Update Maps::getBiomeType and Maps::getBiomeTypeWithRef

[Bumber64]

I've updated the functions getBiomeType and getBiomeTypeWithRef based on reverse-engineering of v50.11.

The only real changes seem to be:
potential_tropical (rather than tropical) is sufficient for OCEAN_TROPICAL
region.vegetation isn't used at all, instead replaced by region.rainfall checks

getBiomeType and getBiomeTypeWithRef now do bounds checking on the region coords by calling getRegionBiome, then doing a CHECK_NULL_POINTER on the region. I'd prefer to return biome_type::NONE here instead of throwing an exception, but that doesn't exist so we'd have to add that to structures if appropriate.

I indented the DFHack and Maps namespaces in Maps.h, and inlined getBiomeType there.

[Bumber64]

Anybody know why getBiomeTypeWithRef exists in the first place? Embark assistant uses it to check a region biome using an adjacent tile's tropicality/latitude (with all other regional parameters remaining the same,) but I'm unsure of the reason for doing so.

[ab9rf]

i'm trying to understand the motivation behind this. is the current implementation incorrect in some way?

for the history, see #1392 and #1394

[Bumber64]

i'm trying to understand the motivation behind this. is the current implementation incorrect in some way?

For one, the old code was reporting a regional ocean tile as OCEAN_TEMPERATE while the new code says it's OCEAN_TROPICAL. (Swapping region.vegetation for region.rainfall checks may or may not have created changes. Haven't tested this since I ran into the ocean issue first, meaning there are changes regardless.)

I used a lua script iterating all region tiles to test, but found it too much a hassle to verify what the game UI says (even in a pocket world,) since there's no keyboard support for embark screen and I'd basically have to use trial and error to embark on the right tile to verify region coords. Just going to assume the current disassembled function (FUN_140bfe460 v50.11 win64 Steam) is correct on oceans, otherwise there's a Bay12 bug.

AFAIK, getBiomeTypeWithRef (where we use region_ref_y (world_ref_coord_y) for tropicality) doesn't exist in Bay12 code, so I'm just swapping the ref coord in where the old DFHack function does and have no way of verifying it without knowing the intent.

[myk002]

do you have a map where I can see the changes in interpretation? I tried to verify on the maps I have, but I'm not sure how to specifically get a biome with potential_tropical set.

[Bumber64]

do you have a map where I can see the changes in interpretation? I tried to verify on the maps I have, but I'm not sure how to specifically get a biome with potential_tropical set.

Using this branch: https://github.com/Bumber64/dfhack/tree/test-biomes
Generate a Pocket Region with North pole and seed test.
Use this Lua script:

function test_biomes()
  local wd = df.global.world.world_data
  for x=0, wd.world_width-1 do
    for y=0, wd.world_height-1 do
      local b_new = df.biome_type[dfhack.maps.getBiomeType(x,y)]
      local b_old = df.biome_type[dfhack.maps.getBiomeTypeOld(x,y)]
      if b_new ~= b_old then
        print(("(%d, %d): %s -> %s"):format(x, y, b_old, b_new))
      end
    end
  end
  print("Done!")
end

Output should be:

(14, 11): OCEAN_TEMPERATE -> OCEAN_TROPICAL
(15, 11): OCEAN_TEMPERATE -> OCEAN_TROPICAL
(15, 12): OCEAN_TEMPERATE -> OCEAN_TROPICAL
(16, 11): OCEAN_TEMPERATE -> OCEAN_TROPICAL
(16, 12): OCEAN_TEMPERATE -> OCEAN_TROPICAL
Done!

These regions should be potential_tropical but not tropical.

[myk002]

nice. thanks!