[wsf-c-dev] [C++] Question on data types

[Uthaiyashankar]

Damitha Kumarage wrote:
> Senaka Fernando wrote:
>
>> Yes, that is a good choice, but don't we miss out in defining boolean
>> types using integers?
>>
>> Because C++ defines bool, we instead can use a subset of types defined
>> in axis, and use the new additions from C++
>>
>> BTW, this only affects the interaction through the API. The internal
>> core will remain the same as C.
>>
>> Therefore the C++ developers will feel more comfortable dealing with
>> types derived from bool, when it is a true-false scenario instead of the
>> AXIS implementation
>>  
>>
> hmmm,
> Say axis2c api function function foo is
> axis2_bool_t foo()
> and it is wrapped in c++ function bool bar()
>
On the other hand,

if api function foo is
void foo(axis2_bool_t status)
{
    if (status == AXIS2_TRUE)
          do something;
}

and bar is
void bar(bool status)
{
    return foo(status);
}

and the call is
bar(true);

then, foo might or might not work. C++ specification tells "true has to 
be non zero value and false has to be zero value". However,  it does not 
specify a specific integer value for true. Most of the compilers are 
using true = 1, but it is not a must. I don't know whether there are any 
compilers which are specifying other values for true. So, for the safe 
side, bar has to be
void bar(bool status)
{
    if (status)
          return foo(AXIS2_TRUE);
    else
       return foo(AXIS2_FALSE);
}

if that is the case, then can use bool. Otherwise, it is better to stick 
to axis2_bool_t.

Shankar.
>
> axis2_bool_t
> foo(void)
> {
>    return AXIS2_FALSE;
> }
>
> bool bar(void)
> {
>    std::cout << "In C++ wrapper function \n";
>    return foo();
> }
>
> Then C++ user experience would be
>
>    bool ret;
>    ret = bar();
>    if(ret == true)
>        std::cout << "Success \n";
>    if(ret == false)
>        std::cout << "Failure \n";
>
> OK. I agree with you
>
> Damitha
>
>
>
>
>> What do you say?
>>
>> On Tue, 2007-10-16 at 16:27 +0530, Damitha Kumarage wrote:
>>  
>>
>>> Senaka Fernando wrote:
>>>
>>>   
>>>> Is it better to proceed with the addition of C++ primitive types
>>>> sparingly, or only with types defined in AXUTIL_..? (for instance, the
>>>> bool type can be re-defined or the existing definition in
>>>> AXUTIL_UTILS_DEFINES.h can be used which is derived from int)
>>>>
>>>> Which would be better and why?
>>>>
>>>>
>>>>     
>>> Since you are going to write a set of C++ wrapper functions over C I 
>>> think it is better to adhere to existing typing instead of doing 
>>> type mapping between C++ and C.
>>> Damitha
>>>
>>>   
>>>> Regards,
>>>>
>>>> Senaka
>>>>
>>>>
>>>> _______________________________________________
>>>> Wsf-c-dev mailing list
>>>> Wsf-c-dev at wso2.org
>>>> http://wso2.org/cgi-bin/mailman/listinfo/wsf-c-dev
>>>>
>>>>
>>>>
>>>>     
>>> _______________________________________________
>>> Wsf-c-dev mailing list
>>> Wsf-c-dev at wso2.org
>>> http://wso2.org/cgi-bin/mailman/listinfo/wsf-c-dev
>>>   
>>
>>
>> _______________________________________________
>> Wsf-c-dev mailing list
>> Wsf-c-dev at wso2.org
>> http://wso2.org/cgi-bin/mailman/listinfo/wsf-c-dev
>>
>>  
>>
>
>
>
> _______________________________________________
> Wsf-c-dev mailing list
> Wsf-c-dev at wso2.org
> http://wso2.org/cgi-bin/mailman/listinfo/wsf-c-dev
>