Comment by 1996

> Some compression algorithms support checkpointing the compression state so that it can be resumed from that point repeatedly ("dictionary-based compression"

Is it some kind of memoization?