Code review comment for lp:~chasedouglas/grail/pivot_type

Removing the moveness threshold values, i can live with that, no biggie

The transform matrix should include the pivot, or it would no be useful for other programs which expect a linear transform, which is also the most efficient.

What you call the center of rotation is precisely the same as the moveness = 0 special case,
so configuration could instead be done by allowing the cutoff radius to vary, instead of having separate "algorithms".

The rest seems to be a reproduction of what is in grail2, only less efficiently coded.

All in all, I can imagine reducing this to a ten-line patch, on top of the other grail2 branches, handling mode switches. We should also think about how/if we should model the degrees of freedom depending on what gesture types are being asked for, or if this should really be controllable by the app, or if simply all data should be sent.

Because the patch should really go on top of the original gesture frames branch, I am rejecting this particular merge request.