Undo of big pastes does not remove all blocks

[PapiCapi]

Server Implementation

Paper

Server Version

1.18.2

Describe the bug

The //undo for big structures does not seem to remove all the blocks pasted.

To Reproduce

1. Copy a big amount of blocks
2. Paste it somewhere ( for example in the sky or in the void )
3. Undo
4. You should see blocks flying around

Expected behaviour

Clear all the blocks pasted.

Screenshots / Videos

https://www.youtube.com/watch?v=POocbUssI9s

Error log (if applicable)

No response

Fawe Debugpaste

https://athion.net/ISPaster/paste/view/1245dfd6feb14454a9f6596869f929f3

Fawe Version

FastAsyncWorldEdit version 2.4.1-SNAPSHOT-237;dbfb521

Checklist

• I have included a Fawe debugpaste.
• I am using the newest build from https://ci.athion.net/job/FastAsyncWorldEdit/ and the issue still persists.

Anything else?

Actually my build isn't the newest but the last two were updates of paperweight and some docs modifications as far as i see.

[PierreSchwang]

Is this just a visual bug, or are the blocks still present after a rejoin?

[PapiCapi]

They are still present after a rejoin

[NotMyFault]

What was the amount of blocks you did this with? With 150 billion blocks I had no issue, it just takes a few secs longer to send all the data to the client

[PapiCapi]

I tried with ~3 millions. I can reproduce it really easily so it maybe comes from another plugin ?
Anyway, i got a schematic to send you if you want to test with it. I can't add it here on github

[anonyco]

Please send the schematic here in GitHub.

There should be a button at the bottom of this text box where you type the comment that says "Attach files by dragging & dropping, selecting or pasting them." If you are unable to find this button, then you can drag and drop the file onto this comment text box.

This doesn't feel like the sort of issue that would be caused by particular blocks (I would expect errors being thrown in the console instead of silence), but it's very worth trying out the schematic.

[PapiCapi]

undoBug.zip
I could't add the simple .schem file so after zipping it, github seems to be okay with that file extension

[anonyco]

Ah. That makes sense. Many thanks for sending the schematic file. I am looking into it.

[PapiCapi]

So i just re-tried with only a few plugins ( FAWE, Multiverse-Core and PlotSquared ) and the bug is not showing at the first //undo.
I needed to paste and undo one more time for my bug to appear.
I also tried to enable the debug output but nothing more in the console.

[CJBlakey]

I'm experiencing a similar issue, but with a relatively small paste (a gateway for a castle, maybe a thousand or so blocks tops)

[Folas1337]

Same has been happening here throughout several months by now, just thought I'd add my experience to this issue at least :)

It's just happened yesterday for a user with a massive schematic (like 300 million blocks) where after using "//undo" small fragments where left over seemingly at random.

[PierreSchwang]

undoBug.zip I could't add the simple .schem file so after zipping it, github seems to be okay with that file extension

I'm able to replicate the issue using that schematic multiple times

[OneLiteFeather]

It doesn't happen when you do //paste without -e and then //undo. It always happens when you //paste -e and //undo (sometimes need to be done twice in a row) on these schematics, I guess it is an issue with entities, some TileEntities break as seen in my debugpaste here.

Debugpaste with log and stacktrace:
https://athion.net/ISPaster/paste/view/dc7b33a8f48d4ad0b46728aeb48f8fbe
Maybe this can be helpful for developers