OSDev.org

I am actually finding this flamefest (which really is what it is, not an actual debate, as those involved - especially Tilde - aren't really paying attention to what the others are saying) interesting, because my own language ideas are in part focused on finding a way to allow both hand-coded and automated machine-specific optimizations in the context of a very-high-level language.

My main thrust has been the use of macros, read macros, code templates, code destructuring, and in-language compiler extensions to allow fine-grained specialization, hopefully in ways that doesn't interfere with each other or with higher-level optimizations. This is part of the reason for my interest in both Synthesis kernel (which templates the system services and can in places generate specialized service operations at run-time) and Slim Binaries (to allow most of the compilation to be done AOT, while still giving the JIT compiler enough information to specialize the program for the specific hardware it is to run on).

I have no idea where this will lead, or if any of it will work. Probably, it won't work at all. That's why I call it experimental.

dseller: the operating system is not just a kernel. the kernel is just a little small piece of the operating system. the definition on wikipedia is also wrong (just like usually everything on wikipedia). operating system is basically the base software package of the computer. its role is to manage the hardware, the resources, control the interaction of the user and the system, and offer the basic tools for the intended handle of the computer.

a text editor, or a file copying software is obviously part of the os too.

i also think that web browser is not part of the operating system, but if somebody thinks so, thats his business.
the same applies for even a jpg or ogg decoder, tcp/ip compatibility, or avi player. i dont think they are part of an operating system, so i didnt added them. but if somebody adds them, thats again, his business, as displaying a jpg file or browsing html can be considered a general need nowdays, so it can be considered as a basic tool.

the question is where the programmer pulls the limit, and says: this is a basic tool, but that one is not.

Geri wrote:dseller: the operating system is not just a kernel.

I know.

Geri wrote:... if somebody thinks so, thats his business

I think this summarizes everything very well. There is no clear distinction. So basically this whole thread is pointless, besides the fact that it was created to call me a troll in the first place.

The more tools you can package in a general way into the base of an OS for proven genuine reasons, the better it's made; the clearer, reusable and modular its code must be.

So far only DOS, Windows, Mac and UNIX/Linux/BSD have emerged because they have proven to be valid architectures that PCs mainly are able to sustain and because they have always been adding the programs they really need. We should do the same with our OSes to prove how valid our architecture is in the executable "opinion" of the machines we are targeting.