Windows CE [obsolete]

[zoomx]

Last visit in this forum of dboling is on Mar 18, 2013.

He has a blog too here
http://bolingconsulting.com/blog/
but last post was published on September 13, 2012

[bbking]

Windows CE has embedded market share of over 40%.

Laughable. CE has nowhere near 40% embedded market (if it has, can I have the evidence please, because EVERYTHING I have seem says CE is a dying/dead). The huge majority of embedded devices (set top boxes, TV's, mobile devices [Android] etc) run Linux.

I believe you are talking about different markets here.
There is, I believe, almost no Windows CE on the end-user devices market, which covers set top boxes, TVs etc.
There is, however, quite a few Windows CE on the corporate devices market (e.g.), those you can see in any big store, and I'm yet to find a similar device that would run Linux.

Statement was "40% embedded market share", which is clearly incorrect. Embedded market encompasses mobile phones, end user equipment, and industrial equipment. The first two are clearly much larger markets than the third, and in them Linux is dominant. That dominance is creeping in to the third area as well, due to cost - cheaper than CE - processor coverage - more processors run Linux than anything.

jamesh, where is your proof for CE not having 40% of the embedded market?
Your statements like "CE on PI wont happen, live with it" are quite strange - what's your pleasure in trying to prevent people to port CE to the PI?! If you are a linux-fan, it's fine - but let other people like whatever is their choice!
BTW, if you'd read my comment (http://www.raspberrypi.org/phpBB3/viewt ... 04#p340704) above more in detail, you'd see that there is a HUGE market for CE devices (I am working in that field, so I have seen enough). They come from, as you pointed out, the third area - industrial equipment. There is only little movement away from CE to linux - exactly because of the costs you mentioned! You can have VxWorks (linux) in that field but one licence is far beyond the price that MS takes for one CE licence! On the other hand NO sane company will deliver a device without commercial OS support behind them - imagine what happens when a bug in the linux kernel pops up and the company is not able to fix it. Whom will they ask to do so? They'd be depending on the good will of the linux community which is clearly not a business model you can build on.

So it would be cool if you could stay fact-bound and objective as well even if people have other opinions on what OS they want to see on a PI.
And two thumbs up for dboling to continue his work on porting CE to the PI from my side!

Just my 2 cents.

[jamesh]

I believe you are talking about different markets here.
There is, I believe, almost no Windows CE on the end-user devices market, which covers set top boxes, TVs etc.
There is, however, quite a few Windows CE on the corporate devices market (e.g.), those you can see in any big store, and I'm yet to find a similar device that would run Linux.

Statement was "40% embedded market share", which is clearly incorrect. Embedded market encompasses mobile phones, end user equipment, and industrial equipment. The first two are clearly much larger markets than the third, and in them Linux is dominant. That dominance is creeping in to the third area as well, due to cost - cheaper than CE - processor coverage - more processors run Linux than anything.

jamesh, where is your proof for CE not having 40% of the embedded market?
Your statements like "CE on PI wont happen, live with it" are quite strange - what's your pleasure in trying to prevent people to port CE to the PI?! If you are a linux-fan, it's fine - but let other people like whatever is their choice!
BTW, if you'd read my comment (http://www.raspberrypi.org/phpBB3/viewt ... 04#p340704) above more in detail, you'd see that there is a HUGE market for CE devices (I am working in that field, so I have seen enough). They come from, as you pointed out, the third area - industrial equipment. There is only little movement away from CE to linux - exactly because of the costs you mentioned! You can have VxWorks (linux) in that field but one licence is far beyond the price that MS takes for one CE licence! On the other hand NO sane company will deliver a device without commercial OS support behind them - imagine what happens when a bug in the linux kernel pops up and the company is not able to fix it. Whom will they ask to do so? They'd be depending on the good will of the linux community which is clearly not a business model you can build on.

So it would be cool if you could stay fact-bound and objective as well even if people have other opinions on what OS they want to see on a PI.
And two thumbs up for dboling to continue his work on porting CE to the PI from my side!

Just my 2 cents.

Well, the plethora of Android and iOS devices greatly outweighs any industrial applications running CE, so that's why the embedded market is not '40%' CE. It interesting - I actually tried to find some exact figures, and there was nothing on CE since about 2008 which is telling in itself. So that's many millions of devices running over the linux kernel. It doesn't seem to worry Samsung, LG, Apple, Google, Amazon, STB makers etc that Linux itself doesn't have commercial support (although in fact, it does as it's supported by all those companies - and there's also VXWorks, Montavista etc). Note that outside the embedded industrial space Linux also runs that majority of servers, settop boxes, TV's, and many other 'white goods'

As to CE on the Pi, as I've always said, anyone can do what they like with their PI. If that includes porting CE they should go for it. I just cannot see it happening - for a number of reasons. CE costs money, CE takes lots of support to port, market is small, end user support is limited (compared with Linux).

So, if someone wants to port CE, they should do so. If no-one does? Well, fairly self explanatory. I seriously doubt the Foundation is in the slightest bit interested in doing it themselves, as it has very little (read, no) impact on their raison d'etre.

My personal opinion is that as Linux gets more mature (it's pretty mature already) it will make faster and faster inroads in to whatever market share CE now has in the industrial sector. (Anyone know that figure? Google fails me, is it that 40% quoted?)

[startrek.steve]

Well my question is:
Why would anyone want to run ANY kind of windows on the Pi when there are many free (Superior) Linux distros for it?

[GSerg]

Why would anyone want to run ANY kind of windows on the Pi when there are many free (Superior) Linux distros for it?

The answer to that question is another question.
Would you undertake porting the millions of lines of already existing, working, correct code to another platform (e.g. Linux) and guarantee that the ported code behaves in the exactly same way?

[jamesh]

Why would anyone want to run ANY kind of windows on the Pi when there are many free (Superior) Linux distros for it?

The answer to that question is another question.
Would you undertake porting the millions of lines of already existing, working, correct code to another platform (e.g. Linux) and guarantee that the ported code behaves in the exactly same way?

Nope. Not really, although people do take on those sorts of projects all the time as platforms become obsolete. But think about what you are saying. Instead of porting that code, you want someone else to port the OS. ie save you the trouble of doing the porting. Until someone does port it, the Raspi is not an option.

The Raspi is a cheap and cheerful *Linux* machine, which has a few other free OS's ported to it (RISCOS, Plan9 etc). If you want a cheap and cheerful Windows CE machine (which I think may be 'licenceingly' impossible) then you are looking in the wrong place. You can continue to complain it doesn't run CE, or you can port CE yourself, or find someone to do it for you. That's going to cost of course, but that may be a fee you are willing to pay. I hope if someone does port it, they make the BSP widely and freely available (excluding MS licence fee) to match the ethos of the Raspi itself.

(Thought for the day. Moving to a new platform, even with the same OS is no guarantee the code will a) compile, b) work the same way.)

[GSerg]

But think about what you are saying. Instead of porting that code, you want someone else to port the OS. ie save you the trouble of doing the porting. Until someone does port it, the Raspi is not an option.

Yes, yes I realize I'm saying that. My points in doing so are:

• It is easier/better/less work/less points of failure to port just one piece of software (the OS) instead of the [numerous pieces of software minus the OS]. It's like choosing between having a NES emulator versus porting&recompiling all existing NES games as exes.
• There are people out there for whom porting an OS is fun (and porting some regular applied software is not), and they will just do it.

I completely appreciate Raspi being a Linux system and not saying that we should instead make it CE-centered, or that the community should concentrate on making me a port. I only want to support the guy who is porting CE (and from whom the public attention was distracted many posts ago) and to assert Windows CE is actually alive and wanted.

[RareHare]

Well, what are you waiting for?

Let's get this done. It's long over due.

I am actually pretty surprised no one has mentioned this operating system for the Raspberry pi, but I suppose I will go ahead and do so...

[RareHare]

I am going to do something better: port ReactOS to ARM.

You interested?

Would it be possible to run a version of Windows CE on the Raspberry Pi using some kind of virtualisation software on top of (e.g.) Linux? Using a VM could bypass the need for a native port of a closed-source OS, while still being ARM-native.

I'm not sure how this would perform, or of possible legal issues, but WinCE runs on much slower ARM-based devices than the R.Pi.

[rpdom]

I am going to do something better: port ReactOS to ARM.

You interested?

Do you mean make the existing ARM port of ReactOS run on the Pi's ARMv6?

[RareHare]

Do you mean make the existing ARM port of ReactOS run on the Pi's ARMv6?

The existing ARM port of ReactOS did not go very far. AFAIK, it barely booted. However, whatever code was written in the original effort to port ARM will certainly help.

[barbagris]

Would you undertake porting the millions of lines of already existing, working, correct code to another platform (e.g. Linux) and guarantee that the ported code behaves in the exactly same way?

+1.

I'm 100% new to Pi. In fact im still in the "try it or not" phase.

Here's what I'm after - a small, cheap and very portable platform to connect to already installed device (Vehicle ECU) on a motorcycle that ONLY comes with Windows programs. As it stands I either have to strap a netbook to the tank or connect only at standstill.

There is a wierd cross of Windows CE and Windows7 called Thin-PC. But it's obviously i386. Now if there is a way to run WINDOWS STARTER as a guest under Rasbian - that's OK. I'm assuming Virtualbox does not run under rasbian as Oracle seem only to list i386 and AMD versions.

Thankfully FTDI does ARM drivers for the interface hardware.

And no - I have neither the time or skills to cross port OS's.

[ghans]

Just to interject :

Forum member dboling has released an initial WinCE 7 BSP
http://www.raspberrypi.org/phpBB3/viewt ... 56&t=58259


ghans

[ideeman1994]

CE on Pi will not happen. Live with it.

CPUid: 0x410fb767
(Something I don't get is the build date though... I compiled it last week, must be Platform Builder acting weird)

[rurwin]

that ONLY comes with Windows programs.

Then Windows CE on the RaspPi probably won't help you. It can not run x86 Windows programs. If you have the source code for the programs and can recompile them for the ARM then you might be able to get it working. But of course Windows CE is not Windows either. It's closer than linux, but you could still have problems.

[RareHare]

Then Windows CE on the RaspPi probably won't help you. It can not run x86 Windows programs. If you have the source code for the programs and can recompile them for the ARM then you might be able to get it working. But of course Windows CE is not Windows either. It's closer than linux, but you could still have problems.

There is a technique that would allow Windows programs written for x86 to run on RPi/Windows CE. It would take less than a year to develop this technique (my estimation), and while it would not work for all Windows programs, it would work for a huge number of them (hundreds of thousands).

But as Microsoft itself wrote, "source code is eternal", so once there is an open platform of Windows on ARM, there will be a massive kick on the market where 25 years of Visual Studio experience will be brought to bear on ARM CPU's. Things will get very interesting very quickly at that point, IMO.

[jamesh]

CE on Pi will not happen. Live with it.

CPUid: 0x410fb767
(Something I don't get is the build date though... I compiled it last week, must be Platform Builder acting weird)

Well I've been proved well wrong there! Glad someone thought it was worthwhile. Not sure what the licencing issues will be. Microsoft will want their pound of flesh at some point.

[ideeman1994]

Well I've been proved well wrong there! Glad someone thought it was worthwhile. Not sure what the licencing issues will be. Microsoft will want their pound of flesh at some point.

I think it's going to be different than the other Raspberry Pi OSes available, as I'm not sure you can't download the binairy images containing Microsoft licensed code.

But you still can get the BSP, which includes the entire source code specific to the Raspberry Pi (and under Mozilla Public License, I've heard)
So if you get the Windows CE 7.0 building environment (I think Eval version works just fine, I've got a full Student version), you can build it without too much trouble...

Let's hope now dboling didn't lose interest in developing USB drivers, it would be a huge leap

[joan]

I'm not sure what has changed. Wasn't this the Window's CE status on the PI in January 2013?

[redhawk]

There is a technique that would allow Windows programs written for x86 to run on RPi/Windows CE. It would take less than a year to develop this technique (my estimation), and while it would not work for all Windows programs, it would work for a huge number of them (hundreds of thousands).

Running x86 code on ARM would require some kind of emulation layer a bit like how NT4 for Alpha was able to run x86 programs.
Even if running x86 programs on CE was possible the speed of execution would be no better than Dosbox or Qmenu (which is pretty slow to say the least).

Richard S.

[RareHare]

Running x86 code on ARM would require some kind of emulation layer a bit like how NT4 for Alpha was able to run x86 programs.
Even if running x86 programs on CE was possible the speed of execution would be no better than Dosbox or Qmenu (which is pretty slow to say the least).

Well, that's true if one uses emulation. I would not use emulation, but decompilation/recompilation.

[jamesh]

Running x86 code on ARM would require some kind of emulation layer a bit like how NT4 for Alpha was able to run x86 programs.
Even if running x86 programs on CE was possible the speed of execution would be no better than Dosbox or Qmenu (which is pretty slow to say the least).

Well, that's true if one uses emulation. I would not use emulation, but decompilation/recompilation.

Decompilation of other people software is against the law in many countries....in the States, check out the DCMA.

What might not be so clear is translation of x86 in to Arm code. i.e. what emulators do, but doing it up front rather than at execute time. This was how FX!32 (X86->Alpha) sort of worked, except its translation was done on the fly, and it stored the results so it only had to do it once. It was also able to optimise the results somehow to improve performance from the brute forced translation. IIRC.

[LeZandre]

After reading pro's and con's of porting WinCE to the RPI, I'd like to express my pro.

GPS Navigation.

Almost a decade ago I bought a MIO168, a WinCE PDA with integrated GPS receiver, and it served me well.

Four years ago I bought an OZON X700, which also runs WinCE with iGO Primo as navigation software.
I'd love to have that program running a RPI.

I haven't been able to find any decent satnav program for Linux, and Android is still as much as a no-go for the PI.

So if the WinCE port to the RPI comes to the point of supporting an USB GPS, TMC and touchscreen in a 800x480 resolution, I now what my next carputer will be!

My 2cents,

LeZandre

[jamesh]

After reading pro's and con's of porting WinCE to the RPI, I'd like to express my pro.

GPS Navigation.

Almost a decade ago I bought a MIO168, a WinCE PDA with integrated GPS receiver, and it served me well.

Four years ago I bought an OZON X700, which also runs WinCE with iGO Primo as navigation software.
I'd love to have that program running a RPI.

I haven't been able to find any decent satnav program for Linux, and Android is still as much as a no-go for the PI.

So if the WinCE port to the RPI comes to the point of supporting an USB GPS, TMC and touchscreen in a 800x480 resolution, I now what my next carputer will be!

My 2cents,

LeZandre

Anything here any good ?

http://tuxmobil.org/linux_gps_navigatio ... tions.html

[LeZandre]

None of them as good as iGO, nor any of them has updated maps.