Seeking reverse-engineers - Apple II VisiCalc

There is no "each time"; this is _live_ backup: writes are mirrored in, ideally, real time. (Obviously, there's a fast-producer-slow-consumer issue here, and dealing with it is where most of the hair comes from. The backup can fall behind, but as long as the aggregate write rate is low enough, it'll get caught up again.) Yes, it's sending just changes. Yes. I got stubborn and did the initial sync-up over the net instead of carrying a disk between locations; it took more than a month. (Would've been less except that I throttled it to avoid rendering my link useless for everything else for the duration.) I have it cooked up and in live use, for NetBSD. :-) My design works at the disk-block level, not the file level, so moving files is cheap to exactly the extent that it invovles only small amounts of writing to disk instead of copying all the data. It certainly could be done at the file level; it just means hooking in at a different layer and redesigning the wire protocol correspondingly. I personally have no interest in supporting Windows, though if someone else does and wants to be wire-compatible with my code I'll be happy to collaborate to the extent that a non-Windows person can. It's fairly easy to use my techniques in anything sufficiently open-source to add hooks into the disk drivers. (I've had someone suggest doing it as a stacked disk driver, a la cgd or ccd, but didn't think of that originally and haven't looked at doing it myself.) /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B