Thanks for y'all inputs and suggestions, but so far none of the suggested workarounds have solved the issue in my particular case. My current workaround is to switch my audio device from speakers to headphones when the bug occurs, and then back to the speakers again, which for some reason magically fixes the sound (although Spotify is still kaput and has to be restarted after).
Happy to provide any additional requested information wrt. my hardware & software setup.
Technical detail: This happens because the shape is not updated until CaveVinesBlock::randomTick() is called, which happens ... well ... randomly. You can see this happen if you bump the random tick speed up to some higher value to accelerate growth ticking.
Source: IDA
From my analysis of the server in IDA, I believe the intended behaviour was to send an empty string if the item name wasn't changed.
For some reason all variants have a hardness value of 0.3, instead of the correct 2.0. I stumbled across this while implementing these blocks in a custom server.
This appears to be fixed as of 1.18.12.
This bug is related to MCPE-80865
I've also encountered this on a third-party server using a synthetic skin generated by the server.
My observations are as follows:
Only 64x32 skins are affected
Only the Steve model is affected (Alex works correctly) - seems like this is only true in the model selection menu; after selecting, it's covered in garbage
A random skin chosen from planetminecraft shows this bug quite easily. I used this one: https://www.planetminecraft.com/skin/boy-with-shading-64x32/
Correction: disabling vsync via Nvidia control panel did not make any difference; however setting `gfx_vsync:0` in options.txt did seem to work.
I have this issue with a Dell XPS 9560, but only when using the dedicated graphics card (nVidia GTX 1050). Using the Intel HD integrated graphics significantly reduces the latency.
Disabling vsync did not make any difference in my case.
Reproduced in 1.16.40. In JE coral blocks take about ~7sec to break by hand, in BE ~35sec. The hardness value in BE is 7.0 instead of 1.5 as per JE.
This is caused by images which don't have an alpha (transparency) channel. The game shouldn't be accepting such images to begin with, since other parts of the game choke on it.
You want to make sure you use PNG32 (32-bit colour) when you save a modified skin image. If your image has 24-bit colour, it doesn't have an alpha channel and the game will choke on it.
The fix for this has apparently been reverted in 1.11.0.8.
In 1.9.0 this issue is now utterly infuriating due to automatic text wrapping. Attaching a screenshot. Notice how the "fubar" ends up on the second line in the UI, while it's on the first line in the actual sign in the background.
[media]
This may relate to https://bugs.mojang.com/browse/MCPE-41671.
Do you have a copy of the world I could examine? I have a suspicion this might be related to https://bugs.mojang.com/browse/MCPE-41671
Please be aware that PocketMine-MP (and forks) are not Mojang-supported software, you might experience issues with them that don't appear in vanilla gameplay. (I notice a PocketMine-MP command in the second screenshot which does not exist in vanilla.)
Reproduced on 1.2.14.3 beta
[media]
I obtained this stack trace from a crashdump. It looks like it has something to do with crafting grid rendering.
https://gist.github.com/dktapps/19546169b482de86635d11eec8edc434
A full crashdump can be provided if interested.
I've been able to work around the issue doing the following:
-> Go to Device Manager
-> Sound, video and game controllers
-> Right-click Realtek USB Audio
-> Update driver
-> Browse my computer for drivers
-> Let me pick from a list of available drivers
-> Choose "USB Audio 2.0" from the list and install it
While there is still an audio stutter at ~40% loading, sound on the computer continues working normally. Seems this driver is provided by Microsoft rather than Realtek, which likely explains why the issue isn't present.
Maybe this helps someone else while we wait for a fix from Realtek. I've tried all drivers as far back as 6.3.9600.2291 (2021) and still had the same issue.