Post your modding questions here

[TenPlus1]

Whoops! wrong thread...

[Don]

Could you please move my "Lucky Block" mod to MOD Releases :) Thanks...

viewtopic.php?f=9&t=13284

...and also my Pie mod also...

viewtopic.php?f=9&t=13285

Wrong place.
https://forum.minetest.net/viewtopic.php?f=11&t=10418

[Sane]

Hi there,

I have noticed that the function registered with minetest.register_on_generated gets called more than once. That seems a bit odd to me.

If that behaviour is intended,
what is the differnce between the calls?
which call is the one that counts? (Most probably the last one, but who knows for sure?)
are earlier calls missing information somehow? If so what is missing?
Isn't it a waste of processing time to have map sections generated several times?

To reproduce run a new minimal word with the following test code:

+ minetest/games/minimal/mods/test/init.lua

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

local poscount = {}

minetest.register_on_generated(function(minp, maxp, seed)
   local vm, emin, emax = minetest.get_mapgen_object("voxelmanip")

   local posText = minetest.pos_to_string(emin)
   poscount[posText] = (poscount[posText] or 0) + 1
   
   print("poscount:"..dump(poscount))
end)


[Sane]

...
I have noticed that the function registered with minetest.register_on_generated gets called more than once. That seems a bit odd to me.
...

Nevermind, just found the solution. That thing bugged me for some days now and as it seems actually posting the question seems to have triggered someting ...

The soulution is that I've set num_emerge_threads = 4 in minetest,conf. Leaving this at 1 solves the issue.

[Don]

I want to place a block every 1000 nodes square. It should be somewhat random. I do not want it to be exactly 1000.
The node will be placed in water on mapgen at surface height.

I am not great with mapgen and was wondering what the best way to do this is?

[Sane]

I want to place a block every 1000 nodes square. It should be somewhat random. I do not want it to be exactly 1000.
The node will be placed in water on mapgen at surface height.

I am not great with mapgen and was wondering what the best way to do this is?

Will this be a new world or an exisiting one?

[Don]

New world

[Sane]

New world

ArguablySane has posted a solution:
viewtopic.php?f=3&t=13132&p=191868#p189480

[Don]

New world

ArguablySane has posted a solution:
viewtopic.php?f=3&t=13132&p=191868#p189480

Thanks. I will try to modify it to make it work. Do you know if this works with all mapgens or just v5 and v7?

[Sane]

New world

ArguablySane has posted a solution:
viewtopic.php?f=3&t=13132&p=191868#p189480

Thanks. I will try to modify it to make it work. Do you know if this works with all mapgens or just v5 and v7?

No I don't. But singlenode does not work, the hightmap is always empty.

As I already was fiddeling with mapgen and hightmaps ...
I've patched up a thingy that places a node on the surface, at a random position in the mg block that contains a grid point defined by cell_size. If the mg block is not within y > -200 and < 200 then the node will be placed completely random.

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

--[[
Node placer - Places a node on the surface, at a random position in the mg block that contains a grid point defined by cell_size

published under Creative Commons CC0 license.
--]]

-- cell_size  must be bigger then mg block size (80) to work.
local cell_size = 150
local item_id = minetest.get_content_id("default:steelblock")

minetest.register_on_generated(function(minp, maxp, seed)
   -- If the next_hotspot is not within this mg block then exit
   local next_hotspot = {
      x = math.ceil(minp.x / cell_size) * cell_size,
      y = math.ceil(minp.y / cell_size) * cell_size,
      z = math.ceil(minp.z / cell_size) * cell_size
   }
   if next_hotspot.x > maxp.x or
      next_hotspot.y > maxp.y or
      next_hotspot.z > maxp.z then return end
   
   -- Place the object somewhere in this block

   local vm, emin, emax = minetest.get_mapgen_object("voxelmanip")
   local va = VoxelArea:new{MinEdge=emin, MaxEdge=emax}
   local vd = vm:get_data()
   
   local r = PcgRandom(seed) -- this ensures that you get the same results each time you generate the world anew.
   local item_pos = {
      x = r:next(minp.x, maxp.x),
      y = r:next(minp.y, maxp.y),
      z = r:next(minp.z, maxp.z),
   }

   -- This is to optimize, because getting the height map is expensive
   if minp.y > -200 and maxp.y < 200 then
      local hm = minetest.get_mapgen_object("heightmap")
      local hz = (item_pos.z - minp.z) * 80
      local hx = (item_pos.x - minp.x) % 80
      local height = hm[1 + hx + hz]
      item_pos.y = height + 1
   end

   vd[va:indexp(item_pos)] = item_id

   --
   vm:set_data(vd)
   --vm:calc_lighting()
   --vm:update_liquids()
   vm:write_to_map()
   --vm:update_map()
   --
end)

[Don]

@Sane - Thanks. I got busy with something else and haven't tried yet. Will do it tomorrow.

I have another issue though. I have a mts schematic that contains a node with a formspec. When the schematic is placed the node does not have the formspec. I created another block and then used an abm to place the one with the formspec. This seems to work fine.
My question is, is that the best way to do it or is there a better way?

[kaadmy]

I have another issue though. I have a mts schematic that contains a node with a formspec. When the schematic is placed the node does not have the formspec. I created another block and then used an abm to place the one with the formspec. This seems to work fine.
My question is, is that the best way to do it or is there a better way?

That's how I do it, I couldn't find a better solution.

[Hybrid Dog]

I have another issue though. I have a mts schematic that contains a node with a formspec. When the schematic is placed the node does not have the formspec. I created another block and then used an abm to place the one with the formspec. This seems to work fine.
My question is, is that the best way to do it or is there a better way?

That's how I do it, I couldn't find a better solution.

There's another way: you could execute the function which usually becomes executed after the node appears (http://dev.minetest.net/minetest.regist ... _construct)

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

local on_construct = minetest.registered_nodes[name].on_construct
if on_construct then
on_construct(pos)
end


You would need to loop through the positions where the schematic added nodes.

[Ferk]

-Deleted-

Yep, I didn't read properly. Sorry.

[GunshipPenguin]

I'm having some trouble with the Voxel Manipulator.

Basically, I want the mapgen to generate nodes up to a certain point, then just have all nodes generated past that point filled with an invisible undiggable block. I'm using the following code:

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

local SIZE = 200;

minetest.register_node("limit:limit_node", {
   diggable = false,
   sunlight_propagates = true,
   drawtype = "airlike",
})

minetest.register_on_generated(function(minp, maxp, seed)
   local c_limit = minetest.get_content_id("limit:limit_node")
   local vm, emin, emax = minetest.get_mapgen_object("voxelmanip")
   local area = VoxelArea:new{MinEdge=emin, MaxEdge=emax}
   local data = vm:get_data(minp, maxp)
   
   for i in area:iterp(minp, maxp) do
      local currPos = area:position(i)
      if math.abs(currPos["x"]) > SIZE/2 or
            math.abs(currPos["y"]) > SIZE/2 or
            math.abs(currPos["z"]) > SIZE/2 then
         data[i] = c_limit;
      end
   end
   vm:set_data(data)
   vm:write_to_map(data)
end)


In the code above, If the x, y or z coordinate of any generated block is > 100, it will be replaced with the node "limit:limit_node" which is simply an undiggable, invisible node. This code does work, but I find that certian nodes are not replaced by limit:limit_node like I want them to be. Here's are 2 screenshots of what I'm talking about:

+ Spoiler

As you can see in the screenshots, the map stops when x, y or z > 100 except for a few scattered unreplaced nodes. The unreplaced nodes seem to usually be liquids, dungeons and a few dirt or dirt_with_grass nodes.

I'm experiencing this using minetest 0.4.13 stable with the latest minetest_game.

Can anybody tell me why these nodes are not being replaced? Many thanks in advance.

Also, I understand that I could use map_generation_limit in minetest.conf for this, but it doesn’t exactly work with what I'm ultimately trying to do.

[rubenwardy]

This is, afaik, due to overgeneration, which causes the c++ mapgen to override your blocks. Paramat is the one to talk to about this.

[Hybrid Dog]

This is, afaik, due to overgeneration, which causes the c++ mapgen to override your blocks. Paramat is the one to talk to about this.

ls overgeneration also the reason for usual trees in the meow heaven?

[paramat]

Hybrid Dog yes probably.

All core mapgens work on the 80^3 node mapchunk plus a 16 node deep shell that extends into / overlaps neighbouring mapchunks. Dungeons, large caves and mgv6 dirt/grass all 'overgenerate' into the shell sometimes.

In your code emin emax are the shell limits, minp maxp are the mapchunk limits try:

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

   local data = vm:get_data(emin, emax)
   
   for i in area:iterp(emin, emax) do

It might help.

[Hybrid Dog]

Hybrid Dog yes probably.

All core mapgens work on the 80^3 node mapchunk plus a 16 node deep shell that extends into / overlaps neighbouring mapchunks. Dungeons, large caves and mgv6 dirt/grass all 'overgenerate' into the shell sometimes.

In your code emin emax are the shell limits, minp maxp are the mapchunk limits try:

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

   local data = vm:get_data(emin, emax)
   
   for i in area:iterp(emin, emax) do

It might help.

it would cause unnecessary lag, l guess, the map would become generated more than one time at the same place

[Ben]

as far as l know the client needs to load the chunk, not the server
there's a function to forceload a chunk: [url]dev.minetest.net/minetest.forceload_block[/url]

Thanks, and sorry for the late reply, but forceload is a mechanism to keep a block loaded "forever", right? So that's not what I'm looking for. Thanks, though!

[Ferk]

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

--[[
Node placer - Places a node on the surface, at a random position in the mg block that contains a grid point defined by cell_size

published under Creative Commons CC0 license.
--]]

-- cell_size  must be bigger then mg block size (80) to work.
local cell_size = 150
local item_id = minetest.get_content_id("default:steelblock")

minetest.register_on_generated(function(minp, maxp, seed)
   -- If the next_hotspot is not within this mg block then exit
   local next_hotspot = {
      x = math.ceil(minp.x / cell_size) * cell_size,
      y = math.ceil(minp.y / cell_size) * cell_size,
      z = math.ceil(minp.z / cell_size) * cell_size
   }
   if next_hotspot.x > maxp.x or
      next_hotspot.y > maxp.y or
      next_hotspot.z > maxp.z then return end
   
   -- Place the object somewhere in this block

   local vm, emin, emax = minetest.get_mapgen_object("voxelmanip")
   local va = VoxelArea:new{MinEdge=emin, MaxEdge=emax}
   local vd = vm:get_data()
   
   local r = PcgRandom(seed) -- this ensures that you get the same results each time you generate the world anew.
   local item_pos = {
      x = r:next(minp.x, maxp.x),
      y = r:next(minp.y, maxp.y),
      z = r:next(minp.z, maxp.z),
   }

   -- This is to optimize, because getting the height map is expensive
   if minp.y > -200 and maxp.y < 200 then
      local hm = minetest.get_mapgen_object("heightmap")
      local hz = (item_pos.z - minp.z) * 80
      local hx = (item_pos.x - minp.x) % 80
      local height = hm[1 + hx + hz]
      item_pos.y = height + 1
   end

   vd[va:indexp(item_pos)] = item_id

   --
   vm:set_data(vd)
   --vm:calc_lighting()
   --vm:update_liquids()
   vm:write_to_map()
   --vm:update_map()
   --
end)

I haven't tested your code, but since you are calculating next_hotspot in all 3 dimensions, wouldn't a new block be created in the same area in the surface when a chunk gets generated further down close to the hotspot below?

I know you have checks to avoid placing it in the surface if it's below -200 and above 200 but with a cell_size of 150 you still can have 2 hotspots in there.

Also, if it's below -200 or above 200, it would place the node in a random location, even if it's in the sky or underground, no?

[Sane]

This code does work, but I find that certian nodes are not replaced by limit:limit_node like I want them to be.

Maybe those nodes where generated after your mod already ran. Do you have default in your depends.txt.

[Sane]

I haven't tested your code, but since you are calculating next_hotspot in all 3 dimensions, wouldn't a new block be created in the same area in the surface when a chunk gets generated further down close to the hotspot below?

The hotspot is a single coordinate so it can only be in one mapgen block.

I know you have checks to avoid placing it in the surface if it's below -200 and above 200 ...

No, the purpose of of the checks is to avoid getting the hightmap unnecessarily. Just a performance thing.

Also, if it's below -200 or above 200, it would place the node in a random location, even if it's in the sky or underground, no?

Yes this is what it does.

[Ferk]

The hotspot is a single coordinate so it can only be in one mapgen block.

Yes, but after each hotspot is found, a node is placed at the surface (using the heightmap) whenever the chunk that is getting generated is between -200 and 200 (the node in the surface might not be part of the chunk generated).

If this is the case then it's possible for 2 nodes to be placed at heightmap level next to each other, even though the goal was to have them separated by 150 nodes. The hotspots might be separated by that much (in the Y axis), but not the nodes you are placing.

I think it would be better to calculate 2D hotspots(only X and Z coordinates) and then after getting the heightmap check that the node is within the chunk that you are generating before placing it. That way you ensure you only place one node once on each X-Z division and only at surface level.

No, the purpose of of the checks is to avoid getting the hightmap unnecessarily. Just a performance thing.

I know, the fact that this also limits the amount of nodes that get placed in the surface is just an after-effect. Perhaps because you added this it's less noticeable, since only 2 nodes would be created on each 150x150 area in the surface. But still there's a chance they would be next to each other (also, even though the area is 150 the randomness in the X and Z axis when placing the node is limited to the currently generating chunk, which might not be that big).

[Don]

I just tested it and it does place 2 nodes close together. I found 3 within a 50 node radius.
The question is how to fix this. I need the nodes to be seperated. If they are close then it will cause an issue with what they do.
Thanks for all your help on this!

[Sane]

The hotspot is a single coordinate so it can only be in one mapgen block.

Yes, but after each hotspot is found, a node is placed at the surface (using the heightmap) whenever the chunk that is getting generated is between -200 and 200 (the node in the surface might not be part of the chunk generated).

If this is the case then it's possible for 2 nodes to be placed at heightmap level next to each other, even though the goal was to have them separated by 150 nodes. The hotspots might be separated by that much, but not the nodes you are placing.

That's right.

The code is a reply to Don, who wanted to place a node at a 1000 nodes spacing. The intend of the source is to give an general impression on how to go about to solve a thing like that. It is not thought to be the solution.

Thank you for your interest. Feel free to optimize and extend the code to a working condition.

[Sane]

I just tested it and it does place 2 nodes close together. I found 3 within a 50 node radius.

That's odd. That should not be possible.
Could you send the coordinates and the grid size you test with?

The question is how to fix this. I need the nodes to be seperated. If they are close then it will cause an issue with what they do.

The code places the node within a mapgen block. That is a 80x80x80 volume. So if you stay with a 1000er spacing you should be safe,

Thanks for all your help on this!

You are very welcome.

[Ferk]

The code is a reply to Don, who wanted to place a node at a 1000 nodes spacing. The intend of the source is to give an general impression on how to go about to solve a thing like that. It is not thought to be the solution.

Yes, I know. And thanks for that, I'm interested as well because something like this would also be useful for my project. I would like to place an entrance to an underground dungeon in the surface in specific X,Z positions at fixed distances.

This is what I'm doing in my register_on_generated for that:

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

if (minp.y > dungeon_rooms.origin.y) then

        -- avoid the calculation when too high up
        if (minp.y > 200) then return end

        local cell_size = {
            x = dungeon_rooms.stairs_distance * dungeon_rooms.room_area.x,
            z = dungeon_rooms.stairs_distance * dungeon_rooms.room_area.z,
        }

        local next_entrance = {
            x = maxp.x - (maxp.x % cell_size.x),
            z = maxp.z - (maxp.z % cell_size.z),
        }

        if (minp.x < next_entrance.x) and (minp.z < next_entrance.z) then

            -- Put the entrance spawner on ground level
            local hm = minetest.get_mapgen_object("heightmap")
            local hz = (next_entrance.z - minp.z) * 80
            local hx = (next_entrance.x - minp.x) % 80
            next_entrance.y = hm[1 + hx + hz]

            -- But only if it's in the generating chunk
            if (next_entrance.y < maxp.y) and (next_entrance.y > minp.y) then
                 minetest.set_node(next_entrance, minetest.get_content_id("dungeon_rooms:entrance_spawner"))
            end
        end
else
   [......]
end


However, for some reason my heighmap is always -31000, for every value of the array... I'm not sure why that's happening.

[Don]

I screwed up. I commented out the part of my code. I still had nodes spawning.
I have just your code running now. I have been looking for around a half hour and can not find one. I will let you know if I do find one. I have to make supper now. I will get back to it this evening.

[Ferk]

I have been looking for around a half hour and can not find one.

Can you print out the value of your coordinates, or at least the item_pos.y after the heighmap calculation?

Your phone or window isn't wide enough to display the code box. If it's a phone, try rotating it to landscape mode.

Code: Select all

minetest.log("action","Placing node at " .. item_pos.x .. ",".. item_pos.y .. ",".. item_pos.z)

Perhaps, like me, your heightmap returns -31000.

I can't find proper documentation regarding minetest.get_mapgen_object("heightmap").

Also, how sure can we be that a chunk is gonna be 80x80? there's a "chunksize" configuration option in minetest.conf. Potentially, a mapgen relying on a hardcoded 80x80 chunk size could be messed up if people changed their settings. Maybe we should use "(maxp.x - minp.x)"