See reply/update/thread reply –>
On 11/22/07, Alberto Moreira wrote:
> If nothing else, such a project would be tons of fun! And they would learn
> a
> great deal.
Couldn’t agree with you more - and if we incorporate we will not only learn
but, if overall successfull (which we will be), then all shareholders will
make a lot of money 
I actually don’t see much more needed than plain vanilla make.exe to build
such a project - if you need
more than that, I’ll tell you, your OS is probably more complicated than it
needs to be.
- completely agree mate ! We are sticking to stevewo’s build.exe which is
found in, well, in one of many places including \nt4\private\sdktools\build<br>
The only problem is we can not get this to work (or any other make.exe/builder
program from the public WDK - vanilla as you say - without the proper MS
internal RAZZLE build environment as there are special differences between
the public and internal build environment used to make NT/Windows/even-DOS)
Would you have any enlightening data on how to convert the public WDK (we
are using Vista-Longhorn 6000) into the proper environment used
internally…
Can you or anyone else (with proper knowledge) explain what contents a
desktop icon RAZZLE.CMD needs in it’s contents so that it will
convert/setup/align the public WDK into the state it needs to be in (RAZZLE)
so it will build proper internal NT/Windows/MS code - that is code just like
the real NT/Win sourcecode ?
I would also take the position that anything undocumented is a prime
candidate for chucking. It takes a competent programmer less time to write a
component than to reverse engineering its counterpart in an OS.
We already have reverse engineered NTOSKRNL.EXE using a control group - when
I say control group that is because we had a group of programmers split in
half and 50% of them were dedicated to those who had the real proper NT
source code (see below regarding non-disclosure agreement) and the rest had
to start from scratch - at the end of the day although, yes you are right in
saying to took longer for the rev-eng-team, but both were able to,
ultimately, patch the real source and work from scratch to produce the same
product - the other half had genuine/legally obtained from MS - both
produced good NTOSKRNL.EXE
Plus, in case of doubt, the Linux source code or even the Minix listing
embody a
wealth of information that could be of great help in such a project.
Yes, along with TinyKRNL source, ReactOS source, Bochs-X86 Source, UEFI
Source, BAMBIOS source, Award BIOS Source, and maybe even some other
low-level-based “firmware” bases or simulators - I believe the IBM based
PowerPC is going to be revived big-time in the next 10 years or less as it
is the number one processor in the supercomputer market and to answer your
next question we would like to work from existing codebases (rather than
starting from scratch) in order to make it easier to provide support for
other platforms such as the POWER architectures. The new games consoles
(Xbox 360 and PS3) both use POWER based processors and these machines could
provide a lot of extra engine on a cluster - the XB360’s processor (Xenon)
is POWER with a triple-core die and double layer - effectively given it 6
processor cores - and the CELL on the PS3 is also POWER yet uses a single
core main CPU with 8 other “synergistic processing elements” on the same
DIE… Furthermore both theses IBM POWER PROCESSORS were designed with
supercomputing in mind - in fact, IBM are making there new supercomputer -
the IBM BLUE GENE /? for IBM BLUE BRAIN project (Whole Brain Emulation) -
which, along with IBM ROAD RUNNER - will be based on a hybrid of AMD
Opteron’s x64 and CELLs in a 1:1 configuration… Anyway the reason I bring
this up is that NT4 and some of WIN2K source had support for POWER - that is
why they incorporated \nt\private\mvdm\softpc.new\ which gives NT it’s own
built-in CPU and, unlike many people realize, BIOS…
Last but not least, if I was using Linux or Solaris, I would write the whole
thing in Java, and develop a back end converter from intermediate code into
machine code prior to downloading it to the Windows target.
I want to keep pure and that is why I want to use stevewo’s BUILD.EXE and
RAZZLE.CMD - I do not want to overly virtualize with psuedo-machines and
would prefer to work directly with the machine code itself - and the BIOS
and Processor - in fact, if need be, we will create a link using some form
of special custom BUS-BUS interlink - maybe even by FGPAs or PRE-PROCESSORs
integrating the slave boxes etc… But at the end of the day I want to keep
this whole project as low-level as possible - closest to the PROCESSOR.
It would take a bunch of talented and committed programmers, but I find it
quite a feasible proposition. Still, if you want to spend all this time and
effort, why not go the whole nine yards and just write something new and
exciting ? When the best operating systems we have are basically a rehash of
40 year old architectures, maybe it’s time for a shakeup!
It is indeed time for a shakeup and I am happy to even consider creating not
just a new operating system but perhaps even a new processor !!!
First things first however - I need to get a proper RAZZLE system open that
can build Windows Source Code so if you can help get things started please
continue to read and/or post me all info you can on how to script WDK into
RAZZLE so I can build any internal source code including the holy grail -
actually build Win/NT sourcecode.
> *
On 11/22/07, xxxxx@hotmail.com wrote:
> > If you’re creating your own kernel, won’t you be creating your own build
> tools as well?
Yes we have built our own kernel but we are not at a stage where such can be
released under any licence because as stated above there is no actual
official “project” or official “member” list - that is why we must
“incoroprate” as a legal entity, and have all persons interested in
contributing, pay “something” - ie even if this is 1 share at $US1 - that
would provide a perfect foundation with legal basis that would allow members
to see such source…
In regards to build tools well, for history’s sake, we want to use the
original BUILD.EXE created by STEVEWO and use it under RAZZLE.CMD - this is
why we want to know EVERYTHING there is about how to build Windows -
specifically - how can one convert/setup/re-set the standard WDK/DDK/SDKs
(we are using Vista RTM 6000+ because it includes IFS, HAL, and normal
NT4/2000/.NET (and Vista/Longhron) build environments…
Now, yes, we have completely reverse engineered NTOSKRNL.EXE and
BUILD.EXEbut can not work out how to bring up a proper window, under
the WDK 6000 RTM
- from connect.microsoft.com - that is genuinely a RAZZLE build
environment…
When I say genuine I mean there are differences between WDK/DDK build
environs and the true MS internal used RAZZLE –>
Can anyone tell us how to setup a RAZZLE.CMD script that will bring up a WDK
window that would be capable of building REAL MS Windows/NT Source Tree Code
???
- the following is appended in ascending order from the first reply -
please continue to read this to to the end:
Dear all interested parties,
First of all my group and I are very happy to see there is a statistical
significant, legally valid, and unified base quorum of parties worthy of
giving support AND constructive criticism toward this project; excuse my
humour but in all seriousness I am very happy to see the positive responses
given my persons from OSR and many other groups/threads/forums which show
support toward the development of a version of Windows/NT/hybrid (ie
including perhaps TinyKRNL & ReactOS bloodlines) which ultimately provides a
genuine Microsoft operating system albeit with a custom Kernel that allows
many boxes (ie say 10 for home/testing/beta use in Phase 1) to be clustered
together in such a way that the end server terminal (and end user) are
presented with a GUI/Shell/Console that “thinks”/“sees”/“realizes” the
resources (processor, RAM, storage et al) of all other slave boxes as being
LOCAL MOTHERBOARD RESOURCES in order to provide a true NT/Windows/MS
operating system (genuine) albeit with REAL clustering - that is, instead of
the orthodox supercomputing , distributed/cluster/grid/matrix, interprocess
communication that we know - but a true/proper single system image.
With such in mind perhaps the best way to go should set up a proper
project whereby such could be managed in a sourcedepot fashion.
Another thing to take into mind is that by asking persons to make a 1 share
(ie US$1 initially) will lead to a more structued/organized/democratic and
ordered project.
Each contributing-coder/member/potential-member (including you OSR members
etc) would make input as to how this is organised and then legally put in to
place.
Once we organise/setup/manage such I feel the project will take off like no
other GPL type project ever has in Internet history - this would be the
first true team studio developed commercial enterprise in Internet history -
especially when it comes to creation of an operating system.
All I am saying is that there should/must/will be a set of
rules/constitution/licensing binding all parties to the
incporated/registered entity; this includes a new domain with membership
and upfront share payment for those who have the faith something will work -
this is because at the end of the day everyone involved from zero hour will
be rewarded (at final RTM) with cash.
Or should we just begin by using standard GPL ?
Personally I believe the above question is rhetorical.
Anyway, with this in mind I must make it clear that until we have enough
members- ie 500+ - and have incorporated in such a way (ie as a co-operative
research centre / co-operative with government guaranteed shares to be
eligible for government funding) then there will be problems down the track
when we decide this operating system hybrid is ready for proper release to
manufacture.
So let’s start brainstorming now - as you already are - thanks OSR et al !
In summary - we need this sort of brainstorming (just like these -emails -
with people on OSR and other sites) in order to decide how the project is to
operate on a legal, sourcdepot, and team-studio-type operation.
Again, my answer to this is that anyone interested in assisting with the
project (as coder/member/investor) should first sign-up for membership to a
pre-organised legal entity (ie the co-operative) so that there is a
constituion in place to protect the project, members, and source.
This membership system would then work on a shareholder basis so that, down
the track and at completion of final phase, all those people involved in the
creation of the operating system, will receive a dividend
(equal/proportianite/respective to their input from code writing,
share-capital-payment, and brainstorming/general-input - as OSR members are
doing now).
This way all persons involved in the project will receive money (cash money
profits) from the sales, distribution, and RTM of the final build.
In summary I think the most important thing to first considering is how such
a project should be organised legally because I want to deviate from
standard “benevelont-free-work” principia and by doing so all people who
contribute will receive, in turn for their coding efforts, well, money…
This can all be worked out by this sort of brainstorming (happening on OSR
right now in these e-mails etc).
Anyway, here are the answers to everyone’s questions - this tradition of
dialectical materialism, where thesis VS antithesis = synthesis,
should produce one single & perfect grand unified theory (GUT) –>
once we have our GUT and democratically voted/survey assessed/quorum
determined set of protocols/constitution/rules, we can proceed - but first
things first: in my humble opinion this is setting up a brand new domain
domain under a brand new entity which all persons of interest will sign-up
to via making a pledge of shares (which can be anywhere from $1 at the start
–> and extended up to any amount - even $1000 + minimum sharehold -
depending on the success of the project as determined by our internal
democratice purpose and perhaps outsied beta testing/evaluation by select
groups of persons/companies - this might also include an external
feasability assesment from an accounting, IT, and legal firm which I am
happy to pay for along with the initial company/association/co-operative
incoropration as yes - I HAVE THE FUNDS AND SEED CAPITAL TO BRING THIS
PROJECT TO LIFE)
Ok, now I will answer the technical answers everyone has asked - I just
wanted to make it clear that if something like this is going to work we need
to setup a complete new project with domain name (and sourcedepot system)
that is managed by a domain name owned by every single member - via an
incorporated entity to which each member is an equal member relative to
their share value paid - this solves the ultimate goal of marketing our
final operating system as I want EVERYONE to get paid for their efforts -
unlike most GPL projects - in the end.
[CONTINUATION FROM xxxxx@hotmail.com]
Another interesting question is how this custom kernel/HAL is going to
interact with the existing drivers (?)
We have reverse engineered the proper and original NT/Windows Kernel(s) in
such a way that there will be no problem having this NTOSKRNL.EXE work with
existing and genuine drivers…After all, the guy made it clear that he
does it for WinNT - apparently, he plans to provide his own implementation
of NTLDR, ntoskrnl.exe and HAL.DLL , but all vital drivers are still going
to be MSFT -
It may or may not be necessary for us to use customized HALs, OSLOADER.EXEs,
or any other components - this is why we are brainstorming with you now - it
may after all be quite possible to do everything by incorporating
everything we desire/need/have-already-done (including our special
unviersal intercommunication system between motherboards) by putting all
this into NTOSKRNL.EXE…
Taking into consideration the fact that most ntoskrnl.exe’s and HAL.DLL’s
exports are undocumented (which means they are meant to be called only by
MSFT-provided drivers), he is going to face HUGE problems when it comes to
implementing those undocumented functions that are callable by MSFT -
provided drivers - if he wants to integrate his custom kernel with the
existing drivers, his custom exported functions have to act exactly the
same way the original functions do, as far as client callers are concerned.
I am afraid that, unless the guy has an access to ntoskrnl.exe’s source, his
task is practically infeasible - ntoskrnl.exe is just a bit too large for
being completely and thoroughly disassembled within a reasonable
timeframe…
To answer your question, without breaking any non-disclosure-agreements
signed, yes we have reverse-engineered the components required to make this
project MORE THAN FEASABILE but we are simply having problems with the
simple side of things - we can not get these things to build under the
standard WDK/DDK !!!
As opposed to the public used WDK build env, we need to know how to setup a
proper RAZZLE build environmnet as used by MS internally - so if someone
could simply explain how to setup a build environment - using the WDK or any
other DDK/SDK - we would be happy; this is all we need at this stage - the
details of how to convert the existing public WDK/DDK/SDKs into RAZZLE.
When I say RAZZLE we mean a build environment capable of building true
Windows Source Tree/Base Code…
So basicaly we are asking outright for anyone’s expertise in order to asnwer
this question - how does one setup a public WDK/DDK/SDK (particularly
Vista/RTM 6000) into RAZZLE - so that it will properly work in the exact MS
way to build real Microsoft Windows/NT sourcecode ?
PS:
Yes, we have signed a non-disclosure agreement with Microsoft but this is
what prevents us from explaining why/why not we have certain things- all I
can say here is that we are authorised to do what we are doing and that
Microsoft have provided us with the Win/NT sourcecode.
