Thread: Can you make new data types, just as efficient as built in (ie int, byte)

Is it possible to create a new data type in java? My idea for how it could be done is to create an object that holds holds a boolean array of n size and translates the number you make only when necessary.

As a background to what I'm talking about, I know boolean is 1 bit and so it holds 2^0=1 or 0 info. Byte is 8 bits (1 byte) and so it holds 2^8=256 unsigned. So my object (called a halfbyte) holds a boolean array 4 in length, and so the max value would be 2^4=16 unsighned which is good if you're doing something with values in a dozen or between 1 and 10 for example.

I'm wondering if this method of making a new data type -could- be just as efficient as the built in types like int and byte or if assembly language would be required to do something like this. Also is the source code for int and byte and other basic data types made in java like any other object you can make and can you view the source code?

Is it possible to create a new data type in java?

Define a create a new class. I'm unsure what exactly you wish to accomplish, but you could define an object with the values you wish (btw according the java API a boolean represents a byte but - and I quote - "its "size" isn't something that's precisely defined"). Using the example you describe, a 4 bit primitive, can be accomplished using Bitwise and Bit Shift Operators on a particular primitive (byte, int, etc...). It's fast and quite powerful.

what do you mean by a four bit primitive? and do you have a pseudo example of your last sentence? And about the boolean, I've always read that it is simply 1 bit and so it's value is either 1 or 0. What I meant by using a boolean array was to string 4 of them together and create a translation method. If you string 8 together, technically it will be the same as byte, but can you make it just as efficient as byte?

Originally Posted by Perd1t1on

what do you mean by a four bit primitive?

I was simply restating what I thought you meant

Here's is an example of what I was alluding to:

	public static void main(String[] args){
		byte one = 1;
		byte ten = 10;
		byte combined = (byte)((one << 4 ) | ten);
		System.out.println("One = " + one);
		System.out.println("Ten = " + ten);
		System.out.println("Combined = " + combined);
 
		byte one2 = (byte)(combined >> 4);
		byte ten2 = (byte)(combined & 0x0f);
		System.out.println("One = " + one2);
		System.out.println("Ten = " + ten2);
 
	}

This example packs two numbers into a single byte, then extracts them back out using bit shifting. This can be used for all sorts of things. One common problem is how to store and transfer a variable number of options (aka flags). Passing them as bitwise values rather than booleans is a great way to do so. Each byte being a zero and a one, and each 'on' byte a power of two, you can set an integer's bits by 'anding' the integer with that value, as follows:

	static int PRINT_YES = 1;
	static int PRINT_NO = 2;
	static int PRINT_HELLO = 4;
	static int PRINT_WORLD = 8;
 
	private static void printer(int vals){
		if ( (vals & PRINT_NO) != 0 ){
			System.out.println("NO");
		}
		if ( (vals & PRINT_YES) != 0 ){
			System.out.println("YES");
		}
		if ( (vals & PRINT_HELLO) != 0 ){
			System.out.println("HELLO");
		}
		if ( (vals & PRINT_WORLD) != 0 ){
			System.out.println("WORLD");
		}
	}
 
	public static void main(String[] args){
		printer(PRINT_HELLO | PRINT_WORLD );
	}

Not sure if this fully addresses your original question, but this does allow you to use primitives to technically store variable bit sized values.