Never heard of operating system

[John1397]

I would like to write a totally new operating system different from anything that is now available, can anyone tell me how to get started?

John

[patio]

Pencil and paper...
Lot's of coffee...
Boatload of laid = off software engineers...

[Geek-9pm]

What is your strong area? What is y0pur weakness? Do you went to use your strong abilities or would you rather work on your weakness?
If you did make somersetting really new, would it matter to others? Do you want others to benefit? What is the goal you want to chive.

For one person an operating system can take several weeks of hard work. But if you can get a group of  five together, you should have something in about one year.

[DaveLembke]

Best thing to do would be to dive head first into looking at the source of the Linux Kernel and then write your own after first altering and compiling and learning how it all works. Its not an easy task! I have altered the Linux Kernel before and it was pretty involved for slackware using 2.4xxx kernel back in college for a project.

You will want to be really good with advanced math and programming in assembly language although you can build one based on the C language as well which is way easier than assembly ( at least it is for me because I can follow C way easier ), but expect something like 100,000+ lines of code, although I think DOS was done in 5000, and Windows I think was something crazy like 2 million lines of code...LOL

Depending on what you want to achieve, it may be better to not reinvent the wheel and make your own Linux Kernel based on an existing kernel, altered and compiled for a special need.

I was tempted to make my own distro flavor, but I just haven't had the time to do it. Since then though I have played with altering Knoppix and making my own software combination Live Distro.

Writing an OS is not easy and I would almost suggest attending MIT first to start from ground zero to advanced computer science/engineering. Then write one if you hadn't already written one for a lab project while attending MIT. Speaking of MIT they have free courses online, to learn from, BUT you benefit most from attending if you can afford to. But if you can follow the course content that is available for free, then there is free information to gain through lectures etc.

And I agree with patio with:

Pencil and paper...
Lot's of coffee...
Boatload of laid = off software engineers...

[Geek-9pm]

I disagree. The OP did not say what his objective is. Star with a clean slate.
We have yet to know if he even intends to write something of a Intel CPU. Noat do we know what he wants  an OS to do.
Maybe he wants an OS for a space sip that searches for signs of life on the planets of out solar system. Linux is not well suited for that  kind of work.

[patio]

It's good enough for the ISS...

http://www.linuxjournal.com/article/3024

[kpac]

The first thing to do is learn C.

[John1397]

The first thing to do is learn C.

That wouldn't help me much do not know c would rather start from scratch.

[Geek-9pm]

Nobody ever has to learn C. You can write in any language you want. There are a set of computer  languages well suited for system stuff. But the rest will work too.
Hee a short history of system languages. Noticer


http://en.wikipedia.org/wiki/System_programming_language
Note that  PL360 was done by     Niklaus Wirth in  1968  based on Algol 60
A year before  C was done by  Dennis Ritchie.

  But why use anything before 1995? 
In 1995. or earlier, MS did Windows 95, which was a really big  improvement  for a system to  run on a desktop. So after that designers began to focus more on the GUI. Old languages like C are the pits for GUI programming. Yes, they did it. But designers wanted better tools. And they came.
Look here:
http://www.javaworld.com/javaworld/jw-01-2012/120103-experimental-programming-languages.html

Or maybe something more recent?
Mozilla, the makers of Fire Fox, want something new.
http://readwrite.com/2010/11/29/mozilla-designing-programming-language-rust

[kpac]

That wouldn't help me much do not know c would rather start from scratch.

The best way to learn is to study existing operating systems and for that you'll need to learn C.

[Geek-9pm]

Why is everybody giving the OP textbook answers? He said he wanted to do something the had never been heard of before.
If the OP has a very high abstract idea, let him  explorer it.

IMHO the answer from Patio was a better fit f to the goals of the OP.
Pencil and paper is very useful.  I would recommend lined paper notepad with spiral binding. Letter size But that is just a personal preference. Alsop,  I like No. 2 hard for a pencil.

And I go for black o for coffee. But in Silicon valley the preference of many  is Jolt COLA. If not available use Coca Cola.

When looking at the work of others, do NOT look at the source code. Instead find a good book that explains what the did, how and why. Look beyond the source code. Source code is gibberish.

[DaveLembke]

If we are going to mention stationary suggestions, I'd say add tabs or sticky notes to the many many pages as GOTO's, GOSUB's, PEEK's, and POKE's...LOL

[DaveLembke]

Got called off to a problem and didnt get to complete my last post... sorry for double post here, but more info I wanted to share based on my experience with modern computer hardware and writing your own OS. With my experience being alteration of Linux Kernel 2.4xxx for slackware and helping a friend in college with his Micro Computer Electronics Degree in which I had to assist him with his projest to make a computer using the Z80 processor similar to this project: http://cpuville.com/Z80.htm.

Regarding the statement of:

That wouldn't help me much do not know c would rather start from scratch.

My suggestion is starting from scratch with C if you plan on making anything that will interface easily, and in a more timely manner, with already existing hardware that is out there. What can be performed in say 20 lines of C could take 200 lines in assembly language or more depending on the complexity of what is to be achieved.

However if you are creating your own language and your own computer hardware that it functions on, you can make anything happen in any way that logically functions as intended. Depending on complexity it can be a simple computer with a small shift register or someing even more powerful.

If you plan on anything to the scale of modern computers, unless you have a team of engineers working for you at your direction, and money is no problem, you will want to use already existing hardware that will work with your language for your OS that you create.

The best way to learn how to interface with modern hardware would be to look at the Linux Kernel and learn from it as for Microsoft will not share the source code with you.

To understand the source code you would need to know C in order to read and understand the Linux Kernel, so thats why I suggest learning C as the foundation to start from if you are planning on making an OS that is for modern computer hardware.

BUT if you are making your own computer that will interpret the language that you create that does not follow the same rules as Intel, AMD, Motorola CPU's for example in (calls), then have at it ... you dont need to conform to anyone elses rules on how their CPUs function you can make your own rules with your own CPU that abides rules you created and understands calls you created.

[BC_Programmer]

That wouldn't help me much do not know c would rather start from scratch.

So are you saying that you literally have no experience with any programming topic?

[patio]

With the stated lofty goal my guess would be...No.

[Geek-9pm]

That wouldn't help me much do not know c would rather start from scratch.

Candidate for quote of the month.   

[Ryuk]

I would like to see a OS like that in .hack but, the first thing would be to flow chart your process.  Learn a language like Java or some form of C.  Coffee or Tea and make sure to have some Aleve near by and plenty of failures

[Geek-9pm]

This point I want to make is that one does not have to duplicate what others have already done. It depends on your objective. And your timeline.
If you want a file serve, use Linux code. But if  you are working with model boats, a very different OS would make sense.

Some years ago an astronomer want to use small computers available at that time. He invented an language and OS now called FORTH.
http://www.forth.org/svfig/osf.html
An there are other OS that are different that anything else out there
Another:
http://en.wikipedia.org/wiki/Pick_operating_system

[Ryuk]

I never heard of that 

[patio]

This is really silly on many levels...

As noble as it sounds the objective of writing a brand spankin new OS from the ground up is even given the great group of minds here an insurmountable task in and of itself...
Secondly while intentions may be good on how to go about doing so advisers have never attempted nor probably won't do so in the near future...
Thirdly...it's now turned into an esoteric discussion of how one should go about it...as it usually does here instead of what it would actually take resource wise to accomplish such a lofty goal...

That being said i'm by no means trying to discourage the OP...however he hasn't been back for a Month.

[Ryuk]

This is all true.  However, I think we showed it isnt easy to do something like that.  Probably gave up on it

[John1397]

So are you saying that you literally have no experience with any programming topic?

Well I could program, but that seems like you are using someone's else's work what I want to do that I never seemed able to was buy a new computer with nothing on the hard drive and by just typing letters, commands or different combination's of key strokes start writing your own operating system.