Hi ! Happy New Year to All !

//Given a POD struct,
struct recordType {
int id; // act as a key field
char name[12];
// more fields if any
};

// prototype: DBT<recordType, keyType, numOfbuckets, sizeOfkeyType>
DBT<recordType, int, 117, sizeof(int)> db;
recordType input;

// fill in data (id, name) for input

// prototype: bool Insert(keyType, recordType), keyType is int for this case.

if(db.Insert(input.id, input)) // { keyType => recordType } ( mapping ! )
cout<<"OK\n";
else
cout<<"ERR\n";

// A programmer is always lazy, at least wants to save typing !
// A dirty solution:
// In coding, after the declaration of 'struct recordType', define a Marco, INSERT(x) for a shortcut.


#define INSERT(x) Insert(x.id,x)

// Then, we could write
if(db.INSERT(input)) // hide the key field, keyType
cout<<"OK\n";
else
cout<<"ERR\n";


So far, so good ! Right ? Thank you for reading this pseudocode.

Now, here is my question coming:
Can we eliminate the Marco above by using trait, or something magic (i.e.,
field extractions) tech. ? because I think the Marco is not flexible enough.
Any input makes a difference & is welcome !!!

just quickly went through, why this can't be defined as a function template for DB?

as:
if it is only applied to the POD record, the function template can be replaced by the specified recordType.

>>why this can't be defined as a function template for DB?
>>//inside class definition
>>template<typename T>
>>bool insert(T x){ return insert(x.id, x); }


My point here is that the template class DB is provided with a generic type
, keyType and that the template class DB is served as a library. Therefore,

if the function template above is defined inside the class, because of its
hardcode, then the class DB can not be served as a library.

The recordType is limited to a POD type, that is the keyType must be one of
(int, long, double, and char*) only.