On Mon, Dec 14, 2009 at 10:47 AM, Ian Clarke <ian at locut.us> wrote: > I don't think you'll gain anything by re-compressing an already > compressed file unless the original compression mechanism was really > dumb.
Typical multimedia formats have repetitive header information used to regain synchronization after a seek or corruption. You should expect a percent or two of compression. It's still not worth the cpu overhead, but it's not none at all. BTRFS (and other filesystems with compression) attempt to compress an initial chunk (64k in btrfs) and disable compression if the size is not reduced by some threshold amount. This is probably a better heuristic than guessing based on extensions or trying to compute histograms.
