Over time due to changes, it's possible that the output file logging can get borked. There's no way to identify this right now, but we ought to add in some verification runs (that perhaps run against real hardware/systems) to fix this.
Ideas are:
building full executables, and running a few full sessions (such as the permutor-cli and benchmark) to catch most issues
verifying that log files contain data (no zeros when not expected)
running this on all supported platforms (Windows, Linux)
have pre-stored verified run outputs (i.e. log files, or expected fps parameters) that are compared (to catch any calculation breakage that affects fps calculations mostly)
Over time due to changes, it's possible that the output file logging can get borked. There's no way to identify this right now, but we ought to add in some verification runs (that perhaps run against real hardware/systems) to fix this.
Ideas are: