Thought I'd give an update of my experience after quite a few TR rides on the .26 beta. I'm using a pair or Vectors as the PM and TR to drive the workout.
Mostly I'm seeing reasonable results. Actual (Vector) power mostly tracks closely with the target power but can go through phases of being either too low or too high. I've found it struggles a bit with ramps such as those found in Keeler Needle, I suppose this is unsurprising as it will take a while for the feedback loop to adjust. Despite all this TSS at the end of a workout is always very close to the target value.
Three times it has completely dropped the ball and turned into a mid workout disaster, but I suppose this is to be expected of beta software.
How are othe unofficial beta testers getting on?
It is mostly working for me. It varies with different TR workouts, so I'm starting to question whether there's some elemts to how the particular TR workout was created that affects behavior. For instance, on Pettit, the first segment has the Kickr reading higher than the Quarq (i.e., as if the Quarq is not controlling it), but the next two segments are spot on. This behavior is consistent on that particular workout. Interestingly, the discrepancy also returns if I got over around 110 rpm.
Other workouts behave differently, but consistently unto themselves. That's why I'm thinking there's a subtlety to TR's control of the erg that I'm not aware of.