Re: [JOS-Kernel] HAY I'm writing a report

[Gilbert Carl Herschberger II]

At 09:32 AM 3/23/00 -0600, [EMAIL PROTECTED] wrote:
>First of all, the current JOS stage is in developing the kernel and JVM.
>We've just really broken ground on going into the actual architecture.  So
>at this point there isn't much of an OS to speak of.

Excellent point. It takes time to break new ground. Unlike the Linux
project, which fully immitated a well-known 30+ year old design for its
first 5 years, our operating system is all new. It has nothing to immitate.

>Since there isn't much of an OS, I feel that the project isn't as
>interested in performance and usability, as it is with creating a new
>paradigm shift in how people think and design OSs.

I feel the same way.

"Do it once." When you're doing something that has never been done before,
it doesn't matter so much how fast it is.

"Do it right." Once we have something to use, we will be increasingly
concerned about usability. We can address the usability of something that
doesn't exist yet.

"Do it again." And then comes speed. Once we have something that's easy to
use, we will be increasingly concerned about efficiency. As more people use
the operating system (and operating systems like it), optimization of the
underlying operating system will have the greatest impact on a
bytecode-based architecture. A little tweak of one function in a virtual
machine might have huge impact on all applications across the board.
Because JOS is meant to be portable, that optimization will be available
across many platforms.

Much of the architecture is volatile at this point, and should be. We are
doing research. We want to find out how many options we have. We want to
pick the option that seems to be the most promising. As real code is
written and used, the architecture will stablize.

I feel that the project has set its priorities around choice, flexibility
and reliability. I want a new OS because my old one offers me no choice. I
want a new OS because my old one is as flexible as, well, concrete. I want
a new OS because my old one crashes two or three times each day, always at
the worst possible moment.

If we have choice, flexibility and reliability, all these other things will
follow. Choice must be reflected throughout the design of an operating
system. Instead of switching from one operating system to another, JOS will
let you plug-in the subsystem that's right for you. While one person might
be willing to risk a system crash and plug-in a super-fast file subsystem,
another might be more conservative and plug-in a slightly slower file
subsystem that almost never crashes. Either way, you don't have to leave
JOS to get what you want.

When it comes to the visual impact of the operating system, JOS is like no
other. While one person might enjoy a shell, browser or desktop that looks
like the operating system they grew up with, another might enjoy the new
universal browser. Either way, you don't have to leave JOS to get what you
want.

This is a research project. We are learning skills by one of the greatest
challenges known to a technician: building an operating system. It is the
most difficult work because most of the work an operating system will do is
unknown and unknowable by the people who build it. An operating system
determines what is easy and what is difficult for other programmers to build.

The product of research is knowledge. What is the worst case scenario? The
worst case imaginable is that nobody learns to appreciate JOS -- except me.
Oh, well. I will have met more interesting people across the globe than
otherwise. I will have more strength in programming, writing, design and
architecture than otherwise. I will have learned more about the strengths
and weaknesses of Java than otherwise. I will have taught myself more about
putting a complete system together than otherwise.

What is the best case scenario? JOS is slow and everybody knows it. Then,
people from across the globe begin to contribute their choice of
optimization. At first, the optimizations trickle in. And then it's a
flood. Everybody celebrates as the most flexible and reliable operating
system becomes the fastest operating system. Look at all the Java
applications it will run!

The opposite has been done. It has been done repeatedly. A stupid operating
system (SOS) is really fast but offers you no choice, no flexibility and no
reliability. I don't want that, do you?


_______________________________________________
Kernel maillist  -  [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel