Lookup structure for handling future events (time based)

Question

I am looking for an efficient data structure, that\'ll allow me to cue events ... that is, i will be having an app, where at any time in execution, it is possible, that an event

Answer 1

How big is N? How often do you have to insert & remove items, compared to everything else going on? If this is more than 10% of total execution time, and if N is typically more than 100 (say) maybe, maybe it's time to be concerned about big-O. I've seen programs with priority queues implemented with fancy container algorithms, allocating iterators, hash maps, heaps, etc. and spending all their time in creating and releasing abstract objects, where the median queue length was like three.

ADDED: OK, since N ~ 10^6, and the frequency is ~ 100hz, you probably want some sort of binary tree or heap with O(log(N)) insertion/removal time. If you're willing to devote 1% of CPU time to this, that is 10^6 microseconds * 1% / 100 = 10^2 microseconds / operation. That should not be at all difficult, because if typical search depth is 20, at ~50ns per comparison, that is ~1 microsecond to do the search. Just make sure to keep it simple, not getting all wrapped up in abstract datatypes. You shouldn't have to worry much about time spent in allocating/freeing tree nodes because you only allocate/free one node per operation. Rebalancing need not be done frequently, like maybe only after every 1000 operations. If you can collect your insertions in batches, and then insert them in random order, that may prevent the tree from getting too unbalanced. If many of your events are simultaneous, you could add a small amount of noise to the time code, to prevent parts of the tree from becoming more like a linear list.