Redundant code

[rparkins999]

Redundant code

[bailuk]

The first loop is to update the access time inside the 1. Level cache. (if the tile is already loaded).
The second loop is to load missing tiles.

I've added some comments to clarify: 222a07a

[rparkins999]

As my comment says, getHandle() always calls through to cache.get() anyway, so you just update the access time twice. I couldn't find any path through getHandle() that didn't call cache.get() before doing anything else.

[bailuk]

The first loop only handles cached tiles. It kind of "reserves" them by setting the access time.

Without this loop it might be possible that, in the second loop, a needed tile gets replaced inside the cache before the call to cache.get(). Because the second loop can put (replace) tiles into the cache.

You are right that cache.get() gets called twice, but that is actually intended here.

[rparkins999]

Without this loop it might be possible that, in the second loop, a needed tile gets replaced inside the cache before the call to cache.get()

I don't see any way in which this is possible. The first thing that is done in the second loop is to call getHandle() on an object of class Tile. getHandle() is private, so it must be the version in TileProvider which is called.

The first thing that getHandle() does is to call getTileHandle() on the same object of class Tile which was passed to it. getTileHandle() is also private, so it must be the version in TileProvider which is called.

The first thing that getTileHandle() does is to call cache.get() on the same object of class Tile which was passed to it. So there is no way that anything else can happen before cache.get() is called.

If you're suggesting that another thread can change tilePositions[i].tile to point to a different object of class Tile, in that case there was no point in calling cache.get() in the first loop to set the access time, because you aren't going to access the object of class Tile that you called it on.