OSDev.org

go for ARM or some other processor etc.

why is that for?

I maybe wrong, but i think if you had the same OS's that have been named in this topic, but ported to say ARM, they would have more chance of success, theres more and more ARM produces coming onto the market that need OS, the same can not be said for x86.

Crazed here, one of the irregulars who has yet to give up,

I'll be happy if, someday, somebody steals an idea from my OS and credits me. Or if I manage to turn this OS-thing into some kind of project I can get credit for.

I'll also be happy if someone ever uses EDI (yes, I'm working on a real sample driver, it'll be ready RSN).

Hi,

Candy wrote:As I said above, I'm aiming too high. That includes only considering stuff that's in full PD as fully open, the rest has at least one string attached, attempting to recreate everything and to understand everything to the lowest level I can comprehend myself. I'm partly satisfied with make, mostly satisfied with gcc (but at the moment not at all with ld), mostly satisfied with yasm and somewhat satisfied with common shells. I intend to recreate them all, plus OS kernel, libraries and so on. All together until it forms a full system again that isn't built layer on layer.

I guess I'm also aiming high, but in steps, where the first step is a working kernel and boot code (followed by generic device drivers and required system code (CLI, GUI, file systems, etc), my own emulator, my own tools on top of that, and finally many more device drivers).

For the first step the difficulty isn't so much the goal, but the details - a modular micro-kernel with formal specifications for all the interfaces between each "boot module", "kernel module" and the rest of the system; allowances to become the foundation of a peer-to-peer distributed system, support for NUMA/SMP/hyper-threading, fully asychronous "everything", provisions for centralised administration of the distributed cluster, support for headless and/or diskless machines, security, self-testing, hardware auto-detection and as much self-configurations as possible; all hopefully done with high quality code, supporting documentation, no "cut & paste" GPL/BSD/PD code and minimum reliance on third party tools.

So far, one thing I've learnt is that the longer it takes the longer it will take. For example, if I worked 60 hours a week on it I could probably finish the first step in 2 years, but if I only work 30 hours a week it might take 6 years, and at 15 hours a week it'd take 18 years. This is mostly due to the underlying hardware changing while you're trying to support it.

The other thing I've learnt is that for me, "aiming low" takes longer, because half-way through I decide to add support for some other feature...

I just wish I had a time machine, so I could go back to when I first started this project and change my initial goals - if I had enough foresight, I would've been finished the first step by now.


Cheers,

Brendan

Brendan wrote: So far, one thing I've learnt is that the longer it takes the longer it will take. For example, if I worked 60 hours a week on it I could probably finish the first step in 2 years, but if I only work 30 hours a week it might take 6 years, and at 15 hours a week it'd take 18 years. This is mostly due to the underlying hardware changing while you're trying to support it.

This is pretty normal, the more hours you put in the less time it takes to complete. I've found that if I put too little hours a week I tend to get it finished faster, however with lack of thought into the work theres big chances I go back to reinvent the wheel which may dramatically makes changes to other things that will become a greater mess. Sometimes though, I feel the more hours I put in, the more excessive thinking and overly design & implementation, or perhaps because of not enough rest.

I often get reflective thoughts about if I were to perish the next day, did I accomplish anything in my life? My goals for my OS project isn't very high, not as high as Brendans anyways. But one can consider theres much accomplishment done even if the OS has not been completed. Life does get in the way with precious time for OS deving but its also necessary that you accomplish the other life as well. I guess what I'm trying to say is, you may end up accomplishing the bigger goal and nothing else or accomplish many smaller goals, as for me I'm down to earth and trying to fulfil many smaller dreams.

Just my thoughts.

For me i dont really care if it gets big. I just have a lot of fun coding my kernel, and learning, and get a feeling of accomplishment from it. I love it. if it did ever get big, then awesome. but i know i'm not doing anything groundbreaking that hasn't already been done before. I am just happy knowing i'm doing what i am doing, and enjoy the understanding i gain.

This discussion is going to philosophy... Whether it's good or bad, I don't know.