← Back to team overview

syncany-team team mailing list archive

Arbitrary Chunk Sizes


I think it would be useful to have arbitrary chunk sizes in Syncany. In all
cases where there is an insertion to the beginning of a file that is not of
length 512, the rest of the file will be offset and all subsequent chunks
will be marked dirty and discarded. I think it would be useful to have a
procedure that tries to find existing chunks in the new file and generates a
minimal changeset. This might result in fragmentation as smaller and smaller
chunks are created, so there should also be a feature to combine fragments
into larger chunks so that filesystem performance isn't degraded.

Anyone have any thoughts on this? Should I file a blueprint for this?

See Chapter 3 of Andrew Tridgell's thesis for more details about how rsync
accomplishes this:

Follow ups