Thanks for reporting this. This should not be happening. It's unlikely to cause any corruption (which would require a strong-sum collision), but it might be indicating a rollsum bug that could be causing delta calculation to miss some matches, resulting in a larger delta than it should.
I was just considering removing the --paranoia setting because I thought the delta code in librsync was pretty mature and we didn't need it any more. I filed an upstream bug against librsync and will investigate/fix; https://github.com/librsync/librsync/issues/155 -- Donovan Baarda <a...@minkirri.apana.org.au>
