Absolute addressing means that you specify the destination and it remains fixed. e.g. goto 1234
Relative addressing means that that it is relative to the original instruction. e.g. goto PC+56.

If the PC was 1178, then the destination address is still 1234.

You may have noticed that most branches are fairly short. Some subroutines may not be very far away. So if you can encode the operand in one byte rather than two, it uses less program memory. Relative addressing for the 8051 uses -127 to +128 and hence fits in one byte.

You may have also noticed that "LCALL 1234" will always execute the subroutine at 1234. If all the opcodes were moved by one byte, the opcodes and operands would have to be re-calculated to account for the changes in PC.

With Relative addressing, you could store your program anywhere in memory and never have to re-calculate. This is never really used with an 8051.

An assembler lets you specify expressions in different ways. You commonly just use the label directly. The assembler then uses whichever opcode is both legal and matches your instruction.

It will tell you if your instruction is the wrong syntax (no opcode matches) or if it cannot evaluate the operand. e.g. the opcode needs the operand to fit in 'PC-127' to 'PC+128' range.

These errors can occur with JC type opcodes or even with AJMP type opcode. (AJMP needs to refer to destinations in the same 'page')

Read your assembler manual, and the opcode descriptions.

David.