Not using the `RecheckCartHack` on >5.9 seems to help with #4670 - not tested on older minetest; chest minecarts might still be empty there when the block is unloaded in the meantime. For <5.9, maybe it helps to decrease the time interval, 3 seconds seems to fairly long.
This also makes the minecarts random: 40% minecart, 40% chest minecart, 20% tnt minecart.
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4684
Reviewed-by: the-real-herowl <the-real-herowl@noreply.git.minetest.land>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
- peaceful structure spawns would not run in peaceful mode (e.g., parrots)
- water structure spawns (e.g., guardians) would not run because the code required air above
- small code improvements
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4607
Reviewed-by: the-real-herowl <the-real-herowl@noreply.git.minetest.land>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
No functional changes, just more vector API, which supposedly is faster?
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4567
Reviewed-by: the-real-herowl <the-real-herowl@noreply.git.minetest.land>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
Sometimes, pools would spawn deep underground and then produce deep holes.
I noticed that such pools would be at +48, at the block boundary.
IMHO this may be an error in minetest surface detection.
Nevertheless, here is a workaround: require air above pool spawns.
Also clean up the pool code
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4571
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
As spawning happens via gennotify anyway, we can omit placing a
structblock right away.
This also avoids certain cases of holes in snow cover or water.
Plus, the code is simpler.
Isolated from the big mapgen overhaul, for the main branch.
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4619
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
As witch huts use flag "liquid_surface", place_on only can be water. If we want on-shore witch huts, this needs to be solved differently.
Also, probably no witch huts in deep ocean swamp water?
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4605
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
- breaking cactus will drop randomly in x=-0.75..+0.75, z=-0.75..+0.75
- breaking cactus will have an initial velocity in this direction
- if a larger cactus break, they break into the same direction
- cactus growth rate reduced 4x, to debuff farm efficiency (will need a larger rebalancing)
- cactus only spawns when there is air surrounding it and above, so it does not immediately break
- slightly increase the frequency of cactus to counter this
The first changes make cactus farms possible.
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4581
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
Mangrove tree variants 4, 5 and bees nest were not post processed with root growth.
Reviewed-on: https://git.minetest.land/VoxeLibre/VoxeLibre/pulls/4584
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: kno10 <erich.schubert@gmail.com>
Co-committed-by: kno10 <erich.schubert@gmail.com>
Some mapgen settings used to crash the game.
Reviewed-on: https://git.minetest.land/MineClone2/MineClone2/pulls/4276
Reviewed-by: Mikita Wiśniewski <rudzik8@protonmail.com>
Co-authored-by: Bram van den Heuvel <bram.vdh1999@gmail.com>
Co-committed-by: Bram van den Heuvel <bram.vdh1999@gmail.com>
This commit adds back the old `set_node` method for the foliage, since the foliage is much more difficult to work with via the VoxelManip method due to them being part of schematics that could span across mapblocks in some cases.
The old method will complement the new one by running after the VoxelManip has done its job, and fixes any foliage that the VoxelManip had missed.
This commit adds some code to the `set_foliage_palette` function which checks for the `param2` of the foliage node in question to see whether or not said `param2` value is `0` upon first being generated.
If it isn't, then it's safe to assume that said foliage is a vine, and therefore needs to use the other method of calculating the final `param2` value.
