Another "maybe" bug.
I have duet 3d wifi on a delta printer, with the new function "M665 L260.0:260.0:260.0" to give the firmware the diagonal rod offsets, if I leave them the same as in the example everything works fine, now, if I change the values after calibrating with the hexagon I get (and thats what I got when I was running Marlin) L267.3:267.1:266.9, a 0.2mm difference between the rod pairs, my printer misses steps while homing (the second bump, but not in the first one) and misses steps while calibrating and printing.
I, then, proceed to experiment, the greater the difference, the more steps it misses.
So I'm now back to the L267.1 (the mean of the tree pairs) until this is fixed, it happens too to others in my printing community.

Here I provide you my homedelta, config override and config in case you want to check them out.

2_1561532460509_homedelta.g 1_1561532460509_config-override.g 0_1561532460509_config.g