that is cross platform into a single OS is pointless as it would just be adding a restriction onto a tool for no reason. Tbh - I think Flux is just massively underestimating the work involved in any variant of these tasks.

 

Well yes except he wants to remove everything from the o/s thats not relevant to game development in hope of greater performance and ease of use.

 

that makes more sense