1. A one-sided fudge factor, that is, an allowance for error but in only one of two directions. For example, if you need a piece of wire 10 feet long and have to guess when you cut it, you make very sure to cut it too long, by a large amount if necessary, rather than too short by even a little bit, because you can always cut off the slop but you can't paste it back on again. When discrete quantities are involved, slop is often introduced to avoid the possibility of being on the losing side of a fencepost error.
2. The percentage of "extra" code generated by a compiler over the size of equivalent assembly code produced by hand-hacking; i.e. the space (or maybe time) you lose because you didn't do it yourself. This number is often used as a measure of the quality of a compiler; slop below 5% is very good, and 10% is usually acceptable. Modern compilers, especially on RISCs, may actually have *negative* slop; that is, they may generate better code than humans. This is one of the reasons assembler programming is becoming less common.