OSDev.org

Kevin wrote:And that's fine. I just disagree that it automaticaly makes these OSes more impressive or at least interesting and "successful".

But other than freedos, you have not shown me one OS's that is not made up of bits of linux or BeOS code.
Anyone can rip others code and call it a new OS.

It was Owen, not me who posted the list of advanced hobby OSes that are so advanced that you don't accept them as a hobby any more.

You asked me for C OSes that you can develop on, so assuming that you wouldn't accept anything else, I made sure to choose OSes of people in this forum and named Pedigree and tyndur. I also looked at some threads in the announcements board yesterday and it seems SeaOS fulfills your requirement as well.

Do you want me to look for even more? If so, is self-hosting a requirement (it's a much larger requirement for C OSes than for ASM ones) or just being able to write and run a program on the given OS?

Many C OSs have started as hobbies but became non-hobby successes. How does this not fit your criteria of "success"? This thread makes no sense.

neon wrote:Many C OSs have started as hobbies but became non-hobby successes. How does this not fit your criteria of "success"? This thread makes no sense.

Many? AFAIK, the only "hobby-OS" that ever turned into a success is Linux.

Which would mean that 100% of all successful hobby OSes are written in C.

But I think your bar is set very high if you look at it this way.

Kevin wrote:Which would mean that 100% of all successful hobby OSes are written in C.

But I think your bar is set very high if you look at it this way.

The conclusion rather is that the number of hobby-OSes that ever become successful is extremely small, and one successful OS doesn't lend statistical support for which language is best.

The next level of successfulness might be defined based on if a OS is used for commercial applications. That would increase the list, but unfortunately for you, this would add at least one OS written in assembly.

I don't think that commercial use is a particularly good measure of success. But at least you're the first one in this thread (if I didn't miss anything) who actually defines what "success" means for him. Now you need only a complete list of commercially used OSes and the language they are written and you can answer whether the OP is right or not. Of course, you wouldn't convince me this way, because I don't agree with your definition of "success" in the first place.

guyfawkes (indirectly) suggested a different measure of success: Can the OS "be use to code them selfs"? (I'm still not sure if he means that you can write and run program on the OS, or if he means complete self-hosting) I think this is much closer to a definition that I would agree with. Other people may argue that OSes can be successful, even if self-hosting wasn't even a goal for them, and I think they have a point. It's quite hard to define universally applicable criteria for successful OSes.

Kevin wrote:I don't think that commercial use is a particularly good measure of success. But at least you're the first one in this thread (if I didn't miss anything) who actually defines what "success" means for him. Now you need only a complete list of commercially used OSes and the language they are written and you can answer whether the OP is right or not. Of course, you wouldn't convince me this way, because I don't agree with your definition of "success" in the first place.

Not commercially used OSes, but comercial hobby-OSes, which is a big difference. I don't count Windows, QNX, MSDOS, various commercial Unix-derivates created by companies, but only OSes that started out as hobby-OSes. My initial list would then contain RDOS and Linux (which makes it even).

Kevin wrote:guyfawkes (indirectly) suggested a different measure of success: Can the OS "be use to code them selfs"? (I'm still not sure if he means that you can write and run program on the OS, or if he means complete self-hosting) I think this is much closer to a definition that I would agree with. Other people may argue that OSes can be successful, even if self-hosting wasn't even a goal for them, and I think they have a point. It's quite hard to define universally applicable criteria for successful OSes.

Self-hosting is a bad requirement for embedded systems that typically are not edited, compiled and linked on the target, but on a host. But if you expand it like this: An OS that is either self-hosted, or provides a full toolkit for program transfer and remote debugging I'll accept it as a usable criteria. OTOH, I'd be tempted to claim that self-hosting is not enough, but also must be combined with at least a usable environment for application debugging.

rdos wrote:This thread will soon be locked, but I think that most C based OSes are just poor Unix/Linux clones.

Add mine to your exception list.

rdos wrote:The conclusion rather is that the number of hobby-OSes that ever become successful is extremely small, and one successful OS doesn't lend statistical support for which language is best.

Try none.

Kevin wrote:I don't think that commercial use is a particularly good measure of success.

Accepted. "Success" itself has a broad meaning. May be you can define 'Success' from your perspective but that doesn't necessary compel everyone to accept it on your behalf. To me, a successful OS is the one which has accomplished it's intended goal, no matter if there isn't a single end-user using it. Accomplishing an intention is being successful. Maybe the rest of world has a different thing to say.

IMHO I would call a OS that has accomplished it's goals a very good qualifier.

This also means Haiku and AROS are pretty close to their own respective goals.


And please better define ASM hobby OS, some parts of just about every OS here is written in ASM simply to make them boot on x86.

The only BeOS code in Haiku is a couple of user-space apps (file browser and menu): http://haiku-os.org/about/faq#3b

My definition of a successful hobby OS

1. ***It needs to be self hosting (you can compile/assemble and code the OS, on the OS its self.

2. It needs some form of basic network

3. ***It needs to be able to load and run app

3. It needs both a text and graphical interface

4. ***It's code must be 100% original

5. ***It must have been available to test for at least 5 years.

6. ***It must have been independently chosen as a good example of a hobby OS.

All the one's marked with *** are must have and at least one of the others.

@Rusky, Haiku OS, other than BeOS, has "derivatives of existing open source software."
Most likely linux code.

So what's the point of the whole topic for arguing on a subjective opinion?
You have your definition of success OS, and people have other definition of success, as some of your point like network or graphical interface is optional for an useful OS(*); this give no meaningful insight other than language flame war.

(*) Do you think an OS operating a router that totally lacks a graphical interface, but instead provide a web-based configuration utility, is successful?
(*) Do you think an OS with no network, but drive the mp3 player, or an old PDA, is successul?

As people point out, the term successful is so subjective that it's hard to judge. In the end, who care?

I can only say this a successful OS has completed it's own design points if it hasn't then it's not yet successful.
Besides i can not call GNU/Linux a hobby OS for it was only the Kernel that was initialy made as a hobby effort.

CrypticalCode0 wrote:I can only say this a successful OS has completed it's own design points if it hasn't then it's not yet successful.

That implies it has a design at all.

When I work with things for fun, I avoid boring things like "design".

A long-term OS project will most likely have different goals as the project becomes more mature. For instance, in 1988 my goal was simply to enter protected mode without tripple-faulting.