Comments:
=========
Anything after '#' on line is a comment. Empty lines are ignored.

Names:
======
The names must consist only of ASCII alphanumerics and '_'.

Labels:
=======
When referencing (but not defining) label, the label is prefixed by '/' (e.g. 'jp /foo.bar').

Labels can be one of the followign forms:

1) <name>

Global label.

2) .<name>

Block-local label. Automatically qualifies with name of the block.

Block-local labels can't be used with LABEL pseudo-instruction.

3) <name>.<name>

Label in another block. The first <name> is the name of the block, and the second is the block-local name of the
label.

Labels in another block can only be referenced, not defined (must be defined as block-local).

Pseudo-instructions valid outside blocks:
=========================================
1) INCLUDE <file>

Include the contents of file <file>. Recursive includes are handled.

2) LABEL <name> <addr>

Define global label <name> to address <addr> (given as four hexadecimal digits).

3) BLOCK <name>

Start a named block of instructions. The name of the block also Automatically becomes a global label.


Pseudo-instructions valid inside blocks:
========================================
1) DATA <block>

Read hexadecimal byte values on line (separated by spaces) as literial bytes.

2) ENDBLOCK

End a block.

3) <name>:  or .<name>:

Define global or block-local label named <name>.

4) NOPS <number>

Expands to <number> (hexadecimal!) NOPs.

Special blocks:
===============
The following block names are special:

- __start: The startup block. Always placed at 0x100, can be at most 4 bytes in size.
- __fixed150: This block is always placed at 0x150.
- __header: The cartridge header. Placed at 0x134, Must be 25 bytes.
- __freestanding_labels: Logically contains labels defined by LABEL pseudo-instruction. Must be empty.


Instructions & operands:
========================
Here are the rules for operands:

- Instructions are only valid inside blocks.
- Following are appended into opcode (no space in between), everything else has space separating it from opcode:
  * Condition codes (z, nz, c or nc)
  * Bit numbers (0, 1, 2, 3, 4, 5, 6 or 7)
  * Reset vectors (00, 08, 10, 18, 20, 28, 30 or 38)
- Basic 8-bit "registers" are denoted 'a', 'b', 'c', 'd', 'e', 'h', 'l' and '(hl)'
- Basic 16-bit "registers" are denoted 'bc', 'de', 'hl', 'sp' and 'af'
- There is no implicit a register operand in arithmetic operations.
- Memory address operands or registers used as pointers are enclosed in parenthesis.
- Increment/decrement hl on access is written as '(hl-)' and '(hl+)'
- One can encode raw byte as instruction xxx<two hex digits>.
- 8-bit unsigned immediate operands are given as two hexadecimal digits
- 8-bit direct address in ldh can also be given as /<label>, but that label must point to FFxx (and is still
  enclosed in parenthesis)
- 16-bit unsigned immediate operands are either given as four hexadecimal digits or /<label>
- Relative jumps (jr) are always given as /<label> (and must be in range of jump).
- The signed 8-bit immediates in 'add sp,<imm>' and 'ld hl,sp<imm>' are given as '+' (optional with first) or '-'
  followed by two hexadecimal digits.