AliCE updated

[Phantom]

Hi all,<br><br>The packages on the Alice page are updated:<br>http://www.pocketmatrix.com/phantom/alice.htm<br>Tell me what you think. <br><br>Sources are available too. It's not a pretty sight, I'm afraid; this is quite old code. The keyboard code is new though.<br><br>Jacco.

[suchiaruzu]

Cool, although I never liked your site's design. But maybe that's just me...

[Phantom]

I am not a site designer. <br>If you would like to redo it, be my guest.

[OLFID]

Perhaps I'm doing something wrong, but the MIPS version I downloaded a few minutes ago still produces an error message...

[Phantom]

I'm getting more complaints about it, and I think I know what's going wrong. I'll see what I can do tonight.

[Digby]

Some of these Pocket PC CPUs will generate an exception if you try to make unaligned accesses to memory (i.e. read/write a WORD to an odd address).  I don't know if that's what is going on here though.  0xC0000005 is an access violation, and you usually see this when you try to read/write to memory that doesn't belong to your process.  I don't know if that same error denotes unaligned accesses.  Just an idea.<br><br>

[Phantom]

Good point, but I'm afraid that's not the problem in this case. I'm very sure the ARM doesn't allow for unaligned access either, so this still doesn't explain why it doesn't work on the Casio. If you wish, we can do a debug session on monday; I'll send you a couple of AliCE executables with exit(0) at various stages of initialization, so we can pinpoint the location of the crash. I've never done that by e-mail though; I usually use this approach when I have a crash that only occurs on my iPaq, and not on the PC (VERY rare).

[Digby]

Just tried this on my iPaq.  If your code causes an alignment exception it will generate the following error message in the debugger:<br><br>First-chance exception in foo.exe: 0x80000002: Datatype Misalignment.<br><br>However, the "Fatal Application Error" dialog that is displayed on the device gives the above info along with the address where the exception occured.  If someone can give you that address, you should be able to find where the error is your code.  It will be a lot easier to find if you give them a build with debug info.  Of course if this died somewhere down in a system DLL you're going to have a tough time without a call stack finding the problem spot in your code.  <br><br><br>

[Digby]

This might make it easier for you than sending someone multiple .exes with exit(0) sprinkled throughout your initialization routines.<br><br>Inside WinMain(), bracket all your code with __try{}, then add an exception handler.  Add a global DWORD variable to your app and init to 0.  At the beginning of each of your functions increment the counter by 1.  In the exception handler, either write the value to a log file, or put up a message box and display the value.  The person running the app and getting the exception needs to give you this value.<br><br>Once you have the value, set a conditional breakpoint on the value of the counter equal to whatever it was when it crashed.  Run your app and it should break in the function that caused the error.  If your functions contain a lot of code, you can add more calls to bump up the counter variable and ask the person to run again.<br><br>Code with exception handler looks like so:<br><br>DWORD g_dwDbgCnt = 0L;<br><br><br>INT WINAPI WinMain (...)<br>{<br>__try<br>{<br>    // all your code in WinMain here<br>}<br>__except (EXCEPTION_EXECUTE_HANDLER)<br>{<br>    // save counter value here<br>    MessageBox(...)<br><br>    // exit the app<br>    return 0;<br>}<br>   return 0;<br>}<br><br><br>I've done things like this before when I couldn't get a call stack and was able to narrow down where the problem was.  What I did was actually write a _penter() routine and compile with /Gh so that the compiler generated a call to _penter() at the entry to every function.  This kept me from having to manually add the code to increment the counter variable, and was easier to remove once I found the problem.<br><br>Good luck!<br><br>

[Phantom]

Cool. I'll add this to AliCE as soon as I find the time. Thanks a lot for the info.

[jongjungbu]

The fatal application error window does not come up for me when AliCE crashes. It just exits the program completely and my Start bar never returns unless I reboot my @migo.<br>

[Phantom]

I hope the @migo problem is caused by the (0,0)=Framebuffer+319 problem. I'll post a version that fixes this on Monday (can't do that from my home). Otherwise, you'll have to wait for the version with the __try blocks.

[Rommel]

How do I turn it off and go back to the today page? Great freeware by the way.

[Phantom]

Type 'exit'.

[jlp]

I seem to have tried this program on my Casio E-125, but it didn't work either. Is it working now on the Cassiopeias?!