On Sun, 28 Jan 2007, Jan Danielsson wrote:

>   Before anyone suggests unison: Yes, I have tried it, but since I only
> have 2GB physical RAM in this system, I would prefer not to use it,
> since it makes my system slow to a crawl and swap like mad.

>   I decided to write a tool which calculates md5 sums on a block basis.
[...]

   If the problem is that unison is bad at handling large files (but 
handles large *collections* of files fine), then a quick and dirty 
solution might be split(1).  Divide-and-conquer...

   I guess the pros/cons are obvious (quick to implement in a script / uses 
twice as much disk), but I just thought I'd mention it.


MAgnus