Error using mex G:\Matlab_​WS\code_te​st\Assignm​ent1\Add_T​wo.c:3:1: warning: return type defaults to 'int' [-Wimplicit-int] Add_Two(U16 a, U16 b)

Trying to understand mex compilation with an example, but getting compilation error. Below are the commands that I used to compile. Mex file is generated but its compilation failed.

def = legacy_code('initialize');

def.SourceFiles = {'Add_Two.c'};

def.HeaderFiles = {'Add_Two.h'};

def.SFunctionName = 'Add_Two_SFun';

def.OutputFcnSpec = 'uint32 y1 = Add_Two(uint16 u1, uint16 u2)';

legacy_code('sfcn_cmex_generate', def);

>> legacy_code('compile', def)

### Start Compiling Add_Two_SFun

mex('-IG:\Matlab_WS\code_test\Assignment1', '-c', '-outdir', 'C:\Users\Shafi\AppData\Local\Temp\tp548d5365_ef4a_45c6_9aec_e200ba5a55ff', 'G:\Matlab_WS\code_test\Assignment1\Add_Two.c')

Building with 'MinGW64 Compiler (C)'.

Error using mex

G:\Matlab_WS\code_test\Assignment1\Add_Two.c:3:1: warning: return type defaults to 'int'

[-Wimplicit-int]

Add_Two(U16 a, U16 b)

^

G:\Matlab_WS\code_test\Assignment1\Add_Two.c:3:1: error: conflicting types for 'Add_Two'

In file included from G:\Matlab_WS\code_test\Assignment1\Add_Two.c:1:0:

G:\Matlab_WS\code_test\Assignment1/Add_Two.h:11:5: note: previous declaration of 'Add_Two' was here

U32 Add_Two(U16 a, U16 b);

^

Error in legacycode.LCT/compile

Error in legacycode.LCT.legacyCodeImpl

Error in legacy_code (line 101)

[varargout{1:nargout}] = legacycode.LCT.legacyCodeImpl(action, varargin{1:end});

***************************************************

Input files:

Add_Two.c:

#include <Add_Two.h>

Add_Two(U16 a, U16 b)

{

U32 add_result;

U32 output;

add_result = (U32)a + (U32)b;

if(THOUSAND > add_result)

{

output = add_result;

}

else

{

output = SUM_DEFAULT;

}

return output;

}

Add_Two.h:

#ifndef ADD_TWO_H_

#define ADD_TWO_H_

typedef unsigned int U16;

typedef unsigned long int U32;

#define THOUSAND (U32)1000

#define SUM_DEFAULT (U32)25

U32 Add_Two(U16 a, U16 b);

#endif

Answer 1

[Andrew Janke]

See this line where the function is actually defined in Add_Two.c?

Add_Two(U16 a, U16 b)

You need to add an explicit "`U32`" as its return type there.

U32 Add_Two(U16 a, U16 b)

[BABANBHAIGARI SHAFIULLAH]

Oh GOD, how did I miss that one ? It worked!!! Thanks a lot, @Andrew Janke.

Actually, I tried by removing the function decalaration from Add_Two.h file. It ran successfully, but still there was datatype warning. Now, its resolved.

[Andrew Janke]

Heheh. Happens to the best of us.

You need to keep that function declaration in there: because it is returning something besides a regular int, you need the explicit prototype in there to tell the C compiler what type of value to expect coming back from the function.