First of all, thanks for putting this app together. It’s exactly what I’m looking for - I need TrainingPeaks imports and Strava exports of structured workouts with a BIG HUD running on my tablet while I watch youtube on the big screen on the wall, without having to subsidize features I don’t like (gamification) and a racing league I’m not participating in.
I’ve been experimenting with importing routes into slope mode. Specifically, Alpe du Huez from RideWithGPS. I also have imported several Alpe du Zwift workouts that include power data from Zwift.
The imported workouts populate the workout editor at a resolution of one line per minute. This creates a reasonably decent facsimile of the route, tied to whatever my effort was for that workout. It mostly works, and I see the logic of it, but it’s not the ideal solution. I have to know my target time before I start, and I lose some of the slope detail, especially around the gloriously flatter hairpins.
Importing a TCX or GPX of just the route exported from RideWithGPS populates the workout editor with one line per KM. This is, frankly, not good enough for my needs. First, it completes a 10k warmup and the climb in about 37 minutes. I wish I was that fast.
Stretching out the interval times for the climb in the GPX import can get me to the +/- 70 minutes I usually need to complete the climb, but then all I have are a handful of 4.5 minute intervals of 7-11% grade - and my tired old legs need those flatter hairpins.
I feel like the ideal solution for the current workout builder would be to increase the resolution of the sample size during GPX import. I realize this is the most computationally expensive and therefore operationally costliest part of the operation. I’d be willing to pay an additional premium to do this on your machines. If that’s not possible, is there a tool I can use to convert GPX to the TrainerDay workout format locally? I know about gpsbabel, etc - but they don’t do any slope calculation.
Also, what is the maximum number of lines per workout?
The ultimate solution would be to import a GPX and build a logical model of the slope that is independent of interval time, the way Zwift does it. I suspect that would entail a lot of reengineering on the backend though.