ABM works unstable/incorrect

[bosapara]

A lot of people already talked about this but no one dared to write

abm works unstable/incorrect, ive checked at servers with 30 players and 5 players, without lags, the same. It works wrong (at single it works)

Instead 5 sec of signal - we have 10, 15 or even 30 seconds (by random)

Any idea how to fix it?

[firefox]

and what abm is that?

[bosapara]

and what abm is that?

Here you read about abm

[Pyrollo]

Yes but can you give the definition of your ABM ? Usually it works.

[bosapara]

Yes but can you give the definition of your ABM ? Usually it works.

Im here to report about problem, isn't about giving definition to you.

[firefox]

Yes but can you give the definition of your ABM ? Usually it works.

Im here to report about problem, isn't about giving definition to you.

so you did. and what shall we do now?

abms work fine in singleplayer and multiplayer, no problems detected.
so it have something to do with the specific abm you are using.
so which abm is it exactly? we need the code.

[bosapara]

Yes but can you give the definition of your ABM ? Usually it works.

Im here to report about problem, isn't about giving definition to you.

so you did. and what shall we do now?

abms work fine in singleplayer and multiplayer, no problems detected.
so it have something to do with the specific abm you are using.
so which abm is it exactly? we need the code.

Code doesnt metter. Works wrong even default lava + water.

Delay that happens by random 'crash' all idea of ABM.

[firefox]

my minetest runs at 30fps
flooding a mountain with lava and water dropped it to 22fps, which was to be expected.
yet the lavacooling worked correctly and there was no significant delay or changes in the flowing speed of liquids.

i cannot reproduce the problem. which version of minetest are you using?
also, did you activate any external mods?

[rubenwardy]

I need a test case so I can try and reproduce this (and also make sure there's no silly mistakes being made)
Lava and water flow isn't an ABM

[Pyrollo]

Yes but can you give the definition of your ABM ? Usually it works.

Im here to report about problem, isn't about giving definition to you.

I was here to try to help you. Forget it.

[texmex]

Lock the thread and redirect this to the Github issues, where bug reporting belongs.

[bosapara]

I need a test case so I can try and reproduce this (and also make sure there's no silly mistakes being made)
Lava and water flow isn't an ABM

Easy way to proof my report - visit 'jozet.duckdns.org:30000' and check lava+water or clockgen from basic_machines mod (it give 1 signal at 5 seconds and can activate something)


About stone and water. I mean stone generator with lava and water

\games\minetest_game\mods\default\functions.lua

Code: Select all

default.cool_lava = function(pos, node)
	if node.name == "default:lava_source" then
		minetest.set_node(pos, {name = "default:obsidian"})
	else -- Lava flowing
		minetest.set_node(pos, {name = "default:stone"})
	end
	minetest.sound_play("default_cool_lava",
		{pos = pos, max_hear_distance = 16, gain = 0.25})
end

if minetest.settings:get_bool("enable_lavacooling") ~= false then
	minetest.register_abm({
		label = "Lava cooling",
		nodenames = {"default:lava_source", "default:lava_flowing"},
		neighbors = {"group:cools_lava", "group:water"},
		interval = 1,
		chance = 2,
		catch_up = false,
		action = default.cool_lava,
	})
end

[Hume2]

Code: Select all

default.cool_lava = function(pos, node)
	if node.name == "default:lava_source" then
		minetest.set_node(pos, {name = "default:obsidian"})
	else -- Lava flowing
		minetest.set_node(pos, {name = "default:stone"})
	end
	minetest.sound_play("default_cool_lava",
		{pos = pos, max_hear_distance = 16, gain = 0.25})
end

if minetest.settings:get_bool("enable_lavacooling") ~= false then
	minetest.register_abm({
		label = "Lava cooling",
		nodenames = {"default:lava_source", "default:lava_flowing"},
		neighbors = {"group:cools_lava", "group:water"},
		interval = 1,
		chance = 2,
		catch_up = false,
		action = default.cool_lava,
	})
end

It's because of the chance = 2. It means that it will be run each second only in 50% chance.

[bosapara]

Code: Select all

default.cool_lava = function(pos, node)
	if node.name == "default:lava_source" then
		minetest.set_node(pos, {name = "default:obsidian"})
	else -- Lava flowing
		minetest.set_node(pos, {name = "default:stone"})
	end
	minetest.sound_play("default_cool_lava",
		{pos = pos, max_hear_distance = 16, gain = 0.25})
end

if minetest.settings:get_bool("enable_lavacooling") ~= false then
	minetest.register_abm({
		label = "Lava cooling",
		nodenames = {"default:lava_source", "default:lava_flowing"},
		neighbors = {"group:cools_lava", "group:water"},
		interval = 1,
		chance = 2,
		catch_up = false,
		action = default.cool_lava,
	})
end

It's because of the chance = 2. It means that it will be run each second only in 50% chance.

Its clear. In result we have isnt 1-2 sec, from 1 to 5, or even more

[texmex]

50% chance doesn’t mean result in 1-2 seconds. You may have a non-result streak of several seconds.

[bosapara]

50% chance doesn’t mean result in 1-2 seconds. You may have a non-result streak of several seconds.

Alright, lets talk about clock gen. There are 1 exact signal at 5 seconds, nothing different means.

But it works incorrect too.

[texmex]

I’ve no idea what clock gen is, sorry.

[bosapara]

I’ve no idea what clock gen is, sorry.

Ok, doesnt metter. Just if we will use stone generator ABM with chance 1, it will work unstable.
The same happens with any ABM.

[texmex]

AFAIK, ABMs are always slightly delayed depending on the workload of the specific server step where they are executed. Depending on the number of other active ABMs running at the same time and other jobs, this delay fluctuates from not noticeable to very laggy. Besides that, to my knowledge ABMs are in general quite unoptimized in them features C++ implementation, which is why all timing depending features should avoid the altogether. Don’t use them if possible.

[Gael de Sailly]

@bosapara: all that you describe in your first post is an expected behaviour, and is explained by the "chance" parameter. If chance = X and delay = Y, your ABM has 1/X chance to be executed every Y seconds, to add randomness.
If you want it to be executed at every time, use chance = 1, or remove the chance parameter (which does the same).

Ok, doesnt metter. Just if we will use stone generator ABM with chance 1, it will work unstable.
The same happens with any ABM.

Stop crying foul, man. Test it and then give feedback. You should also understand that ABMs were first made to handle randomly occurring events, consider node timers if you want a more precise timing.

[bosapara]

Stop crying foul, man.

Wipe your eyes and try to read this topic again.

[texmex]

You don’t seem to understand what ABMs do. Read all the answers, again.

[bosapara]

AFAIK, ABMs are always slightly delayed depending on the workload of the specific server step where they are executed. Depending on the number of other active ABMs running at the same time and other jobs, this delay fluctuates from not noticeable to very laggy. Besides that, to my knowledge ABMs are in general quite unoptimized in them features C++ implementation, which is why all timing depending features should avoid the altogether. Don’t use them if possible.

it's like the truth, + random delay, even without lags on the server

[paramat]

"Code doesn't matter" :D

Clock generator has chance 1: https://github.com/ac-minetest/basic_ma ... .lua#L1424
ABMs aren't meant to be reliable, use a node timer, globalstep timer or maybe 'minetest.after'

[bosapara]

ABMs aren't meant to be reliable

Ouch:/ it explains all my bad expectations. Thanx for explaining.

se a node timer, globalstep timer or maybe 'minetest.after'

Alright, I'll know