Old Forum messages

[admin]

Dee Sharpe 4.21.2004

NT now uses a sub-system to emulate some of os/2's behavior. For one thing, this is mostly
based on the early versions of os/2, back when ms & ibm were getting along. It is emulating
what os/2 was doing natively; internally it's converting os/2's way of doing things into the NT
way. These are things that an os/2 clone would need to do natively. Also, that code is based
on code that was shared before warp 3 & 4 were started. This means that any compatibility
from NT very outdated considering that it does not contain any of the advancements created
after ms & ibm split. We know that the code took very different directions in the hands of the
2 parent companies. There needs to be a kernel that does os/2 things that are updated and
done natively, rather than a kernel with an emulation layer.

[admin]

Dee Sharpe 4.21.2004

I agree with Sergey, OS/2's internal structure is vastly different from NT's. So, an NT
clone's structure would be vastly different from an OS/2 clone's. Considering the fact
that they would be cloning systems designed with different design objectives.

[admin]

Dee Sharpe 4.25.2004

Which leads me back to another question. Is the goal here to recreate OS/2 Warp 4
with the 16bit portions as well? If so, then I think I will need to research the thunking
of 16bit addresses into 32bit addresses & vise versa...

[admin]

Yuri Prokushev 4.28.2004

2Dee Sharpe Well, having 16bit is required. At least if we want to have support
for 'old' OS/2 applications.

[admin]

Dee Sharpe 4.28.2004

Yuri, you've got a point. The dos compatibility should not be too difficult, there are
plenty of open source dos clones that could be modified. The problem would be
rewritting the win16 portions. However, this rewrite could be in our favor; a chance
to get windows compatibility right --ie a complete win16/32 layer, with the aid of
the odin source. Making all window's support native. Since the central part of OS/2's
kernel is the Control Program, which is 32-bits, then we should probably look at
thet 32bit dos clones. Does anybody else on the list have another suggestion?

[admin]

Yuri Prokushev 4.29.2004

I'm talking not about dos support at all. dos isn't primary goal. But still lot of CPI
(At least VIO/MOU/MON and some DOS) still 16bit only. If we want not just another
OS, but OS/2 clone then we need binary compatability. So we need to support 16bit
calls. According kernel. In general I agree about we need to have out own kernel
(based on some microkernel or written from the scratch), but at the present point
we can quickly have base CPI to allow developers work on other parts of system
(like command line tools, PM, SOM, etc.). And NT/ReactOS is good choice for that.
Or we can directly start working with our own kernel (i.e. based on L4/Mach/etc.).
BTW, currently Sergey works on OS/2 API (DOS API) review (unfortunately, in russian
only) to understand how CPI must be implemented on top of an kernel and his work
can be great basis for developers.

[admin]

Dee Sharpe 4.30.2004

Currently I am in the process of trying to load Warp 4 onto a Compaq Presario 1200TA
laptop. Not really going too well, because of driver issues for the video chip & LCD and
the PCMCIA & eithernet. As soon as I get this resolved, I will try to install OpenWatcom
and see what I can come up with. The OS/2 API Project's documentation on the Control
Program's functions seem to be reasonibly complete. However, my job and its relation to
this war thing makes me a poor candidate for real member of the development team
(for now). But since this seems to be an area that has not been started, I'll try my hand
at it. I guess that all of the different API functions, including the Control Program's,
could be implemented in a dll that plugs into a generic kernel. I have been lurking around
various os projects taking notice of progress. There does not seem to be alot of OS/2 clone
project, so as a big fan of OS/2 I felt leftout and let down. I thought that there was not
enough support from the community to get a good one started. The main problem with
alot of the projects that I saw is the abondance of talk and the lack of any work getting
done. I really believe in the goals of this project, so in order to try to help out I will do
my best to be productive.

[admin]

Dee Sharpe 4.30.2004

I try to be good at accepting critisism (spelling?), so feel free to critique any work
that I submit and any idea I present. I am all about trying to do what it takes to make
this all work.

[admin]

Dave Johnson 5.27.2004

I am working on a small OS that has a modified mach kernal that runs the top layer of functional
hardware. (my english no good.) My home brew os hot swaps kernal code from what i like to call
a non-emu emulayer. if i want to run a unix prog and then a win32 prog the #1 kernal overseas the
swap of code for that. I use a g5 PPC @1.25ghz.

[admin]

long-term storge 5.29.2004

where can I find OS/2 installation d/l? bez $.