Geoffreylikespie

I haven't done (and don't know how to do) any code analysis myself, but I've heard from someone more knowledgeable that the bonemeal pattern follows the first configured feature of type flower defined in a biome. forest_flowers is defined before flower_default in these biomes' json files, but forest_flowers is a simple_random_selector feature rather than a flower feature — seemingly because then it can then choose between patches of flowers and have a consistent type between them on world generation (as opposed to default flowers, where each flower is randomly a poppy or dandelion).

I think grouping the tall flowers (already renewable in themselves) in one feature and the short flowers in another would be beneficial, though I believe the short flower feature would either need to be a gradient in its own right, or lilies of the valley would be able to generate in bonemeal applied in forests regardless of whether they were there initially.

Alright, thank you for the clarification!

I'm confused how this is considered a duplicate. The linked bug, MC-123739, refers to the order of items within any given category of the recipe book. My bug report here is that I believe some items were put in the wrong category altogether. These are two distinct things, with their only similarity being that they're both bugs with the recipe book. (Admittedly, I used confusing language when I wrote "surrounded by", which I apologize for if that made my report unclear.)

With the new /tick command's functionality, I think this bug could finally be fixed relatively easily. I believe the issues at the source of this problem are that previously, having mobs with AI inside jigsaw blocks wouldn't have been possible, and the mobs would have needed to be contained by barrier blocks while their structures are being saved. This would have allowed some mobs to move towards the edge of their 1x2x1 barrier enclosure, which causes slight overlap between their hitboxes and the adjacent blocks.

Now, however, someone can run /tick freeze and then recreate these jigsaws pieces containing entities. You can summon mobs with AI sitting at the same height as the jigsaw block, since they won't be forced out while ticks are frozen. Likewise, mobs can't move to the edge of their barrier enclosures while ticks are frozen (barriers wouldn't even be needed), thus preventing any overlap with adjacent blocks. If entities are summoned at the bottom of the jigsaw block and at the block's horizontal center, and then those jigsaw pieces are saved to replace the old ones, this should fix both villagers getting stuck in the ceiling and villagers getting stuck in walls. I believe this would also fix issues with village-generated zombie villagers, pigs, cows, and sheep. It might not fix issues with any village-generated iron golems, horses, or camels, as those entities all have hitboxes that pass the bounds of a 1x2x1 volume.

I have tested a tiny bit in creative: structure blocks can still be saved and loaded while ticks are frozen, and after unfreezing ticks these mobs loaded from structure blocks have AI and act normally. (I have not confirmed that my proposed fix would actually fix the issue, but if I should do that I can try.)

Can confirm in 1.20.3-pre3. It seems like for both of these inconsistencies the issue is that redstone smelting and blasting recipes have their "category" field set to be "blocks" (where recipes such as smelting cobblestone into stone are) rather than "misc" (where all the other ores and blastable items are). Assuming the game directly references the .json files that one gets from extracting the jar version file, changing the "category" field to "misc" in all 4 redstone smelting/blasting recipes' .json files should fix this.