There's a lot to explain in Elite, and some of it is pretty challenging stuff. Before getting stuck in, it's probably wise to take a brief look at some of the terminology I've used in this commentary.
Let's start with some general terms.
- Given a number X, ~X is the number with all the bits inverted
- Given a number A, |A| is the absolute of that number - i.e. the number with no sign, or just the magnitude of the number
- Given a multi-byte number, (S R) say, the absolute would be written |S R| (see below for more on multi-byte numbers and terminology)
- Coordinates are shown as (x, y), both on the screen and in space, so the centre of the space view is at screen coordinate (128, 96), while our trusty Cobra Mk III is at space coordinates (0, 0, 0)
- Vectors and matrices are enclosed in square brackets, like this:
[ 1 0 0 ] [ x ] [ 0 1 0 ] or [ y ] [ 0 0 -1 ] [ z ]
We also need some terminology for multi-byte numbers, but that needs its own section, particularly as Elite has quite a few variations on this theme.
Not surprisingly, Elite deals with some pretty big numbers. For example, the cash reserves are stored as big-endian 32-bit numbers, space coordinates are stored as 24-bit sign-magnitude little-endian numbers, and the joystick gives us two's complement signed 16-bit numbers. When you only have the 8-bit bytes of 6502 assembly language to play with, things can get confusing, and quickly.
First, let's recap some basic definitions, so we're all on the same page.
- Big-endian numbers store their most significant bytes first, then the least significant bytes. This is how humans tend to write numbers.
- Little-endian numbers store the least significant bytes first then the most significant ones. The 6502 stores its addresses in little-endian format, as do the EQUD and EQUW operatives, for example.
- Sign-magnitude numbers store their sign in their highest bit, and the rest of the number contains the magnitude of the number (i.e. the number without the sign). You can change the sign of a sign-magnitude number by simply flipping the highest bit (bit 7 in an 8-bit sign-magnitude number, bit 15 in a 16-bit sign-magnitude number, and so on). See below for more on sign-magnitude numbers.
- Two's complement numbers, meanwhile, are the mainstay of 6502 assembly language, and instructions like ADC and SBC are designed to work with both negative and positive two's complement numbers without us having to worry about a thing. They also have a sign bit in the highest bit, but negative numbers have their bits flipped compared to positive numbers. To flip the sign of a number in two's complement, you flip all the bits and add 1.
Elite uses a smorgasbord of all these types, and it can get pretty confusing. Given this, let's agree on some terminology to make it easier to talk about multi-byte numbers and how they are stored in memory.
If we have three bytes called x_sign, x_hi and x_lo, which contain a 24-bit sign-magnitude number, with the highest byte in x_sign and the lowest in x_lo, then we can refer to their 32-bit number like this:
(x_sign x_hi x_lo)
In this terminology, the most significant byte is always written first, irrespective of how the bytes are stored in memory. So, we can talk about 16-bit numbers made up of registers:
So here X is the high byte and Y the low byte. Or here's a 24-bit number made up of a mix of registers and memory locations:
(A S S+1)
Again, the most significant byte is on the left, so that's the accumulator A, then the next most significant is in memory location S, and the least significant byte is in S+1.
Or we can even talk about numbers made up of registers, memory locations and constants, like this 24-bit number:
(A P 0)
or this constant, which stores 590 in a 32-bit number:
Just remember that in every case, the high byte is on the left, and the low byte is on the right.
When talking about numbers in sequential memory locations, we can use another shorthand. Consider this little-endian number:
(K+3 K+2 K+1 K)
where a 32-bit little-endian number is stored in memory locations K (low byte) through to K+3 (high byte). We can also refer to this number like this:
K(3 2 1 0)
Or a big-endian number stored in XX15 through XX15+3 would be:
XX15(0 1 2 3)
where XX15 is the most significant byte and XX15+3 the least significant. We could also refer to the little-endian 16-bit number stored in the X-th byte of the block at XX3 with:
To take this even further, if we want to add another significant byte to the 32-bit number K(3 2 1 0) to make a five-byte, 40-bit number - an overflow byte in a memory location called S, say - then we might talk about:
K(S 3 2 1 0)
or even something like this:
XX15(4 0 1 2 3)
which is a five-byte number stored with the highest byte in XX15+4, then the next most significant in XX15, then XX15+1 and XX15+2, through to the lowest byte in XX15+3. And yes, Elite does store one of its numbers like this - see the BPRNT routine for the gory details.
With this terminology, it might help to think of the digits listed in the brackets as being written down in the same order that we would write them down as humans. The point of this terminology is to make it easier for people to read, after all.
Many (but not all) of Elite's multi-byte numbers are stored as sign-magnitude numbers.
For example the x, y and z coordinates in bytes #0-8 of the ship data block in INWK and K% (which contain a ship's coordinates in space) are stored as 24-bit sign-magnitude numbers, where the sign of the number is stored in bit 7 of the sign byte, and the other 23 bits contain the magnitude of the number without any sign (i.e. the absolute value, |x|, |y| or |z|). So an x value of &123456 would be stored like this:
x_sign x_hi x_lo + &12 &34 &56 0 0010010 00110100 01010110
while -&123456 is identical, just with bit 7 of the x_sign byte set:
x_sign x_hi x_lo - &12 &34 &56 1 0010010 00110100 01010110
There are also sign-magnitude numbers where the sign byte is only ever used for storing the sign bit, and bits 0-6 are ignored, and there are others where we only ever care about the top byte (a planet's distance, for example, is determined by the value of x_sign, y_sign and z_sign, for example). But they all work in exactly the same way.