Thread: Trouble changing from Arrays to dynamic pointers

Oh so this is getting more complicated that I thought.
I dont know too much about that technical side so excuse if my answers dont get the point.
Im running Windows 7 and the Wascana/Eclipse IDE, using MinGW.
Is that what you asked?

Thanks

I ran the example and malloc and free work fine. I checked the addresses and they match up.
You have an issue with defining CC CC[2];. I'm surprised it compiled at all. It believe it should be CCDATA CC[2].

Thanks nonoob. You are right its CCDATA CC[2], but I had it originallycorrect in my program as well. This is more an error that ocurred when I copied the code to this board.
Strange that it works. So it must be because of parts I didnt include here because I thought they were fine.
For example the TBuf and nBuf get sorted before being freed. I'll check that and be back.

@Salem: I guess what you proposed is some kind of diagnose tool. Is there anything that'll work with my setup?

So I went thorugh the whole code again. I commented out everything apart from the malloc and free commands and it worked fine (I kept checking the addresses and if they were freed correctly). Then step by step I activated the program. So I found the line which seems to cause the problem. If its comment, the program runs. If the line is active it wont. And I dont understand why (I dont get error messages or warnings before). Some names have changed to the example above but the idea is the same. Here it is:

Code:

typedef struct{
	double *hcc;
	double *tcc;
	double *fw_tot;
	double *nStr;
	double *intgrl;
} COMPCURV;

int arrn[2]

int main(void){
        ...
	int i,err=0;
	COMPCURV CC[2];
	for(i=0;i<2;i++){
		err = calculateCompositeCurve(ft,CC,i);
		if (err == 1){
			printf("error in calculateCompositeCurve\n");
			goto back;
		}
	}
back:
	return err;
}

int calculateCompositeCurve(TASKDATA tdata,COMPCURV CCdata[], int HorC){
	int i,err=0;

	CCdata[HorC].hcc = malloc(2*arrn[HorC] * sizeof(double));
	if (!CCdata[HorC].hcc){
		printf("Out of memory: CCdata.hcc\n");
		err = 1;
		goto back;
	}    // this part is repeated for all pointers of COMPCURV CCdata
        ...
	CCdata[HorC].hcc[0]=0;
	CCdata[HorC].intgrl[0]=0;
	for (i=1;i<2*arrn[HorC];i++){
		CCdata[HorC].nStr[i]+=CCdata[HorC].nStr[i-1];
		CCdata[HorC].fw_tot[i]+=CCdata[HorC].fw_tot[i-1];
		CCdata[HorC].hcc[i]=CCdata[HorC].hcc[i-1]+(CCdata[HorC].tcc[i]-CCdata[HorC].tcc[i-1])*CCdata[HorC].fw_tot[i-1];
		CCdata[HorC].intgrl[i] = CCdata[HorC].intgrl[i-1] + (CCdata[HorC].hcc[i] - CCdata[HorC].hcc[i-1]) * (CCdata[HorC].tcc[i]+CCdata[HorC].tcc[i-1]) / 2;
	}
back:
	return err;
}

So its the last formula "CCdata[HorC].intgrl[i] = ...". Strange thing is, that it works when I change the pointers from the COMPCURV structure to arrays with fixed length.

Did you malloc for intgrl and all the others: hcc, tcc, fw_tot, nStr ?
Perhaps it would be easier if you change the structure:

Code:

typedef struct{
    double hcc;
    double tcc;
    double fw_tot;
    double nStr;
    double intgrl;
    } COMPCURV;

COMPCURV CC[2][10];
or
COMPCURV *CC[2];
CC[0] = malloc(sizeof(COMPCURV) * N);
CC[1] = malloc(sizeof(COMPCURV) * N);

That way you don't have to malloc separately for each element. Then you access elements by CC[0][0].hcc for example.

Originally Posted by nonoob

Did you malloc for intgrl and all the others: hcc, tcc, fw_tot, nStr ?

Yes I casted malloc for all the pointer. And they are filled with the necessary data as well.

Your other suggestions would save me some mallocs thats true. Ill try that later home, thanks

@ nonoob: Youre the man. It works like this. Thanks a lot. As you said it saves me some malloc and free casts. The disadvantage is that now I cant simply memcopy for example the *hcc pointer but have to do it with a loop. I just had to change how the different bytes are addressed etc but concerning the calculations nothing has changed. Just how the memory is managed.
Do you also have an explanation for this behaviour? Its just that I'd like to understand it as well.

I don't know what you want to memcpy. But whatever, I think it should be easier this way. Less loops. Just assign... You can copy entire struct contents or struct member.
What behaviour do you wish explained?

What I mean is that before I could write

Code:

memcpy(dblPointer,CC[0].hcc,n)

which now is

Code:

for(i=0;i<n;i++){
    dblPointer(i)=CC[0][i].hcc;
}

In the case of the CC strucutres its more likely that I want to copy for example all the hcc Values in a row and not a Structure with its different values. But I'm fine with the loop above. As long as it works its okay. What I meant with the mysterious behaviour is that before, when my Structure CC had pointers (of dynamic length) it didn't work but now - with the Structure pointers being of dynamic length - it does.

That's because all the hcc values are not contiguous any more. They are interleaved with tcc, fw_tot, and the rest of the structure members.