Mobkit - Entity API [mobkit][alpha]

[u18398]

crash in wildlife mod

2019-09-29 15:12:34: ERROR[Main]: ServerError: AsyncErr: environment_Step: Runtime error from mod 'wildlife' in callback environment_Step(): .minetest/mods/mobkit/init.lua:180: attempt to index field 'node_box' (a nil value)
2019-09-29 15:12:34: ERROR[Main]: stack traceback:
2019-09-29 15:12:34: ERROR[Main]: /.minetest/mods/mobkit/init.lua:180: in function 'get_node_height'
2019-09-29 15:12:34: ERROR[Main]: /.minetest/mods/mobkit/init.lua:211: in function </home/mars/.minetest/mods/mobkit/init.lua:207>
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: .minetest/mods/wildlife/init.lua:116: in function '?'
2019-09-29 15:12:34: ERROR[Main]: /usr/local/share/minetest/builtin/game/register.lua:419: in function </usr/local/share/minetest/builtin/game/register.lua:399>
2019-09-29 15:12:34: ERROR[Main]: stack traceback:

and:

2019-09-29 15:22:27: ERROR[Main]: ServerError: AsyncErr: ServerThread::run Lua: Runtime error from mod 'wildlife' in callback luaentity_Step(): .minetest/mods/mobkit/init.lua:180: attempt to index field 'node_box' (a nil value)
2019-09-29 15:22:27: ERROR[Main]: stack traceback:
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:180: in function 'get_node_height'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:211: in function </home/mars/.minetest/mods/mobkit/init.lua:207>
2019-09-29 15:22:27: ERROR[Main]: (tail call): ?
2019-09-29 15:22:27: ERROR[Main]: (tail call): ?
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:411: in function 'is_neighbor_node_reachable'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:1110: in function 'func'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:676: in function 'execute_queues'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:894: in function </.minetest/mods/mobkit/init.lua:785>

[Edit] Or maybe it is mobkit itself ?

2019-09-29 15:30:51: ERROR[Main]: ServerError: AsyncErr: ServerThread::run Lua: Runtime error from mod 'sharks' in callback luaentity_Step(): .minetest/mods/mobkit/init.lua:180: attempt to index field 'node_box' (a nil value)
2019-09-29 15:30:51: ERROR[Main]: stack traceback:
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:180: in function 'get_node_height'
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:211: in function </home/mars/.minetest/mods/mobkit/init.lua:207>
2019-09-29 15:30:51: ERROR[Main]: (tail call): ?
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:1380: in function 'okpos'
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:1403: in function 'aqua_radar_dumb'
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:1444: in function 'func'
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:676: in function 'execute_queues'
2019-09-29 15:30:51: ERROR[Main]: .minetest/mods/mobkit/init.lua:894: in function </home/mars/.minetest/mods/mobkit/init.lua:785>

[runs]

crash in wildlife mod

2019-09-29 15:12:34: ERROR[Main]: ServerError: AsyncErr: environment_Step: Runtime error from mod 'wildlife' in callback environment_Step(): .minetest/mods/mobkit/init.lua:180: attempt to index field 'node_box' (a nil value)
2019-09-29 15:12:34: ERROR[Main]: stack traceback:
2019-09-29 15:12:34: ERROR[Main]: /.minetest/mods/mobkit/init.lua:180: in function 'get_node_height'
2019-09-29 15:12:34: ERROR[Main]: /.minetest/mods/mobkit/init.lua:211: in function </home/mars/.minetest/mods/mobkit/init.lua:207>
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: (tail call): ?
2019-09-29 15:12:34: ERROR[Main]: .minetest/mods/wildlife/init.lua:116: in function '?'
2019-09-29 15:12:34: ERROR[Main]: /usr/local/share/minetest/builtin/game/register.lua:419: in function </usr/local/share/minetest/builtin/game/register.lua:399>
2019-09-29 15:12:34: ERROR[Main]: stack traceback:

and:

2019-09-29 15:22:27: ERROR[Main]: ServerError: AsyncErr: ServerThread::run Lua: Runtime error from mod 'wildlife' in callback luaentity_Step(): .minetest/mods/mobkit/init.lua:180: attempt to index field 'node_box' (a nil value)
2019-09-29 15:22:27: ERROR[Main]: stack traceback:
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:180: in function 'get_node_height'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:211: in function </home/mars/.minetest/mods/mobkit/init.lua:207>
2019-09-29 15:22:27: ERROR[Main]: (tail call): ?
2019-09-29 15:22:27: ERROR[Main]: (tail call): ?
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:411: in function 'is_neighbor_node_reachable'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:1110: in function 'func'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:676: in function 'execute_queues'
2019-09-29 15:22:27: ERROR[Main]: .minetest/mods/mobkit/init.lua:894: in function </.minetest/mods/mobkit/init.lua:785>

I think it is related to https://forum.minetest.net/viewtopic.php?f=9&t=22112&start=150#p355442

In my mobkit version I've added a line (line 180) to avoid this:

if node.drawtype == 'nodebox' then
if node.node_box == nil then return nil end --added by petz
if node.node_box.type == 'fixed' then

[u18398]

In my mobkit version I've added a line (line 180) to avoid this:

if node.drawtype == 'nodebox' then
if node.node_box == nil then return nil end --added by petz
if node.node_box.type == 'fixed' then

thanks, looks like that helps

[Termos]

attempt to index field 'node_box' (a nil value)

Yeah, there must be a mod running that defines a nodebox type node without an actual node_box.

This happens in get_node_height() - the function is still very 4.17 and doesn't understand some new 5.0 stuff.
I've been reluctant to fix it because of a fun bug that makes mobs jump in snow - maybe you seen it. But I will eventually.

[runs]

attempt to index field 'node_box' (a nil value)

Yeah, there must be a mod running that defines a nodebox type node without an actual node_box.

This happens in get_node_height() - the function is still very 4.17 and doesn't understand some new 5.0 stuff.
I've been reluctant to fix it because of a fun bug that makes mobs jump in snow - maybe you seen it. But I will eventually.

Yes, mobs jumping in snow is cool and appropiate. :-D

[Termos]

Would it be possible to have mobs defend their owners when attacked, or attack anything their owner punches? I've tried accomplishing it using mobkit functions but I've had no luck so far.

This is a very good question.

First, the only reliable place for punch detection is callbacks, so we need a function that will run in every mob's on_punch and on_punchplayer as well.

The function would:

• - determine if a player takes part
  - yes: do get_objects_inside_radius()
  - iterate over all luaentities to see if they have a memory entry indicating the player:get_player_name() is their owner
  - for matching entities: do something to them that would trigger an attack.
  One way would be to add a custom luaentity field that would contain a reference to the culprit's objectref, that would then be interpreted by the brainfunc to kick off an attack behavior.

Please note: I'm storing objectrefs and luaentities in direct violation of RW's guide, but there are no good alternatives.
Just remember to do mobkit.is_alive() first thing in every step if you're going to use a stored reference, and to clear the reference as soon as it's not needed anymore.
Also never place references to objects and luaentities in mobs memory, or it will crash the engine because memory is serialized.

That said, it became clear to me that the api needs some generic event facility to handle such cases and more, so this becomes a priority and the next thing I'm going to try and implement.

[runs]

Some request to complete aquatic behaviour:

- Sharks (my dolphins) do not submerge. Maybe a min and max ocean deep set?
- Follow, run away and predator behaviours (so my dolphins could be smart enough and pursue fishes)

[Termos]

Sharks (my dolphins) do not submerge. Maybe a min and max ocean deep set?

Probably because they have no reason to do so, and maybe their buoyancy is positive. Neutral is 1.0 try setting it to like 1.02, they'll slowly sink.
I don't use specific world coordinates ever, so sharks should work in layered worlds etc, but you can if you're coding for a specific game.

- Follow, run away and predator behaviours (so my dolphins could be smart enough and pursue fishes)

Predator behavior is ready, hq_aqua_attack takes an objectref as an argument, it can be used to attack anything that floats.

As to follow and run away, I might do these at some point, but builtin behaviors are really just examples, the main reason I'm doing them is to figure out what utility functions are needed.
I encourage all to try and write some custom behaviors.

Also if you don't mind my opinion, dolphins just have to jump out of water occasionally, so you need a behavior for that - if it doesn't jump out of water, it's not really a dolphin ;)

[Robsoie]

After a quick test, the less i can say is that i am extremely impressed by the sharks, i never thought i would see in minetest a mob moving so naturally, that's brillant.

On my quick test so far i met mostly zombies that were all "mario-jumping" on me, it felt rather hilarious coming from zombies type of enemies (but very damaging juding on much health i was losing).

Oddly i saw some kind of antilopes on the ice/snowy areas, when i approached them they slowly ran away (though very slowly) from me , a behaviour that really felt believable (out of the very slow speed of course).

In some forest i saw what i think was some kind of lion (i think as the sound was rather odd at first i thought i was some more noisy version of a zombie from the sound) that i managed to kill due to being positionned on the high ground (sorry anakin) and striking it everytime the beast was trying to jump on upper nodes.
When it died it waited a few minute to suddenly disappear.

On the fight with the zombies and the "lion" i noticed there wasn't a noticable "feedback" on my sword strikes on them, i had in fact no idea if i was really hitting them or not, as they all jump a lot on their attack i couldn't make a difference between them actually jumping or if it was a "jump" as a sword strike feedback.

Anyways, that shark mob is incredible and its moves feel so natural in the environment.

[Termos]

Thanks for the review!

Just to clarify things up a bit:
What you've seen is the fruit of accidental collaboration of a few people: Gundul (server), runs (Petz) and me.
This thread is concerned primarily with mobkit which is a humble api and as such doesn't do anything by itself.

[ElCeejo]

I tried doing what you suggested

Code: Select all

if mobkit.is_alive(self) then
      if puncher:is_player() then -- determine if a player takes part
         local objs = minetest.get_objects_inside_radius(puncher:get_pos(), self.view_range) -- get objects inside radius
         local obj = nil
   
         for n = 1, #objs do

            obj = objs[n]:get_luaentity()
            if obj and obj.owner == puncher:get_player_name() then -- look over all entities to see if the puncher is their owner
                    mobkit.clear_queue_high(obj)
               mobkit.hq_hunt(obj,10,self) -- for all matching entities, do something that would trigger an attack
            end
   

In theory this should work, but the game crashes.

It gives an error message saying it failed to call method "get_pos" which it says is a nil value

Code: Select all

function mobkit.hq_hunt(self,prty,tgtobj)
	local func = function(self)
		if not mobkit.is_alive(tgtobj) then return true end
		if mobkit.is_queue_empty_low(self) and self.isonground then
			local pos = mobkit.get_stand_pos(self)
			local opos = tgtobj:get_pos() <<< Code that's causing the crash
			local dist = vector.distance(pos,opos)
			if dist > self.view_range then
				return true
			elseif dist > 3 then
				mobkit.goto_next_waypoint(self,opos)
			else
				mobkit.hq_attack(self,prty+1,tgtobj)					
			end
		end
	end
	mobkit.queue_high(self,func,prty)
end

I probably could have described that better but I'm tired. Hopefully I was clear enough that you get what I'm saying :P

[u18398]

How can I slowdown speed on entities.

I copied the sharks and added new b3d and texture for fish
but when I set max_speed = 0.5 and use
mobkit.hq_aqua_roam(self,10,0.1)

the fishes still panicing through the water

[Termos]

In theory this should work, but the game crashes.
It gives an error message saying it failed to call method "get_pos" which it says is a nil value

What you did is more or less it, given the mod is very simple and all affected mobs use same behaviors and priority scheme, there are just some errors, in order of importance:

mobkit.hq_hunt(obj,10,self)
This is the showstopper, the function takes objectref as an argument, try passing self.object

mobkit.clear_queue_high(obj)
Clearing the queue means they abandon whatever they've been doing, even if they're busy dying, they'll get up and come to the rescue. Better is to simply queue the behavior without deleting the queue, it'll get executed when it's time comes.
Better yet, do get_queue_priority and only queue the behavior if prty < 10

if obj and obj.owner == puncher:get_player_name()
There's special memory structure for this kind of thing. Do mobkit.remember(self,'owner',playername) when setting ownership, then here you could do mobkit.recall(obj,'owner').
Added benefit is mobs having memory entries won't timeout.
Calling a luaentity 'obj' is misleading btw.

if mobkit.is_alive(self) then
Ths check is not crucial here. It's not a stored reference, because it's just been retrieved from callback so we know it's active, and behaviors already check if their object is alive. However it might've just been killed by the punch so it's ok I guess, given you call the function after mobkit.hurt() call

[Termos]

How can I slowdown speed on entities.

but when I set max_speed = 0.5 and use
mobkit.hq_aqua_roam(self,10,0.1)
the fishes still panicing through the water

Works ok on my end, I'd have to take a look at your code.

[ElCeejo]

Code: Select all

	if mobkit.is_alive(self) then
		if puncher:is_player() then
			local objs = minetest.get_objects_inside_radius(puncher:get_pos(), self.view_range)
			local obj = nil


			for n = 1, #objs do
				local prty = mobkit.get_queue_priority(self.object)

				obj = objs[n]:get_luaentity()
				if obj and puncher:get_player_name() == mobkit.recall(self, "owner") then
				    if prty < 12 then 
					    mobkit.hq_hunt(self.object,12,self)
					    return
					end
				end
			end
        end
    end
end

Now when I punch the mob I get

Code: Select all

..\mods\mobkit\init.lua:604: attempt to get length of field 'hqueue' (a nil value)

[Termos]

You've changed it the wrong way around
Remember here 'self' is the punched one, its priority and memory are of no importance
The one who's to defend his master I called 'luaent' instead of misleading 'obj'

Something along these lines:

[...]
local pname = puncher:get_player_name()

for n = 1, #objs do

• local luaent = objs[n]:get_luaentity()
  if luaent and pname == mobkit.recall(luaent, "owner") then
  local prty = mobkit.get_queue_priority(luaent)
  • if prty < 12 then
    mobkit.hq_hunt(luaent,12,self.object)

[..]

[u18398]

How can I slowdown speed on entities.

but when I set max_speed = 0.5 and use
mobkit.hq_aqua_roam(self,10,0.1)
the fishes still panicing through the water

Works ok on my end, I'd have to take a look at your code.

I checked again, but I do not know what I am doing wrong.
This time I took sharks mod and adjusted max_speed and
mobkit.hq_aqua_roam(self,10,X)

But sharks seem to move as usual when I start game again.
https://github.com/berengma/sharks


I think these funstion are still missing in your documentation :)
mobkit.pos_translate2d(pos,yaw,dist)
mobkit.go_forward_horizontal(self,yaw,speed)
mobkit.hq_aqua_roam(self,prty,speed)
mobkit.hq_aqua_turn(self,prty,tyaw,speed)
mobkit.hq_aqua_attack(self,prty,tgtobj,speed)

[Termos]

I think these funstion are still missing in your documentation :)

Now that's an important hint, because they aren't missing.
Their entries were added in the last commit 190927. Also in the same commit, I fixed some hardcoded speed values.

Your mobkit seems outdated.

[u18398]

Now that's an important hint, because they aren't missing.
Their entries were added in the last commit 190927. Also in the same commit, I fixed some hardcoded speed values.

Your mobkit seems outdated.

LOL I can't believe it. You are right ! I mixed up the konsoles of my remote servers (which are up to date) with my
local computer directory (which was not up to date)

All good now, sry for wasting your time.

[runs]

Also if you don't mind my opinion, dolphins just have to jump out of water occasionally, so you need a behavior for that - if it doesn't jump out of water, it's not really a dolphin ;)

I did it :-D but I have to tweak the API again cos this line forces fishes to stay in water:

self.object:set_acceleration({x=-vel.x,y=buoyacc-vel.y*abs(vel.y)*0.7,z=-vel.z})

[Termos]

I did it :-D but I have to tweak the API again cos this line forces fishes to stay in water:
self.object:set_acceleration({x=-vel.x,y=buoyacc-vel.y*abs(vel.y)*0.7,z=-vel.z})

No, please, don't do that ;)
Just when they're close to the surface, give em strong enough initial velocity impulse at an angle forward and upwards, and let physics do the rest.
And don't forget to dir_to_rot() them.

[ElCeejo]

I'm doing even more complicated stuff, namely trying to make mobs grow in increments, and I've found a puzzling issue.

Code: Select all

paleotest.progressive_growth = function(self)

	self.base_size = self.visual_size
	self.base_colbox = self.collisionbox
	self.base_selbox = self.selectionbox

	if self.growth_stage == 1 then
		self.object:set_properties({
			textures = textures,
			visual_size = {
				x = self.base_size.x / 5,
				y = self.base_size.y / 5
			},
			collisionbox = {
				self.base_colbox[1] / 5,
				self.base_colbox[2] / 5,
				self.base_colbox[3] / 5,
				self.base_colbox[4] / 5,
				self.base_colbox[5] / 5,
				self.base_colbox[6] / 5
			},
		})
	end
end

So what's supposed to happen is if the mob is hatched from an egg or a result of breeding 2 adult mobs it starts off at growth stage 1, and grows from there (I probably could find a better way to do it than what I've shown above but I'm lazy :P) . The problem is, if you use set_properties to lower a mobs size it seems to cause some behavior issues, like if it's set it to the size I have in the above code, it won't roam or play animations at all, only moving to attack the player.

[runs]

I did it :-D but I have to tweak the API again cos this line forces fishes to stay in water:
self.object:set_acceleration({x=-vel.x,y=buoyacc-vel.y*abs(vel.y)*0.7,z=-vel.z})

No, please, don't do that ;)
Just when they're close to the surface, give em strong enough initial velocity impulse at an angle forward and upwards, and let physics do the rest.
And don't forget to dir_to_rot() them.

I did it :-D

A question:
does mobkit.is_in_deep(target) check the water deep (below) or the deep where the mob is (above)?

[Termos]

does mobkit.is_in_deep(target) check the water deep (below) or the deep where the mob is (above)?

Both;) At least one of them has to be liquid source as well.

I'm doing even more complicated stuff, namely trying to make mobs grow in increments, and I've found a puzzling issue.
[...]
The problem is, if you use set_properties to lower a mobs size it seems to cause some behavior issues, like if it's set it to the size I have in the above code, it won't roam or play animations at all, only moving to attack the player.

That's probably on me.
The core of the api was done in 4.17, where changing colbox on the fly wasn't possible.
I'll try to do a quick fix soon. (edit, here, the new version)

Another issue, I've noticed you do big mobs, however built in land behaviors don't go well with collisionboxes bigger than 1 node. Please make sure x and z collisionbox dimensions never exceed 1. Support for huge collisionboxes is to come later.

[Termos]

Quick update 191005

Changes:

• Improved compatibility with set_properties() objectref method
  Now it should be safe to alter collisionbox at any time.