I'm using the HPSTrackingBunchedMC.lcsim steering file with my own analysis driver, along with a .slcio file of simulated events. The input file uses a coordinate system with z as the beamline, so when my analysis driver accesses the MCParticles, they're using z as beamline, as expected. However, the reconstructed tracks being passed to my analysis driver appear to be using x as beamline, which means the reconstruction process is rotating the frame at some point, but never rotating it back. Is this a fixable bug? Of course, the alternative is to simply send z to x, x to y, etc. for the tracks, but it is definitely more confusing when the MCParticles use a different set of coordinates!