[Experimental] Wrong type of village in the desert

This may be intentional, plains biome border can be just over sand blocks where village generated. Plains/forest doesn't have to have grass blocks in some places. Hope this helps.

I might be wrong here, but I've done some seed searches (just preempting requests for speed running seeds when 1.18 hits) and I don't think I've seen a non plains village.  So it could be the village type selection isn't working due to the 3d biomes or something. (edit: turns out I was wrong, ether that or it changed in 1.17.30.25)

Still a problem in the experimental world at 1.17.40

Seed 2019014307
Coordinates 10915 134 91230

This is possibly because of microbiomes. If a village well (the origin of the spawning) generates on at least 1 block of plains biome, it will generates a plains village even if the surroundings are not plains.

In the process of adjusting my seed finding tools for 1.18 I'm pretty sure I've discovered what's going on.  

When its doing the biome check to place a village its using the world height (ie y=320), when placing the village it's choosing the village type from the biome at y=0.  If the biome  at y=0 doesn't have a village type you get a plains village.

I was able to run a search for villages that would be affected and that search seems to correctly identify when the problem will occur.

I've attached screenshot of 3 examples.

Seed 54278 /tp -208 98 -293

Seed 89360 /tp -267 100 -312

Seed 556420 /tp -224 114 -379

I can provide a list of search results if that would be useful.

Can someone provide an example seed (with coordinates) for 1.17.40.21 Beta where this bug occurs? Since most of the information and screenshots here are from previous betas and it seems that this beta has changed around some of the world gen again. Also, I haven't been able to find any occurrences of this bug so far in this beta.

Resolving temporarily as Awaiting Response. Could someone please provide a seed and coordinates that can be used to reproduce this issue in the 1.17.40.21 Beta?

1.17.40.21 was mistakenly added to this report from [MCPE-142512]. ([MCPE-142512] is actually a duplicate of [MCPE-41332]) So it seems to me that the problem has been fixed in version 1.17.40.21 🙂

It looks like villages now generate in the correct type when the biome they generate in has a village type.

However, 1.17.40.21 villages are generating in biomes they haven't traditionally.  

Seed 4345832 (/tp 131 125 149) has a plains village in a jungle with the closest non jungle biome 96 blocks away.  

I searched for seeds with a temple start and village start in the same chunk to demonstrate its definitely a jungle.

Seed 16198093 (/tp 126 106 126) has a plains village in a swamp with the closest non swamp biome 84 blocks away.

And like before witch hut to confirm the start Chunk biome.

I'm not sure if this is actually the same issue tho, and may even be intentional.  These biomes do have distinct villager clothing, so might be nice to leave in as a rare village find.

I tested the seeds you specified and indeed. In these villages, there is no village slice in its home biome. This looks good, but it shouldn't be (for consistency with the Java version). Since this is a slightly different problem, it is better to create a new report (since these biomes should not contain villages at all, and it seems that only plains villages can be generated in these biomes).

I created MCPE-142840