Nick Bastin
- nbastin
- nbastin
- Europe/Stockholm
- Yes
- No
MC-168488
Normal
Some (? only one so far) end cities are generating in ways that look weirdly corrupt (screenshot attached). These are completely newly generated chunks in vanilla 1.16.1 (server), with vanilla 1.16.1 client. The server was upgraded from 1.15.2, and 1.14.4 before that (just in case something about generation is inherited).
As you can see in the image, a section of a structure has generated on the top of another structure that should have nothing on top of it (at the very least the top structure should probably have a floor).
2024-07-11_03.39.56.png
Nick Bastin: This is a bug tracker, not a discussion forum, that is what we have Discord and Reddit for (see below).
John Lucas: Are you absolutely certain that there is no datapack or command block constantly resetting the gamerule?
Quick Links:
📓 Issue Guidelines – 🛠 Community Support – 📧 Customer Support – ✍️ Feedback and Suggestions – 📖 Game Wiki
💬 Mojira Subreddit – 💬 Mojira Discord
@Nick Bastin, flint and steel, lava bucket are not regarded as player kills, just because the game can't record which block is placed by which player.
However, it can be done for TNT and end crystal.
In what world is that bug a duplicate of this one? This happens all the time, and it's not related to lag (it happens on single player as well).
What I am typically doing when this happens is chewing up ore blocks that were silk touched and converting them to their real pieces - I don't know if it matters that xp and such are at play here. I can reliably reproduce by:
It usually doesn't take long before you get slid over on the block you're falling onto (the camera never changes for me, unlike some of the other reports, so I don't know if that's indicative of something or not). If you were right in the middle of the block you probably won't fall, but if you keep going you will.
Also fwiw, this definitely happened in 1.14.4 - 1.15.1 doesn't change anything about the behaviour.
That is really interesting - I've never seen that particular behaviour. When I fall, I hit the ground (whether I die of course depends on normal factors).
So, I got a new PC and can confirm at least anecdotally that system specs appear to matter. I have a Threadripper 3960X and this never happens in single player. It still happens on my laptop which is an i7-8565U (which is otherwise still plenty for playing minecraft).
We have had this problem. It would be nice if the game did not create maps that were never used (e.g. when you use the cartography table to go from 1:1 to 1:16, you waste a bunch of IDs that can never be created again). One mechanism would be to allow users to create zoomed out maps before they used them - let the cartography table make empty 1:2-1:16 maps that could be used on demand at the requested zoom level. Alternatively support an internal map defragmenter that removes map IDs that don't exist in your world and rewrites the existing ones indexed from 0 (we have a custom script that effectively does this).
I'm not sure why "awaiting response" is "resolved", that completely screws up people who are searching for the same issues such that you can't aggregate them properly, plus it's flat out rude.
We are having the exact same problem as shown in this ticket - Zombified Piglins will attack you forever regardless of the status of the gamerule. UniversalAnger seems to work properly, but this is almost worse because it means there's no incentive for people to clean up angry pigmen, and the new anger spreading almost always means that new pigmen have spawned when you're trying to deal with the old ones, making an endless stream of hell that can only be resolved if you manage to despawn them without unloading the chunk.
I have reproduced this with 2 portals in vanilla 1.16, upgraded from 1.15.2 from a world that was started in 1.14. Most portals are ok, but we have 2 that create phantom portals when you go through them after you try to move them (the new locations are fine from the nether to the overworld, but when you try to come back to the nether you appear at the old location, with no generated portal).
If you are "awaiting response", the status is NOT RESOLVED. You are obviously waiting for something. If you would like someone to respond, it would help if you wouldn't also tell them that their ticket is resolved.
The server is vanilla 1.16.1, the client is vanilla 1.16.1. I'm not sure what other "vital" information you could need. You do understand that by nature of the problem in this report, capturing it with a screenshot at all is non-trivial? (Since you'll be, well, getting your face smashed in while you try to take it). There's nothing in F3 other than this information that is relevant to this ticket.
Then the bot should set "Awaiting Response" to resolved after an appropriate amount of time (which I think it actually does). Setting it to resolved immediately ends up in an offensive feedback loop where people don't find tickets, report new ones, and then get dumped on for filing duplicates and not searching. I have no problem with asking for a response and then dumping the ticket after a few days as a policy. Setting a ticket to "Awaiting Response" and then setting it to "Resolved" immediately is nonsensical - it's obviously not resolved, since someone asked for more information.
You understand that the set of "resolved" reports in this tracker is more than 100k items, even when restricting to a single project? Sure, you can search for "resolved" and "won't fix" or "works as intended", and indeed I do, but there is no way a reasonable person thinks that they need to search for "resolved" and "awaiting response". Those are mutually exclulsive states.
I can't seem to find the text on the wiki that says "your portal can dump you in a location where there is no portal, and it will not generate a portal".
So, I'm inferring from your comment that you're only supposed to get subtitles for sounds that are on? This is not the case for 1.15, so maybe my ticket is really that the behaviour has changed. I'm attaching some screenshots - one from 1.15.2 where you can see that all the sounds are off, but subtitles are still appearing, while that behaviour has disappeared in 1.16.
Unfortunately, if this wasn't the case there'd be no way to get from Fancy to Fast on systems that don't support Fabulous. This UI control is very annoying in general as you can't select the mode you want, and are forced to go through this warning every single time you change from Fancy to Fast.
Screenshot attached.
Is there some place where we can find some clarity on the intentions here, other than filing tickets and getting the form response? (e.g. for what is a bug and what is intended as a limitation of the nether)
Brewing stands not being craftable apparently was a bug (after being marked as intended), while stone cutters and grindstones are not?
I have managed to figure this out after tinkering with our world for a few weeks on and off. It is not a game bug, but rather a filesystem data corruption issue, although it would be nice if there was some kind of server command / utility to fix the problem.
The issue (for us, at least) is that at some point in the past (during 1.14, actually, the only reason we found it in 1.16 is we were moving a lot of portals around) a few poi files got out of sync with their region files (probably actually a bunch due to an L2 arc panic, but the nether portal ones manifest in an obvious way to users), such that nether_portal locations in the poi existed, but the portals weren't actually there in the region. Once that happens, there's no way to get rid of those portal locations in the POI file from the game, because making a new portal even at the same location just makes new entries, and deletes those entries when you remove it, but you're still left with the stale ones (and they get used when you use a portal, and don't generate a portal when you come through on the other side since it thinks one is already there, thus creating this "phantom portal" experience).
I managed to fix one up by editing the POI mca with NBTExplorer, although I don't claim to understand enough about the format of that file to be sure I didn't create new problems (I just deleted the chunk branches entirely, rather than trying to edit their values). Are these files safe to delete and just have the game regenerate them (which would obviously be the best solution right now from a server admin point of view)?
EDIT: Just to be clear, the files themselves are not corrupt individually, they're just in a state where there is a logical de-sync due to one of them being written and one of them not at some point in the past.
This has been the case since 1.16 - it breaks our smelter that has a configurable number of furnaces (by having a piston move the rail up to provide a new endpoint after each bank of furnaces). We've just had to adapt to either not using the system or dealing with placing the rails back.
Edit: in our environment this happens 100% of the time, with unpowered rail. From our perspective it seems to be related to any rail connect to movement, not just powered rail.
Just to note, of course this is also true of every other distance statistic. On any long-lived server it's pretty easy for this to happen even without elytra.
For what it's worth, indirect tnt kills being credited in stats is the exception rather than the rule. Killing something with flint and steel, lava bucket, pushing it off a cliff, etc., all are not regarded as player kills. This is very useful for disposing of villagers, iron golems, pillagers, etc. without having to work around problematic side effects.
@Dl if you did it by recording who placed the block, now you'll be getting player kills from magma blocks, wither roses, etc. which wouldn't really make much sense. Still, the question isn't really whether it can be done, it's a question of whether it makes any sense. Attaching kills to the person who placed a block would, I think, be quite odd (or at least make stats even less useful - I didn't really kill all of the creepers in the creeper farm, but I did place the magma blocks....why should that count?). Also of course end crystals are entities when you place them, which is quite different from tnt which is placed in block form.
This is evident in the jfr output as well - there are zero minecraft.PacketSent events recorded:
Neither my game, my launcher, nor my server is modified. I presume the bot is complaining that the level was once loaded in Paper, and so there are some NBT tags on entities that are not vanilla. That is not causing this problem... If Mojang is going to reject these types of bug reports, there's really no point in many people trying to help test new builds.
Same message here:
This is terrible - it ruins hundreds of hours spent on building out big farms with substantial decorations. We were told we could upgrade our old worlds, but that was apparently not really true.
This is definitely a problem for upgraded worlds - upgraded chunks in plains biomes will never generate a tulip (at least, we haven't found a place we can bonemeal for tulips yet). Unfortunately,
MC-244716indicates that they are aware of and will not fix this issue. Yet another case of "You can't really upgrade your old worlds, despite what we told you" for 1.18.I bonemealed about 60000 grass blocks in a new (1.18.1) single-biome world and did find some lily of the valley, although not much. There were 19 lily of the valley, in two separate "clusters" - one cluster is 18, and one of....one. They are still inside cornflower teardrops, but you need to find a very large cornflower area to get any reasonable number of lily of the valley spawn points (some cornflower areas are too small to have any, and of course some have such a small number - like one - that making a farm isn't really viable).
I've added 3 screenshots to the ticket of the area, and the two places where I found lily of the valley. While it does spawn, it certainly seems like it is much too infrequent. Also because Y values matter for the flower gradient now, your chances of finding them are ridiculously small.
This happens to me on linux - there's no MS store, Xbox app, etc. to be involved.
This is a locale problem, not a translation problem (it should use the list separator from the locale database, although they are not universally available but in that case falling back to comma seems fine).
This is another locale issue - the dots should be replaced by an actual ellipsis character, which then should be properly replaced based on the LC_COLLATE definitions in your current locale.
I can't reproduce this when using the nvidia X server on linux - I suspect it's related to some driver/compositor behaviour that is specific to the OP configuration.
There is no language list separator - it's a regional configuration specified in locale. It is not universally available, so that can create some issues (although a value is typically available from the underlying OS at LOCALE_SLIST or similar). If you wanted a per-language setting (not sure why, this seems much worse than using the configuration of the local host), you'd have to do something like carve out some PUA space for a custom "list separator" codepoint and have each translation supply it (but again, this would be as unexpected to users as using a comma all the time).
You could use the LC_COLLATE rules, but they've never been as complex as required to really solve the issue. I'm not sure Mojang really wants "Pink, Lime, and Red", versus "Pink, Lime, Red", or, "Pink, Lime and Red", all of which are variants that you'd get in english switching between en_US and en_UK even (and of course you wouldn't want to redo the sort, as the order does matter for these values).
"twitter" is an english word with verb usage from the 14th century and noun usage from the late 17th century. The grammatical usage in the splash has nothing to do with the website.
The java minecraft window (nearly) always accepts keyboard events (even if you clear all the keybindings it still captures some non-configurable ones like F3), thus a dynamic OSK would appropriately show up when the window is focused.
This is a bug in the way that wayland OSKs interact with XWayland, not with Minecraft.
This has been true (and relied upon) for years - bells illuminate illagers (and witches) at any time, not currently participating raid mobs. Changing this would break a lot of server games that rely on this feature.
Seems like a feature request, not a bug.
Was able to get zombie into minecart on wholly powered rails in survival 1.21. They don't want to walk on rails so it takes a bit, but it will eventually pick them up (curves on normal rails are obviously more efficient). The example appears to show unpowered rails, and getting a mob into a minecart that isn't moving is impossible (regular rails allow a mob to bump a minecart and then be captured when it moves).
This change completely sucked - ruined whole decorated builds of farms. PLEASE DONT DO IT AGAIN. Just leave the current map where it is. This (and the ice/snow elevation biome change) completely screwed people who had old worlds that had built up full decorated farm areas which were no longer useful farms when we upgraded. Please take a lesson from this and don't change it again, leave it well enough alone.
Possibly/probably a duplicate/related to MC-254614.