And those terms may be used not to refer to the system as a whole, but particular tasks within it. Hard constraints on tasks that can never miss a (often nanosecond resolution) time point, firm real-time that can't be missed but which can be delayed, soft real time that can be missed (but only a certain number or percentage of times, or can't be missed twice in a row, etc). Different systems and teams may use those categories differently. For example, the AGC alarms that nearly caused the Apollo 11 Eagle landing to abort were for tasks in the last category, so they were able to continue. There were other tasks on that same real-time system for which missing the timing would have required an abort (or triggered a lot of very loud noises followed by silence).