[Mod] Experimental mapgen [mg]

[Sokomine]

Is there any way to "scout" what materials are/will be availalbe in that area before the village is created? For example, if it's going to be a log-cabin type village - is there a way to find out what wood is "naturally" in the area? Or even base it on what biome you're in (and make some assumptions about what's available)? Just curious. Haven't checked out the pack yet, but will be very soon.

Not really. The mods installed (especially tree and wood types) determine what will be considered as building material. When it comes to placing a town, random materials are selected. Basing that decision on the biome present wouldn't be trivial at all, and it might not work very well anyway. Most biomes are pretty small. Their borders will in many cases be inside the village. Ground types (grass, snow, dirt, dirt from ethereal, ...) are kept, so you can see where those borders wander through the village. Giving each house individual materials would be possible, but it won't be a good idea. The villages need some common style in order to look well. With towns, that might be diffrent.

[Fritigern]

So for the past weeks, I was stumped why my static_spawnpoint wasn't being used. Finally figured out that mg was to blame for this.

From init.lua, lines 714-739

Code: Select all

local function spawnplayer(player)
	local noise1 = minetest.get_perlin(12345, 6, 0.5, 256)
	local min_dist = math.huge
	local min_pos = {x = 0, y = 3, z = 0}
	for bx = -20, 20 do
	for bz = -20, 20 do
		local minp = {x = -32 + 80 * bx, y = -32, z = -32 + 80 * bz}
		for _, village in ipairs(villages_at_point(minp, noise1)) do
			if math.abs(village.vx) + math.abs(village.vz) < min_dist then
				min_pos = {x = village.vx, y = village.vh + 2, z = village.vz}
				min_dist = math.abs(village.vx) + math.abs(village.vz)
			end
		end
	end
	end
	player:setpos(min_pos)
end

minetest.register_on_newplayer(function(player)
	spawnplayer(player)
end)

minetest.register_on_respawnplayer(function(player)
	spawnplayer(player)
	return true
end)

This is so not cool, all mods should honor minetest.conf's settings, unless the user somehow indicates that they don't wish to use the static_spawnpoint, for instance by leaving static_spawnpoint empty, or by setting a config option on the mod's settings but no such provision was made in this mod's code!!!

Not cool guys, seriously not cool.

[Sokomine]

This is so not cool, all mods should honor minetest.conf's settings, unless the user somehow indicates that they don't wish to use the static_spawnpoint, for instance by leaving static_spawnpoint empty, or by setting a config option on the mod's settings but no such provision was made in this mod's code!!!

The reason for a spawn point beeing set by the mod is that prior to that change, people used to spawn embedded in stone or dirt when joining a world. Which is very confusing and looks like something's broken. Now players are placed at the center of the next available village - which is a far better starting point.

Beeing able to override that automatic spawning may indeed be intresting for servers. Perhaps I'll consider it in my mg_villages mod. A /spawn command couldn't hurt in singleplayer either...

[Fritigern]

Here's an idea in pseudo-code, because i don't know Lua:

Code: Select all

trigger_on_spawning_new_player
{
  while(node.above.head != air && node.at.head.level != air )
  {
     player_pos.z = player_pos.z + 2
  }
}

This will not prevent players from spawning in caves or tiny pockets of air, but at least it will prevent them from spawning inside solid nodes and be trapped there.
Better yet, make sure there's air all around so that the player can dig out if needed, but as long as the spawn point is obstructed, move them up by 2 nodes because moving up by a single node may still lead to the player spawn with their head buried.

[Sokomine]

Here's an idea in pseudo-code, because i don't know Lua:

IMHO the spawning in the next village is a very good thing. That way, you know what this mapgen can do, and have a more intresting starting point than those random ones. It's just when servers change the spawn point which ought to be honored. Of course, all this beeing open source, you can change it all according to your liking and don't have to wait upon what someone else likes :-)

[Fritigern]

IMHO the spawning in the next village is a very good thing. That way, you know what this mapgen can do, and have a more intresting starting point than those random ones.

There are two big issues with your statement
ISSUE #1
Before you can spawn in a village, the following two conditions have to be met:
1 - The world must have generated far enough to have spawned a village, or you will simply spawn at 0,0,0
2 - The user must have died *after* condition 1 became true.

ISSUE #2
I will illustrate this issue with an example. Imagine, if you will, the following (single player) scenario:
Jimmy creates a new world using a random seed and found that this time he did not spawn in water, a rare occasion indeed. He waits a little and sees that 0,0,0 is actually a great spot to set up his base.
So Jimmy starts building and building, he remembered to do /sethome so that he is certain to spawn inside his awesome base should anything happen. He also put down a bed (he has the beds mod) and even set his home in Unified Inventory. Nothing can go wrong!
And Jimmy builds higher and higher, he keeps going until he accidentally falls off his building. No sweat, he has set a home so he should spawn right there!
But hey! A village? Why did I spawn here, Jimmy wonders, I was supposed to spawn inside my base! This is when Jimmy finds out the difference between a spawnpoint and a home. And which way is my home again? And how far?
So okay, Jimmy goes and sets 0,0,0 as his static_spawnpoint in minetest.conf because that should work.
Only, it doesn't, because Sokomine thought it was a good idea to override the static_spawnpoint, causing unexpected behavior in the game.

See the problem with an overridden static_spawnpoint?

[Nore]

ISSUE #1
Before you can spawn in a village, the following two conditions have to be met:
1 - The world must have generated far enough to have spawned a village, or you will simply spawn at 0,0,0

Wrong. The code looks for where a village will spawn and spawns the player there.

Only, it doesn't, because Sokomine thought it was a good idea to override the static_spawnpoint, causing unexpected behavior in the game.

See the problem with an overridden static_spawnpoint?

Overriding static_spawnpoint was unintended behaviour, as I never use this feature. It has been fixed in the latest version.

[Sokomine]

So okay, Jimmy goes and sets 0,0,0 as his static_spawnpoint in minetest.conf because that should work.

If your Jimmy's experienced enough to get that done, he'll maybe know about other ways of getting where he wants to.

Only, it doesn't, because Sokomine thought it was a good idea to override the static_spawnpoint, causing unexpected behavior in the game.

And I still think it is a far better idea than to spawn players embedded in dirt :-)

It has been fixed in the latest version.

Thanks! I've added that fix to mg_villages as well.

[Krock]

There's an issue with the leafdecay:



Please add

Code: Select all

trunk = "mg:pinetree",

to the pine (and savanna tree) leaves definition to support the subgames BASE and Voxelgarden.

[Nore]

Done :) https://github.com/Ekdohibs/mg/commit/0 ... 25bacb2906

[Fritigern]

Hey guys, the latest Git version of minetest checks for uninitialized globals and the mg mod has proven to have a lot of them. I have taken the liberty of fixing them on my system, see my fixes below.
These fixes have stopped the warnings and all is honky dory now. In fact, it seems to have improved the speed of /mg_regenerate, but that could be either coincidence or even the power of suggestion.
P.S. This took me all day, i barely knew what i was doing, but i think i have learned a thing or two about Lua...
P.P.S. If i knew how to make a diff, i probably would have...

+ Spoiler

MESSAGE : WARNING: Undeclared global variable "DEBUG" accessed at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:747
MESSAGE : WARNING: Undeclared global variable "DEBUG" accessed at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:759
WHEN : At server startup, repeated several times.

OLD

Code: Select all

line 2 : <empty line>

NEW

Code: Select all

:
line 2 : DEBUG = false

Notes: DEBUG needs to be a global, but globals need to be initialized. This does exactly that.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "wseed" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:118.
WHEN : At server startup, shown once.

OLD

Code: Select all

Line 117 : <empty line>

NEW

Code: Select all

Line 117 : wseed = 0

NOTES : Declared wseed as a global in line 117 because i am not sure how else to allow the two funtions that need wseed to use the variable. There is most likely a better way and this fix feels like a very dirty hack.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "biome" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:458.
MESSAGE : WARNING: Assignment to undeclared global "biome_humidity" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:459.
MESSAGE : WARNING: Assignment to undeclared global "biome_temperature" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:460.

WHEN : After startup, after user logs in and starts exploring

OLD

Code: Select all

		biome = get_nearest_biome(biome_table, x, z)
		biome_humidity = biome.h
		biome_temperature = biome.t

NEW

Code: Select all

		local biome = get_nearest_biome(biome_table, x, z)
		local biome_humidity = biome.h
		local biome_temperature = biome.t
		

NOTES : It makes sense to make these vars local because they are defined inside functions, also none of the variables are being used outside the mg_generate function

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "stk" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/init.lua:649.
WHEN : Upon village generation
OLD

Code: Select all

	stk = ItemStack({name=items[ii]:get_name(), count=pr:next(1, stacksz), wear=items[ii]:get_wear(), metadata=items[ii]:get_metadata()})

NEW

Code: Select all

	local stk = ItemStack({name=items[ii]:get_name(), count=pr:next(1, stacksz), wear=items[ii]:get_wear(), metadata=items[ii]:get_metadata()})

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "pos0" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/ores.lua:12.
MESSAGE : WARNING: Assignment to undeclared global "p0" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/ores.lua:35.
MESSAGE : WARNING: Assignment to undeclared global "new_minp" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/ores.lua:94.
MESSAGE : WARNING: Assignment to undeclared global "new_maxp" inside a function at .../Minetest/bin/../worlds/Pacata/worldmods/mg/ores.lua:95.

OLD

Code: Select all

Line 12 : pos0={}
Line 35 : p0=round_pos(p)
Line 94 : new_minp={x=minp.x+xblocksdiff*80,y=minp.y+yblocksdiff*80,z=minp.z+zblocksdiff*80}
Line 95 : new_maxp={x=maxp.x+xblocksdiff*80,y=maxp.y+yblocksdiff*80,z=maxp.z+zblocksdiff*80}

NEW

Code: Select all

Line 12 : local pos0={}
Line 35 : local p0=round_pos(p)
Line 94 : local new_minp={x=minp.x+xblocksdiff*80,y=minp.y+yblocksdiff*80,z=minp.z+zblocksdiff*80}
Line 95 : local new_maxp={x=maxp.x+xblocksdiff*80,y=maxp.y+yblocksdiff*80,z=maxp.z+zblocksdiff*80}

NOTES : It makes sense to make these vars local because they are defined inside functions.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "bsizez" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:164.
MESSAGE : WARNING: Assignment to undeclared global "bsizex" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:164.
MESSAGE : WARNING: Assignment to undeclared global "rotation" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:164.
MESSAGE : WARNING: Assignment to undeclared global "btype" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:164.
WHEN : When a village is being generated

OLD

Code: Select all

 (line 158, 159 and 204, 205)
			local tries = 0
			while true do

NEW

Code: Select all

 (line 159-163 and 207-212 line 204 transposed to 207 because of earlier insertion at 159)
			local tries = 0
			local btype = 0 -- Initialize with zero. Counting on the vars being redefined in line 168
			local rotation = 0
			local bsizex = 0
			local bsizez = 0
			while true do

+ Spoiler

From here, line numbers are the numbering AFTER applying the fix for line 164 and 204. Original line numbers will be put in (parentheses).

MESSAGE : WARNING: Assignment to undeclared global "rzmin" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:252.
MESSAGE : WARNING: Assignment to undeclared global "rzmax" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:253.
MESSAGE : WARNING: Assignment to undeclared global "rxmin" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:254.
MESSAGE : WARNING: Assignment to undeclared global "rxmax" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:255.

OLD

Code: Select all

 (line 245, 246)
	mz = mmz or rdz*math.max(rdz*mz, rdz*m2z)
	if rdx == 0 then
	

NEW

Code: Select all

 (line 245-250)
	mz = mmz or rdz*math.max(rdz*mz, rdz*m2z)
	local rxmin = 0 -- Initialize with zero. Counting on the vars being redefined later
	local rxmax = 0	
	local rzmin = 0
	local rzmax = 0		
	if rdx == 0 then

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "calls" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:282. (270)
WHEN : Upon village generation
OLD

Code: Select all

 
Line 8 : <empty line>

NEW

Code: Select all

Line 8 : calls = 0

NOTES : calls is used as a global, another function counts on this variable being defined as well. So seting as a global at line 8. This causes /mg_regenerate add empty lines on the console when showing feedback, but at least the warning is gone.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "i" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:284. (272)

OLD

Code: Select all

	i = 1

NEW

Code: Select all

	local i = 1
	

NOTES : This is just an iterator and should be local unless the value is used elsewhere. In which case "i" would be a horrible name for a global.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "az" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:306. (294)
MESSAGE : WARNING: Assignment to undeclared global "ay" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:306. (294)
MESSAGE : WARNING: Assignment to undeclared global "ax" inside a function at ...etest/bin/../worlds/Pacata/worldmods/mg/villages.lua:306. (294)
WHEN : Upon village generation
OLD

Code: Select all

 
Line 305 :	for z = 0, pos.bsizez-1 do
Line 306 :		ax, ay, az = pos.x+x, pos.y+y+binfo.yoff, pos.z+z

NEW

Code: Select all

Line 305 :	for z = 0, pos.bsizez-1 do
Line 306 :	    local ax = 0
Line 307 :	    local ay = 0
Line 308 :	    local az = 0
Line 309 :		ax, ay, az = pos.x+x, pos.y+y+binfo.yoff, pos.z+z

NOTES : Defining the vars as local fixed the problem.

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "value" inside a function at ...am/Minetest/bin/../worlds/Pacata/worldmods/mg/we.lua:15.
WHEN : Upon village generation
OLD

Code: Select all

	value = f:read("*a")

NEW

Code: Select all

	local value = f:read("*a")

+ Spoiler

MESSAGE : WARNING: Assignment to undeclared global "new_scm" inside a function at ...inetest/bin/../worlds/Pacata/worldmods/mg/rotate.lua:22.
MESSAGE : WARNING: Assignment to undeclared global "new" inside a function at ...inetest/bin/../worlds/Pacata/worldmods/mg/rotate.lua:43.
WHEN : Upon village generation
OLD

Code: Select all

line 22 :	new_scm = {}
line 23 :	for i=1, ysize do

NEW

Code: Select all

line 22 :	local new_scm = {}
line 23 :	local new = 0
line 24 :	for i=1, ysize do

[Nore]

Ok, those leaking globals should be fixed now.

[wcwyes]

I'm having what seems to be a bug in the mapgen code farming:cotton_seed seems to be referenced as farming:seed_cotton when placed in a chest I can't seem to find the reference though

[Nore]

These are in the we schematics of the mods; I will see if I can fix that.

[Wayward_One]

Hello :) I'm not sure how hard it would be to implement this, but I would really like to see abandoned mines in addition to villages, like the ones from this mod: viewtopic.php?f=11&t=6307 or this one: viewtopic.php?f=9&t=10225

Maybe they could even be made to spawn under and/or near villages as well :D

[Sokomine]

Hello :) I'm not sure how hard it would be to implement this, but I would really like to see abandoned mines in addition to villages, like the ones from this mod: viewtopic.php?f=11&t=6307 or this one: viewtopic.php?f=9&t=10225

The trouble with the mines is that it's hard to find them if they're entirely underground.

Maybe they could even be made to spawn under and/or near villages as well :D

A miner's village might be something. The entrance could either be a drop-shaft with a small tower on top, or a ground-level entrance with some mining shafts going straight through the next mountain. Making that look good and working would require quite some work I'm afraid.

[wcwyes]

It has been fixed in the latest version.

It still a little buggy

[Nore]

It has been fixed in the latest version.

It still a little buggy

You mean the static_spawnpoint thing?

[swordpaint12]

I just got this error:

14:13:46: ERROR[main]: Server: Failed to load and run C:\Users\<my user>\Desktop\minetest-0.4.12-win64-msvc\minetest-0.4.12\bin\..\mods\mg_biomes\init.lua
14:13:46: ERROR[main]: ModError: ModError: Failed to load and run C:\Users\<my user>\Desktop\minetest-0.4.12-win64-msvc\minetest-0.4.12\bin\..\mods\mg_biomes\init.lua

[Nore]

Could you try to rename the folder to "mg" please ?

[swordpaint12]

Thanks, I did and it cleared up. Thank you!

[Nore]

Hello all!
The mod was moved to minetest-mods, so you can expect more frequent updates or bugfixes now ;), even if mg is starting to get deprecated...

[redblade7]

Hello all!
The mod was moved to minetest-mods, so you can expect more frequent updates or bugfixes now ;), even if mg is starting to get deprecated...

Is it safe to remove this mod from a server which used it from the beginning? Are the villages and biomes already in existence or does this generate them on the fly?

[Nore]

Hello all!
The mod was moved to minetest-mods, so you can expect more frequent updates or bugfixes now ;), even if mg is starting to get deprecated...

Is it safe to remove this mod from a server which used it from the beginning? Are the villages and biomes already in existence or does this generate them on the fly?

If you remove this mod from a server, this will not affect the part of the map which has already been generated. However, there will be discontinuities at chunk borders where the map was not previously generated and is generated by another mapgen instead of mg if you remove it.

[redblade7]

Hello all!
The mod was moved to minetest-mods, so you can expect more frequent updates or bugfixes now ;), even if mg is starting to get deprecated...

Is it safe to remove this mod from a server which used it from the beginning? Are the villages and biomes already in existence or does this generate them on the fly?

If you remove this mod from a server, this will not affect the part of the map which has already been generated. However, there will be discontinuities at chunk borders where the map was not previously generated and is generated by another mapgen instead of mg if you remove it.

I don't understand, can you please explain?

I just want to make sure that this mod won't become/isn't broken with new versions of Minetest and then everything everyone's built has to be erased.