List of generic references to mixed types (Part 2)

[NorseMN]

I have managed to get a few things to work.
Maybe some example code will help to focus the conversation.
(Sorry to paste my code in-line. I can't find any way to create a "text block" for code entry like I see used in other posts. And I can't find any instructions on how to use this forum's editing capabilities.)

Header file:

#include "StdTypDefs.h"   // U8, S8, U16...

template<typename T, size_t SIZE>
class DataTyp
{
private:
	static	T Data[SIZE];
	static	const size_t size = SIZE;
	U16		num;

public:
	DataTyp(U16 elemNum)
	{
		num = (elemNum < (size - 1)) ? elemNum : size;
	};

	U16 GetNum() { return num; };
	void SetVal(U16 const& val) { Data[num] = val; };
	U16 &GetVal() { return Data[num]; };

	T& operator[] (U16 i)
	{
		return (i < size) ? Data[i] : num;	// This use of "num" only for testing.
	};

	const T& operator[] (U16 i) const
	{
		return (i < size) ? Data[i] : num;	// This use of "num" only for testing.
	};

};

template<> U16 DataTyp<U16, 16>::Data[16];
template<> S16 DataTyp<S16, 16>::Data[16];
template<> U32 DataTyp<U32, 16>::Data[16];

CPP file:

#include "stdafx.h"
#include<iostream>
#include "StdTypDefs.h"
#include "RefArray_01.h"

using namespace std;

int main()
{
	U16 A_U16 = 123;
	U16 B_U16 = 456;
	U16 C_U16 = 789;

	DataTyp<U16, 16> val_U16_A(0);
	DataTyp<U16, 16> val_U16_B(5);
	DataTyp<U16, 16> val_U16_C(9);

	val_U16_A.SetVal(A_U16);
	val_U16_B.SetVal(B_U16);
	val_U16_C.SetVal(C_U16);

	cout << "val_U16[" << val_U16_A.GetNum() << "] = " << val_U16_A.GetVal() << endl;
	cout << "val_U16[" << val_U16_B.GetNum() << "] = " << val_U16_B.GetVal() << endl;
	cout << "val_U16[" << val_U16_C.GetNum() << "] = " << val_U16_C.GetVal() << endl << endl;

	DataTyp<U16, 16> val_U16_tmp(0);	// Second reference to same element as val_U16_A.
	val_U16_tmp.SetVal(777);

	cout << "val_U16[" << val_U16_A.GetNum() << "] = " << val_U16_A.GetVal() << endl;
	cout << "val_U16[" << val_U16_B.GetNum() << "] = " << val_U16_B.GetVal() << endl;
	cout << "val_U16[" << val_U16_C.GetNum() << "] = " << val_U16_C.GetVal() << endl << endl;

	U16 tmp;
	tmp = val_U16_B[5];
	val_U16_A[0] = tmp;
	cout << "val_U16[" << val_U16_A.GetNum() << "] = " << val_U16_A.GetVal() << endl << endl;


	int i;
	cin >> i;
	
	return 0;
}

This runs on Visual Studio Express 2015 and produces this consul output:

val_U16[0] = 123
val_U16[5] = 456
val_U16[9] = 789

val_U16[0] = 777
val_U16[5] = 456
val_U16[9] = 789

val_U16[0] = 456

So I have something very ragged working that comes close to part of what I want.

This template instantiation format seems to work:
template<> U16 DataTyp<U16, 16>::Data[16];
But:
- It would be nice if I could avoid the redundant type and size values.
- Browsing the web, I found a wide variety of opinions on the proper format for this and the proper file (H or CPP). Some of it conflicting.

This value instantiation format is NOT so acceptable:
DataTyp<U16, 16> val_U16_A(0);

I want to be able to declare DataTyp instances using a constructor with only a type specifier. In this case, the class would have a static "cnt" member that's initialized to zero when the Data array is allocated and "size is set.
Then, when a class instance is declared, "num" member is set to cnt++ so each declared value is associated with the next element of Data[].
- I couldn't figure out how to get "static const cnt" initialized.

The operator[] members in this example allows external selection of random subscript numbers. This is NOT desired behavior.

my original concept war to create a value reference type composed of:
{typeNumber, instanceNumber }
That could act a a reference to an element in an array of a specific type.
Then I could make lists of that type and have them reference any value of any type (within the allocated arrays).

I found no solution that efficiently uses "typeNumber" to select the correct array and return a reference value for that type.

I was hoping that templates for this example DataTyp class would automatically use the correct DataTyp instantiation based on the function arguments of the member functions. Then I need only the instanceNumber.

So I think I'm headed down a dead-end road right now.
I'm in over my head and I need some help digging my way out.

P.S. Preview doesn't work for me (IE or Firefox) and the topic editor tells me that this post is 67 characters long.
P.P.S. I got an error when I tried to post. Maybe from experimenting with format buttons. I was able to cut-and -paste the content into a new post, and this one worked.

[TheIdeasMan]

Hi,

Welcome to cplusplus :+D

(Sorry to paste my code in-line. I can't find any way to create a "text block" for code entry like I see used in other posts. And I can't find any instructions on how to use this forum's editing capabilities.)

Have a read of this:
http://www.cplusplus.com/articles/z13hAqkS/

There is the format menu on the right of the posts. But there is also a bug in that the formatting options don't work when making a new topic. One can either type the codes in manually, or use the format menu after the fact :+)

Also, just continue with the one topic, don't create a new one for the same subject :+)

Cheers.

[NorseMN]

TheIdeasMan,
Thanks for the link to the "How to use tags" article.
I searched for something like that, but failed to find it.
I was also confused by the formatting buttons not working when creating a new topic. So I lost faith in them. It also didn't help that the Preview feature didn't work on my company (Windows 7) laptop.

I found that the format buttons work on my (Windows XP) home PC when editing a post within a reply. But they fail when directly editing a post. Same for the Preview feature.

Of course, you're correct that I shouldn't have started a second topic. (Doh!)
I did so because I was significantly expanding the scope of the issue.
At this point, the subject title describes only a piece of the issue.

Would it be appropriate to paste the initial content of this post as a reply to the original topic and then close this one? (Mark as solved? Or if possible, delete it?)
Thanks,
Reid