Opersting System and Kernel Features

[DavidVO]

Greeting to whoever sees this,

        I am in the beginning process of creating a(n) libresoftware and opensource (FOSS) operating system for critical infrastructural systems, embedded systems, and scientific computing. A good analogy for a use case would be something like this: "My operating system will drive a Mars rover or a satellite and the computers which manage and analyze or host the data received in real time..." Or "the operating system which powers the robots that produce cars or run power plants." A basic rundown is that I want to be a Linux competitor. I figured, in order to make good competition, it would be amazing for the community if my software would reflect wants or specific requests for such operating system by fulfilling the wants of the community as a whole, that Linux may not be able to achieve from the get-go due to its already massive code base making major additions or features quite difficult to implement.

        It would be of great service you (the LTT community) would be interested in providing feature requests.

Thank you all,

       -David

 

P.S. If anybody has any questions or wants updates I will respond accordingly and publicly if enough people are interested.

 

P.P.S. I am currently setting up a server with the soon to be repos and website so hold tight on that ;).

[Sauron]

Hmmm... writing a Linux competitor that actually stands a chance, completely on your own, is basically impossible - if it's a passion project, don't worry about that. Getting it to work at all is a already a pretty nice goal.

4 hours ago, DavidVO said:

that Linux may not be able to achieve from the get-go due to its already massive code base making major additions or features quite difficult to implement.

 

Major additions such as...? Many of the features that would actually be useful but are (for the time being) missing in the mainline Linux kernel are hugely complex and would rely on much of that preexisting code base to work.

 

Why don't you try and finish HURD? It's been a meme for the last 30 years, it's about time someone actually put in the effort to turn it into something release worthy...

[tux_peng]

Contribute to https://www.haiku-os.org/ or ReactOS

 

Solo OS attempts don't always go great, ever hear of SKYOS?

[DavidVO]

Thank you both so far for your criticisms regarding the product. I do wish to clarify that my goal is not to eliminate or replace Linux in the first place, it is to create a competing product in the rather neesh market of scientific computing (i.e. computers that control satellites up in space or the 'Marianas trench' scientific computing servers) as my primary market, then infrastructural systems as my secondary (which is a not so neesh market).

 

I really appreciate how both of you took the time on giving me this feedback. I really appreciate it.

[Guest]

In my opinion the most valuable thing I have to suggest is: Please use a microkernel, or maybe hybrid, creating another "monolithic unix flavor" or following what *nix operating systems are today is a very old approach, In my opinion this is the most important thing needed judging from your goals, linux is showing it's age from his ancestors. Plan 9 is being discontinued which should be the successor.

Also, while maintain a Stable ABI, and maybe also a C compatible interface making life very easier for developers since you have in mind a FOSS OS so other people may have interest joining in . (Yea speaking of that before you even create the kernel is early but whatever) 

C is never old! 

Also, compatibility between arm and x86_64 would be the best choice for compatibility (x86) and future (arm) 
 

[wasab]

99.99999999999999999999............% chance you will fail, much less so for competing against something as robust as Linux. 

[Sauron]

On 12/1/2018 at 5:30 PM, DavidVO said:

scientific computing (i.e. computers that control satellites up in space or the 'Marianas trench' scientific computing servers)

Linux is just fine for all of these, in fact that's what's used right now for many of these projects. It's robust, scalable and lightweight enough. If you have a specific workload in mind that could be optimized, you should probably make a patch for Linux instead of starting from scratch.

 

Also, it's "niche"