IRC log for #minetest-dev, 2013-11-01

Time	Nick	Message
00:16		IceCraft joined #minetest-dev
00:30		IceCraft joined #minetest-dev
00:39		IceCraft joined #minetest-dev
01:03		IceCraft joined #minetest-dev
01:11		IceCraft joined #minetest-dev
01:21		IceCraft joined #minetest-dev
02:39	VanessaE	are there any active devs on right now?
02:44	* VanessaE	random-highlights... sfan5 celeron55 thexyz proller kahrl
02:47	thexyz	VanessaE: that's not our problem why bother with it?
02:48	VanessaE	eh?
02:48	VanessaE	oh the android thing? not talking about that :)
02:48	VanessaE	however as long as you mention it, I was thinking from a name/publicity standpoint.
02:48	VanessaE	but anyway
02:49	VanessaE	#967
02:49	VanessaE	what's the word on this?
02:53	VanessaE	the only change this pull needs is that the image scale of the sun and moon should be made the same, and made abotu twice as big as they are now. Compensate for this by altering the actual images (add some empty padding).
02:54	VanessaE	(this allows some extra space in texture pack images for a corona around the sun)
02:54	RealBadAngel	this is subject for another pull
02:54	VanessaE	yes
02:54	VanessaE	one pull at a tyime
02:54	VanessaE	time*
02:54	RealBadAngel	that will add settings
02:55	RealBadAngel	ie moon_size = x, sun_size = y
02:56	VanessaE	oh hell, got my two windows swapped around.
02:56	VanessaE	for now, let's merge this patch
02:56	RealBadAngel	i havent done this because im not sure about how it should be done
02:57	VanessaE	I use it on my client, there seem to be no performance issues that I can spot, and it works exactly as one would expect it to.
02:57	VanessaE	(this patch == #967)
02:57	RealBadAngel	imho such settings shall be set by texture pack itself
02:57		darkrose joined #minetest-dev
02:58	RealBadAngel	but then, theres no way currently to let texture pack set anything
02:59	RealBadAngel	C55 started to code api for sky.cpp but this is a mod approach
03:03	RealBadAngel	VanessaE, if any (performance) issues, patch imho should be faster than previous code
03:03	RealBadAngel	before it called 3 times irrlicht routine to draw primitive quad
03:03	RealBadAngel	now it draws only one texture
03:03	VanessaE	perhaps. either way, it doesn't slow anything down and it works exactly like it should
03:04	RealBadAngel	(4 times in case of moon)
03:05	RealBadAngel	anyway main reason of keeping this thing alive was the fact of inabilitly to tint properly the sun
03:06	RealBadAngel	there was some weird code in sky.cpp to calculate colors that were never used
03:06	RealBadAngel	i got rid of that thx to tonemaps
03:07	RealBadAngel	and i can say just i was observing dawns and the sunsets IRL lately
03:07	RealBadAngel	colors shifting is just accurate
03:08	VanessaE	thanks to my tweaks ;)
03:08	RealBadAngel	i have tweaked your tweaks ;)
03:09	RealBadAngel	but yes, it was team work
03:12	VanessaE	guys come on, what's the delay here
03:12	VanessaE	yes or no?
03:15		mrtux_ joined #minetest-dev
06:18		darkrose joined #minetest-dev
07:02		nore joined #minetest-dev
07:19		Ritchie joined #minetest-dev
08:11	nore	I have a question: should the grow grass/remove grass/grow saplings ABMs left in core, or moved to Lua?
08:34	celeron55	RealBadAngel: they are not alternative approaches
08:34	celeron55	RealBadAngel: what you are doing is extra functionality for the "regular" sky
08:36	celeron55	RealBadAngel: does the default sun and moon look the same as currently? looking at the image files, they don't seem to have such functionality
08:36	celeron55	i made the sun be colored in a very specific way which is much better than coloring it with just one color
08:37	celeron55	i don't want any regressions here
08:37	nore	celeron55, do you think the grow grass/remove grass/grow saplings ABMs should be moved to Lua?
08:37	celeron55	nore: it depends on whether their speed is roughly equivalent there
08:38	nore	that's what I am testing right now...
08:38	celeron55	saplings at least should be
08:38	nore	yes, I reckon it is
08:48	nore	strange bug:
08:48	nore	I get that:
08:48	nore	09:47:04: ERROR[main]: ServerError: LuaError: error: attempt to call a nil value
08:48	nore	09:47:04: ERROR[main]: stack traceback:
08:48	nore	and then nothing
08:48	nore	(no stack traceback, no indication of where it crashed, nothing)
08:49	celeron55	sometimes that happens; dunno why
08:50	nore	yes, but there it is not sometimes
08:50	nore	it is each time I try to start the server
08:50	celeron55	i don't know in what set of cases it can't get a traceback
08:50	celeron55	there's probably some logic for that though 8)
08:52		yno_ joined #minetest-dev
08:56		dzho_ joined #minetest-dev
08:59		ecube joined #minetest-dev
08:59		ecube joined #minetest-dev
09:07	nore	celeron55, I may have found where it comes from
09:07	nore	(testing right now)
09:08	nore	but I reckon it is a callback that I defined in another branch, and then I switched branches
09:08	nore	and forgot to recompile
09:08	nore	to it looks like when C calls Lua functions, and there is an error, there is no traceback
09:09	nore	ok, it was indeed that
09:10		proller joined #minetest-dev
09:27	celeron55	but isn't that what always happens?
09:28	celeron55	the only time when lua code is executed without C calling a function is at load time
09:29	nore	no, I mean, C tried to call a nil value
09:29	nore	instead of calling of function that called a nil value
09:29	celeron55	ah, yes, that makes sense
09:29	nore	anyway, I managed to code this ABMs in Lua
09:30	nore	and I don't see more server load than before, even with both running at once
09:30	nore	moreover, those ABMs don't execute very often
09:31	nore	(average time are 500, 400 and 40s)
09:45	nore	celeron55, https://github.com/Novatux/minetest/commit/c656b6ecff88338ed6c181d33ab230a420fcce26
09:46	nore	I also put the sapling growing functions in default.grow_tree and default.grow_jungletree so a mod can override them if wanted
09:46	proller	need snow remover abm
09:47	nore	that can be added in the game too, no need to put it in the engine
09:47	proller	https://github.com/minetest/minetest/pull/882/files#diff-2e849c6a20ae0b361d0acb000da3ae02R105
09:48	celeron55	really the grass stuff should probably be considered a game thing too...
09:48	celeron55	and tree
09:48	celeron55	and well, that's what that does
09:48	nore	I will make a pull for minetest_game too if it is ok
09:49	nore	anyway, the trees should grow exactly as before with that pull, I changed nothing
09:49	RealBadAngel	celeron55, the patch contains textures that look exactly like old sun and moon
09:49	nore	another advantage is that now a mod can override, so if a mod doesn't want saplings to grow in deserts, it can do that...
09:52	celeron55	RealBadAngel: so does it look like this? http://imgur.com/a/rCb4k
09:53	celeron55	i wonder if there would be some sane way of handling this stuff that isn't really engine stuff and isn't really game stuff where it doesn't need to be decided which it is
09:53	nore	well, perhaps there should be no objects at all by default
09:54	nore	and each game would have to register sun and moon
09:54	nore	as objects moving in the sky
09:55	RealBadAngel	celeron55, http://i.imgur.com/CdBomHE.png http://i.imgur.com/QWWivwV.png
09:55	celeron55	RealBadAngel: the sun is all the same color; it looks bad
09:55	RealBadAngel	its the texture
09:56	RealBadAngel	it can look way different if you want to
09:56	celeron55	you can't make a sun like the current one with your system
09:56		ImQ009 joined #minetest-dev
09:56	RealBadAngel	why dont you try the commit instead of talkin what can be done?
09:57	RealBadAngel	because youre wrong
09:57	celeron55	because i already know and if it could be done, you'd have already done it
09:57	RealBadAngel	no you dont know
09:57	RealBadAngel	do i need to make another screenshots? :)
09:57	celeron55	yes, take screenshots of a sun made with that that resembles the current one
09:58	celeron55	or at least doesn't look worse
09:58	nore	celeron55, https://github.com/minetest/minetest/pull/976 https://github.com/minetest/minetest_game/pull/209
09:59	nore	(both pulls go together)
10:01	nore	^ any thoughts on this?
10:02	celeron55	my thought is that it's a pain in the ass for other games...
10:02	proller	how fast it is? when you fast moving and world generating?
10:04	celeron55	this goes back somewhat to the "common mods" thing, but would somthing like "builtin mods" work? a game could enable or disable them and they would be reserved mostly for this kind of "common sense" stuff, also for weather and things like that
10:04	nore	proller, I did not see difference
10:04	celeron55	and they would reside in the engine repo
10:04	celeron55	and they would default to being like what the engine has previously done by itself
10:05	nore	well, about other games, they only have to copy the code...
10:05	nore	but perhaps what would be possible would be a "mod library"
10:06	proller	i think core must have tree grow api (with step-by-step grow), and mods must define schemes and conditions
10:06	celeron55	if they want to run on multiple minetest versions, they need to check the version, so this change should be added to minetest.features
10:06	nore	and games that want e.g. sapling growth would do dofile(minetest.get_library_path().."treegrowth.lua")
10:06	celeron55	nore: i think they should be packaged exactly like any mod, because that system works
10:06	celeron55	but in a special location
10:07	nore	yes, but a game would need to enable/disable them
10:07	nore	i.e., for sapling growth, the code need to be run after the nodes are registered
10:07	celeron55	the enabling and disabling them would need to be special though... uhm, well that's ugly
10:07	RealBadAngel	http://i.imgur.com/DsuNv1Q.png
10:08	nore	whereas with a dofile(minetest.get_library("treegrowth"))
10:08	RealBadAngel	http://i.imgur.com/UIXDU51.png
10:08	nore	it would be easy
10:08	nore	or even minetest.load_library(treegrowth)
10:08	celeron55	nore: that cannot be made to be default behavior with non-updated games
10:08	celeron55	so it's useless for the purpose i wanted it for
10:09	nore	I did not understand the purpose... is it to make current games able to run without modifications?
10:09	RealBadAngel	celeron55, and now?
10:09	celeron55	RealBadAngel: those are completely different and don't have good varying colors at all
10:10	celeron55	RealBadAngel: you don't understand; good colors are important, plain textures are just one way of visualizing things
10:10	RealBadAngel	have you actually tried how the colors change during time?
10:10	celeron55	nore: imagine your patches being put in 0.4.8, and imagine yourself running some independent subgame in it
10:11	celeron55	nore: it won't have sapling or grass growth at all
10:11	celeron55	which is bad
10:11	celeron55	unless it's updated
10:11	RealBadAngel	they are defined using a tonemap, and the tinting is made with it
10:12	celeron55	RealBadAngel: i want separate tints for multiple parts of the sun; that's what the current sun does
10:12	RealBadAngel	huh? its the very same colour and different aplha
10:13	RealBadAngel	*alpha
10:13	celeron55	i don't understand why you try to do this visual stuff when you don't have good judgement on what looks good
10:14	RealBadAngel	this does look good. and is textureable
10:14	celeron55	it doesn't, and you apparently can't understand it
10:15	RealBadAngel	whats wrong with it then?
10:15	celeron55	i have already explained it
10:16	nore	celeron55, I just had an idea:
10:16	nore	games need to do minetest.register_alias("sapling", sapling_name) to have sapling growth
10:16	RealBadAngel	i dont get it. everybody around says its cool lookin but just you
10:17	nore	so minetest.register_alias could be changed to get that and register the tree growth ABM
10:17	RealBadAngel	with the textures i made it does look nearly the same as before
10:17	RealBadAngel	only tinting changes the visual
10:17	nore	of course, that would be marked as deprecated and removed in a future version
10:17	celeron55	RealBadAngel: it looks like a square-shaped orange fruit on the sky; not like a bright sun
10:18	celeron55	to visualize brightness, you need to "overflow" the brightness to white
10:18	RealBadAngel	it looks exactly the same as before without other textures
10:18	celeron55	but show the border as some other color
10:18	RealBadAngel	youre blaming me for original look??
10:18	RealBadAngel	put there round texture with halo
10:18	celeron55	nore: hmm, what about grass?
10:19	nore	same thing, IIRC
10:19	RealBadAngel	http://i.imgur.com/IFLGLay.png
10:19	RealBadAngel	like this one
10:19	RealBadAngel	its textureable for christ sake
10:19	nore	ah, no
10:20	RealBadAngel	more, texture pack maker can provide his own tinting file
10:20	nore	it is mapgen_dirt, mapgen_dirt_with_grass, but dirt_with_snow
10:20	nore	and snow too
10:21	celeron55	RealBadAngel: it's like saying "hey i made a car; it doesn't have front tires but you can have any kind of rear tires you want"
10:21	RealBadAngel	no it does have tires that you have designed
10:21		addi1 joined #minetest-dev
10:21	nore	celeron55, perhaps detect the dirt_with_snow alias
10:21	RealBadAngel	to not change original feel
10:21	nore	and if so, register the ABM
10:22	nore	(it is getting more and more hacky)
10:22	RealBadAngel	i made square textures to keep it
10:22	celeron55	nore: maybe compatibility should just be dropped...
10:22	nore	I don't mind about that...
10:22	nore	moreover, it is a game thing, not a mod thing
10:22	nore	and there are max. 10 games out there
10:22	celeron55	yeah, i guess it's fine
10:23	nore	ok, 30
10:23	celeron55	it probably should be implemented as a separate mod in minetest_game so that it's easy to take into use anywhere
10:23	celeron55	or should it?
10:23	nore	I don't know
10:24	celeron55	my opinion on minetest_game is that nothing should be added to default anymore
10:24	nore	well, sapling growth is default
10:24	celeron55	it's too bloated and if adding stuff to it continues, it cannot be used as an universal based for games anymore because of it's bloatness
10:24	RealBadAngel	celeron55, take a look at this vid http://www.youtube.com/watch?v=3jV7Oxw-IJo
10:24	celeron55	base*
10:24	nore	so perhaps leave it here...
10:24	nore	celeron55, almost any game tweaks default anyway
10:24	RealBadAngel	actual tinting file in commit is a bit different, but its pretty close
10:24	celeron55	RealBadAngel: i think i'll use my time more productively than argue with you
10:26	nore	celeron55, however, should I change something to warn the games that there is no sapling growth?
10:27	celeron55	nore: i don't know; we really should have some regular way of handling things like these, as this kind of things are going to happen in the future
10:28	nore	so, I should add it to features.lua?
10:29	nore	this can be thought as a regular way to handle that
10:29	celeron55	yes, because it's an undetectable feature
10:29	celeron55	or, change
10:30	celeron55	or... umm... well, it can be detected by checking if default.grow_tree exists, but that's quite arbitrary
10:30	nore	Is no_legacy_abms a good name for that?
10:30	celeron55	yes
10:30	RealBadAngel	celeron55, so you dont like it to be plain colours, ok. but what about shape? shall it be squarish as before or more round?
10:30	nore	celeron55, no since it is for a game
10:31	nore	celeron55, pushed
10:31	celeron55	RealBadAngel: shape is not an issue with textures
10:31	nore	RealBadAngel, about tinting
10:32	nore	it could be possible to make the image to be tinted use RGB
10:32	nore	and have 3 tintmaps, one for each channel in the image to be tinted
10:32	celeron55	that would allow what i want
10:33	RealBadAngel	that could only be done using shaders
10:33	nore	why not? how is tinting done right now?
10:33	RealBadAngel	emissive light of a material
10:33	celeron55	not if you separated r, g and b to different images and overlaid them with different emissive lights
10:33		Akien joined #minetest-dev
10:34	celeron55	or had them separate from the start
10:34	nore	in that case, 2 colors should be enough, no?
10:35	celeron55	enough for something that looks like the current sun, yes; the amount of channels is kind of arbitrary though
10:35	nore	that's why it is perhaps best to have a single image
10:35	celeron55	this really is a thing in which a client-side script would be immensely more good than anything else
10:36	nore	so it would be less heavy to send textures
10:37	RealBadAngel	current moon texture is 286 bytes...
10:37	nore	celeron55, if you have got some time, could you check #966 please?
10:37	nore	RealBadAngel, yes, you may be right that it is not needed
10:37	RealBadAngel	its tone map is 260 bytes
10:38	celeron55	nore: did someone look at it already?
10:39	celeron55	i really need to do something else right now
10:39	nore	RealBadAngel did it, and PilzAdam did too but had no time to test the latest version
10:39	nore	but if you have no time, it isn't really a problem, I will wait
10:41	celeron55	maybe i'll throw you and shadowninja or someone in github members now so you can maybe do something
10:41	celeron55	altough no matter who i put in there, you're going to be on your own for today and not be able to do anything 8D
10:41	celeron55	(proably)
10:41	celeron55	+b
10:42	celeron55	i'm very hesitant in agreeing with this ABM thing, because there have been many things recently that have turned out not ideally
10:43	celeron55	it feels like it would be best to halt all development and say it's done and will only go worse
10:43	celeron55	and it sucks
10:43	RealBadAngel	i have altered the look of the sun a bit: http://i.imgur.com/vrTlUjv.png http://i.imgur.com/uOjgKNL.png http://i.imgur.com/vwukEre.png
10:43	proller	101% of servers now have HUGE lags
10:44	proller	part of lag from lot of slow abm
10:44	nore	proller, those ABMs are very rarely triggered
10:44	celeron55	proller: coding an army of ABMs in C++ is not a sustainable solution
10:44	nore	and they don't use much time
10:44	celeron55	there will always be more content and having always more content in the engine is just bad
10:45	proller	celeron55, but coding good fast c++ api may help
10:45	nore	the one that is slowest is sapling growth, but that one triggers only once every 500s
10:45	nore	and once it has triggered, it won't do it anymore
10:45	proller	nore, but always on generate?
10:45	nore	what do you mean, on generate
10:45	nore	these are not called on generate
10:45	proller	on map generate
10:46	proller	2 ways to row tree?
10:46	proller	grow
10:46	nore	sapling growth is not called on map generate
10:46	nore	it is only for ABM, the map generate one is still C++
10:46	celeron55	would it be possible to expose the engine's tree grow code to lua for that? it probably is
10:46	proller	trees grow code must be in one place
10:47	proller	and lua is slow place for it
10:47	nore	celeron55, there are 2 possibilities
10:47	celeron55	altough, we could just as well just change both to some reasonably similar L-system tree
10:47	nore	1. tree growth code is in C++, but then no mod can override it
10:47	celeron55	(assuming it doesn't have performance problems)
10:47	nore	2. it is in Lua, but if called by the mapgen, it gets slow
10:48	proller	L-system + good conditions (heat, humidity. neibhours)
10:48	nore	and L-system trees are slow
10:48	celeron55	well fuck
10:48	nore	proller, yes, but what if a mod wants its own humidity functions?
10:48	proller	nore, for what?
10:49	nore	proller, for everything... (mapgen, etc)
10:49	celeron55	maybe because the engine doesn't have proper ones for a kind of world it wants to do
10:49	nore	for example, when coding mg, I use custom temperature and humidity noises
10:49	proller	maybe define condition callback
10:50	nore	remember that callbacks are slow
10:50	proller	and mod can deside about growing here and ask engine to start grow
10:50	nore	on the other hand, it is possible to do a fast lua-only mapgen
10:51	nore	with the perlin maps and voxel manips, it is possible to do almost as fast as C++ mapgen
10:52	nore	tests by hmmmm:
10:52	nore	minetest.env:set_node: 6854.4ms
10:52	nore	VoxelManip: 153.5ms
10:52	nore	Mapgen in core: 60ms (estimated)
10:52	celeron55	should we switch the mapgen to completely lua? 8)
10:52	proller	i want to make step-by-step growing and leaves recovery - its maybe too slow in lua
10:52	nore	no, but we don't need to change everything...
10:52	celeron55	proller: then it won't be done
10:53	celeron55	proller: unless it's generic enough
10:53	nore	proller, if you use voxelmanip, I reckon it can be done fast enough
10:56	celeron55	i added nore, ShadowNinja and sapier to core devs now
10:56	celeron55	hopefully stuff doesn't completely explode
10:57	proller	nore, its was set only benchmark? growing tree needs lot of get_node
10:57	nore	celeron55, could we be added in the credits tab too please?
10:57	celeron55	just add yourself
10:57	celeron55	lol
10:57	nore	proller, get_node is done with voxelmanip...
10:57	nore	one call, and you have all nodes
10:59	celeron55	i'm basing this addition on the hope that if people follow what i wrote here, it's possible to avoid huge conflicts in direction and unmergeable work: http://c55.me/blog/?p=1491
10:59	celeron55	maybe this is the worst decision of the year, but that i'll see next year
10:59		jin_xi joined #minetest-dev
10:59	celeron55	8D
11:02	celeron55	(well yeah, that was inappropriate; but i'm full of inappropriateness)
11:13		nore_ joined #minetest-dev
12:05		Zeitgeist_ joined #minetest-dev
12:09		Jordach joined #minetest-dev
12:13		hmmmm joined #minetest-dev
12:27	proller	nore, commiting without approve and waiting few months prohibited !
12:29	nore	what is prohibited? waiting not enough or too long?
12:30	nore	anyway, that commit was approved... ;)
12:30	nore	proller, ^
12:32	VanessaE	celeron55: While I understand your desire to keep the different parts of the sun/moon tinted as they are now, does it truly matter if they change a little bit for the sake of overall improvement? I know the type of change that is necessary to get exactly the effect you want, and it's not hard, but is it really THAT important right now?
12:33	proller	nore, it was 51% joke
12:33	VanessaE	the type of change being a 2d tonemap - the X axis being time, the Y axis being distance from the center of the image
12:35	VanessaE	(whether it would be particularly fast, or how it would compare to your idea of overlaying multiple images, I'm not sure)
12:35	kahrl	about the sometimes missing debug traceback
12:35	kahrl	I'm pretty sure that comes from all the lua_pcall calls that pass 0 for the last parameter
12:35	nore	proller, yes I know
12:35	kahrl	I asked about that some weeks ago but nobody seemed interested at the time
12:36	nore	kahrl, what does the last parameter mean in lua_pcall?
12:37	kahrl	http://www.lua.org/manual/5.1/manual.html#lua_pcall
12:38	nore	kahrl, does is work if the function is nil?
12:38	kahrl	especially "Typically, the error handler function is used to add more debug information to the error message, such as a stack traceback. Such information cannot be gathered after the return of lua_pcall, since by then the stack has unwound."
12:38	kahrl	nore, pretty sure no
12:39	nore	because I had missing traceback because C++ tried to call a nil value
12:39	kahrl	oh in that case
12:39	nore	(in fact, I had not recompiled and the callback was not in builtin anymore)
12:39	kahrl	no idea what happens then
12:47	kahrl	if I understand the lua code correctly it's supposed to raise a type error
12:48	kahrl	and it calls the error handler passed to lua_pcall
12:50	celeron55	VanessaE: a 2d tonemap like that won't be good; i'm more with something like nore suggested
12:51	celeron55	VanessaE: and i think it shouldn't be merged until it can do something resembling the current sun.
12:52	celeron55	people will surely find a lot of use for the functionality other than remaking what already exists
12:55	VanessaE	celeron55: I don't see nore's suggestion? (all I see is his commentary on the tree stuff)
12:56	nore	VanessaE, it is two use RGB in the image to stone 3 different channels
12:56	nore	and each channel is tinted
12:56	VanessaE	oh yes
12:57	celeron55	10:33 UTC
12:57	VanessaE	yeah I saw it, just couldn't find it again on the second scrollback
12:57	VanessaE	nore's idea could work
12:59	celeron55	one alternatively really also is to assume the only purpose of the multi-tonemap is to allow visualizing brightness, and somehow do an overexposure effect for the end result to allow it with a single tonemap
12:59	celeron55	alternative*
12:59	celeron55	but it's not properly doable without a shader
13:00	celeron55	(and it's very inflexible)
13:00	celeron55	(but, almost everything falls in the category of "not properly doable with a shader")
13:00	celeron55	without*
13:01	VanessaE	then the solution is obvious: get as close as it can without shaders, and without a performance penalty
13:02	VanessaE	(and add a shader later on if necessry)
13:02	VanessaE	necessary*
13:05	VanessaE	all I know is, aside from the visible sizes of the sun and moon relative to the space given to their corona and halo, what we have now works well enough from a texture author's standpoint.
13:07	celeron55	RBA would be a much better graphics developer if he understood these kind of artistic standpoints
13:08	pitriss	RealBadAngel: I just read backlog.. and sorry, but I think your shaders and sun modifications doesn not fit of game theme..
13:08	VanessaE	well I can't speak for him directly of course, but I would guess his point is "good enough" is better than "not done at all". At that point, "good enough" is dependent on the person.
13:08	VanessaE	pitriss: then don't use them.
13:09	pitriss	VanessaE: of course, it will be option when not using them don't change default look..
13:17		proller joined #minetest-dev
13:39		jojoa1997 joined #minetest-dev
13:41		salamanderrake joined #minetest-dev
13:47		jojoa1997 joined #minetest-dev
14:15	ShadowNinja	As it is pretty simple and has already been agreed to by multiple other core devs I will merge #975 in 15 minutes, unless there are objections.
14:16	ShadowNinja	nore: Minetest schematics were designed for placing trees, both by the core and Lua. See PilzTest.
14:16		iqualfragile joined #minetest-dev
14:17	ShadowNinja	They should be very fast too.
14:18	ShadowNinja	nore: Better not to add a ABM on default:dirt, as there is a lot if it. Perhaps change dirt below on snow dig/place.
14:19	ShadowNinja	nore: And core treegen should be removed if it is no longer used.
14:21	ShadowNinja	#942 is purely a fix, can someone else check it?
14:22	nore	ShadowNinja, I am ok for#975
14:23	nore	ShadowNinja, core treegen is used currently
14:23	celeron55	it's not a problem for the engine to have an ABM for "dirt with air as a neighbor"
14:23	celeron55	as it's so infrequent one
14:23	nore	and that ABM on default:dirt has chance = 200, so it runs without any problem
14:24	celeron55	it has always been that way and should be in the future too
14:24	nore	and celeron55, it is not possible to put air as neighbor
14:24	celeron55	oh, it's not?
14:24	nore	yes, because snow or other sunlight_propagate node will make grass too
14:24	ShadowNinja	As for #972, "#def ~= 0" is definitely wrong. Lua just looks for the largest integer index.
14:24	celeron55	ah
14:24	celeron55	well whatever
14:24	nore	but adding a sunlight_propagates group would work
14:25	celeron55	sunlight_propagates isn't a group
14:25	nore	ShadowNinja, you are right for #972
14:25	ShadowNinja	(Which nodedefs shouldn't contain)
14:25	celeron55	(and grass should grow in less than sunlight too)
14:25	nore	yes, so I mean adding a group sunlight_propagate
14:25	celeron55	i don't think anything like that is needed, it's fine as-is
14:25	nore	celeron55, grass should be able to grow below a ladder, for example
14:26	nore	but anyway, since chance = 200, the ABM runs fine
14:27	nore	and do you haev any thoughts on #966? (craft callback)
14:28	nore	I have some code to test it: https://gist.github.com/Novatux/7179966
14:29	nore	celeron55, I have an idea that would go with #960
14:29	nore	it would be a function that adds another object in the sky for a player (like sun, moon)
14:31	nore	ShadowNinja, did you see #966?
14:31	ShadowNinja	How about #942? Previously falling nodes would stop at unknown nodes(because of collision detection), but stay in entity form because they didn't think it was a solid node.
14:32	ShadowNinja	nore: I haven't tested it.
14:32	nore	I'm ok for #942
14:32	nore	if it works, of course
14:32	celeron55	i really have forgotten how crafting currently works even though i have made most of it
14:32	ShadowNinja	nore: #966 needs squashing.
14:33	nore	ShadowNinja, yes I will squash it for when it is merged
14:33	celeron55	but your scriptapi code is missing some "const"s in function parameters that are supposed to be constant references
14:33	nore	ehh, what are "const"s?
14:33	nore	and where should they go?
14:33	celeron55	you code C++ and don't know what is "const"? 8)
14:33	celeron55	maybe google a bit
14:34	nore	const is a parameter that does not change, is it?
14:35	celeron55	yes; it's very recommended to use const for parameters that are pointers or references that aren't going to be modified by the function
14:35	nore	and for the inv, for example
14:35	nore	the lua callback can modify it
14:35	nore	so should I const it?
14:36	celeron55	i think only the item is modified in those
14:37	nore	I mean, the callback can change the content of the inventory
14:37	nore	so I should put a const before the InventoryList and the InventoryLocation, am I right?
14:37	celeron55	it doesn't matter because it doesn't happen there... i guess
14:38	nore	and why not before the player?
14:38	celeron55	that too
14:38	nore	(the other callbacks don't have ot for the player)
14:38	celeron55	but
14:38	celeron55	it looks like the other callbacks don't have it either
14:38	celeron55	i wonder if there's some reason for this or if it's just lazyness
14:38	proller	nore, we have 3+ places with light calculations, you want to make 2 places with tree grow
14:39	nore	yes... (but mapgen tree grow and sapling tree grow are not the same...)
14:39	celeron55	nore: leave the ServerActiveObject*s as not const as that seems to be the convention there
14:40	celeron55	but eg. some PointedThing&s are const so maybe the inventory and inventorylocation can be const like them
14:40	proller	nore, but they produce same trees
14:40	nore	celeron55, it doesn't want to compile now
14:40	celeron55	what does it say
14:41	nore	ah no, I forgot the header... ;)
14:41	nore	ok, it works
14:43	celeron55	but what is the point of this pull
14:43	celeron55	is there something useful that can be done with it
14:43	nore	and the code still works
14:43	nore	celeron55, tons of things
14:44	nore	for example, tools that wear out when crafting
14:44	nore	see the two examples I did
14:44	nore	https://gist.github.com/Novatux/7179966
14:44	nore	first one crafts steelblocks from ingots, but with only 1 stack of ingots, taking 9 each time
14:45	nore	second one crafts axe+stone to get stone slabs, and the axe a bit worn out
14:45	celeron55	hmm... i guess it could also do some kind of fine-grained permission or ability things
14:45	nore	that can allow player-dependent recipes
14:45	nore	achievements for crafting/statistics
14:45	nore	etc
14:45	celeron55	can it nullify the craft result?
14:46	celeron55	that is, disable a recipe that exists in the normal system
14:46	nore	yep, and if craftpredict return ItemStack(""), there is no need to add the on_craft function
14:46	celeron55	good
14:47	nore	craftpredict changes the craft prediction on client too
14:47	celeron55	i'm okay with it
14:47	nore	so, I can merge it?
14:48	celeron55	that's how it goes 8) just remember to not use github's floody merge commits
14:49	nore	so, I make a patch file
14:49	nore	and then git am patch_file
14:49	nore	is that right?
14:50	ShadowNinja	Since it's local you can just run "git cherry-pick 1234567890abcdef"
14:51	ShadowNinja	But for other pulls "wget <url>/123.patch -O ../patches/123.patch && git am ../patches/123.patch"
14:54	celeron55	making a shell script for that is useful
14:54	celeron55	less copypasting of numbers
14:56	ShadowNinja	#972 seems good, anyone else to agree?
14:56	nore	ok, checking one last time if it compiles before pushing...
14:56	nore	I'm ok for it too
14:57	ShadowNinja	nore: Ok for #972?
14:57	nore	yes, but wait before pushing
14:58	nore	ok, now you can
14:58	celeron55	it seems correct; i wonder why what it has even ended up there
14:59	celeron55	apparently i committed it as a pull request made by some unknown person
14:59	nore	someone that didn't know Lua must have coded that ;)
14:59	celeron55	...in 2012-08
15:01	celeron55	oh, not unknown person but matttpt
15:03		iqualfragile joined #minetest-dev
15:03	nore	btw, UI needs to be updated...
15:03	nore	sry, wrong channel
15:06	ShadowNinja	What about minetest_game/#201? I have seen numerous servers(including mine) add it manually.
15:07	ShadowNinja	s/manually/locally/ would be better.
15:07		PilzAdam joined #minetest-dev
15:10	nore	what is it?
15:11	ShadowNinja	nore: SamII + cape.
15:11	nore	I'm ok with it too
15:11	nore	ShadowNinja, and #856?
15:13	ShadowNinja	nore: Seems good to me of course, but I'm not sure if everyone else agrees, better wait.
15:13	nore	I agree with it (I'm already using it for my mods anyway)...
15:14	nore	#968, #971 and #973 look like fixes, so perhaps merge them too
15:14	nore	#965 too
15:15	ShadowNinja	celeron55: /msg ChanServ template #minetest-dev dev !+oir ? See /msg ChanServ help flags/template
15:18	nore	ShadowNinja, what is that?
15:18	ShadowNinja	I don't know about #968 or #971. #973 has a ugly way of copying VERSION_MINOR, and I can't test #965.
15:18	nore	I can't test any of those too
15:18	ShadowNinja	nore: It gives devs the ability to op, invite, and unban.
15:19	nore	ShadowNinja, #963 is good too
15:19	ShadowNinja	+via ChanServ
15:19	nore	I tested it
15:20	ShadowNinja	I don't understand it. Does it keep the game background in menus like "Configure"?
15:20	nore	it keeps it for create world menu, configure menu, etc
15:21	nore	(to know that, I tested it, it was simpler...)
15:21	ShadowNinja	Ok, that sounds good, but better ask someone else too.
15:21	nore	I am ok for #202 and #203 in minetest_game too
15:22	nore	and what do you think of #198 and #200?
15:22	PilzAdam	#200 is a no from me
15:23	nore	what would be needed for #200 to be merged?
15:23	PilzAdam	since that goes against what celeron55 said about not changing minetest_game too much anymore
15:24	nore	and #198?
15:24	ShadowNinja	#198 looks good. #200 shouldn't just repeat the function over and over, leaving a area for a long time could cause the server to lock up when it is loaded.
15:24	nore	ShadowNinja, there is a max number of times it can be repeated
15:25	nore	so it doesn't lock up
15:26		iqualfragile joined #minetest-dev
15:26	ShadowNinja	Still, it should exit the loop when everything is burned, and there are other optimiztions that can be done.
15:27	nore	yeah, I guess you're right
15:27	nore	I may change that
15:28	nore	but are you ok for #198?
15:28	ShadowNinja	I guess #206 is good, unless someone says otherwise.
15:28	ShadowNinja	Yes.
15:29	nore	I don't have anything to say against #206
15:29		zat joined #minetest-dev
15:37	nore	anyone else agreeing on #963?
15:40	ShadowNinja	PilzAdam: ^
15:41	zat	^.^ <3
15:42	PilzAdam	nore, works for me
15:43	nore	are you ok for it?
15:43	nore	if so, I will merge it
15:43	PilzAdam	if it doesnt break anything else
15:45	nore	PilzAdam, it doesn't
15:45	nore	if it worked before, it should still work now
15:49	PilzAdam	what about https://github.com/PilzAdam/minetest_game/commit/d5c1105d8fa8ef362707bfd538d29673db215da8 ?
15:50	nore	I'm ok for it too
15:51	ShadowNinja	Sounds good.
15:54	PilzAdam	ShadowNinja, do you want your e-mail be listed in minetest.net/contributors?
15:57	thexyz	what's this stupidity, for some reason I don't have a button to "publicize membership" for sapier and ShadowNinja
15:58	ShadowNinja	thexyz: You can have secret menberships I beleive, good for private repos I guess, or something like that.
15:58	thexyz	nore is Novatux, right?
15:58	nore	yes
15:59	thexyz	okay
15:59	ShadowNinja	PilzAdam: Sure, but I would like to have a @minetest.net one there(If possible). I beleive they are just forwards, right?
15:59	PilzAdam	ask thexyz about that
15:59	ShadowNinja	Sapier has another reason to register now. ;-)
16:00	thexyz	yeah pm me the address you want
16:02	ShadowNinja	thexyz: Maybe you can /msg ChanServ template #minetest-dev dev !+ovir ?
16:02	thexyz	maybe
16:03	thexyz	can you explain what do those mean?
16:04		Akien joined #minetest-dev
16:04	Exio4	it will update the "dev" template, adding +o, +v, +i and +r to all of the users with that group
16:04	Exio4	+ = access to op command, +v to "voice", +i, to invite, and +r to unban/quiet
16:05	Exio4	s/^../+o/
16:05	celeron55	wait, what
16:05	celeron55	thexyz: i can't either
16:05	celeron55	have they enabled some setting in github to not allow public memberships or something
16:05	thexyz	dunno
16:05	celeron55	we definitely don't want that
16:06	thexyz	> The user can publicize their own membership. (A user cannot publicize the membership for another user.)
16:06	thexyz	from http://developer.github.com/v3/orgs/members/#publicize-a-users-membership
16:07	ShadowNinja	My membership is public now.
16:07	celeron55	okay, i wonder if that has changed because i have never stumbled upon this before
16:07	celeron55	it probably has
16:13	ShadowNinja	Well, after receiving a email in russian, I think forwarding is set up, and I added the email to the contributor page.
16:15	ShadowNinja	thexyz: I beleive there is a developer group on the forum. Could you add me to it? My username is "Shadow" (ShadowNinja was taken).
16:15	nore	Ah, and me too please...
16:17	ShadowNinja	thexyz: And "Your signature is not displayed ... not spent enough time at the forum." Can the limit be lowered? I remember it being lower.
16:18	ShadowNinja	And I don't like how some users flood the forum with messages and get privileges like ^ even though they haven't contributed sygnificantly to any discussions.
16:18	nore	thexyz, can I be in the "Developper" group too please?
16:20	thexyz	yeah, sorry, I forgot about forums, done now
16:22	ShadowNinja	It isn't showing for me, is there a delay? I tried ignoring the cache, so it isn't that...
16:23	thexyz	what isn't showing for you?
16:24	ShadowNinja	The "Developer" line.
16:24	thexyz	https://forum.minetest.net/userlist.php?username=&show_group=5&sort_by=username&sort_dir=ASC&search=Search
16:25	ShadowNinja	thexyz: "Shadow", not "ShadowNinja". (Someone else took that a while ago)
16:25	thexyz	oh right, can't read, sorry
16:26	thexyz	okay, done now
16:26	ShadowNinja	thexyz: Could you swap those usernames or something like that?
16:27	ShadowNinja	thexyz: The developer group allows me to post in "News", does it give any other abilities?
16:27	thexyz	hm.. dunno what should our policy regarding that issue be, celeron55?
16:29	ShadowNinja	He only posted a few times a long time ago, and the username is pretty close.
16:29	thexyz	ShadowNinja: http://i.imgur.com/Qw3aa4P.png
16:29	thexyz	you can change your title!
16:31	ShadowNinja	Title == username I guess. But I probably can't change it to the same as another user...
16:32	ShadowNinja	One of us needs to take a temp username.
16:32	thexyz	no, title == that thing below your username (it's set to "Developer" now)
16:32	ShadowNinja	Oh, hmmm, that's interesting. But I think I will leave it.
16:34	thexyz	yeah, I'm not sure what to do about usernames, ask celeron55
16:36		IceCraft joined #minetest-dev
16:47	proller	nore, can you make logic for lua dirt like here https://github.com/minetest/minetest/pull/882/files#diff-2e849c6a20ae0b361d0acb000da3ae02R63 ?
16:50	proller	if weather enabled - remove grass if too cold, add snow if very cold, remove snow if too hot, remove grass if too hot
16:56	nore	proller, how do you get block heat from Lua?
16:57		rubenwardy joined #minetest-dev
16:58	proller	if minetest.setting_getbool("weather") then ...... if minetest.get_heat(p0) < 500 then .....
16:59	proller	here is burning on hot and flower grow stop - https://github.com/proller/minetest_game/compare/weather
17:00	nore	proller, I will change that if your pull is merged too...
17:00	nore	if mine is merged before, you will have to change yours
17:02	proller	get_heat now available but sometimes return garbage
17:02	ShadowNinja	thexyz: (Seems to have been forgotten) /msg ChanServ template #minetest-dev dev !+ovir ?
17:03	thexyz	ask celeron, there was a discussion about giving ops to developers a while ago and he seems to be against it; and why should devs have op anyway?
17:05	celeron55	i see no benefit from that; devving and moderating are completely different things
17:06	ShadowNinja	So that they can moderate the channel. If you only have a few ops then someone can, eg, flood the channel for a few hours.
17:06	celeron55	maybe it could be said that this channel should be ruled by developers but... it's not like there's even anything to moderate here ever
17:07	celeron55	oh whatever, i guess that could be done if people insist... but giving power to people who insist having power is rarely a good idea
17:10	celeron55	i can promise to give ops to developers at the instant when someone starts flooding this channel repeatedly
17:10	ShadowNinja	Fair enough. Does #974 look ok?
17:13	rubenwardy	81 pull request :O
17:13	rubenwardy	*s
17:13	ShadowNinja	rubenwardy: It was 85 earlier today.
17:13	rubenwardy	lol
17:14		werwerwer_ joined #minetest-dev
17:14	nore	ShadowNinja, it looks ok
17:14	ShadowNinja	Anyone else to agree with #965?
17:15	nore	ShadowNinja, I can't say what #965 does
17:15	nore	so I can't agree with it, and I can't test it either
17:16	nore	is there anyone to agree on #976?
17:17	ShadowNinja	nore: Use MTSMs instead. PilzTest has one.
17:17	nore	what is MTSMs?
17:17	rubenwardy	Dissecting the Minetest source code is fun(!)
17:17	nore	schematics?
17:18		Calinou joined #minetest-dev
17:18	nore	ShadowNinja, schematics do not make good trees
17:19	nore	and that code is fast enough now... (the one I am wondering if it is fast enough is the grass removing code)
17:20	PilzAdam	schematics should be used for trees
17:21	rubenwardy	If send_formspec packets are handled here: https://github.com/minetest/minetest/blob/master/src/client.cpp#L1977
17:21	rubenwardy	where are they taken from m_client_event_queue
17:22	rubenwardy	found it, i think: https://github.com/minetest/minetest/blob/master/src/game.cpp#L2253
17:24	rubenwardy	Wow, I talk and it goes silent. Am I realy that noob?
17:24	rubenwardy	*really
17:25	Exio4	rubenwardy: yes, because we all hate you
17:25	rubenwardy	8)
17:25	celeron55	obviously!
17:35	ShadowNinja	Does #957 look good?
17:37	nore	ShadowNinja, I trying to use the ABM profiler to test #976
17:37	nore	however, git does not want to apply the patch: corrupted patch at line 86
17:37	nore	what does that mean?
17:37		rubenwardy joined #minetest-dev
17:37		rubenwardy joined #minetest-dev
17:38	ShadowNinja	nore: It probably needs to be rebased.
17:40	rubenwardy	Are there any logical situations where the formspec string would be "", except for minetest.send_formspec("player","","")?
17:44	nore	ShadowNinja, are there commands merge a patch file?
17:45	ShadowNinja	nore: "git am" loads a patch file. Not sure what it does about conflicts.
17:50	celeron55	it doesn't handle conflicts well by default
17:52	celeron55	i used this to deal with it succesfully once: http://www.pizzhacks.com/bugdrome/2011/10/deal-with-git-am-failures/
17:53	celeron55	(it's much more robust by adding a remote and doing it that way instead of git am)
18:10	ShadowNinja	Should I add sapier's email to the mainmenu? It wasn't there before, but he uses it in his commits.
18:10	rubenwardy	https://github.com/minetest/minetest/blob/5dce44ec5e5f7231dba3cdbec3f1fdc683ce5a34/src/game.cpp#L2253
18:10	rubenwardy	If current_formspec = 0, does that mean that their is no formspec open?
18:11	rubenwardy	and where is the current_formspec / current_textdest change intercepted?
18:11	celeron55	why is my e-mail in the menu? i think i wanted it not to be there
18:12	ShadowNinja	I will remove it if you like.
18:14	celeron55	whatever, just leave it
18:19		werwerwer joined #minetest-dev
18:28	rubenwardy	Hello?
18:31	rubenwardy	Can someone help me with that :P At least point me in the right direction
18:32	celeron55	what are you trying to find?
18:33	rubenwardy	The bit where the send_formspec is intercepted, so I can cancel it if string == ""
18:34	rubenwardy	I have seen the packet recieve in client.cpp, and the event handle from list in game.cpp
18:34	celeron55	uh, so server side?
18:34	rubenwardy	client side
18:34	rubenwardy	I want minetest.send_formspec("player","","") to close open formspec
18:35	rubenwardy	Basically what happens after here: https://github.com/minetest/minetest/blob/5dce44ec5e5f7231dba3cdbec3f1fdc683ce5a34/src/game.cpp#L2274
18:36	rubenwardy	Just noticed m_form_src
18:37	rubenwardy	found it
18:39	rubenwardy	lol
18:40	rubenwardy	SegFault when I try to close the formspec :P
18:41	celeron55	how and where do you close it
18:43	ShadowNinja	Does #197 look good?
18:43	rubenwardy	here: https://github.com/minetest/minetest/blob/master/src/guiFormSpecMenu.cpp#L1760
18:43	rubenwardy	with: quitMenu()
18:43	rubenwardy	I suppose it is because it is not deleted from what ever list you use to store it, so when it tries to update bad things happen
18:43	celeron55	you did add return; after it?
18:43	rubenwardy	yeah
18:44	rubenwardy	if (newform.compare("")==0){quitMenu();
18:44	rubenwardy	return;}
18:45	ShadowNinja	Or if (newform == "")...
18:45	celeron55	can't really help; this menu stiff is kind of convoluted
18:47	rubenwardy	Thats not the problem, shadowninja, if I remove the quitMenu, there is no crash.
18:48	rubenwardy	How ever, the form is invisible, and the mouse is captured
18:48	ShadowNinja	Yes, not the problem, but a better way to code it.
18:49	rubenwardy	Does one instance of guiFormspecMenu handle all formspecs?
18:49	celeron55	yes, but that instance is deleted when there is nothing to show
19:08		proller joined #minetest-dev
19:17		IceCraft joined #minetest-dev
19:19		werwerwer joined #minetest-dev
19:44		damiel joined #minetest-dev
19:44		damiel joined #minetest-dev
19:54		damiel joined #minetest-dev
19:54		damiel joined #minetest-dev
19:57		mkn joined #minetest-dev
19:57		sapier joined #minetest-dev
20:03		mkn joined #minetest-dev
20:15		ecube_ joined #minetest-dev
20:20		Zeitgeist_ joined #minetest-dev
20:24	nore	Ok, I tested the ABM speed:
20:24	nore	dirt and dirt_with_grass growth: always below 1/4ms
20:25	nore	every 2 seconds
20:25	nore	and for each sapling that grows: C++: 130ms, Lua: 150ms
20:25	nore	so I reckon ABM speed is not a real problem even in Lua
20:26	nore	celeron55, ^
20:26	sapier	wait 130 --> 150? per sapling?
20:26	ShadowNinja	sapier: Do you find that surprising?
20:27	sapier	thats 15% ? I'd not call this "no problem"
20:27	nore	sapier, yes
20:27	ShadowNinja	The VM is quite fast, and LuaJIT is comparable to C++ iin terms of speed.
20:27	sapier	are your mesurements done with lua or luajit?
20:27	nore	15%, knowing that under normal conditions, a sapling takes 500s to grow
20:27	ShadowNinja	That sounds good to me.
20:28	sapier	I'd consider 130ms a problem not talking about 150ms
20:28	ShadowNinja	sapier: You have been made a core dev, you will have to register with NickServ if you want +V.
20:28	nore	it is with LuaJIT enabled, I guess (since it is installed on my system)
20:29	sapier	so if 10 saplings grow same time you add 1.5 lag am I correct?
20:29	ShadowNinja	sapier: Are you OK with me adding your email(from our commits) to the main menu?
20:29	ShadowNinja	+y
20:29	sapier	no plz don't add it I removed it on purpose
20:30	ShadowNinja	Ok.
20:30	sapier	I don't even know what +V is ... do I need it?
20:31	ShadowNinja	sapier: Auto-voice.
20:31	sapier	plz give me more information I still have no clue ;-)
20:32	sapier	wait did you say core dev? ... so I can't issue unfinished pull requests any longer? ;-)
20:32	ShadowNinja	sapier: See how all the other core devs are voiced? (A + in front of the nick usualy)
20:33	sapier	talking about unfinished ... https://github.com/sapier/minetest/tree/asynchronous_events_2 fixes the "hang" issue with modstore
20:34	sapier	it's a quite more generic solution than thexyz's one
20:34	nore	sapier, about your 1.5s lag, before it was 1.3s lag
20:34	nore	so it is quite the same
20:34	thexyz	sapier: what's my solution?
20:35	thexyz	sapier: also, please register your nickname
20:35	sapier	yes ... but I'd prefere 1.3 being reduced instead of increased ;-)
20:35	thexyz	(when you have time)
20:35	thexyz	I don't remember anything like that, but whatever
20:35	sapier	ok ok as everyone demands ... where do I register?
20:35	sapier	did I mix you with kahrl again? :)
20:35	thexyz	http://freenode.net/faq.shtml#userregistration
20:36	thexyz	dunno, those days I don't do anything
20:36	nore	and if you have 100 saplings, the probability that at least 10 of them grow at once is 0.0000344168, so...
20:36	sapier	I was talking about the httpfetch patches
20:36	nore	if that happens, you're really unlucky
20:36	sapier	ok ok :-) I guess it'll be fine at that chances ... still in general 15% performance reduction IS an issue ;-)
20:37	proller	sapier, animal_* broken again
20:38	nore	sapier, the performance is even worse for dirt... but since it is 250Âµs instead of 50Âµs every 2 seconds, nobody minds
20:38	nore	(I mean, the performance change)
20:39	sapier	what happens this time proller?
20:39	proller	i commented to pull
20:39	proller	ops, issue
20:40	nore	so, sapier, do you think performance increase is really an issue?
20:40	sapier	performance is always an issue ;-)
20:40	ShadowNinja	s/increase/decrease/
20:40	nore	btw, for saplings, remember that they take 500s average to grow
20:40	ShadowNinja	That seems like good performance.
20:41	proller	duplicated tree code is issue
20:41	nore	to 20ms difference = 4*10^-5 % slower than before
20:43	nore	no, sorry, 4*10^-3 % slower
20:43	ShadowNinja	sapier: What does #962 fix?
20:43	nore	and for dirt, 0.1% slower
20:43	sapier	fixes broken textures on windows
20:44	sapier	in main menu
20:46	PilzAdam	nore, can you test the speed when using schematics?
20:47	PilzAdam	using schematics would also fix what proller mentioned: the duplicated tree code
20:47	ShadowNinja	sapier: Do you mean the tiling splash?
20:48	PilzAdam	we would just remove the tree flag and place the schematics with the decoration API
20:48	ShadowNinja	I gues #937 can be closed.
20:50	sapier	yes shadow
20:51	ShadowNinja	Alright, looks good, you can merge it.
20:53	sapier	proller can you try new version
20:56	hmmmm	noo
20:56	hmmmm	the decorations aren't good enough to replace trees
20:56	hmmmm	don't do that yet
20:57	hmmmm	see the thing is, trees created by decorations aren't of variable height so they're really not as good
20:57	hmmmm	i wasn't able to figure out a good way on how to do that, and I sorta forgot about it when I started working on other things
20:57	celeron55	it's easy to make them plenty variable enough for nobody to notice
20:57	PilzAdam	anything else that needs to be improved?
20:58	celeron55	the L-system trees of moretrees are like clones of each other too
20:58	hmmmm	yeah
20:58	celeron55	at least some of them
20:58	hmmmm	one of these days I'm sure someone will fix them......
20:58	hmmmm	one of these days..............................
20:58	celeron55	8D
20:58	hmmmm	lol
20:58	hmmmm	I never thought working full time would cramp my style this much
20:59	sapier	:-) I could've told you hmmmm if you'd have asked
20:59	hmmmm	alright, that's it
20:59	hmmmm	I am going to finally finish the lua menu thing
20:59	celeron55	i recommend a 4-day workweek
20:59	sapier	hmmmm could you have a look at that branch I just mentioned first
21:00	hmmmm	oh boy async jobs
21:01	sapier	forget about it the conflicting work was kahrls ;-)
21:01	sapier	but you could have a look at those things too if you have time
21:01	hmmmm	are you using spaces instead of tabs?
21:01	hmmmm	do you have a death wish or something??
21:02	sapier	I don't think so what file are you talking about?
21:02	celeron55	what kind of an environment od the threads have?
21:02	celeron55	do*
21:02	hmmmm	oh that's just github
21:02	sapier	each asynchronous thread has it's own lua stack
21:03	celeron55	what functions can they call?
21:03	sapier	any function not required to access the gui (by now it's main menu only)
21:03	sapier	e.g. settings favourites file download
21:03	sapier	things like that
21:04	sapier	but I made settings read only by now ... didn't have time to check if it's locked correctly
21:04	hmmmm	is there a queue or something to carry out thread unsafe lua functions?
21:04	sapier	there aren't unsafe functions as all of them are in their own environment
21:04	hmmmm	oh that's sorta lame
21:05	celeron55	what filters the unsafe functions to not be visible?
21:05	celeron55	or is there a completely new redundant API for those jobs?
21:05	sapier	no it isn't
21:05	sapier	there are initilizer functions for all major components
21:06	celeron55	maybe you could simply write a description about how this thing works so that nobody would need to poke more and more answers from you
21:06	sapier	only those initializers need to be modified to not add the "unsafe" functions
21:06	sapier	ok I'm gonna write it down give me an hour ;-)
21:07	celeron55	good
21:13	ShadowNinja	I have rebased #751 and it works nicely. Any comments on it?
21:14	ShadowNinja	And I need someone else to agree with #197 before I push (squashed).
21:14	ShadowNinja	minetest_game/#197 that is.
21:24		Miner_48er joined #minetest-dev
21:26		proller joined #minetest-dev
21:30	ShadowNinja	minetest/#583 too.
21:41		jin_xi joined #minetest-dev
21:45	sapier	https://github.com/sapier/minetest/wiki/Minetest-Asynchronous-Lua I guess this won't answer all questions so I'll add my answers to it
22:05		werwerwer joined #minetest-dev
23:12		Zeitgeist_ joined #minetest-dev

← Previous day | Channels | #minetest-dev index | Today | Next day → | Google Search | Plaintext