ReactOS: Execute Natively or Virtually?

[J. C. Jones]

Hi,

I had forgotten all about BOCHS. It is amazing how many emulators there are on the market
these days. 

I think that, for mass appeal, we should try to get ReactOS to run natively. The problem
is that there are basically two types of users of ReactOS:

1. Those who enjoy OS development.
2. Those who do not enjoy OS development. :)

#1 is the type that might run ReactOS inside an emulator.
#2 is the type that would probably never run ReactOS inside an emulator, even if ReactOS
were stable.

Unfortunately, for all the people who might run ReactOS someday [millions and millions of
people] the ratio #1/#2 is probably less than 1/100, conservatively speaking.

Therefore, it makes sense to get ReactOS to run on as many of the devices that users of
type #2 would use. That leads to the following reasoning:

There are several popular CPU's on the market. X86_32/x86_64 rules the desktop. ARM is
90% of mobile devices. And there is the "other" category, which is less than 10%
of all new devices, including mobile.

Since it is not (yet) likely that a user of type #2 will remove Microsoft Windows from an
existing device and replace it with ReactOS, even if ReactOS were stable, one must
conclude that the best way to get ReactOS onto as many devices as possible is to make
ReactOS the first OS on the device before any other OS has a chance to be on the device.
That means new devices.

New devices come with Microsoft Windows, iOS, or Android. Getting HP, Dell, ASUS, Acer,
etc..to start using ReactOS instead of Microsoft Windows on their hardware is an up-hill
battle and not likely to succeed. Apple hardware, of course, is out of the question. That
leaves Android hardware, which is more vulnerable, IMO, than people might think. Google
does not own these devices. As Timo showed, there are manufacturers in China/Taiwan/Korea,
etc. that make Android hardware very cheaply. Those manufacturers choose Android, but they
could choose whatever they desire.

So I think that getting ReactOS to run on just one, commonly-available, mobile device,
like an ARM-based SmartPhone, that is made by a hardware manufacturer who is not committed
to Android or any other OS, would open the flood-gates.

I would not worry about the number of available "apps" on ReactOS on ARM,
because there are many, many developers who know how to write applications for Microsoft
Windows  on x86_32/x86_64, and therefore, for ReactOS on ARM, and they will be immediately
attracted by the open nature of ReactOS. Also, I suspect that the following technique for
converting Windows-on-x86_32 applications to ReactOS-on-ARM will work:

http://www.reactos.org/forum/viewtopic.php?f=9&t=12262

Cheers!

-John


-----Original Message-----
From: ros-dev-bounces(a)reactos.org [mailto:[e-mail]] On Behalf Of
HeChi-Lau
Sent: Saturday, January 19, 2013 10:20 AM
To: ros-dev(a)reactos.org
Subject: Re: [ros-dev] Ros-dev Digest, Vol 101, Issue 30

Hey，Do you know the BOCHS？An x86 emulator.Its android version works fine through the JNI
library.Reactos also runs okay.But I configured the net card in conf file,nothing
effected.
In my opinion if anyone could develop a similar,but advanced software:it included full
hardward support and released with reactos.Or just add the x86 vm or  in reactos's
kernel,and then reactos can work with arm cpus.After one of I said above done,I believe
more people will know reactos and use it.
reactos has some advantages that other os can't compared with.such as free,fully
opensource,pe compatible,low memory usage and fast boot and shutdown.If application
experience be better，it will be finer.

[Bernd Blaauw]

Op 19-1-2013 18:46, J. C. Jones schreef:
>   Hi, 
Per email thread subject, I'd say it would be nice if ReactOS loads at 
all, critically only depending on a minimum set of requirements and 
having everything else loaded/working if possible, without ever hanging 
or stalling the loading process.

All that's necessary is system memory and finding ReactOS files. Having 
output devices like screen/printer is usefull ofcourse, but not critical 
for loading ReactOS. Same for input devices (mouse/keyboard) and the 
subsystem they are on (USB).

However, currently ReactOS is stalling easily when certain system 
hardware or external (USB) devices are present. Not even working on x86 
in a general way, might not encourage going beyond x86.

>   I had forgotten all about BOCHS. It is amazing how
> many emulators there are on the market these days. 
Each ReactOS LiveCD ISO-file is inside a 7z-archive, such an archive 
supports including multiple files (unlike XZ for example) so adding 
Bochs would be easy and limited overhead (bochs.exe + vgabios + bios 
together below 1MB, if Bochs.exe gets UPX-compressed).
Doubleclicking bochs.exe or a batchfile to run the LiveCD inside Windows 
would be most convenient.

>   I think that, for mass appeal, we should try to get
> ReactOS to run natively. The problem is that there are basically two types of users of
> ReactOS: 
Would you require proof of ROS working as good as possible inside 
emulators first? Or try anyway on native x86/x64 and other architectures?

>  
> http://www.reactos.org/forum/viewtopic.php?f=9&t=12262 
Linux/Android have the advantage of source-level drivers that can be 
compiled/ported for multiple architectures. Windows seems to have 
x86/x64 closed-source drivers, on which ReactOS would also depend. Thus, 
porting to other architectures is impossible. That is, unless ReactOS is 
intended to develop their own (accelerated) drivers for hardware on 
multiple architectures.

Anyway, if you manage to get ReactOS on mobile (ARM?) devices, all the 
better ofcourse. It can only benefit the project.

Bernd