Printable View
I'm trying to use ReturnFloat() but for some reason, it crashes. Are there any specific reasons it would crash? Probably something common that I don't know about.
If you want some of the code I got plenty here. I get no warnings or errors on compile.
You need to set the flag first. I'm using this:
// Set the "Float" flag
rdPtr->rHo.hoFlags |= HOF_FLOAT;
// Return the float
float fResult = 0.2123;
return *((long*)&fResult);
rSDK ReturnFloat() sets the float flag.
Yes, I am using the rSDK, sorry I forgot to mention that.
I'm using the rSDK as well.
So you mean I should avoid ReturnFloat() and instead do it the way you posted?
Mokhtar, I did it the way you suggested, but it did not make any difference. It still crashes, same way, same spot.
I made a test expression that was just a ReturnFloat(0.123456789); and it did not crash; it worked fine. (It returned 0.123457)
So why else would it crash?
Maybe it's not the return that's causing this. Perhaps you're accessing an array index out of bounds?
I think it must be corruption. I put debug messages and it doesn't crash until the ReturnFloat() line, so I must be giving it some bad data. I haven't changed the code much since the pastebin I linked above..
Is this the PC of yours with the bad ram? Because bad ram can cause some really fun errors...Quote:
Originally Posted by LB
Instead of doing debug messages, have your tried actually debugging your extension? It is described in the extension SDK help file how you do it. It really helps finding errors.
He has, but he doesn't get it to work for whatever reason...
Hi,
looks like a index out of bounds or casting
try float RetVal = (float)((*rdPtr->Points)[At].X);
just in case you are using VS2010 debug the previous line to check if a float value is present
Regards
@Dynasoft: No it got so bad I couldn't use MMF2 anymore. I am on the other computer in our house (It's an annoying vista laptop, but it works and it's fast)
@Andos: I did read the instructions and followed everything to a tee, but any breakpoints I set give me a message saying they won't be met and then they indeed never get met. I tried every file, both CPP and headers. Not a single line would let me set a working breakpoint. So, I had the application running from the debugger mode but that was it.
@FVivolo: Why do I need to cast it to a float if it is already a float? I'll try anyway, just in case.
If you are using VS2010, it did nasty behavior to me so i cast all the variable i need and it works, if you ask me why i donīt have a straight answer but it works for me, hope solve you situation too
Regards
Were you building in release mode?Quote:
Originally Posted by LB
No, I was building with debug. I made sure of it! ;)
Is the debug configuration set to build to the right place?
LB,
below some code you may add to help you to get some info during execution
in Common.h
extern void Msgx(char *szFormat, ...);
in General.cpp or Main.cpp
void Msgx(char *szFormat, ...)
{
char szBuffer[512];
va_list pArgs;
va_start(pArgs, szFormat);
vsprintf(szBuffer, szFormat, pArgs);
va_end(pArgs);
MessageBox(NULL, szBuffer, "Testing", MB_OK);
}
and finally use like this in your code
Msgx("My Result: %10d "%s %4.2f hex=0x%x", decimal, string, float1, hexa);
so you may track the index and the corresponding value
hope its help
This message appears if you are not building in debug mode.Quote:
Originally Posted by LB
If you are, you might not have set up debugging correctly
It will also appear if you try debugging before the MFX is loaded. I find it easiest to debug by attaching to edrt on a dummy frame and then advancing.
Just use CopyRun.bat and the /f command line parameter to edrt.exe..Quote:
Originally Posted by Retriever2
I fixed the crash by doing this another way, but now I have a new problem.
I switched to using a map instead of a vector, and I used a struct (ActionpointID) for the key, and a struct (Actionpoint) for the value. This works great and ReturnFloat() no longer crashes (nothing is crashing).
But, when I try to get the X position, it returns the Y position; and when I try to get the Y position, it returns 0. I can't figure out why it's doing this, so here's what I have:
Actionpoints.h
http://paste.pocoo.org/show/255700/
This file is included in Common.h
In Common.h I included <map> and am no longer including <vector>
In Data.h, I changed the declaration of Points to:In Runtime.cpp, I changed the creation of Points to:Code:map<ActionpointID, Actionpoint>* Points;
Deleting it is the same.Code:rdPtr->Points = new map<ActionpointID, Actionpoint>;
And in Main.cpp I have:
http://paste.pocoo.org/show/255703/
(for reference, here is ParamHelp.h which is included in Main.cpp)
You are retrieving the coordinates wrong. When you retrieve a rotated coordinate the resulting X position will depend on both the original X and the original Y coordinate. Try using this instead:
You might have to change some signs though, I'm not 100% sure I got it right.Code:// X
float RetVal = (*rdPtr->Points)[ID].X * cos(Angle);
RetVal += (*rdPtr->Points)[ID].Y * -sin(Angle);
RetVal *= Scale;
RetVal += HotSpot;
// Y
float RetVal = (*rdPtr->Points)[ID].X * sin(Angle);
RetVal += (*rdPtr->Points)[ID].Y * cos(Angle);
RetVal *= Scale;
RetVal += HotSpot;
I was just using angle 0 and scale 1.0 and hotspot 0.0 for testing. But your explanation would explain why sin() and cos() were taking radians and returning degrees...
Upon more testing, I found that it's not just the X and Y that are messed up, it's everything! It's all in reverse order!
For some reason, they are still being set and accessed as the correct types, but the order is reversed! I can't figure it out. Why would it flip the variables around like that?
So, I have made an expression to get all the actionpoints as string. Then,I added two actionpoints.
The expression SHOULD return this:
1\2\3\4\5\6.7,8.9|10\11\12\13\14\15.16,17.18
BUT! Instead, it returns this:
5\4\3\2\1\8.9,6.7|14\13\12\11\10\17.18,15.16
As you can see, it seems the be flipping around the content of the structs I made (ActionpointID and Actionpoint)
Why is it doing this?
OK, really weird thing. when I use the following code:
The parameters get reversed in order.Code:ActionpointID ID ((unsigned long)xlParam(), (unsigned long)xlParam(), (unsigned char)xlParam(), (unsigned long)xlParam(), (unsigned long)xlParam());
So, I have to make temporary values to hold the parameters like this:
And it works fine.Code:unsigned long Object = (unsigned long)xlParam();
unsigned long Animation = (unsigned long)xlParam();
unsigned char Direction = (unsigned char)xlParam();
unsigned long Frame = (unsigned long)xlParam();
unsigned long Id = (unsigned long)xlParam();
ActionpointID ID (Object, Animation, Direction, Frame, Id);
I'll let you guys know if I have any more problems, but thanks for being so helpful!
Is xlParam() ultimately reading off of MMF's parameter queue? You are assuming that the parameters are going to be evaluated left-to-right, but I believe, as you have just observed, that the parameters are evaluated right-to-left. However I don't believe there is any hard gaurentee provided by the language about the order in which parameters are evaluated and may be different in some circumstances, so to be safe you should _never_ write code that depends on assumptions about the order of parameter evaluation in C/C++ functions.
The order of parameters to functions getting evaluated is dependent on the calling convention. The standard one is right-to-left, but there are others.