Some file types seem to change in ways such that rsync's algorithm
wouldn't be any better that just straight copying the file. For
instance, if a gzipped file is uncompressed, changed even by 1 char at
the beginning, and recompressed, the result can have very little in
common with the original (in terms of identical blocks).
But then again the gzipped files might have a lot in common if the
change occurs near the end, as it would for a log file that was
updated.
So, in practice, which file types it is not worth running the
rsync algorithm on? Does rsync ever look at file extensions, etc, and
from that try to figure out if it running the algorithm would be worth
it? Thanks.
--
Ben Escoto