There seems to be some inaccuracies with selecting entities at coordinates at the center of blocks due to some issue with positioning.
Test:
0. Teleport to a coordinate with no decimals at the center of a block such as by /tp 0 0 0 or /tp 18 32 12.
1. /summon Pig ~ ~ ~ {NoAI:1b}
2. /tp ~1 ~ ~
3. /summon ArmorStand
4. /say @e[type=!Player,c=1]
5. Notice the result is the pig, rather than the armor stand which is clearly closer and you are standing inside it directly at its location.
6. /say @e[type=!Player,r=5]
7. Notice that the first result is the pig rather than the armor stand which if it was selecting correctly, should be displaying the armor stand first in the list of results due to it being closer. see MC-97316
8. Repeat test with various entities to see the same results.
Linked issues
Attachments
Comments 15
Related to MC-97316
97316 rather for discussion about selector "r" and here for "c", to differentiate between both.
Would that be okay @unknown?
Should probably mention that c-execution got an directional bias, thus executing it directly at the entity would be more consistent, accurate, easier.
(Correct me if I'm mistaken, you're better at that topic };])
Maybe it's selecting entities from oldest instead of nearest? I can't test it for now, can someone?
@unknown The way it works is that it detects firstly nearest or farthest (depending on whether c got a positive or negative value), and if there are at least 2 entities at the very same position/distance, then it'll fall back to the second detection method, the age(youngest/oldest, dependant on +/-).
ok changed it to focus on c=
Directional bias of c visualized in an image by @unknown.
CB outputs the closest entity. As you can see, it outputs the AS whereas it should output the player, because it starts to measure from its edge/corner, not from its middle.
If the player in this image were to move a little bit more to the right, the CB would finally detect him as the closest entity.
Now it's a dupe of MC-95352
Yeah, better to have one combined post with all the selector stuff in it. Thanks gamerguppy =d
This bugpost must stay open, as it apparently is still running from its edge in pre-4 according to @unknown as noted here
I cannot reproduce this at all on 1.9-pre4. As far as I can tell it is fixed.
Still directional bias for c-selector confirmed, 2 screenshots attached.
Seems to have been fixed in 1.9.1 pre-1.
anybody else still able to reproduce?
Seems fixed! 1.9 pre-1
[Bot] Arisa seems to be different opinion Ö.ö
Seems also fixed/well-changed for me in 1.9 pre-1, detection is now with really tiny values, awesome }=)
Confirmed