DrMorphism
- unblock
- unblock
- America/Vancouver
- Yes
- No
REALMS-11027
MCPE-168925
This bug has completely broken the game, has been out in public for more than a month and it the second highest voted issue. Cannot believe that it is not patched yet. I have stopped playing and it is getting very frustrating to keep coming to this page and seeing no updates 😥
It has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the gameIt has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the game
It has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the game.It is unclear if the game not saving consistently when the last player signs out or if the saves
It has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the game.It is unclear if the game not saving consistently when the last player signs out or if the saves
It has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the game.It is unclear if the game not saving consistently when the last player signs out or if the saves are silently failing. Below is another instance that many hours of gameplay was not saved even though players were active on the server, but after everyone signed out and someone signed back in, the game was reverted and then an old snapshot is saved (the "28m ago" save is when someone signed back in).
It has recently happening frequently to us that when everyone signs out of the realm and then we come back online the work in the last couple of hours are reverted without the game having crashed.
The screenshot above is one of the examples when the last save is 2+ hours before the last person signing out of the game.UPDATE: It is unclear if the game not saving consistently when the last player signs out or if the saves are silently failing. Below is another instance that many hours of gameplay was not saved even though players were active on the server, but after everyone signed out and someone signed back in, the game was reverted and then an old snapshot is saved (the "28m ago" save is when someone signed back in).
image-2022-12-14-18-11-30-513.png
The reason explained for marking this "As intended" seems to use circular reasoning fallacy https://bugs.mojang.com/browse/MCPE-28890?focusedCommentId=449680&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-449680
We have had a discussion about this and it appears that it's working as currently intended. It's true that a hopper always pushes an item from the first non-empty slot, however this doesn't necessarily translate to preserving the order in a whole series of hoppers. Intuition would suggest that it should, but intuition is wrong in this case.
The problem occurs because all the hoppers are pushing on the same tick simultaneously, but the game can only update them one at a time, and it updates them in random order. Consider this scenario: Hopper A points to hopper B and each hopper has a single item in slot 1. When a redstone tick happens, the game can update either hopper first. If it updates hopper B first, hopper B pushes its item from slot 1 to somewhere, then hopper A pushes its item from slot 1 into hopper B, where it goes into slot 1 because it's now empty. That's the way you expected it to work.
It explains that because hoppers are being updated randomly, this bug happens and so it is as intended. But the bug is that the hoppers are being updated randomly. If the hoppers do not get updated on a random order and instead based on the order they are connected to each other, the bug will not exist as evident by how Java does not have this bug.
For a hopper chain:
A->B->C->D->ChestIf the hoppers are processed in the inverse order in the chain
D, C, B, A then there will not be any issue.In a general case, if a directed graph is created based on how the hoppers are connected to each other, as long as there is no hopper cycle in the connections, the sequence in which the hoppers need to be updated can be easily calculated.
For cycles it might not seem trivial
A->B->C->A
but also logically there is no need for a predeterministic order in a cycle, as long as the hoppers in each cycle are processed in a fixed order of the cycle starting from any point it is fine.
The code can keep the directed graph in the memory when the corresponding chunks are loaded so the order could be preserved.
-----------------
Even if the hoppers are processed in a random order, as long as the order does not change every tick, again there will not be an issue in most cases. I imagine the code is not literarily using a random function to determine the order and most likely is using a data structure like a set that does not preserve orders when new items are added or removed, so if the data structure holding the hoppers change to something that keeps the orders stable, then this issue will not persist.
Repro steps:
- Place 3 hoppers facing north
- Place a hopper one block higher, facing north, but one block south of the other three
- Place a chest boat on one of the lower hoppers
- Hop in the boat and ride it so it is on top of all three hoppers
- Sign out/In
- Boat moves a little backwards
After reload, you can ride the boat forward again and if you log out again and log back in, it goes a little backwards again.
Once the boat has gone backwards enough that can't unload items into the first hopper, it stops moving
. Seems like a floating point rounding issue.
Sample world file attached, you can ride the boat forward, log out/in and see the boat going a bit backwards from where it was left
Repro steps:
- Place 3 hoppers facing north
- Place a hopper one block higher, facing north, but one block south of the other three
- Place a chest boat on one of the lower hoppers
- Hop in the boat and ride it so it is on top of all three hoppers
- Sign out/In
- Boat moves a little backwards
After reload, you can ride the boat forward again and if you log out again and log back in, it goes a little backwards again.
Once the boat has gone backwards enough that can't unload items into the first hopper, it stops moving:
Sample world file attached, you can ride the boat forward, log out/in and see the boat going a bit backwards from where it was left. It seems like a floating point or rounding issue.
Indeed, many thanks to DrMorphism! If you hadn't discovered the OneDrive connection, this might never have been solved because the problem is likely irreproducible in the office network environment Mojang uses. (OneDrive is designed for and used almost exclusively by home users.)
As it happens, I upgraded my PC a couple of months ago and chose as part of Windows 11 setup to start using OneDrive to back up my Documents folder, which causes files in that folder to be redirected to OneDrive. I have already exported several of my worlds, including my 4 year old survival world, only to discover today that those backup copies are useless. I exported a small world today and was about to delete it from my world list with the intent to re-import it later, only to discover that the backup was of size 0, which is how I found out this report exists.
This is a serious problem that can lead to the total loss of worlds that users have put a lot of time into developing. Backups should never fail silently for this very reason! I have asked Mojang to review this report and I hope they give it the priority it deserves. I'm sorry it's been neglected so long, but please understand that Mojang can't fix a problem they can't reproduce, and until it was discovered that OneDrive was somehow implicated in the failure, reproducing it would probably have been impossible for them.
Same here, playing on 1.16 on Win 10 bedrock in a Realm.
I lost many villagers. I even used name tags on some of them to stop it, but they are disappearing regardless.
Also in the Nether, I got killed by a Piglins right by the portal so I went back immediately by some of my items despawned.
The issue with the villagers disappearing happens even when I do not go to the nether. I have closed the game multiple times (but waited till save completed) and also went more than 1024 blocks away from the village multiple times.
Happened again (v1.16 on Windows), this time I had downloaded the world so it was not on a realm and I did not go to the nether, I just went around 800 blocks away from the village. I have been building a village with people from different races, it is a lot of work to go different villages and bring people which makes it too heartbreaking to keep seeing them disappear, I need to stop playing till this issue is fixed.
I am experiencing the same issue as well. Mobs sometimes freeze in one location
I have been having this issue for a while now, but I noticed something new today. Sometimes the exps are already collected, but the orb do not disappear. Eg: I have a guardian farm which uses tridents to kill the mobs and then the loots and exp floats towards me. What I noticed is that I do not see any orbs coming towards me, but my exp is going higher which means I am collecting them, at the same time, I see the orbs being rendered as stuck where the guardian is killed. When I go to where they show up, none of them are collectable (which makes sense as I have already collected them), but they do appear till I move them somewhere and run away from them till they eventually stop being loaded by the client. I am running client (1.17.2) and server (1.17.1.01) on Windows 10 and tried multiple tick distances. It was not great, but since 1.17 the magnitude seem to have increased
Please do not fix this and instead make it a feature! This is a great way of making bubble elevators that look great. Once powder snow is dispensed or placed (using bucket) and washed away with water we can see a water less bubble elevator which works the same as bubble elevators inside water (moves items and players up and down), but look fantastic and are great for building in survival.
The slowness is not only impacting the player whose chunk is being loaded. but the entire server is impacted that makes the lag very bad and the activity on community realms have been severely impacted as some people can't bare the amount of lag.
It could be useful to have a couple of specific objects do this as a part of game mechanic (by design) way of detecting loading. But most of the current list could make redstone designs unnecessarily more complicated.
It happenings to me on a small world (50MB) as well. I downloaded my world from my realm and when I try to export, it creates 0MB file
I tried a different folder and it worked. The original folder was part of my OneDrive and it seems that something about the way that Windows is syncing OneDrive folders is not being handled in Minecraft.
This change seems very counter-intuitive (literal fire, not causing fire damage and instead normal damage). It also breaks farms that use campfire to clean up leftover xp. It does not seem to be a right approach for intended game mechanics.
Still happening on 1.19.50 as well
Happened again and lost another hour of unsaved gameplay
Again happened with multiple hours of game play not saved after everyone signed out
We are having a similar issue:
REALMS-11027but have they same root cause or might just be related. But sometimes the game does not saveI have had this happening to me multiple times and every time that it happened was when the workstation was broken/moved by a piston while the villager was just leveled up and was emitting particles. Does not happen 100% of times, but has happened a handful of times. Seems like a timing issue.
Not sure if it is the same bug or a different bug, but color of the signs also reset. Sometimes signs dyed with a color and a glow ink become non-glowing (keeping the dye) or loosing both the dye color and the glow
Essentially hopper-minecart is not able to remove music discs when from a jukebox that is playing a song.
This is also a parity bug:
What if you first put the boat, put the hopper on the top layer one block off and then push it to the place in the photo using a piston? I think that is how I placed them