I would guess an incremental backup option would be easier than a resume option... the problem with a resume function would be knowing what has already been done especially if the download failed and ended the tar halfway through a file... this problem is exacerbated (sp?) by the possibility that data on the empeg may have changed between the initial backup starting and the resume happening... though I suppose you could ignore this possibility if you know the empeg hasn't mounted its disks rw in the mean time (check ext2fs mount count...?) and then assumed that the same tar command would tar in exactly the same way and throw away the equivalent amount of bits from the start of a new tar before piping the rest across to the desktop machine...

Just guessing at some of the problems...

Kim