OSDev.org

I assume most of you have seen the
OS Decisions Q&A page
(http://www.execpc.com/~geezer/osd/design.htm).
I was wondering what different people's answers were.

Why are you writing an OS?
What do you have in mind for your OS?
Do you have a plan laid out? What steps do you
mean to break the process down to?
What language(s) are using? What specific tools
(NASM vs. gas, gcc vs. lcc, DJGPP vs. Cygwin)?
What hardware are you aiming at (Intel PC,
PowerMac, whatever)?
What development platform are you using?
Do you expect to use your OS regularly
once it is working?
Do you hope others will use your OS?
What parts are you working on now?

Inquiring minds want to know.

>On 2002-01-11 15:32:24, Schol-R-LEA wrote:

I may as well give my own answers, too.

>Why are you writing an OS?

For a number of reasons. First off, I want to
learn how to. Second, there's a matter of ego
satisfaction involved; at the same time, there's
a fair amount of ego riding on it's success or
failure, as I've been talking about doing this
for over ten years. Finally, I hope that I really
can create something new and innovative, and see
it spread to other users and programmers. We all
have our dreams, don't we?

>What do you have in mind for your OS?

I'm not entirely sure yet. Much of what I have in
mind isn't really the OS proper, but the UI and
so forth.

>Do you have a plan laid out? What steps do you
>mean to break the process down to?

I actually see this as a process fo writing
*several* smaller OSes and OS toolkits,,
each building on the previous work. I'll probably
start with a simple real-mode file boot-strap/
file loader/debugger combination, a sort of
stripped down DOS. From there I'll re-write the
file system code and debugger to work in
protected mode, and write a second stage
bootloader to handle setting up the GDT, opeing
the A20 line, etc.

My next step would be to create an integrated
multiboot loader and debugger (I've already got a
name for it: Baron Munchausen Utilities). This
would be released seperately, and would be meant
to support OS development by providing a secure
boot-up environment in which to test and debug OS
code. That, at least, is the idea; I'm still too
new at this to say if it is practical.

The next step, built on top of BMU, would be the
most ambitious yet: a virtualizing multiplexer,
similar in part to an exokernel, in part like the
virtual machine in VM/CMS, and in part to the HAL
in VMS and NT. By providing a minimal abstract
'system' over which you could run other OSes, it
would make it easy to experiment with various
designs. Not only is the machine interface easier
in each case, but you could run both you
development environment and your experimental
system on the same machine without risk of
intefering with each other (hopefully).

This would be the *real* starting point of my
work, of course. But that's all off many years in
the future.

>What language(s) are using? What specific tools
>(NASM vs. gas, gcc vs. lcc, DJGPP vs. Cygwin)?

I will almost certainly use NASM and perhaps
Linux gcc for the earliest work. Later
work I'll probably use languages and tools
of my own design.

>What hardware are you aiming at (Intel PC,
>PowerMac, whatever)?

Stock Intel hardware, mostly, though if
I ever do build a wearable I'll probably use
a StrongARM or a PPC to reduce heat issues.

>What development platform are you using?

Currently, an 800 Mhz Pentium III, 128 MB RAM,
dual-booting Linux and Windows 98. I use both
systems for development, depending on which is
more convenient for a given job.

>Do you expect to use your OS regularly
>once it is working?

Probably.

>Do you hope others will use your OS?

Eventually, but not for a very, very long time.

>What parts are you working on now?

The low-level basics: boot-loader, etc. While
this is probably a bad way to go, it does mean
that I'll have a solid basis to work on when I
get to the more serious complex parts of the
design work.

>On 2002-01-11 15:32:24, Schol-R-LEA wrote:

I may as well give my own answers, too.

>Why are you writing an OS?

For a number of reasons. First off, I want to
learn how to. Second, there's a matter of ego
satisfaction involved; at the same time, there's
a fair amount of ego riding on it's success or
failure, as I've been talking about doing this
for over ten years. Finally, I hope that I really
can create something new and innovative, and see
it spread to other users and programmers. We all
have our dreams, don't we?

>What do you have in mind for your OS?

I'm not entirely sure yet. Much of what I have in
mind isn't really the OS proper, but the UI and
so forth.

>Do you have a plan laid out? What steps do you
>mean to break the process down to?

I actually see this as a process fo writing
*several* smaller OSes and OS toolkits,,
each building on the previous work. I'll probably
start with a simple real-mode file boot-strap/
file loader/debugger combination, a sort of
stripped down DOS. From there I'll re-write the
file system code and debugger to work in
protected mode, and write a second stage
bootloader to handle setting up the GDT, opeing
the A20 line, etc.

My next step would be to create an integrated
multiboot loader and debugger (I've already got a
name for it: Baron Munchausen Utilities). This
would be released seperately, and would be meant
to support OS development by providing a secure
boot-up environment in which to test and debug OS
code. That, at least, is the idea; I'm still too
new at this to say if it is practical.

The next step, built on top of BMU, would be the
most ambitious yet: a virtualizing multiplexer,
similar in part to an exokernel, in part like the
virtual machine in VM/CMS, and in part to the HAL
in VMS and NT. By providing a minimal abstract
'system' over which you could run other OSes, it
would make it easy to experiment with various
designs. Not only is the machine interface easier
in each case, but you could run both you
development environment and your experimental
system on the same machine without risk of
intefering with each other (hopefully).

This would be the *real* starting point of my
work, of course. But that's all off many years in
the future.

>What language(s) are using? What specific tools
>(NASM vs. gas, gcc vs. lcc, DJGPP vs. Cygwin)?

I will almost certainly use NASM and perhaps
Linux gcc for the earliest work. Later
work I'll probably use languages and tools
of my own design.

>What hardware are you aiming at (Intel PC,
>PowerMac, whatever)?

Stock Intel hardware, mostly, though if
I ever do build a wearable I'll probably use
a StrongARM or a PPC to reduce heat issues.

>What development platform are you using?

Currently, an 800 Mhz Pentium III, 128 MB RAM,
dual-booting Linux and Windows 98. I use both
systems for development, depending on which is
more convenient for a given job.

>Do you expect to use your OS regularly
>once it is working?

Probably.

>Do you hope others will use your OS?

Eventually, but not for a very, very long time.

>What parts are you working on now?

The low-level basics: boot-loader, etc. While
this is probably a bad way to go, it does mean
that I'll have a solid basis to work on when I
get to the more serious complex parts of the
design work.

>Why are you writing an OS?

To prove that I can
Partly for educational reasons, partly because I want
an OS for myself. I want the distinction of being
able to run an OS written by myself, with programs
written by myself. No OS/Program combo has done
_exactly_ what I want (and, of course, never will...
that's impossible, unless you write it yourself).

Linux rules, but it's still not _mine_

>What do you have in mind for your OS?

OO Design, flat memory, fast, minimal disign,
microkernel, and other stuff

Oh, and _NO_ backwards compatibility.

>Do you have a plan laid out? What steps do you
>mean to break the process down to?

I've got a plan in my head... some written down.
Steps:
boot loader
minimal kernel
v86 access to BIOS
BIOS drivers for hardware
file system implemenation
cli
driver loading mechanism
compiler, assembler

For this point on the OS is bootable, and can load
drivers, so I simply need to create drivers to
expand abilities. Everything is handled by drivers
(objects).

>What language(s) are using? What specific tools
>(NASM vs. gas, gcc vs. lcc, DJGPP vs. Cygwin)?

The OS is being written in my own language, based
on a subset of C++, but with a different low-level
philosophy.

To develop the language, I'm using GCC and Nasm.

>What hardware are you aiming at (Intel PC,
>PowerMac, whatever)?

x86 for now, soon Itanium (once I get my emulator
from HP). I'd _LOVE_ to develop for a PPC... Motorola's
assembly is _SOOO_ much better than Intel's...

>What development platform are you using?

Linux/X86.

>Do you expect to use your OS regularly
>once it is working?

Every day... 'course, mostly adding features and
writting programs for it, so it actually is useable
every day

>Do you hope others will use your OS?

At this point... yes, and no. I don't really care...
I just want my own OS, completely structured for
me. However, if others find it useful... Sweet!

>What parts are you working on now?

My programming language. Other than that, design.
I haven't yet gotten the balls to rewrite all the
code I once had, that was lost in a HD crash.

>Inquiring minds want to know.

Always save your work!

Jeff

>On 2002-01-11 15:32:24, Schol-R-LEA wrote:
>I assume most of you have seen the
>OS Decisions Q&A page
>(http://www.execpc.com/~geezer/osd/design.htm).
>I was wondering what different people's answers were.
>
>Why are you writing an OS?
>What do you have in mind for your OS?
>Do you have a plan laid out? What steps do you
>mean to break the process down to?
>What language(s) are using? What specific tools
>(NASM vs. gas, gcc vs. lcc, DJGPP vs. Cygwin)?
>What hardware are you aiming at (Intel PC,
>PowerMac, whatever)?
>What development platform are you using?
>Do you expect to use your OS regularly
>once it is working?
>Do you hope others will use your OS?
>What parts are you working on now?
>
>Inquiring minds want to know.

In the first place I'm writing an OS for Fun.
I've in mind robustness,speed & effeciency
I had no prelaid plans as such.
Tools : GCC, as
Hardware : Intel 32-bit
Host OS : Linux
Well if it meets my requirements , definitely
If I can use why not others
Basically the kernel? Iput some time even for other parts

1.Because I love programming
2.Written only in 32-Bit assembly,microkernel,
good looking gui
3.I'll write the kernel first (I already have a bootloader),
then make the gui and finally write some apps
4.32-Bit assembly ONLY;NASM and a86
5.Only for Intel PCs for now
6.Win 98 (although I have linux)
7.Yes.
8.Yes,I hope that other people will use my OS
9.On the kernel/gui

>On 2002-01-11 15:32:24, Schol-R-LEA wrote:
>Why are you writing an OS?

Fun. Education.

>What do you have in mind for your OS?

I would like to do all of these with one project,
if possible:
- something like the QNX demo disk
- a series of kernels to teach OS theory (OSD)
- an OS "kit" like Flux, but with a BSD license
and buildable under DOS or Windows

>Do you have a plan laid out?

I have a wish-list

>What steps do you mean to break the process down to?

Chose modest goals for each release of my OS.
Try not to bite off more than I can chew. Write
DOS programs to test things before I put them in
the OS.

>What language(s) are using? What specific tools

DJGPP and NASM. Bochs and Mtools. DOS and Win95.

>What hardware are you aiming at

PC

>Do you expect to use your OS regularly once it is working?

It will be a long time before it does everything I
want to do.

>Do you hope others will use your OS?

Yes.

>What parts are you working on now?

Bug fixes. Device architecture. Mixed text- and
graphics-mode virtual consoles. Improved
bootloader. Kernel modules.