Platform::COMException ^ at memory location 0x0669FAB0. HRESULT:0x80070490 Element not found.

OK, I'll try that. Uh, it didn't throw the usual exception this time. It said it had a stowed exception. Should I until, clean, & run?

Also, I will submit with Unity Bug Tool & post the number I get here.

 

Case Number is 661462

 

Well that's one reason I clean installed Windows 8.1, used 64 bit Windows and have only installed iTunes. You know how every SW maker just has to have their own terminate & stay resident helper app but...

I have 4 GB RAM of which Windows has always said from the beginning only 3.79 GB are available because of the shared memory of the Intel HD Graphics on board I guess. I have paging active and I have hibernation active.

Only thing running that I would have started manually are Unity Free 4.6.1p2 and VS Ultimate 2013. I get the same results if I close Unity or I close VS 2013 or close both and only run installed app.

I think low virtual memory is actually is a stack is being blown. If I were to 'continue' with VS2013' I'd get the blown stack. There seems to be a recursion that doesn't have it's stop condition met.

I can run the app (it's Blank) with IE and everything closed too if you like, that would give it about 3.25 GB to use and send that crash dump to you.

I get these now with it from the event log with only IE & the App 'Blank' (Blank will cycle crash') running.

Code (csharp):

1.  
2. Faulting application name: Template.exe, version: 1.0.0.0, time stamp: 0x54ad4416
3. Faulting module name: combase.dll, version: 6.3.9600.17031, time stamp: 0x53086d7c
4. Exception code: 0xc000027b
5. Fault offset: 0x000fb1d7
6. Faulting process id: 0xbd0
7. Faulting application start time: 0x01d02e7f00735d15
8. Faulting application path: C:\Users\Jack\Downloads\Blank\Blank\Blank\bin\x86\Debug\AppX\Template.exe
9. Faulting module path: C:\Windows\SYSTEM32\combase.dll
10. Report Id: 41968213-9a72-11e4-8257-c48508a74fc1
11. Faulting package full name: Blank_1.0.0.0_x86__pzq3xp76mxafg
12.  

Code (csharp):

1.  
2. Fault bucket 98906562423, type 5
3. Event Name: MoAppCrash
4. Response: Not available
5. Cab Id: 98926025917
6. Problem signature:
7. P1: Blank_1.0.0.0_x86__pzq3xp76mxafg
8. P2: praid:App
9. P3: 1.0.0.0
10. P4: 54ad4416
11. P5: combase.dll
12. P6: 6.3.9600.17031
13. P7: 53086d7c
14. P8: 80070490
15. P9: 00053bff
16. P10:
17. Attached files:
18. C:\Users\Jack\AppData\Local\Temp\WER7420.tmp.WERInternalMetadata.xml
19. C:\Users\Jack\AppData\Local\Temp\WER7654.tmp.appcompat.txt
20. C:\Users\Jack\AppData\Local\Temp\WER76D2.tmp.dmp
21. C:\Users\Jack\AppData\Local\Temp\WER7924.tmp.WERDataCollectionFailure.txt
22. These files may be available here:
23. C:\Users\Jack\AppData\Local\Microsoft\Windows\WER\ReportArchive\AppCrash_Blank_1.0.0.0_x8_4b6c807e3506ee9b86f361dc42cff2a8aba16f1_ccba75d4_cab_3ce27e32
24. Analysis symbol:
25. Rechecking for solution: 0
26. Report Id: 41968213-9a72-11e4-8257-c48508a74fc1
27. Report Status: 8
28.  

 

OK, here is mine and I currently have more 'currently allocated' although I doubt very much I have more thing running than you currently. The only non-windows thing that is TSR are the Bonjour (0.8MB), MobileDeviceService (0.8MB), iPodServiceModule (0.8MB), iTunes Helper (0,4MB),ThrottleStop (0.6MB), & VirtualCloneDriveDaemon (32 bit) (0.2MB).

As you see not much at all and all but Throttle Stop & Virtual Clone Drive are part of iTunes. All the other background process came via Windows 8.1 or VS 2013 install and Windows Update.

Among all those windows updates are a slew of ASUS supplied services which I don't have much confidence in given the speed at with they abandoned support of this Microsoft Signature HW, e.g,:

There is a 'hacky looking' TSR that was installed by Microsoft Update supplied by ASUS: "Fix Win8 TP issues-final version (32 bit)" (at 0.6MB)


Does maybe now unity & VS 2013 together require 8GB RAM? However I hope plain Windows Store apps can handle only 1 GB RAM for customer with WinRT machines.

 

Hey,

That is OK if you can't fix it. I may be the only person with this HW actively using Unity. I plan to replace it with new HW later in 2015 and then this was going to be a test only tablet but since I've seen WinRT tablets for less than $100 that shouldn't give me this problem I'll probably buy one of those too.

I have all updates installed via Windows Update.

I have .Net 4.5.2 installed from Windows Update too.

Thanks.

 

My EP121 has the same issue, see
https://github.com/MSOpenTech/cocos2d-x/issues/116

Regards

 

Interesting. Maybe Unity can make an easy kludge now.

Is the Chipmunk test supplied as example by MS? Link?

I will be clean installing the latest Windows 8.1 x64 + SP1 early next week. I will try to test situation again before the ASUS supplied drivers to MS are installed.

I think the HW drivers are wrong but I'm not keen to download 10GB HW test suite from MS to test the EP121.

 

The Chipmunk test is part of the Cocos2d-x test suite.
I recommend the MS example
https://code.msdn.microsoft.com/windowsapps/Accelerometer-Sensor-Sample-22982671
for quick tests.

 

Thanks. I got that one.

No surprise there:

Code (csharp):

1.  
2. System.Exception was unhandled by user code
3.   HResult=-2147023728
4.   Message=Element not found. (Exception from HRESULT: 0x80070490)
5.   Source=Windows.Devices
6.   StackTrace:
7.   at Windows.Devices.Sensors.Accelerometer.put_ReportInterval(UInt32 value)
8.   at Microsoft.Samples.Devices.Sensors.AccelerometerSample.Scenario1.ScenarioEnable(Object sender, RoutedEventArgs e)
9.   InnerException:
10.  
11. ...
12.  

 

Yes, and changing the ReportInterval = calls like Dale suggested will make it work. (see e. g. https://github.com/MSOpenTech/cocos2d-x/issues/116#issuecomment-70337804)

 

Thanks. After I clean install I will check the generated files from Unity / Visual Studio and see if I can change it or tack it in. Maybe Unity can add that in so it will be generated at the proper depth.

But before I do that I'll do Dale's fix so you will know if that 2nd problem is a device driver problem or the coco2d API port problem.

 

OK, I need some help with Dale's pseudocode:

Code (csharp):

1.  
2. if (_accelerometer !=null)
3.    {
4.    // Establish the report interval
5.  
6.    try {
7.         uint minInterval =0;
8.         uint reqInterval =0;
9.  
10.         minInterval = _accelerometer.MinimumReportInterval;
11.         reqInterval = _desiredReportInterval;
12.         _accelerometer.ReportInterval = reqInterval < minInterval ? minInterval : reqInterval;
13.         // _accelerometer.ReportInterval = _desiredReportInterval;
14.         }// try
15.    catch (Platform::COMException^)
16.        {
17.  
18.        /* Something went wrong, continue */
19.        CCLOG("Device::setAccelerometerInterval not supported on this device");
20.        }// catch
21.  
22.    Window.Current.VisibilityChanged += newWindowVisibilityChangedEventHandler(VisibilityChanged);
23.    _accelerometer.ReadingChanged += newTypedEventHandler<Accelerometer, AccelerometerReadingChangedEventArgs>(ReadingChanged);
24.    ScenarioEnableButton.IsEnabled = false;
25.    ScenarioDisableButton.IsEnabled = true;
26.    }// if
27. else
28.    {
29.  
30.    rootPage.NotifyUser("No accelerometer found", NotifyType.ErrorMessage);
31.    }// else
32.  

1) You send a float for interval to that cocos2d function what value did you set it too? Is it a constant, more or less?

 

For your 2nd problem don't forget to check your 'Rotation Lock' button on the top, closest to center, in landscape mode. I never have interest in it til recently. Changing that doesn't stop the exception but it might solve your second problem.

https://ep121.wordpress.com/2011/01/08/the-ep121-product-tour/

 

OK, I clean installed Windows 8.1 * VS 2013 from latest MSDN ISOs to be sure...

And as you see above I added Dale's changes but that have no effect in the example:

https://code.msdn.microsoft.com/windowsapps/Accelerometer-Sensor-Sample-22982671

From the example I could 'Enable' 'shake' without an exception but not 'polling' or 'data events' - those both caused the same error.

By the way, although I could enable shake (no external monitor, USB devices, or external power supply) the example program did not count any shakes when I tested. It stayed at 0.

I can tell you having an external monitor connected disables Autorotate as you'd expect. Not so with external PS or USB devices because I suppose it with be too limiting to one using the tablet.

There are several problems it seems they aren't handling some of them on EP121 and similar HW :

1) External Monitor connected: Autorotate (Accelerometer) off regardless of Autorotate button but touch can stay on if on.

2) No external monitor - Autorotate on depends on the position of the autorotate button but touch can stay on if on.

3) HW show software keyboard - would be nice if such a HW /SW button could be used in Unity app like with iOS & Android? Can it yet?

4) HW / SW disable touch - makes no sense really - if you have touch keep it on. Exceptions should be done by administrator.

5) 1 - 4 with external power supply (safety improvement - poll and warn if trying to use autorotate)

6) 1 - 4 with connect USB devices (safety improvement - poll and warn if trying to use autorotate)

So the drivers have to handle these cases to give the proper perspective of the OS state to Unity or CoCos2D. I'm not sure if they are, they seem not to completely.

I'm not sure why the code Dale gave fixed the example for you but not me. Did you use only in cocos2d Chipmunk example or both examples?

 

This is for the Unity guys - I'm supposing if I was to buy a WindowRT tablet and publish to that then this problem would go away but how to publish to Windows Store for Windows 8.1 x86 without buying a new convertible tablet PC?

Create & Save to VS 2013 Project in Unity using Windows Store -> Universal 8.1, test on the WinRT tablet, and then exclude the EP121 from the compatible HW list?

Maybe I need to talk to MS support instead.

 

The code applies to the C++ version of the MS example, and it worked for me. I just applied it to the relevant instances of "ReportInterval =", e. g.:

try { // added try
accelerometer->ReportInterval = desiredReportInterval; // this is the original line
} catch (Platform::COMException^) {} // added catch

 

I downloaded the C# version...what I've done should work too as I looked at Dale's code in cocos2d tree. Seems there is a subtle difference in the way C++ & C# are compiled to binaries and they don't behave the same. I will download the C++ version of the project.

 

OK, I downloaded and tried the try/catch code you gave and it didn't work for me.

Now I'm not current on state of the art compilers and build in test initializations, object and memory management, and diagnostics but both the try/catch solution you gave and Dale's if/else C# cocos2d solution seem to be redundant tests and both fail on my computer.


a. accelerometer object must be created

&

b. accelerometer->ReportInterval object must be created after the accelerometer object is created

but since accelerometer->ReportInterval is just a simple uint wouldn't assignment be enough once the accelerometer object was created?

I also AND-ed a check of accelerometer->ReportInterval with accelerometer to no effect in the C# code.

I'm not sure really except the C++ conveniently cause VS to bring up the header containing the core exception code in a switch statement & I can see this error as VS reports falls through to default for the user to handle.

 

OK, I've done a bit more work and again it point to access permission or lack of them.

look at this code:

Code (csharp):

1.  
2. // Establish the report interval
3. try
4.      {
5.          
6.      rootPage->NotifyUser("accelerometer->ReportInterval = "+accelerometer->ReportInterval, NotifyType::ErrorMessage);
7.      accelerometer->ReportInterval = desiredReportInterval;
8.      }// try
9. catch (Platform::COMException^)
10.      {
11.      /* Do Nothing */;
12.      }// catch
13.  

The exception here still gets thrown at:

accelerometer->ReportInterval = desiredReportInterval;

not at:

rootPage->NotifyUser("accelerometer->ReportInterval = "+accelerometer->ReportInterval, NotifyType::ErrorMessage);

because that is read access to the field, the exception gets thrown on an attempt to write to that field.

I guess I need to figure out how to access this definition:

accelerometer->*

and wonder if ASUS made it super restrictive in their implementation of the sensor drivers???

I will look through my old emails and see if I can find the email from the ASUS engineer. They don't want to implement it wrong again on their new HW.

I will try the Win32 build now & see if that works.

 

OK I tried Win32 and no surprise same problem.


I did check the MS definition and here it is:

Code (csharp):

1.  
2. public : property unsigned int ReportInterval { get; set; }
3. Member of Windows::Devices::Sensors::IAccelerometer
4.  

and it is public with get & set methods.

Unless you or Unity has suggestions I will contact ASUS and see what they say. I still don't understand why it works for you now but not me.

 

You're right. The exception happens on "set" (ReportInterval=), not on "get" (x=ReportInterval). And yes, this seems to violate the current MS specs.
Regarding the fix:
You should be aware that in Debug mode the exception is still reported as "1st Chance Exception", which is purely informational. After continuing (ignoring the exception) it should be swallowed by the catch and the example should work. That's what it does for me.

 

OK, before I send an email to ASUS I commented out the assignment line so the code looks like this:

Code (csharp):

1.  
2. if (accelerometer != nullptr)
3. {
4. // Establish the report interval
5. try
6. {
7. rootPage->NotifyUser("accelerometer->ReportInterval = "+accelerometer->ReportInterval, NotifyType::ErrorMessage);
8. // accelerometer->ReportInterval = desiredReportInterval;
9. }// try
10. catch (Platform::COMException^)
11. {
12. ;
13. }
14.  

and now the code runs and reports

accelerometer->ReportInterval = 125

while once the 'Data Events' are Enabled the X, Y, & Z vary swiftly from [-0.0[0-9]*, -0.0[0-9]*], [-1, -0.9*], [-0.1[0-9]*, -0.2[0-9]*

I guess I'll ask ASUS now to look. I could be the ASUS decided to lock down that report interval in their driver as that seems reasonable to keep from polling more than the HW/SW can handle but excludes cases were you want to decrease polling rates but then that's handled by MinimumReportInterval.

I think the definition was changed after ASUS published their driver.


Thanks for your help.

 

Thanks. I will continue then after I uncomment the commented line.

OK, so that works until you 'Disable' and I'm now downloading symbols from MSDN as that seems to be another symptom of the same problem.

I'd like to just comment out that line or test for HW but that's not really practical for Unity to do. I will write ASUS.

 

"Disable" simply does another failing call: it sets ReportInterval = 0. Same problem, same workaround.

 

Thanks I'll look at that after I finish my eMail to ASUS since I'm not too hopeful of Unity including a set of workarounds for 2 1/2 year old HW drivers. I need to get into the VS/Unity generated code and see if I'll have access to it.

 

How to you do

try / catch in .Net & C#?

I have but it's wrong:

Code (csharp):

1.  
2. try
3. {
4.  
5. _accelerometer.ReportInterval = _desiredReportInterval;
6. }//if
7. catch // (IOException e) <- Error
8. {
9.  
10. // Something went wrong, continue - Platform::COMException^
11. rootPage.NotifyUser("Platform::COMException - _accelerometer.ReportInterval = "+_accelerometer.ReportInterval, NotifyType.ErrorMessage);
12. }// else
13.  

I just did a catch but I'd like something more refined in C# too.

 

OK, thanks again.

It makes more sense now, maybe I'll get the hang of tracing in VS / C# / .NET. One thing that it's takes some getting used to as one that primarily programmed in C in the past is remembering these odd error messages with private, public, and so on in these OO languages. I only ever did one major OO implementation in the past and I decided to do that with porting C code to C++ templates rather than convert to classes. So not much to that and not having the source code takes getting used to too.

I can't fix the exception in any Unity/VS generated code I have access to but one of the places the exception happens is:

MTSensorInput<Windows:evices::Sensors::Accelerometer,Windows:evices::Sensors::AccelerometerReadingChangedEventArgs>::InternalUpdateIntervalImpl(void)

and here is the set:

07BCE8AF E8 CE 92 F9 FE call Windows:evices::Sensors::IAccelerometer::ReportInterval::set (06B67B82h)

That's only one, there are likely more calls.

Unity knows about this already and I opened a request for a new G-Sensor driver with ASUS a few hours ago. The best would be for ASUS to update the driver but in your case at least you have access to update the offending code with a try / catch.

 

The ASUS technical support says ASUS does not support the EP121 past Windows 8.0 so they supported that tablet approximately 1 1/2 years which is extremely poor track record.

 

No, thanks. I've done everything I can do. I just need to check the Windows 8.0 definition of

Code (csharp):

1.  
2. public : property unsigned int ReportInterval { get; set; }
3. Member of Windows::Devices::Sensors::IAccelerometer
4.  

I will buy a cheap Windows RT tablet and then decide later if I want to buy a cheap new model laptop or still buy an new 2015 model ASUS PC tablet.

MSDN says that the ReportInterval should be read/write in Windows Store 8 / Windows Phone 8 clients as well.

 

OK, if anyone cares to advise on using the iView 7" Tablet 16GB Quad Core Windows 8.1 as I'll be buying one soon

http://www.walmart.com/ip/40703435

with Unity 4.6.x and Windows 8.1 Universal builds as it looks to be the cheap WinRT equivalent of the cheap Android 4.4.2 NextBook tablet I bought recently

http://www.walmart.com/ip/Nextbook-7-Tablet-16GB-Quad-Core/38334381

The CPU is this: 1.33GHz Intel Atom Z3735G-CR Quad-Core processor so when I deploy to test directly on the device what do I choose? How does an Atom Z3735G-CR Quad-Core compare to the i5-470UM in my EP121? It seems they are both quad core and 64 bit with the Atom using less power.

x86 I suppose for now but what of when Unity 5 comes along? Can I choose to deploy as x64?

If any GPU knowledgeable folks could comment on the Intel Graphics HD GPU in the iView vs the MP450 Quad Core GPU performance wise I'd appreciate it. Are the 2 roughly equivalent handling openGL ES 2.x applications?

It looks like it turns out that this iView is a very cheap replacement of my EP121 as a gaming tablet.

Thanks.