#131 From: "JMA" <mail@...>
Date: Thu Feb 21, 2002 2:07 pm
Subject: Re: Re: Work to do - osFree/CMD - Sign up ! mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 00:18:45 -0000, pinoozzyid wrote:
>>
>> If he agrees about helping osFree, and to make his tools opensource,
>> this would be a very good start.
>>
>> His address is <J.deBoynePollard@t...>. JMA??
>>
>> Bye
>>
>> Cris
>
>I know that the author was contacted and all but let me say that you
>shouldn't get into writting a CMD replacement, because it has been
>done already. It's based on the 32 bit extentions that were developed
>for OS/2 for PPC and I have it installed on and old Warp 3.0 and it
>works like a charm.
>
>Let's not reinvent the wheel. The URL to the page is:
>
>http://homepages.tesco.net./~J.deBoyneP ... s/clu.html
>
Hey, thats the same person
But unless he is willing to release it as opensource his tools are as
useless as the IBM tools.
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Part 5 - Feb 21 2002
Re: Part 5
#132 From: "JMA" <mail@...>
Date: Thu Feb 21, 2002 2:12 pm
Subject: Re: osFree/CMD mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 07:40:55 +0100, Herwig Bauernfeind wrote:
>JMA schrieb:
>
>> Go ahead. All the GNUxxxx packages I have seen on hobbes have source code in
them and as long as
>> we respect their GNU copyright we should use them.
>>
>> This is important, I think it would be positive if we mixed in a lot of GNU
sources in our project.
>> It would stop others from reusing it in a way we may not want them to.
>>
>> Are you able to sign up for developing an app ?
>
>No, sorry, I don't speak C, only REXX and (a little) Pascal.
>
How about dooing some HTML pages for the site on os2world.com ??
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Date: Thu Feb 21, 2002 2:12 pm
Subject: Re: osFree/CMD mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 07:40:55 +0100, Herwig Bauernfeind wrote:
>JMA schrieb:
>
>> Go ahead. All the GNUxxxx packages I have seen on hobbes have source code in
them and as long as
>> we respect their GNU copyright we should use them.
>>
>> This is important, I think it would be positive if we mixed in a lot of GNU
sources in our project.
>> It would stop others from reusing it in a way we may not want them to.
>>
>> Are you able to sign up for developing an app ?
>
>No, sorry, I don't speak C, only REXX and (a little) Pascal.
>
How about dooing some HTML pages for the site on os2world.com ??
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Re: Part 5
#133 From: "zwitska" <richard@...>
Date: Thu Feb 21, 2002 2:29 pm
Subject: JdeBP replacements etc. zwitska
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
Hi,
I'm one of the testers for JdeBP's tools and I think I can clear up a
thing or two :
The CLU just released are all 32-bit replacements and addons to OS/2.
If osFree is compatible with W4, these tools will run.
There's indeed work done on a replacement CMD : 32-bit too !
Don't compare it with anything that's also/already out there, he has
written it from the ground up, conforming to the official specs, and
stripped out everything that's not absolutely needed in there meaning
that commands like DIR, CD etc. are _external_ like in *nix.
Forget about asking him to opensource it : he's a professional
programmer, writing this in his spare time, and he does not want this,
period.
That does not mean it's 'unuseable' !
I think he does a very good job of what he delivers, and whatever is
released will come to hobbes eventually (or his own site).
I think it might be better to concentrate on parts that do not exist
yet (whatever those may be).
wrt. GNU-stuff : there's also a mailinglist, run by John Poltorak,
that is working on getting a *nix-like subsystem on OS/2.
See http://www.unixos2.org for details.
This is a work in progress too but there are skilled ppl. involved to
make it happen.
Besides these 'free' efforts I would also like to point out that I
personally am willing to purchase 'components' as well.
It might be that SDD/2 can be used and maybe NetDrive for coping with
FileSystems (I have/use both).
Maybe it's an idea (if it doesn't exist already) to set up a simple
website where the relevant files can be d/l and links to additional
components ?
Just my E 0.02
rg,rg
Date: Thu Feb 21, 2002 2:29 pm
Subject: JdeBP replacements etc. zwitska
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
Hi,
I'm one of the testers for JdeBP's tools and I think I can clear up a
thing or two :
The CLU just released are all 32-bit replacements and addons to OS/2.
If osFree is compatible with W4, these tools will run.
There's indeed work done on a replacement CMD : 32-bit too !
Don't compare it with anything that's also/already out there, he has
written it from the ground up, conforming to the official specs, and
stripped out everything that's not absolutely needed in there meaning
that commands like DIR, CD etc. are _external_ like in *nix.
Forget about asking him to opensource it : he's a professional
programmer, writing this in his spare time, and he does not want this,
period.
That does not mean it's 'unuseable' !
I think he does a very good job of what he delivers, and whatever is
released will come to hobbes eventually (or his own site).
I think it might be better to concentrate on parts that do not exist
yet (whatever those may be).
wrt. GNU-stuff : there's also a mailinglist, run by John Poltorak,
that is working on getting a *nix-like subsystem on OS/2.
See http://www.unixos2.org for details.
This is a work in progress too but there are skilled ppl. involved to
make it happen.
Besides these 'free' efforts I would also like to point out that I
personally am willing to purchase 'components' as well.
It might be that SDD/2 can be used and maybe NetDrive for coping with
FileSystems (I have/use both).
Maybe it's an idea (if it doesn't exist already) to set up a simple
website where the relevant files can be d/l and links to additional
components ?
Just my E 0.02
rg,rg
Re: Part 5
#134 From: "JMA" <mail@...>
Date: Thu Feb 21, 2002 2:40 pm
Subject: Re: JdeBP replacements etc. mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 11:29:02 -0000, zwitska wrote:
>Hi,
>
>I'm one of the testers for JdeBP's tools and I think I can clear up a
>thing or two :
>
>The CLU just released are all 32-bit replacements and addons to OS/2.
>If osFree is compatible with W4, these tools will run.
>
>There's indeed work done on a replacement CMD : 32-bit too !
>Don't compare it with anything that's also/already out there, he has
>written it from the ground up, conforming to the official specs, and
>stripped out everything that's not absolutely needed in there meaning
>that commands like DIR, CD etc. are _external_ like in *nix.
>
>Forget about asking him to opensource it : he's a professional
>programmer, writing this in his spare time, and he does not want this,
>period.
>
>That does not mean it's 'unuseable' !
>
By all means, he has done a splendid job !
But an opensource cannot mix binaries with sources. That will not work !
If we in the future need to cahnge things (maybe tweak an API) and we
dont have the sources what is the use of it ?
We could just as easy donate a used Warp4 package to everyone that
wants it. They are binaries, cannot be changed but works !
>I think it might be better to concentrate on parts that do not exist
>yet (whatever those may be).
>
NO.
Without the groundwork it all will be futile.
What use is a set of opensource GNU/UNIX tools when the OS
they run on is no longer updated nor sold ?
Noones telling you to drop OS/2 now and run some crude 5%
compatible kernel until enough code gets written.
Our proposal is to gradually replace the whole OS with opensource
components. I say this again -
I choose cmd tools since a) any C developer can help and b)
whatever kernel there will be they will be needed.
If you rather want to start building a PM clone go right ahead and
start the osFree/PM group but thats something I assume not
everyone can help with !
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Date: Thu Feb 21, 2002 2:40 pm
Subject: Re: JdeBP replacements etc. mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 11:29:02 -0000, zwitska wrote:
>Hi,
>
>I'm one of the testers for JdeBP's tools and I think I can clear up a
>thing or two :
>
>The CLU just released are all 32-bit replacements and addons to OS/2.
>If osFree is compatible with W4, these tools will run.
>
>There's indeed work done on a replacement CMD : 32-bit too !
>Don't compare it with anything that's also/already out there, he has
>written it from the ground up, conforming to the official specs, and
>stripped out everything that's not absolutely needed in there meaning
>that commands like DIR, CD etc. are _external_ like in *nix.
>
>Forget about asking him to opensource it : he's a professional
>programmer, writing this in his spare time, and he does not want this,
>period.
>
>That does not mean it's 'unuseable' !
>
By all means, he has done a splendid job !
But an opensource cannot mix binaries with sources. That will not work !
If we in the future need to cahnge things (maybe tweak an API) and we
dont have the sources what is the use of it ?
We could just as easy donate a used Warp4 package to everyone that
wants it. They are binaries, cannot be changed but works !
>I think it might be better to concentrate on parts that do not exist
>yet (whatever those may be).
>
NO.
Without the groundwork it all will be futile.
What use is a set of opensource GNU/UNIX tools when the OS
they run on is no longer updated nor sold ?
Noones telling you to drop OS/2 now and run some crude 5%
compatible kernel until enough code gets written.
Our proposal is to gradually replace the whole OS with opensource
components. I say this again -
I choose cmd tools since a) any C developer can help and b)
whatever kernel there will be they will be needed.
If you rather want to start building a PM clone go right ahead and
start the osFree/PM group but thats something I assume not
everyone can help with !
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Re: Part 5
#135 From: "khaverblad" <kim@...>
Date: Thu Feb 21, 2002 2:47 pm
Subject: Re: OSFree and our future khaverblad
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
--- In osFree@y..., "JMA" <mail@j...> wrote:
> On Thu, 21 Feb 2002 00:58:57 +0100 (CET), Adrian
Gschwend wrote:
> >Even if a lot of us don't like IBM anymore it's
still the only company
> >which can help us. IBM changed a lot the past
years (regarding open
> >source) and because of this I still think it's
not impossible to get at
> >last some parts of OS/2 one day (we would need
at least PM source for
> >example). But if some OS/2 freaks start to
release illegal binaries
> >based on sourcecode they don't own IBM won't
trust the community
> >anymore at all in the future.
> >
> Get real, why did Serenity manage to do what they
did ?
Well, first of all IBM rather early killed their
own belive in the community. Take Team OS2 for
example, organised and maintained by IBM in each
country and killed way back by IBM. When they
figured out that they would never manage to replace
Windows as the desktop OS... well, we all know that
main focus was to be their corporate customers.
IBM also killed OS2 inside the company when they
went to use Windows instead of OS2. It got harder
to talk to customers; try Your self to sell Adidas
if You are using Nike your self.... bad idea. Only
reason why OS2 actually manage to keep alive inside
IBM was that there still was some slow headed
(lucky for us) customers that wanted to keep OS2.
But, was forced by IBM to move over to NT. For IBM
sales rep it was great deal; selling new software
upgrades from OS2 to NT. Lots of new hardware as
well. Migrate their old apps to NT etc.
Don't expect IBM to give us the helping hand. Only
reason that we still get some help from IBM is that
not all OS2 enthusiastic has left the building,
yet. There are lots of names that could be listed
here on people inside IBM that really wanted to
belive in OS2. I even got myself a t-shirt that
says "I leave IBM together with OS2!". And to be
remembered there are still 3rd party code in OS2
that IBM can't be turned over to OpenSource.
This train has left the station without IBM
already!
Kim Haverblad
Date: Thu Feb 21, 2002 2:47 pm
Subject: Re: OSFree and our future khaverblad
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
--- In osFree@y..., "JMA" <mail@j...> wrote:
> On Thu, 21 Feb 2002 00:58:57 +0100 (CET), Adrian
Gschwend wrote:
> >Even if a lot of us don't like IBM anymore it's
still the only company
> >which can help us. IBM changed a lot the past
years (regarding open
> >source) and because of this I still think it's
not impossible to get at
> >last some parts of OS/2 one day (we would need
at least PM source for
> >example). But if some OS/2 freaks start to
release illegal binaries
> >based on sourcecode they don't own IBM won't
trust the community
> >anymore at all in the future.
> >
> Get real, why did Serenity manage to do what they
did ?
Well, first of all IBM rather early killed their
own belive in the community. Take Team OS2 for
example, organised and maintained by IBM in each
country and killed way back by IBM. When they
figured out that they would never manage to replace
Windows as the desktop OS... well, we all know that
main focus was to be their corporate customers.
IBM also killed OS2 inside the company when they
went to use Windows instead of OS2. It got harder
to talk to customers; try Your self to sell Adidas
if You are using Nike your self.... bad idea. Only
reason why OS2 actually manage to keep alive inside
IBM was that there still was some slow headed
(lucky for us) customers that wanted to keep OS2.
But, was forced by IBM to move over to NT. For IBM
sales rep it was great deal; selling new software
upgrades from OS2 to NT. Lots of new hardware as
well. Migrate their old apps to NT etc.
Don't expect IBM to give us the helping hand. Only
reason that we still get some help from IBM is that
not all OS2 enthusiastic has left the building,
yet. There are lots of names that could be listed
here on people inside IBM that really wanted to
belive in OS2. I even got myself a t-shirt that
says "I leave IBM together with OS2!". And to be
remembered there are still 3rd party code in OS2
that IBM can't be turned over to OpenSource.
This train has left the station without IBM
already!
Kim Haverblad
Re: Part 5
#136 From: "Adrian Gschwend" <ktk@...>
Date: Thu Feb 21, 2002 3:56 pm
Subject: Re: OSFree and our future netlabsorg
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>This talk is dangerous.
>Its as illegal to posses these sources as to build and release them.
This discussion is absolutely pointles.
>We have to accept the source are there but dont go around saying,
>"I have them but I dont use them, so I'm no crook".
>That would NOT hold in court.
no but most probably noone is gonna sue you if you just own it and
don't tell anyone.
>There is a difference that almost completly destroys your arguments:
>
>When OS/2 2.0 was designed and a few years ahead the team had one
>of the largest companies in the world backing them up. Also OS/2 was
>at thet time extremly important for IBM.
>I heard quotes that IBM spent almost $1 billion on OS2PPC !
so what? Linux wasn't done like this and Be neither I guess (with money
but with much less).
>Any open source project the OS/2 community could ever pull through
>might get donations of some 10.000 of dollars.
>
>We cannot use IBM's method since there is neither time nor money !!!!
no doubt about that. But that doesn't mean we can't design an OS before
we write it.
>Should we do a 1:1 copy of the OS/2 kernel ?
>No need !
>Should we do a 1:1 copy of the CPI/PM/WPS API ?
>As close as possible, or we could start thinking about Linux or Windows.
yes so I don't understand why you want to redo the current kernel.
>As a former member of the FreeOS team (a one that did find things out):
>A microkernel is capable of almost nothing !
Hmm I wonder why it's called "micro" and not maxi...that's the concept
of it.
>Me must use the time and resources we have !
I can garantee you that not much will happen the way you go because
most professional OS/2 programmers I know won't join *this* effort.
Just read some source at netlabs.org and check the names of the
developers. Quite a lot of those people are willing to help but not
based on this idea we talk about. I'm not gonna spend more time on
this, it's worthless IMHO. As I said everyone is free to do whatever
he/she wants, I just talk about the people I know.
>Just implementing drivers is not that easy. If it was we could implement a
>Win32 driver layer for OS/2 and be done with driver problems. Its lots
>of work and - a kernel is intimatly connected to its drivers model.
>You dont design a kernel without a drivers model !
that's why I say it does not make sense to work on the current kernel
design.
>You know the work that has been put down on Odin. To date Odin is
>the biggest opensource project seen on OS/2. It has a full comercial
>OS with similar origin as Windows. Still Odin is far from complete...
>Dont missunderstand me, Odin is great, but it shows how long it
>would take to do that, a linux layer, supporting drivers from these
>OS'es and giving the GUI apps a complete environment to run in.
Linux was started more than 10 years ago. No doubt that it will take
long but I prefer to have something that works instead of something
that works fast but does not perform in the future.
>Since in part I have to agree with you I suggested that we start with
>the command line tools. These tools will be required undependant
>of what plattform (kernel) they will live upon. They all use the CPI
>and (kbd/mou/vio) API's that MUST exist in a near 100% compatible
>way if we want to be able to run any OS/2 app untop of anything
>else than IBM OS/2.
I personaly doubt that it makes sense to invest time in tools before we
think about a design. Beside this the JdeBP tools and also GNU utils
were mentioned already. That's enough about tools IMHO.
>Get real, why did Serenity manage to do what they did ?
>
>Money !
Believe me you are wrong. Money definitely counts as well but without
*very* good relationships to IBM eCS wouldn't exist.
>What you are saying is as illegal as recompiling and
>releasing unmodified IBM sources !
The source is available for quite some time now. Just because noone
talked about this yet IBM hasn't done anything so far. But I wouldn't
count on this in the future anymore...
>We set out people to do XFree stuff, BSD methods, Everblue, debugging
>Odin and such. All this requires IBM OS/2 to run on.
>Updates to the closedsource IBM OS/2 is no more !
>My recomendation is quite clear:
>
>*** Join a project that have have at least one foot in reality. ***
yeah but I don't find that in the current efforts, sorry.
>Btw: I would love to be proven wrong.
>I would love to see a opensource mk based OS that runs unmodified
>OS/2 and Linux binaries with a mixed X and PM/WPS interface.
>So please, prove me wrong.
you don't write that in two days. As I said, doing a design might suck
for some people but it has to be done first.
cu
Adrian
--
Adrian Gschwend
@ OS/2 Netlabs
ICQ: 22419590
ktk@...
-------
The OS/2 OpenSource Project:
http://www.netlabs.org
Date: Thu Feb 21, 2002 3:56 pm
Subject: Re: OSFree and our future netlabsorg
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>This talk is dangerous.
>Its as illegal to posses these sources as to build and release them.
This discussion is absolutely pointles.
>We have to accept the source are there but dont go around saying,
>"I have them but I dont use them, so I'm no crook".
>That would NOT hold in court.
no but most probably noone is gonna sue you if you just own it and
don't tell anyone.
>There is a difference that almost completly destroys your arguments:
>
>When OS/2 2.0 was designed and a few years ahead the team had one
>of the largest companies in the world backing them up. Also OS/2 was
>at thet time extremly important for IBM.
>I heard quotes that IBM spent almost $1 billion on OS2PPC !
so what? Linux wasn't done like this and Be neither I guess (with money
but with much less).
>Any open source project the OS/2 community could ever pull through
>might get donations of some 10.000 of dollars.
>
>We cannot use IBM's method since there is neither time nor money !!!!
no doubt about that. But that doesn't mean we can't design an OS before
we write it.
>Should we do a 1:1 copy of the OS/2 kernel ?
>No need !
>Should we do a 1:1 copy of the CPI/PM/WPS API ?
>As close as possible, or we could start thinking about Linux or Windows.
yes so I don't understand why you want to redo the current kernel.
>As a former member of the FreeOS team (a one that did find things out):
>A microkernel is capable of almost nothing !
Hmm I wonder why it's called "micro" and not maxi...that's the concept
of it.
>Me must use the time and resources we have !
I can garantee you that not much will happen the way you go because
most professional OS/2 programmers I know won't join *this* effort.
Just read some source at netlabs.org and check the names of the
developers. Quite a lot of those people are willing to help but not
based on this idea we talk about. I'm not gonna spend more time on
this, it's worthless IMHO. As I said everyone is free to do whatever
he/she wants, I just talk about the people I know.
>Just implementing drivers is not that easy. If it was we could implement a
>Win32 driver layer for OS/2 and be done with driver problems. Its lots
>of work and - a kernel is intimatly connected to its drivers model.
>You dont design a kernel without a drivers model !
that's why I say it does not make sense to work on the current kernel
design.
>You know the work that has been put down on Odin. To date Odin is
>the biggest opensource project seen on OS/2. It has a full comercial
>OS with similar origin as Windows. Still Odin is far from complete...
>Dont missunderstand me, Odin is great, but it shows how long it
>would take to do that, a linux layer, supporting drivers from these
>OS'es and giving the GUI apps a complete environment to run in.
Linux was started more than 10 years ago. No doubt that it will take
long but I prefer to have something that works instead of something
that works fast but does not perform in the future.
>Since in part I have to agree with you I suggested that we start with
>the command line tools. These tools will be required undependant
>of what plattform (kernel) they will live upon. They all use the CPI
>and (kbd/mou/vio) API's that MUST exist in a near 100% compatible
>way if we want to be able to run any OS/2 app untop of anything
>else than IBM OS/2.
I personaly doubt that it makes sense to invest time in tools before we
think about a design. Beside this the JdeBP tools and also GNU utils
were mentioned already. That's enough about tools IMHO.
>Get real, why did Serenity manage to do what they did ?
>
>Money !
Believe me you are wrong. Money definitely counts as well but without
*very* good relationships to IBM eCS wouldn't exist.
>What you are saying is as illegal as recompiling and
>releasing unmodified IBM sources !
The source is available for quite some time now. Just because noone
talked about this yet IBM hasn't done anything so far. But I wouldn't
count on this in the future anymore...
>We set out people to do XFree stuff, BSD methods, Everblue, debugging
>Odin and such. All this requires IBM OS/2 to run on.
>Updates to the closedsource IBM OS/2 is no more !
>My recomendation is quite clear:
>
>*** Join a project that have have at least one foot in reality. ***
yeah but I don't find that in the current efforts, sorry.
>Btw: I would love to be proven wrong.
>I would love to see a opensource mk based OS that runs unmodified
>OS/2 and Linux binaries with a mixed X and PM/WPS interface.
>So please, prove me wrong.
you don't write that in two days. As I said, doing a design might suck
for some people but it has to be done first.
cu
Adrian
--
Adrian Gschwend
@ OS/2 Netlabs
ICQ: 22419590
ktk@...
-------
The OS/2 OpenSource Project:
http://www.netlabs.org
Re: Part 5
#137 From: Jason Filby <jasonfilby@...>
Date: Thu Feb 21, 2002 4:54 pm
Subject: Fwd: [freeos] ReactOS jasonfilby
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
To the OSFree people (not FreeOS).. heh.. here's a proposal for
ReactOS as the host kernel.
- Jason
Note: forwarded message attached.
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
To: freeos@yahoogroups.com
From: Jason Filby jasonfilby@...
Mailing-List: list freeos@yahoogroups.com; contact freeos-owner@yahoogroups.com
Date: Thu, 21 Feb 2002 05:38:52 -0800 (PST)
Subject: [freeos] ReactOS
Hi all
I understand that FreeOS is also looking to develop OS/2
compatibility (as is/was OSFree). I've attached a whitepaper
describing the ReactOS architecture (to some degree). If you have any
questions then just post them to this list.
BTW -- ReactOS 0.0.19 is going to be released quite soon (next few
days). Don't be scared off by the 0.0.x thing -- there's a lot of
code in there and we're always improving.
Website: http://www.reactos.com/
Tutorials: http://www.reactos.com/rosdocs/tutorials/
- Jason
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
To Post a message, send it to: freeos@eGroups.com
To Unsubscribe, send a blank message to: freeos-unsubscribe@eGroups.com
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
TABLE OF CONTENTS
1 Introduction
2 The Executive
2.1 Hardware Abstraction layer
2.2 Device Drivers
2.3 Kernel
2.4 System Services
3 Protected Subsystems
4 Native API Architecture
5 Compatibility Targets
1 INTRODUCTION
The ReactOS architecture is based on that of Microsoft Windows NT 4.0. Microsoft
technical specifications define said architecture a modified microkernel (that
is, combining aspects of both microkernels and layered operating systems).
However, the research done by the ReactOS Kernel Team in this field showed that
the adoption of the microkernel paradigm in the Microsoft Windows NT 4.0
operating system is very limited, and therefore that this definition is
questionable. The design was found to be more appropriately described as a
modular and layered architecture.
At the lowest layer is the Executive. The executive includes everything that
runs in kernel mode. Above the executive are the Protected Subsystems. These
subsystems provide implementations of different Operating System personalities.
2 THE EXECUTIVE
The Executive is all the code that runs in kernel mode. The executive can
roughly be broken up into the following components:
o Hardware Abstraction layer (HAL)
o Device Drivers
o The Kernel
o System Services (including the Win32 subsystem)
These components all run in kernel mode. The HAL, Kernel, System Services and
Device Drivers are collectively referred to as the Executive.
2.1 HARDWARE ABSTRACTION layer
The HAL makes it possible for the x86 ReactOS kernel and HAL to run on different
x86 motherboards. The HAL abstracts motherboard specific code from the kernel,
so that different motherboards do not require changes in the kernel. Examples
for different hardware designs are the standard PC, the Japanese NEC PC98 or x86
SGI workstations.
2.3 DEVICE DRIVERS
Device drivers are hardware specific extensions to the ReactOS Executive. They
allow the Operating System to interact with certain devices and visa versa.
ReactOS currently aims to implement the Windows NT 4.0 device driver model. The
Windows Driver Model (WDM) is also a concern for the immediate future. WDM is a
set of rules for writing portable Windows drivers.
Communication:
Device drivers use packets to communicate with the kernel and with other
drivers. Packets are sent via the IO Manager (System Service) and make use of
IRPs (IO Request Packets).
2.4 KERNEL
The kernel is based on that of Microsoft Windows NT 4.0. It implements kernel
mode Asynchronous Procedure Calls (APCs), Deferred Procedure Calls (DPCs),
processes, threading, mutexes, semaphores, spinlocks, timing code and more.
2.5 SYSTEM SERVICES
System services include:
o IO Manager
o Configuration Manager
o Plug & Play
o Power Manager
o Memory Manager
o Executive Support
o object Manager
o security reference monitor, process structure, local procedure call [?]
o Win32 Subsystem
3 PROTECTED SUBSYSTEMS
The Protected Subsystems allow differnt Operating System personalities to run on
top of the ReactOS Executive. The initial target for ReactOS was the Win32
subsystem -- however, the Win32 subsystem runs in kernel mode as part of the
Executive and is not featured here.
User mode subsystems in the works:
o POSIX
Potential Protected Subsystems for the future:
o DOS (Possibly a port of the FreeDOS Operating System)
o Java Operating System (JOS: Open Source Java Operating System)
o OS/2 (FreeOS: Open Source OS/2 clone)
o Many more
Graphical Interface for Subsystems via the Win32 Subsystem:
The Windows NT graphics device drivers are tightly integrated in design with the
Win32 subsystem. Due to this it is impractical for a user mode subsystem to
interact directly with the graphics drivers. For this reason, a subsystem should
make use of the kernel mode Win32 subsystem for a graphics interface. Such a
subsystem need not depend on the Win32 Window Manager, but can instead just use
the graphics primitives provided by the Win32 subsystem.
4 NATIVE API ARCHITECTURE
The Native API Architecture is a set of functions that interface device drivers
and applications to system calls, that is calls to functions exported by the
Executive. This concept is similar to the concept behind the System Call
Interface found in some UNIX operating systems.
The Native API is:
o Portable across hardware architectures. Machine-specific details, such as the
method used to perform system calls, software interrupts, traps, callgates,
etc.), are conveniently hidden behind a standardized calling convention. To
developers who are not familiar with system calls, the Native API will appear
like any other programming interface, such as
the Win32 API.
o Consistent across different implementations of the operating system, as
different Executives may expose incompatible system calls.
The Native API of ReactOS, despite the praiseworthy efforts of the ReactOS
Kernel Team, may present slight incompatibilites with the Native API of
Microsoft Windows NT, since only a small part of the latter is documented
appropriately. Nevertheless, no such incompatibility is known, at the moment.
The Native API is implemented in NTDLL.DLL. This library also
implements:
o Basic process startup code.
o The PE (Portable Executable) Loader, a set of functions to load and manipulate
PE binary images in memory.
o the RTL (RunTime Library), a set of functions covering, among others, string
handling, heap management and process environment management. The RTL is
particularly useful to the developers of protected subsystems and language
runtime libraries.
o A small subset of the libc (C language runtime library). This currently
includes basic string handling and formatting.
5 COMPATIBILITY TARGETS
The original target for ReactOS, with regards to driver and application
compatibility, was Microsoft Windows NT 4.0. Since then, Microsoft Windows 2000
and Windows XP have been released.
Microsoft Windows 2000 and Windows XP are both descendants of Windows NT. As
such we can gradually shift our compatibility target without worrying about the
architecture changing too much. In fact, internally, Windows 2000 is versioned
as Windows 5.0 and Windows XP as Windows 5.1.
The ReactOS team have decided to maintain Windows NT 4.0 as the official
compatibility target. This is because most of the resources, articles and books
on Windows NT/2000/XP technology are written for Windows NT 4.0. This does not
mean that features present in later versions of Windows NT based operating
systems will not be implemented in ReactOS.
Date: Thu Feb 21, 2002 4:54 pm
Subject: Fwd: [freeos] ReactOS jasonfilby
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
To the OSFree people (not FreeOS).. heh.. here's a proposal for
ReactOS as the host kernel.
- Jason
Note: forwarded message attached.
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
To: freeos@yahoogroups.com
From: Jason Filby jasonfilby@...
Mailing-List: list freeos@yahoogroups.com; contact freeos-owner@yahoogroups.com
Date: Thu, 21 Feb 2002 05:38:52 -0800 (PST)
Subject: [freeos] ReactOS
Hi all
I understand that FreeOS is also looking to develop OS/2
compatibility (as is/was OSFree). I've attached a whitepaper
describing the ReactOS architecture (to some degree). If you have any
questions then just post them to this list.
BTW -- ReactOS 0.0.19 is going to be released quite soon (next few
days). Don't be scared off by the 0.0.x thing -- there's a lot of
code in there and we're always improving.
Website: http://www.reactos.com/
Tutorials: http://www.reactos.com/rosdocs/tutorials/
- Jason
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
To Post a message, send it to: freeos@eGroups.com
To Unsubscribe, send a blank message to: freeos-unsubscribe@eGroups.com
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
TABLE OF CONTENTS
1 Introduction
2 The Executive
2.1 Hardware Abstraction layer
2.2 Device Drivers
2.3 Kernel
2.4 System Services
3 Protected Subsystems
4 Native API Architecture
5 Compatibility Targets
1 INTRODUCTION
The ReactOS architecture is based on that of Microsoft Windows NT 4.0. Microsoft
technical specifications define said architecture a modified microkernel (that
is, combining aspects of both microkernels and layered operating systems).
However, the research done by the ReactOS Kernel Team in this field showed that
the adoption of the microkernel paradigm in the Microsoft Windows NT 4.0
operating system is very limited, and therefore that this definition is
questionable. The design was found to be more appropriately described as a
modular and layered architecture.
At the lowest layer is the Executive. The executive includes everything that
runs in kernel mode. Above the executive are the Protected Subsystems. These
subsystems provide implementations of different Operating System personalities.
2 THE EXECUTIVE
The Executive is all the code that runs in kernel mode. The executive can
roughly be broken up into the following components:
o Hardware Abstraction layer (HAL)
o Device Drivers
o The Kernel
o System Services (including the Win32 subsystem)
These components all run in kernel mode. The HAL, Kernel, System Services and
Device Drivers are collectively referred to as the Executive.
2.1 HARDWARE ABSTRACTION layer
The HAL makes it possible for the x86 ReactOS kernel and HAL to run on different
x86 motherboards. The HAL abstracts motherboard specific code from the kernel,
so that different motherboards do not require changes in the kernel. Examples
for different hardware designs are the standard PC, the Japanese NEC PC98 or x86
SGI workstations.
2.3 DEVICE DRIVERS
Device drivers are hardware specific extensions to the ReactOS Executive. They
allow the Operating System to interact with certain devices and visa versa.
ReactOS currently aims to implement the Windows NT 4.0 device driver model. The
Windows Driver Model (WDM) is also a concern for the immediate future. WDM is a
set of rules for writing portable Windows drivers.
Communication:
Device drivers use packets to communicate with the kernel and with other
drivers. Packets are sent via the IO Manager (System Service) and make use of
IRPs (IO Request Packets).
2.4 KERNEL
The kernel is based on that of Microsoft Windows NT 4.0. It implements kernel
mode Asynchronous Procedure Calls (APCs), Deferred Procedure Calls (DPCs),
processes, threading, mutexes, semaphores, spinlocks, timing code and more.
2.5 SYSTEM SERVICES
System services include:
o IO Manager
o Configuration Manager
o Plug & Play
o Power Manager
o Memory Manager
o Executive Support
o object Manager
o security reference monitor, process structure, local procedure call [?]
o Win32 Subsystem
3 PROTECTED SUBSYSTEMS
The Protected Subsystems allow differnt Operating System personalities to run on
top of the ReactOS Executive. The initial target for ReactOS was the Win32
subsystem -- however, the Win32 subsystem runs in kernel mode as part of the
Executive and is not featured here.
User mode subsystems in the works:
o POSIX
Potential Protected Subsystems for the future:
o DOS (Possibly a port of the FreeDOS Operating System)
o Java Operating System (JOS: Open Source Java Operating System)
o OS/2 (FreeOS: Open Source OS/2 clone)
o Many more
Graphical Interface for Subsystems via the Win32 Subsystem:
The Windows NT graphics device drivers are tightly integrated in design with the
Win32 subsystem. Due to this it is impractical for a user mode subsystem to
interact directly with the graphics drivers. For this reason, a subsystem should
make use of the kernel mode Win32 subsystem for a graphics interface. Such a
subsystem need not depend on the Win32 Window Manager, but can instead just use
the graphics primitives provided by the Win32 subsystem.
4 NATIVE API ARCHITECTURE
The Native API Architecture is a set of functions that interface device drivers
and applications to system calls, that is calls to functions exported by the
Executive. This concept is similar to the concept behind the System Call
Interface found in some UNIX operating systems.
The Native API is:
o Portable across hardware architectures. Machine-specific details, such as the
method used to perform system calls, software interrupts, traps, callgates,
etc.), are conveniently hidden behind a standardized calling convention. To
developers who are not familiar with system calls, the Native API will appear
like any other programming interface, such as
the Win32 API.
o Consistent across different implementations of the operating system, as
different Executives may expose incompatible system calls.
The Native API of ReactOS, despite the praiseworthy efforts of the ReactOS
Kernel Team, may present slight incompatibilites with the Native API of
Microsoft Windows NT, since only a small part of the latter is documented
appropriately. Nevertheless, no such incompatibility is known, at the moment.
The Native API is implemented in NTDLL.DLL. This library also
implements:
o Basic process startup code.
o The PE (Portable Executable) Loader, a set of functions to load and manipulate
PE binary images in memory.
o the RTL (RunTime Library), a set of functions covering, among others, string
handling, heap management and process environment management. The RTL is
particularly useful to the developers of protected subsystems and language
runtime libraries.
o A small subset of the libc (C language runtime library). This currently
includes basic string handling and formatting.
5 COMPATIBILITY TARGETS
The original target for ReactOS, with regards to driver and application
compatibility, was Microsoft Windows NT 4.0. Since then, Microsoft Windows 2000
and Windows XP have been released.
Microsoft Windows 2000 and Windows XP are both descendants of Windows NT. As
such we can gradually shift our compatibility target without worrying about the
architecture changing too much. In fact, internally, Windows 2000 is versioned
as Windows 5.0 and Windows XP as Windows 5.1.
The ReactOS team have decided to maintain Windows NT 4.0 as the official
compatibility target. This is because most of the resources, articles and books
on Windows NT/2000/XP technology are written for Windows NT 4.0. This does not
mean that features present in later versions of Windows NT based operating
systems will not be implemented in ReactOS.
Re: Part 5
#138 From: "poldi42" <poldi42@...>
Date: Thu Feb 21, 2002 5:09 pm
Subject: peace. [was:Re: OSFree and our future] poldi42
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
--- In osFree@y..., "Adrian Gschwend" <ktk@d...> wrote:
> On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>
> >This talk is dangerous.
>
> This discussion is absolutely pointles.
>
and further discussion between the 2 of you will most likely lead to
flaming of some sort and be dangerous as well as pointless.
a common ground you both stand on are your aim and intentions, don't
get into a "if you don't do it the right way, you will fail and I as
well as others won't be part of it"-mood too fast.
while I personally (not that it mattered) think Adrian & Achim are
right, listening to the freeos ML for some months lets me easyly
understand why JMA will probably want to _do_ __anything__ to get sth.
going.
most importandly he does _not_ advocate to start one building a cloned
kernel or using the "cloned" one from .ru, but gathers people to code
the documented high-level APIs that will be needed anyway. I fail to
see how this could be in vain. the kernal-question remains open and
unanswered as I understand it.
following JMAs approach a lot can be aquired before this question has
to be answered.
I also think that any suggestion for a specific kernel-architecture
will be taken into account be members of this group - with strong
preferrence for a model that does not only exist on paper but has some
meat already done.
a lot of what could be said regarding the k-question has already been
said on the freeos ML, along with a lot of usefull referrences and
documentation. anybody interested in discussing this further better
took a weekends time out to cross-read the stuff there.
anyhow a decision will be made be the very small circle of people that
are capable to do this part of the work (including for sure not me and
probably neither of you as well [no offense meant!]).
good luck to all!
regards,
Carsten
Date: Thu Feb 21, 2002 5:09 pm
Subject: peace. [was:Re: OSFree and our future] poldi42
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
--- In osFree@y..., "Adrian Gschwend" <ktk@d...> wrote:
> On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>
> >This talk is dangerous.
>
> This discussion is absolutely pointles.
>
and further discussion between the 2 of you will most likely lead to
flaming of some sort and be dangerous as well as pointless.
a common ground you both stand on are your aim and intentions, don't
get into a "if you don't do it the right way, you will fail and I as
well as others won't be part of it"-mood too fast.
while I personally (not that it mattered) think Adrian & Achim are
right, listening to the freeos ML for some months lets me easyly
understand why JMA will probably want to _do_ __anything__ to get sth.
going.
most importandly he does _not_ advocate to start one building a cloned
kernel or using the "cloned" one from .ru, but gathers people to code
the documented high-level APIs that will be needed anyway. I fail to
see how this could be in vain. the kernal-question remains open and
unanswered as I understand it.
following JMAs approach a lot can be aquired before this question has
to be answered.
I also think that any suggestion for a specific kernel-architecture
will be taken into account be members of this group - with strong
preferrence for a model that does not only exist on paper but has some
meat already done.
a lot of what could be said regarding the k-question has already been
said on the freeos ML, along with a lot of usefull referrences and
documentation. anybody interested in discussing this further better
took a weekends time out to cross-read the stuff there.
anyhow a decision will be made be the very small circle of people that
are capable to do this part of the work (including for sure not me and
probably neither of you as well [no offense meant!]).
good luck to all!
regards,
Carsten
Re: Part 5
#139 From: "JMA" <mail@...>
Date: Thu Feb 21, 2002 5:10 pm
Subject: Re: OSFree and our future mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 13:56:07 +0100 (CET), Adrian Gschwend wrote:
>On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>
>>This talk is dangerous.
>>Its as illegal to posses these sources as to build and release them.
>
>This discussion is absolutely pointles.
>
Others started it and screamed "illegal" and "bloddy murder" while they
were in fact doing (almost) the same thing.
I did not know how the binaries were built and I still have no real
evidence except from people that tells me they have the source
so they know...
To hell with the distro if its illegal.
>>There is a difference that almost completly destroys your arguments:
>>
>>When OS/2 2.0 was designed and a few years ahead the team had one
>>of the largest companies in the world backing them up. Also OS/2 was
>>at thet time extremly important for IBM.
>>I heard quotes that IBM spent almost $1 billion on OS2PPC !
>
>so what? Linux wasn't done like this and Be neither I guess (with money
>but with much less).
>
Linux - that is a point but I'm affraid we will never be able to gain that
momentum. Why would any Linux developer move to a OS that
might run Linux apps in a future ?
Some people seem to think OS/2 has this magic aspect, just tell
people it exists and everyone will convert. Why should you work on
a OS/2 project when there are so many better projects ?
You know why you would want to, I know why I would want to but
the OS/2 community is not that many people.
Be - It was a comercial project costing many millions.
Give me that money and I'll and others would work 10 hours a day to
build a clean opensource OS/2 clone.
>>Should we do a 1:1 copy of the OS/2 kernel ?
>>No need !
>>Should we do a 1:1 copy of the CPI/PM/WPS API ?
>>As close as possible, or we could start thinking about Linux or Windows.
>
>yes so I don't understand why you want to redo the current kernel.
>
Strange, I cannot remember saying it that way.
>
>
If you agree with me with this then whats the problem
The only reason for an OS/2 compatible kernel is to be able to reuse the
drivers we have.
OK, another reason might be that the people that hears about this knows
much more about the OS/2 kernel than any other kernel
(They do agree they have or have seen it
Personally, I DONT CARE ABOUT WHAT KERNEL THERE IS !
But, we cannot wait three years for the kernel nor use a kernel that
will not be able to run OS/2 binaries (thats not just a LX loader).
You put it like the FreeOS group has for years.
We cannot start since we dont really know what we want and next
year may bring some better mk so we better wait.
Lets start where we can start: Cmd line tools/CPI API/PM (?).
Then graft that ontop the right kernel when we have it.
Until then we could use the our allready installed IBM kernel
(Warp/MCPeCS) as an interim solution.
>>As a former member of the FreeOS team (a one that did find things out):
>>A microkernel is capable of almost nothing !
>
>Hmm I wonder why it's called "micro" and not maxi...that's the concept
>of it.
>
The concept is to keep the plattformdependant code as small and separate
as possible. The microkernel is micro since its supposed to be as small
as possible and support only a small but well defined interface.
>>Me must use the time and resources we have !
>
>I can garantee you that not much will happen the way you go because
>most professional OS/2 programmers I know won't join *this* effort.
>
OK, these guys may not want to join.
But you are very wrong if you think they are the only capable developers
in the (OS/2) world...
>Just read some source at netlabs.org and check the names of the
>developers. Quite a lot of those people are willing to help but not
>based on this idea we talk about. I'm not gonna spend more time on
>this, it's worthless IMHO. As I said everyone is free to do whatever
>he/she wants, I just talk about the people I know.
>
So, what is the idea I purpused ???
Steal IBM:s latest code (wait I'll book a trip to Austin) ?
Write an 100% compatible kernel down to the old 16bit code ?
Use the Linux kernel ?
Build a wizbang mk based solution ?
This is how it is:
Spending time discussing what kernel to use or build has show futile
to many times. So,
Lets start with something most can do. If we keep talking and never
start coding were will this take us ?
This does not IN ANY WAY stop testing of possible kernel solutions.
But the people capable of building/porting a kernel already knows
what the alternatives are. Let them download the alternatives, build
them and prototype an OS/2 layer atop.
Let us do something in the mean while !
Lets show everybody we want a opensource OS/2 and let everybody participate.
>>Just implementing drivers is not that easy. If it was we could implement a
>>Win32 driver layer for OS/2 and be done with driver problems. Its lots
>>of work and - a kernel is intimatly connected to its drivers model.
>>You dont design a kernel without a drivers model !
>
>that's why I say it does not make sense to work on the current kernel
>design.
>
And thats why you cant just take kernel X and think it will work !
>>You know the work that has been put down on Odin. To date Odin is
>>the biggest opensource project seen on OS/2. It has a full comercial
>>OS with similar origin as Windows. Still Odin is far from complete...
>>Dont missunderstand me, Odin is great, but it shows how long it
>>would take to do that, a linux layer, supporting drivers from these
>>OS'es and giving the GUI apps a complete environment to run in.
>
>Linux was started more than 10 years ago. No doubt that it will take
>long but I prefer to have something that works instead of something
>that works fast but does not perform in the future.
>
Then lets use the the current OS/2 kernel we all have on our machines.
Serenity will sell it to you for $200 or you can get it from ebay for $15.
Why should we all sit doing nothing while some good people selects
the kernel. Whatever kernel get choosen it must still run OS/2
binaries and be able to support a OS/2 subsystem. Now lets start
building the subsystem ontop the existing kernel.
Or do you think we should sit with our arms crossed and wish the
the CPI layer, the cmd line utils, PM and so on gets built magicly
by themselves ??
>>Since in part I have to agree with you I suggested that we start with
>>the command line tools. These tools will be required undependant
>>of what plattform (kernel) they will live upon. They all use the CPI
>>and (kbd/mou/vio) API's that MUST exist in a near 100% compatible
>>way if we want to be able to run any OS/2 app untop of anything
>>else than IBM OS/2.
>
>I personaly doubt that it makes sense to invest time in tools before we
>think about a design. Beside this the JdeBP tools and also GNU utils
>were mentioned already. That's enough about tools IMHO.
>
OK,
Let me remove all your command line utils including cmd.exe from your
machine. Would you mind.
Hey, we are talking about opensource and OS/2 compatible.
If you go for closedsource Warp/MCP/eCS already have it all, why bother.
Also, how do you think the GNU tools work ?
They either call DosXxxx call or uses a compiler that calls DosXxxx calls.
DOSCALLS1 is mainly a translation layer down to the kernel services.
THere is no doubt that you couls build a quite generic layer that can
be bound to most kernels. A kernel must have VMM, threads and so
on. Why not build the CPI layer now and do the glu when the kernel
is getting more ready.
No waiting for the kernel select will not save us tons of work, ask
anyone that knows how kernel/usermode works.
Remember, the CPI API's must still work for binary OS/2 apps.
>>Get real, why did Serenity manage to do what they did ?
>>
>>Money !
>
>Believe me you are wrong. Money definitely counts as well but without
>*very* good relationships to IBM eCS wouldn't exist.
>
For a project (not a dev project), 2001, that Mensys cohosted with me,
I asked John Thompson (the head of IBM eBusOS division) was would be
required for a warp 5.
Has answer was quite plain, at least 200.000 licences.
Now that was the OS/2 boss. What do you think people above him would
ask for ??
OK, Kim and Bob's contacts helped ut without money they would get nothing.
>The source is available for quite some time now. Just because noone
>talked about this yet IBM hasn't done anything so far. But I wouldn't
>count on this in the future anymore...
>
So what, Achim says the sources are old, to old. Why care about them ?
>>We set out people to do XFree stuff, BSD methods, Everblue, debugging
>>Odin and such. All this requires IBM OS/2 to run on.
>>Updates to the closedsource IBM OS/2 is no more !
>
>>My recomendation is quite clear:
>>
>>*** Join a project that have have at least one foot in reality. ***
>
>yeah but I don't find that in the current efforts, sorry.
>
Then give me a project that has !
FreeOS (ha, ha)
Achims eledged mk project noone ever heard about before
ReactOS (I have seen the sources, its FAR from anywhere)
Come up with something better then, we dont have lots of choices !
>>Btw: I would love to be proven wrong.
>>I would love to see a opensource mk based OS that runs unmodified
>>OS/2 and Linux binaries with a mixed X and PM/WPS
Date: Thu Feb 21, 2002 5:10 pm
Subject: Re: OSFree and our future mailjmase
Offline Offline
Send Email Send Email
Invite to Yahoo! 360° Invite to Yahoo! 360°
On Thu, 21 Feb 2002 13:56:07 +0100 (CET), Adrian Gschwend wrote:
>On Thu, 21 Feb 2002 12:01:46 +0100, JMA wrote:
>
>>This talk is dangerous.
>>Its as illegal to posses these sources as to build and release them.
>
>This discussion is absolutely pointles.
>
Others started it and screamed "illegal" and "bloddy murder" while they
were in fact doing (almost) the same thing.
I did not know how the binaries were built and I still have no real
evidence except from people that tells me they have the source
so they know...
To hell with the distro if its illegal.
>>There is a difference that almost completly destroys your arguments:
>>
>>When OS/2 2.0 was designed and a few years ahead the team had one
>>of the largest companies in the world backing them up. Also OS/2 was
>>at thet time extremly important for IBM.
>>I heard quotes that IBM spent almost $1 billion on OS2PPC !
>
>so what? Linux wasn't done like this and Be neither I guess (with money
>but with much less).
>
Linux - that is a point but I'm affraid we will never be able to gain that
momentum. Why would any Linux developer move to a OS that
might run Linux apps in a future ?
Some people seem to think OS/2 has this magic aspect, just tell
people it exists and everyone will convert. Why should you work on
a OS/2 project when there are so many better projects ?
You know why you would want to, I know why I would want to but
the OS/2 community is not that many people.
Be - It was a comercial project costing many millions.
Give me that money and I'll and others would work 10 hours a day to
build a clean opensource OS/2 clone.
>>Should we do a 1:1 copy of the OS/2 kernel ?
>>No need !
>>Should we do a 1:1 copy of the CPI/PM/WPS API ?
>>As close as possible, or we could start thinking about Linux or Windows.
>
>yes so I don't understand why you want to redo the current kernel.
>
Strange, I cannot remember saying it that way.
>
>
If you agree with me with this then whats the problem
The only reason for an OS/2 compatible kernel is to be able to reuse the
drivers we have.
OK, another reason might be that the people that hears about this knows
much more about the OS/2 kernel than any other kernel
(They do agree they have or have seen it
Personally, I DONT CARE ABOUT WHAT KERNEL THERE IS !
But, we cannot wait three years for the kernel nor use a kernel that
will not be able to run OS/2 binaries (thats not just a LX loader).
You put it like the FreeOS group has for years.
We cannot start since we dont really know what we want and next
year may bring some better mk so we better wait.
Lets start where we can start: Cmd line tools/CPI API/PM (?).
Then graft that ontop the right kernel when we have it.
Until then we could use the our allready installed IBM kernel
(Warp/MCPeCS) as an interim solution.
>>As a former member of the FreeOS team (a one that did find things out):
>>A microkernel is capable of almost nothing !
>
>Hmm I wonder why it's called "micro" and not maxi...that's the concept
>of it.
>
The concept is to keep the plattformdependant code as small and separate
as possible. The microkernel is micro since its supposed to be as small
as possible and support only a small but well defined interface.
>>Me must use the time and resources we have !
>
>I can garantee you that not much will happen the way you go because
>most professional OS/2 programmers I know won't join *this* effort.
>
OK, these guys may not want to join.
But you are very wrong if you think they are the only capable developers
in the (OS/2) world...
>Just read some source at netlabs.org and check the names of the
>developers. Quite a lot of those people are willing to help but not
>based on this idea we talk about. I'm not gonna spend more time on
>this, it's worthless IMHO. As I said everyone is free to do whatever
>he/she wants, I just talk about the people I know.
>
So, what is the idea I purpused ???
Steal IBM:s latest code (wait I'll book a trip to Austin) ?
Write an 100% compatible kernel down to the old 16bit code ?
Use the Linux kernel ?
Build a wizbang mk based solution ?
This is how it is:
Spending time discussing what kernel to use or build has show futile
to many times. So,
Lets start with something most can do. If we keep talking and never
start coding were will this take us ?
This does not IN ANY WAY stop testing of possible kernel solutions.
But the people capable of building/porting a kernel already knows
what the alternatives are. Let them download the alternatives, build
them and prototype an OS/2 layer atop.
Let us do something in the mean while !
Lets show everybody we want a opensource OS/2 and let everybody participate.
>>Just implementing drivers is not that easy. If it was we could implement a
>>Win32 driver layer for OS/2 and be done with driver problems. Its lots
>>of work and - a kernel is intimatly connected to its drivers model.
>>You dont design a kernel without a drivers model !
>
>that's why I say it does not make sense to work on the current kernel
>design.
>
And thats why you cant just take kernel X and think it will work !
>>You know the work that has been put down on Odin. To date Odin is
>>the biggest opensource project seen on OS/2. It has a full comercial
>>OS with similar origin as Windows. Still Odin is far from complete...
>>Dont missunderstand me, Odin is great, but it shows how long it
>>would take to do that, a linux layer, supporting drivers from these
>>OS'es and giving the GUI apps a complete environment to run in.
>
>Linux was started more than 10 years ago. No doubt that it will take
>long but I prefer to have something that works instead of something
>that works fast but does not perform in the future.
>
Then lets use the the current OS/2 kernel we all have on our machines.
Serenity will sell it to you for $200 or you can get it from ebay for $15.
Why should we all sit doing nothing while some good people selects
the kernel. Whatever kernel get choosen it must still run OS/2
binaries and be able to support a OS/2 subsystem. Now lets start
building the subsystem ontop the existing kernel.
Or do you think we should sit with our arms crossed and wish the
the CPI layer, the cmd line utils, PM and so on gets built magicly
by themselves ??
>>Since in part I have to agree with you I suggested that we start with
>>the command line tools. These tools will be required undependant
>>of what plattform (kernel) they will live upon. They all use the CPI
>>and (kbd/mou/vio) API's that MUST exist in a near 100% compatible
>>way if we want to be able to run any OS/2 app untop of anything
>>else than IBM OS/2.
>
>I personaly doubt that it makes sense to invest time in tools before we
>think about a design. Beside this the JdeBP tools and also GNU utils
>were mentioned already. That's enough about tools IMHO.
>
OK,
Let me remove all your command line utils including cmd.exe from your
machine. Would you mind.
Hey, we are talking about opensource and OS/2 compatible.
If you go for closedsource Warp/MCP/eCS already have it all, why bother.
Also, how do you think the GNU tools work ?
They either call DosXxxx call or uses a compiler that calls DosXxxx calls.
DOSCALLS1 is mainly a translation layer down to the kernel services.
THere is no doubt that you couls build a quite generic layer that can
be bound to most kernels. A kernel must have VMM, threads and so
on. Why not build the CPI layer now and do the glu when the kernel
is getting more ready.
No waiting for the kernel select will not save us tons of work, ask
anyone that knows how kernel/usermode works.
Remember, the CPI API's must still work for binary OS/2 apps.
>>Get real, why did Serenity manage to do what they did ?
>>
>>Money !
>
>Believe me you are wrong. Money definitely counts as well but without
>*very* good relationships to IBM eCS wouldn't exist.
>
For a project (not a dev project), 2001, that Mensys cohosted with me,
I asked John Thompson (the head of IBM eBusOS division) was would be
required for a warp 5.
Has answer was quite plain, at least 200.000 licences.
Now that was the OS/2 boss. What do you think people above him would
ask for ??
OK, Kim and Bob's contacts helped ut without money they would get nothing.
>The source is available for quite some time now. Just because noone
>talked about this yet IBM hasn't done anything so far. But I wouldn't
>count on this in the future anymore...
>
So what, Achim says the sources are old, to old. Why care about them ?
>>We set out people to do XFree stuff, BSD methods, Everblue, debugging
>>Odin and such. All this requires IBM OS/2 to run on.
>>Updates to the closedsource IBM OS/2 is no more !
>
>>My recomendation is quite clear:
>>
>>*** Join a project that have have at least one foot in reality. ***
>
>yeah but I don't find that in the current efforts, sorry.
>
Then give me a project that has !
FreeOS (ha, ha)
Achims eledged mk project noone ever heard about before
ReactOS (I have seen the sources, its FAR from anywhere)
Come up with something better then, we dont have lots of choices !
>>Btw: I would love to be proven wrong.
>>I would love to see a opensource mk based OS that runs unmodified
>>OS/2 and Linux binaries with a mixed X and PM/WPS
Re: Part 5
#140 Re: [osFree] Fwd: [freeos] ReactOS
Expand Messages
JMA
Feb 21, 2002
On Thu, 21 Feb 2002 05:54:23 -0800 (PST), Jason Filby wrote:
>To the OSFree people (not FreeOS).. heh.. here's a proposal for
>ReactOS as the host kernel.
>
>- Jason
>
>Note: forwarded message attached.
>
Great !
Now, you people that know how to build a kernel.
Download and build this.
Tell us how complete it is and how easy it would be to
put a OS/2 layer ontop.
Everyone else, who dont know how to build a kernel and
dont know what the CPI API needs - keep quiet !
I will !
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================
Expand Messages
JMA
Feb 21, 2002
On Thu, 21 Feb 2002 05:54:23 -0800 (PST), Jason Filby wrote:
>To the OSFree people (not FreeOS).. heh.. here's a proposal for
>ReactOS as the host kernel.
>
>- Jason
>
>Note: forwarded message attached.
>
Great !
Now, you people that know how to build a kernel.
Download and build this.
Tell us how complete it is and how easy it would be to
put a OS/2 layer ontop.
Everyone else, who dont know how to build a kernel and
dont know what the CPI API needs - keep quiet !
I will !
Sincerely
JMA
Development and Consulting
John Martin , jma@...
==================================
Website: http://www.jma.se/
email: mail@...
Phone: 46-(0)70-6278410
==================================