In 14w33 and up, FallingSand that is summoned using an execute command relative to a player, is actually summoned relative to the player (e.g. If the player was standing at x=0.5,z=0.5). This didn't used to happen in the snapshots below this, and causes quite a problem.
The problem with this, is if that a block tries to land on a block, but mostly lands on a block that's an airspace, the block will break and fail to solidify.
In the first picture with stone falling down, you can see the FallingSand entities overlap. These blocks will pop, since they land on only a quarter stone, and 3 quarters air, which means that it will break.
In the picture with the sand riding Item frames, it shows how and why it breaks, and how it sometimes prefers to break on one side rather than the other.
EDIT:
Relative summoning apparently doesn't only affect FallingSand, but also every other summonable entity.
Linked issues
Attachments
Comments 9
1. Place a fence down
2. Stand against fence and type this command
/execute @a ~ ~ ~ summon FallingSand ~ ~10 ~ {Block:sand,Time:1}
Notice how if you execute command on top of fence or on block beside fence, the drop doesn't occur.
The main problem is how the FallingSand doesn't snap on to the nearest block grid, like it did before 14w33a and earlier.
Yes, I have been experiencing that with another project I've been working on.
EDIT: Misread, currently, because of this bug, summoning entities relative to anything, will summon it exactly at those relative location. There is not a single way to make it snap on.
fix one thing and another breaks... personally i think snapping to midblock was a non-issue...pretty sure that it followed the same rules as teleporting players, in that specifying a decimal value makes it exact, but i never needed to test it.
Still, I just jumped into 14w32d, and in there both FallingSand snaps properly and i was able to make an item summon another (diffrent) item to its exact location... so clearly its not impossible.
The big issue though for me is that Falling sand replacing a block doesn't move to the deleted block's grid location. i don't know if i should list that as a different bug though...
I'm assuming this was the result of fixing MC-64188