Weird bug that I can't figure out

[KGS]

So I have a block called a Stopper that whenever it is placed registers its own position within some datastructure called the Tracker. The tracker itself keeps track of the number of stoppers in a chunk. When a Stopper is placed, the Stopper tells the Tracker where the Stopper was placed. The Tracker takes the block coordinates AND the worldID and converts them to chunk coordinates, and then sticks the chunk coordinates and the world id in a custom class (ChunkLocation) which is used as a key in a hash table.

 

The key is used to identify an integer in the Tracker's hash table representing how many stoppers are currently in a particular chunk in a particular dimension. The reason for this is I want some way to quickly look up whether or not there is a stopper in any particular chunk. I suspect this is not the ideal way to keep track of this.

 

// This code is in the Stopper
public void onBlockAdded(World world, int x, int y, int z) {
KGBlocks.tracker.addStopper(x, y, z, world.provider.dimensionId);
}

 

// This is the code in the Tracker
public void addStopper(int x, int y, int z, int w) {
ChunkLocation t = getLocation(x, y, z, w); // This creates a new ChunkLocation from the given coordinates.
if (chunk.containsKey(t)) {
	int counter = chunk.get(t);
	counter++;
	chunk.put(t, counter);
} else{
	chunk.put(t, 1);
}

this.markDirty();
}

 

The bug that occurs is that if I place multiple blocks in quick succession, it seems to register more Stoppers than I place. I get duplicate updates to the hash table. The integer sometimes updates twice instead of once.

 

I currently don't have code to remove stoppers from the tracker, but that will be added when I have this code working.

 

Edit: Maybe it would be safer to store lists instead of integers, and then make the lists contain coordinates for stoppers within a chunk. Checking whether or not there is a stopper within a chunk would then only require checking whether or not that list exists within the tracker OR if the list length is zero.

[GotoLink]

You can probably

synchronize(chunk) {
chunk.put(//stuff)
}

before putting stuff in it.

That would get rid of duplicates.

[KGS]

diesieben07: My ChunkLocation has both hashCode and equals properly implemented. I don't have the code at hand at the moment, but hashCode implementation is based on the one used in Effective Java.

 

The Multiset looks interesting. This is probably going to be my approach.

 

Stopper doesn't have a tile entity, but your solution proposal is interesting.