--failsafe_threshold in ml_optimiser is no longer used.

[huwjenkins]

The option --failsafe_threshold:

relion/src/ml_optimiser.cpp

Line 1010 in 9085250

failsafe_threshold = textToInteger(parser.getOption("--failsafe_threshold", "Maximum number of particles permitted to be handled by fail-safe mode, due to zero sum of weights, before exiting with an error (GPU only).", "40"));

Was added to the original ‎src/gpu_utils/cuda_ml_optimiser.cu in db46c45 after the refactoring of this code was started in f160b82.

The failsafe method itself had already been removed from src/acc/acc_ml_optimiser_impl.h in 3ab12c1 by the time the option was added to ‎src/gpu_utils/cuda_ml_optimiser.cu(!) and so both this option and ERRNUMFAILSAFE:

relion/src/error.h

Line 285 in 9085250

#define ERRNUMFAILSAFE ("Relion had to use extra-precision fallbacks too many times.\n\n\

were no longer used in the code (45a8ed0) before RELION-4 was released . It probably would be good to remove the unused option?

[biochem-fan]

Thank you very much for pointing this out. Could you please send us a pull request?