mex: Compilation error with mxArray
in [Matlab]
|
Prev: matlab code for speed control of 3 phase induction motor using fuzzy logic controller
Next: saving a png with a fixed size
From: Ajay on 29 Mar 2010 01:35 Using MATLAB 2009b and Microsoft Visual C++ 2008 (Express Edition), I created the DLL (first_lib.dll, first_lib.h and first_lib.lib) from MATLAB function (first.m) using mcc and I would like to call it in a C code. I am posting the complete code here with compilation erros while compiling the C code: Reported Errors & warnings are: 1) error C2275: 'mxArray' : illegal use of this type as an expression c:\program files\matlab\r2009a\extern\include\matrix.h(292) : see declaration of 'mxArray' 2) warning C4047: '=' : 'int' differs in levels of indirection from 'void *' 3) warning C4047: '=' : 'int' differs in levels of indirection from 'mxArray *' 4) warning C4047: 'function' : 'const mxArray *' differs in levels of indirection from 'int' 5) warning C4024: 'mxGetPr_proxy' : different types for formal and actual parameter 1 6) error C2172: 'memcpy' : actual parameter is not a pointer : parameter 2 I am posting my complete code here, where mlfFirst is the function call and are files which were created using mcc command. #include <stdio.h> #include <math.h> #include "mex.h" #include "first_lib.h" int main() { float a,b,c, Ans; a = 1; b =2; c=3; mxArray *pmX, *pmY, *pmZ; mxArray *pmAns = NULL; mclInitializeApplication(NULL,0); pmAns = mxCreateDoubleMatrix(1,1,mxREAL); pmX = mxCreateDoubleMatrix(1,1,mxREAL); pmY = mxCreateDoubleMatrix(1,1,mxREAL); pmZ = mxCreateDoubleMatrix(1,1,mxREAL); first_libInitialize(); memcpy(mxGetPr(pmX),a, sizeof(double)*1); memcpy(mxGetPr(pmY),b, sizeof(double)*1); memcpy(mxGetPr(pmZ),c, sizeof(double)*1); mlfFirst(1, &pmAns, pmX, pmY, pmZ); Ans = mxGetPr(pmAns); first_libTerminate(); printf("%f %f\n", Ans); mxDestroyArray(pmX); mxDestroyArray(pmY); mxDestroyArray(pmZ); mxDestroyArray(pmAns); mclTerminateApplication(); return 0; }
From: James Tursa on 29 Mar 2010 02:13 "Ajay " <ajaydashora(a)gmail.com> wrote in message <hope68$ds8$1(a)fred.mathworks.com>... > > float a,b,c, Ans; : > mxArray *pmX, *pmY, *pmZ; > mxArray *pmAns = NULL; : > pmAns = mxCreateDoubleMatrix(1,1,mxREAL); > pmX = mxCreateDoubleMatrix(1,1,mxREAL); > pmY = mxCreateDoubleMatrix(1,1,mxREAL); > pmZ = mxCreateDoubleMatrix(1,1,mxREAL); : > memcpy(mxGetPr(pmX),a, sizeof(double)*1); > memcpy(mxGetPr(pmY),b, sizeof(double)*1); > memcpy(mxGetPr(pmZ),c, sizeof(double)*1); The error message means exactly what it says , the 2nd argument to memcpy needs to be a pointer. So these calls should have &a, &b, &c as arguments instead of a, b, c. Also, you have declared a, b, and c as float (32-bit single precision floating point) but are using them as double (64-bit double precision floating point). You need to declare a, b, and c as double. e.g., double a, b, c; And an easier way to do this is as follows: *mxGetPr(pmX) = a; *mxGetPr(pmY) = b; *mxGetPr(pmZ) = c; > > Ans = mxGetPr(pmAns); Ans is a double, not a pointer to double. mxGetPr needs to be dereferenced. e.g., Ans = *mxGetPr(pmAns); This all assumes that the function you are calling is expecting doubles as inputs and returns a double. Is that the case? James Tursa
From: Ajay on 29 Mar 2010 03:09 Hi James, Many thanks......Once again I repeat that I am using the VC++ 2008 (Express) downloaded from internet as free s/w. Following your advice, reduces the errors. However, first error is still present: error C2275: 'mxArray' : illegal use of this type as an expression c:\program files\matlab\r2009a\extern\include\matrix.h(292) : see declaration of 'mxArray' Due to this error, C compiler is not recognizing either of pmX, pmY, PmZ. I added the pragmas to my code to avoid any library to be missing in compiler settings and I changed my code as: #ifdef _MSC_VER // if MS Visual C++ #pragma warning(push) #pragma warning(disable:4996) #pragma comment(lib, "libdfblas.lib") #pragma comment(lib, "libdflapack.lib") #pragma comment(lib, "libemlrt.lib") #pragma comment(lib, "libeng.lib") #pragma comment(lib, "libfixedpoint.lib") #pragma comment(lib, "libmat.lib") #pragma comment(lib, "libmex.lib") #pragma comment(lib, "libmwblas.lib") #pragma comment(lib, "libmwlapack.lib") #pragma comment(lib, "libmwmathutil.lib") #pragma comment(lib, "libmwservices.lib") #pragma comment(lib, "libmx.lib") #pragma comment(lib, "libut.lib") #pragma comment(lib, "mclcom.lib") #pragma comment(lib, "mclcommain.lib") #pragma comment(lib, "mclmcr.lib") #pragma comment(lib, "mclmcrrt.lib") #pragma comment(lib, "mclxlmain.lib") #pragma comment(lib, "first_lib.lib") #endif #include <stdio.h> #include <math.h> #include "mex.h" #include "first_lib.h" int main() { double a,b,c, Ans; a = 1; b =2; c=3; mxArray *pmX, *pmY, *pmZ; mxArray *pmAns = NULL; mclinitializeApplication(NULL,0); *mxGetPr(pmX) = a; *mxGetPr(pmY) = b; *mxGetPr(pmZ) = c; first_libInitialize(); memcpy(mxGetPr(pmX),&a, sizeof(double)*1); memcpy(mxGetPr(pmY),&b, sizeof(double)*1); memcpy(mxGetPr(pmZ),&c, sizeof(double)*1); mlfFirst(1,&pmAns,pmX,pmY,pmZ); Ans = *mxGetPr(pmAns); first_libTerminate(); printf("%f %f\n", Ans); mxDestroyArray(pmX); mxDestroyArray(pmY); mxDestroyArray(pmZ); mxDestroyArray(pmAns); mclTerminateApplication(); return 0; } "James Tursa" <aclassyguy_with_a_k_not_a_c(a)hotmail.com> wrote in message <hopgdk$grd$1(a)fred.mathworks.com>... > "Ajay " <ajaydashora(a)gmail.com> wrote in message <hope68$ds8$1(a)fred.mathworks.com>... > > > > float a,b,c, Ans; > : > > mxArray *pmX, *pmY, *pmZ; > > mxArray *pmAns = NULL; > : > > pmAns = mxCreateDoubleMatrix(1,1,mxREAL); > > pmX = mxCreateDoubleMatrix(1,1,mxREAL); > > pmY = mxCreateDoubleMatrix(1,1,mxREAL); > > pmZ = mxCreateDoubleMatrix(1,1,mxREAL); > : > > memcpy(mxGetPr(pmX),a, sizeof(double)*1); > > memcpy(mxGetPr(pmY),b, sizeof(double)*1); > > memcpy(mxGetPr(pmZ),c, sizeof(double)*1); > > The error message means exactly what it says , the 2nd argument to memcpy needs to be a pointer. So these calls should have &a, &b, &c as arguments instead of a, b, c. Also, you have declared a, b, and c as float (32-bit single precision floating point) but are using them as double (64-bit double precision floating point). You need to declare a, b, and c as double. e.g., > > double a, b, c; > > And an easier way to do this is as follows: > > *mxGetPr(pmX) = a; > *mxGetPr(pmY) = b; > *mxGetPr(pmZ) = c; > > > > > Ans = mxGetPr(pmAns); > > Ans is a double, not a pointer to double. mxGetPr needs to be dereferenced. e.g., > > Ans = *mxGetPr(pmAns); > > This all assumes that the function you are calling is expecting doubles as inputs and returns a double. Is that the case? > > James Tursa
From: AJ on 29 Mar 2010 09:20 "Ajay " <ajaydashora(a)gmail.com> wrote in message <hopjmk$4c0$1(a)fred.mathworks.com>... > Hi James, > > Many thanks......Once again I repeat that I am using the VC++ 2008 (Express) downloaded from internet as free s/w. > > Following your advice, reduces the errors. However, first error is still present: > > error C2275: 'mxArray' : illegal use of this type as an expression > c:\program files\matlab\r2009a\extern\include\matrix.h(292) : see declaration of 'mxArray' I think see the problem - in "C" all the variable declarations need to be at the top of a code segment. Move the assignment of a, b, and c to below the mxArray* declaritors. [snip] > double a,b,c, Ans; //a = 1; b =2; c=3; > > mxArray *pmX, *pmY, *pmZ; > mxArray *pmAns = NULL; a = 1; b =2; c=3; > > mclinitializeApplication(NULL,0); > You will still need these lines: pmX = mxCreateDoubleMatrix(1,1,mxREAL); pmY = mxCreateDoubleMatrix(1,1,mxREAL); pmZ = mxCreateDoubleMatrix(1,1,mxREAL); > *mxGetPr(pmX) = a; > *mxGetPr(pmY) = b; > *mxGetPr(pmZ) = c; > HTH, AJ
From: Ajay on 29 Mar 2010 14:46
Hi, Many thanks.....it is working now !!! "AJ" <aj(a)johnson.lmco.com> wrote in message <hoq9es$fml$1(a)fred.mathworks.com>... > "Ajay " <ajaydashora(a)gmail.com> wrote in message <hopjmk$4c0$1(a)fred.mathworks.com>... > > Hi James, > > > > Many thanks......Once again I repeat that I am using the VC++ 2008 (Express) downloaded from internet as free s/w. > > > > Following your advice, reduces the errors. However, first error is still present: > > > > error C2275: 'mxArray' : illegal use of this type as an expression > > c:\program files\matlab\r2009a\extern\include\matrix.h(292) : see declaration of 'mxArray' > > I think see the problem - in "C" all the variable declarations need to be at the top of a code segment. Move the assignment of a, b, and c to below the mxArray* declaritors. > > [snip] > > > double a,b,c, Ans; > //a = 1; b =2; c=3; > > > > mxArray *pmX, *pmY, *pmZ; > > mxArray *pmAns = NULL; > a = 1; b =2; c=3; > > > > mclinitializeApplication(NULL,0); > > > > You will still need these lines: > pmX = mxCreateDoubleMatrix(1,1,mxREAL); > pmY = mxCreateDoubleMatrix(1,1,mxREAL); > pmZ = mxCreateDoubleMatrix(1,1,mxREAL); > > > *mxGetPr(pmX) = a; > > *mxGetPr(pmY) = b; > > *mxGetPr(pmZ) = c; > > > > HTH, > AJ |