You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This function breaks a number of invariants, and it is dangerous to call in parallel with iteration over the Ephemeris. It may be a remnant of a time when we were concerned about memory usage. One option would be to not serialize the deep past, so when reloading a save memory would be reclaimed, but without ill-effects on the actual computations.
The text was updated successfully, but these errors were encountered:
Better idea: have a lock system that ensures that the interval [t, +∞[ is not affected by ForgetBefore when computing a prediction that starts at t. Acquire at the beginning of the prognostication loop, release at the end (probably using an RAII object). ForgetBefore can only discard the unlocked part of the timeline. Locking can fail if t has already been forgotten.
This function breaks a number of invariants, and it is dangerous to call in parallel with iteration over the Ephemeris. It may be a remnant of a time when we were concerned about memory usage. One option would be to not serialize the deep past, so when reloading a save memory would be reclaimed, but without ill-effects on the actual computations.
The text was updated successfully, but these errors were encountered: