pointer

shamjs · 11-24-2011, 01:23 AM

Hi All,

i have some problem in pointers

here is my sample code.

Code:

/***predict.cpp********/
Array1D<REAL> CUnscXPredictor::CopyToArray1D(TXMatrix* inmatrix)
{
.
.
Array1D<REAL> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData); //line 1661	
	return amatrix;
}

Code:

/********olux.h*********/
typedef struct tagTMatrix1
{
	long   NumColumns;
	long   NumRows;
	long   NumPlanes;
	float*  aData;            /
	
} TXMatrix;

Code:

/*****tnt_array1d.h****/
template <class T>
		Array1D<T>::Array1D(long n, T *a) : v_(a), n_(n) , data_(v_.begin())
	{
#ifdef TNT_DEBUG
		std::cout << "Created Array1D(int n, T* a) \n";
#endif
	}

compiler is reporting error in line 1661 because of 2nd argument passed to the object.

predictor.cpp:1661: error: no matching function for call to âTNT::Array1D<double>::Array1D(long int, float*&)â
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:107: note: candidates are: TNT::Array1D<T>::Array1D(const TNT::Array1D<T>&) [with T =
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:135: note: TNT::Array1D<T>::Array1D(long int, T*) [with T = double]
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:125: note: TNT::Array1D<T>::Array1D(long int, const T&) [with T = doub
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:117: note: TNT::Array1D<T>::Array1D(long int) [with T = double]
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:104: note: TNT::Array1D<T>::Array1D() [with T = double]

1> im having doubt at 2nd argument
inmatrix->aData
imatrix will finally points to address of aData
and this address will be passed to 2nd parameter of the below constructor which is a pointer type

Array1D<T>::Array1D(long n, T *a) : v_(a), n_(n) , data_(v_.begin())
{
}

hence address is passed to pointer but why compiler is raising error

2> im having doubt at 2nd parameter of error message
predictor.cpp:1661: error: no matching function for call to âTNT::Array1D<double>::Array1D(long int, float*&)â
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:107

but when i replace the line 1661 with
float *ptr=inmatrix->aData;
Array1D<REAL> amatrix(inmatrix->NumRows*inmatrix->NumColumns,*ptr);

compiler wont report error.

plz guide what's the concept behind it and help me to fix this error..............

waiting for your reply.........

JohnGraham · 11-24-2011, 03:06 AM

Quote:

Originally Posted by shamjs

compiler is reporting error in line 1661 because of 2nd argument passed to the object.

predictor.cpp:1661: error: no matching function for call to âTNT::Array1D<double>::Array1D(long int, float*&)â
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:107: note: candidates are: TNT::Array1D<T>::Array1D(const TNT::Array1D<T>&) [with T =
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:135:../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array3d.h:115: note: TNT::Array3D<T>::Array3D(long int, long int, long int, T) [
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array3d.h:94: note: TNT::Array3D<T>::Array3D(long int, long int, long int) [with
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array3d.h:83: note: TNT::Array3D<T>::Array3D() [with T = double]
TNT::Array1D<T>::Array1D(long int, T*) [with T = double]

You seem to have missed a vital portion of your error message - please post it in full.

shamjs · 11-24-2011, 04:19 AM

Quote:

Originally Posted by JohnGraham

You seem to have missed a vital portion of your error message - please post it in full.

Sorry i have pasted wrong one..

please look at the same post,which i have edited now.

millgates · 11-24-2011, 04:26 AM

Quote:

Originally Posted by g++

TNT::Array1D<T>::Array1D(long int, T*) [with T = double]
predictor.cpp:1661: error: no matching function for call to "TNT::Array1D<double>::Array1D(long int, float*&)"

Quote:

float* aData;

so... is it an array of float or double?

shamjs · 11-24-2011, 04:39 AM

its a float
but i dont why compiler is taking T=double

millgates · 11-24-2011, 04:41 AM

Quote:

Originally Posted by shamjs

its a float
but i dont why compiler is taking T=double

Code:

Array1D<REAL> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData); //line 1661

how is REAL defined?

shamjs · 11-24-2011, 04:46 AM

#define REAL float

millgates · 11-24-2011, 04:53 AM

Quote:

Originally Posted by shamjs

#define REAL float

Are you sure it doesn't get redefined somewhere?
Do you still get the error if you change the line 1661 to

Code:

Array1D<float> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData);

shamjs · 11-24-2011, 05:26 AM

when i changed to Array1D<float> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData); this

im getting error as
predictor.cpp:1662: error: conversion from âTNT::Array1D<float>â to non-scalar type âTNT::Array1D<double>â requested

similarly i checked by giving Array1D<double> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData);
when i executed this line
i got the same previous error, what i have posted earlier.........

thanx for your instant replies...............

millgates · 11-24-2011, 05:53 AM

Quote:

Originally Posted by shamjs

when i changed to Array1D<float> amatrix(inmatrix->NumRows*inmatrix->NumColumns,inmatrix->aData); this

im getting error as
predictor.cpp:1662: error: conversion from âTNT::Array1D<float>â to non-scalar type âTNT::Array1D<double>â requested

Oh, and you would have to also change the function to return array of float (obviously, float and REAL are not the same type):

Code:

Array1D<float> CUnscXPredictor::CopyToArray1D(TXMatrix* inmatrix)

so the CUnscXPredictor::CopyToArray1D has the same return type as amatrix

by the way I downloaded the tnt library and the following minimal example compiles without errors

Code:

#include "tnt.h"

class test {
	public:
	float *a;
	test(float* a_ = 0) : a(a_) {}
};

int main(){
	float f;
	test T(&f);
	long n(1);
	TNT::Array1D < float > (n, T.a);
	return 0;
}

shamjs · 11-24-2011, 06:37 AM

Quote:

Originally Posted by millgates

by the way I downloaded the tnt library and the following minimal example compiles without errors

you plz download this http://math.nist.gov/tnt/download.htmlwhich is source code(30kb)

the main problem is in
no matching function for call to âTNT::Array1D<double>::Array1D(long int, float*&)â of 2nd argument
means it is passing float*& argument at caller.

but at callee it needs
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:135: note: TNT::Array1D<T>::Array1D(long int, T*) [with T = double]

T* type parameter.....

millgates · 11-24-2011, 06:46 AM

Quote:

Originally Posted by shamjs

you plz download this http://math.nist.gov/tnt/download.htmlwhich is source code(30kb)

yes, I know. That's what I downloaded

Quote:

Originally Posted by shamjs

the main problem is in
no matching function for call to âTNT::Array1D<double>::Array1D(long int, float*&)â of 2nd argument
means it is passing float*& argument at caller.

but at callee it needs
../inc/../inc/../../corenumeric/inc/utils/../jama/../tnt/tnt_array1d.h:135: note: TNT::Array1D<T>::Array1D(long int, T*) [with T = double]

T* type parameter.....

yes, we already know that, too. The problem is, that the compiler believes that REAL is double. Therefore, amatrix is declared as an Array1D<double>
and the function CUnscXPredictor::CopyToArray1D returns an Array1D<double>, too, while inmatrix->aData is a pointer to float.