[MC-238073] Decorators are independent of world seed

Description:
Decorators are inconsistent in worlds with the same seed

Steps to Reproduce:
1. Generate a world and copy its seed.
2. Make a new world with the same seed
3. Repeat as many times as you like

Expected Behavior:
Each world seed should generate with the same features down to the decorators

Observed Behavior:
Each instance of a seed has a different placement of decorators

World Seed: -8718622960759467126

Notes:
Possible link to MC-219200

Linked issues

relates to 2

MC-240271 Dripstone Cave decorators are different each time ignoring world seed Open

MC-219200 Lush Cave decorator is different each time ignoring the world seed Reopened

Attachments

Comments 21

syarumi 2021-10-02T06:37:39Z

Can confirm, seems to be happening basically everywhere, even dripstone caves.

ampolive 2021-10-02T11:59:35Z

Relates to MC-219200.

ampolive 2021-10-02T18:31:45Z

Actually this is a duplicate of MC-55596.

Brent101 2021-10-02T18:53:41Z

I wouldn't say so, as MC-55596 deals with chunk generation and block data. That issue is quite rare and difficult to reproduce but appears smaller in scale, whereas this issue and MC-219200 deal with all decorators and is relatively easy to mass-produce. MC-55596 does, however, lead to similar instances but on a smaller scale, usually one or two chunks. But if compounded could lead to a more significant difference within decorator or possibly terrain generation.

Ceresjanin123 2021-10-03T10:42:39Z

This is not a duplicate of MC-55596 that bug occurs in rare cases and only affects a small area. The developers have probably made some change to the world generation that unintentionally made the surface builders use random seeds, this means this bug affects every seeds and every chunk.
I will see when this started happening

11 more comments

Ceresjanin123 2021-10-03T11:52:22Z

After doing some experiments I have gathered this information: The bug MC-219200 was actually introduced in 20w49a (the same snapshot the first underground biome was added, dripstone caves) but I only noticed and reported it in 21w10a when lush caves were added. So MC-219200 doesn't affect just lush caves but dripstone ones too altough for some reason it appears to be much more noticable for lush caves compared to dirpstone caves as those keep some parts of the cave unchanged.
But MC-238073 bug did not occur above ground from the first 1.17 snapshot up untill the first 1.18 experimental snapshot ,the first snapshot 3d biomes were introduced (I also tested 1.17.1 with C&C preview), meaning the random surface builders are probably caused by 3D Biomes added in 1.18 Exp Snapshot 1.
My theory is that MC-219200 and MC-238073 are caused by different bugs since 219200 is a problem with underground biomes in general and 238073 is (probably) caused by 3D Biomes

Moesh 2021-10-05T07:14:50Z

Dupe of MC-55596

user-f2760 2021-10-05T07:18:31Z

Not meaning any disrespect, but are you sure? They used to be consistent in older versions; and 55596 is more a rare event rather than being there every single time.

ampolive 2021-10-05T22:35:23Z

This involves the entirety of decorators in all chunks, but MC-55596 only says "some chunks". I think this is distinct enough for it to be not considered a duplicate.

Ceresjanin123 2021-10-06T13:11:52Z

Good thing this bug report was reopened 🙂

Ceresjanin123 2021-10-07T15:09:30Z

Can confirm in 21w40a

Ceresjanin123 2021-10-13T16:57:51Z

Can confirm in 21w41a

MrSpike 2021-10-17T04:36:47Z

The reason this happens appears to be that the second parameter to "WorldgenRandom.setFeatureSeed()" is not consistent. I observed this while reversing the setFeatureSeed() parameters, from a feature that generated in-game. Each time I regenerated a world with the same seed, and reversed the second parameter using the same feature, the second parameter would be different. The first (seed) and third (GenerationStep) parameters are consistent. Due to the way the seeding is done, this often results in the first "WorldgenRandom.nextFloat()" when using "LegacyRandomSource" as the underlying PRNG (or just "WorldgenRandom.nextFloat()" in earlier versions), having a similar value, so the chance decorator will have a similar outcome, but the rest of the feature will be completely different.

Ceresjanin123 2021-10-17T11:40:07Z

Also this first appeared in Exp 1.18 snapshot 1 but affected only lush caves since 21w10a

ampolive 2021-10-20T20:05:10Z

Can confirm in 21w42a.

Ceresjanin123 2021-10-23T10:23:33Z

MC-239247 is a duplicate but I encourage the devs to give it a read as it contains some usefull information even though most of it is already in the comment above by MrSpike

Brent101 2021-10-27T17:24:30Z

Seems to be somewhat fixed with less discrepancy as a whole but still major inconsistencies.

[media][media]

syarumi 2021-10-27T18:29:23Z

@unknown check if what you're experiencing is not MC-55596.

Brent101 2021-10-27T19:29:45Z

It appears that this issue is solved. The discrepancies noted appear to be MC-55596.

Ceresjanin123 2021-10-27T19:41:12Z

I don't think so the discrepancies now are much more severe than before this bug was introduced

ampolive 2021-11-01T17:05:44Z

MC-219200 and MC-240271 lead me to believe that this bug wasn't fully fixed.