Time Nick Message 00:01 Taoki So hardware lighting is still to be done... but it should be easier now. Anyway, I soooo gotta look for bloom and water reflection / reflection shaders now *fangirl squeal* 00:01 Taoki xD 00:02 celeron55 (actually, i did add some fancier calculations to work in linear color space and use an exponential exposure function to not burn out stuff like snow but whatever) 00:03 Taoki nic 00:03 Taoki **nice 00:03 celeron55 it's odd how powerful the GPU is in even the suckiest cards these days 00:04 celeron55 you can do fairly complex calculations for every pixel on the screen in every frame 00:05 Taoki yeah. Was just going to say I see absolutely no performance decrease with shaders. Which makes sense, since I have a very modern video card. But for things as simple as this, even old cards should have no problem 01:52 VanessaE celeron55: nice work on the shaders code. light-emitting objects are too dark, but otherwise it seems to work well. No apparent performance loss here. 02:17 VanessaE celeron55: recommendations: light the north, south, east, and west sides of the cubes slightly differently (not much, maybe mas 5% between darkest and brightest). Reduce the top-side lighting by 25%. Automatically move signlike and raillike objects that are placed on water (e.g. flowers:waterlilies) down to meet the newly-positioned water surface. 02:18 VanessaE s/mas/max/ 02:19 VanessaE s/waterlilies/waterlily/ 02:31 SitDog i've been catching up with the lighting code discussion. my 2 cents is that we shouldn't go overboard with them. i love high end stuff, but the simple graphics of minetest is pleasing to me. 02:31 SitDog it's a good break :D 02:52 hmmmm what on earth.... 02:52 VanessaE ? 02:54 hmmmm i don't understand... 02:54 hmmmm you know how i was having issues with lighting 02:54 hmmmm all of a sudden i'm not 02:54 hmmmm i didn't change anything at all 02:55 hmmmm i just added in a printf statement in between propogatesunlight and unspreadlight 02:55 hmmmm which would mean "aha, race condition" - except when i removed it, it still worked 02:55 VanessaE yes of course. Odd thing to cause it to "just work"? 02:55 hmmmm something really weird is up 02:55 VanessaE maybe you made some other change you just forgot about? 02:55 hmmmm nope 02:56 hmmmm maybe it was a combination of changes 02:56 VanessaE possible 02:58 hmmmm okay i figured out what the problem is 02:58 hmmmm but i'm not sure how to fix it 02:59 VanessaE celeron55: recommendation: page flipping in creative turns the screen black while the display is being regenerated, if the item preload thing is turned off. Suggest defaulting it to on to restore the old behavior until a suitable workaround is found. I suggest, in that regard, that the display either be double-buffered (with a "Rendering next/previous page..." message at the bottom), or that the page be re-painted an ite 02:59 VanessaE m at a time without the black screen (draw an empty inventory, then fill it) 03:05 hmmmm i am looking at the size of the light_sources and unlight_from... generating from full_node_min.Y to full_node_max.Y causes odd variance between sizes of light_sources, namely 512000, 716800, and 385280. also, i must force overtop_has_sunlight = true for this to occur 03:05 hmmmm without overtop_has_sunlight, there is nothing at all added to light_sources 03:06 hmmmm using the combination of generating from node_min.Y to node_max.Y and overtop_has_sunlight = true being forced, however, produces darkness 03:07 hmmmm i've already thought about CONTENT_IGNORE blocking the sunlight, so i set that to sunlight propogates = true in nodedef for a second, and that didn't do what i expected it to 03:09 hmmmm i'm getting somewhere at least 03:29 SitDog i am not having problems with lighting if that was directed at me. i was just saying that the simplicity of the project is what attracted me to it :) 03:56 RealBadAngel SitDog, its an option and some just love it 03:56 RealBadAngel you can still stick to simplest settings 04:02 SitDog i guess that is true. i just need a setting called "drunk & pixelated" 04:02 VanessaE haha 04:04 SitDog I have finished up a tempfile class. it automatically handles writing data to a temporary file. that is you write to the temporary file until you close the object. it then attempts to rename the temp file to the target file -- and creates a backup of an existing target file. 04:06 SitDog so, very little changes as far as the code flow, but if you run out of disk space or the file system becomes corrupt, you will not trash a valid file. 04:16 SitDog celeron55 mentioned an issue with the far rendering distance. i don't recall exactly what that was, but i would like to take a look at it. 04:20 VanessaE the automatic FPS control I think he meant. 04:20 VanessaE the rendering distance is supposed to vary closer or farther to keep the client FPS at whatever's configured in minetest.conf I guess 04:21 VanessaE instead it is..well "badly broken" I think he said, but I don't recall why. 05:00 SitDog i will take a look at it. 05:01 SitDog I didn't commit the temp file thing yet. 05:23 hmmmm alright back to working on minetest 05:24 hmmmm so i got everything fully working - turns out the actual problem was that nodes underneath the overtop were CONTENT_IGNORE 05:24 hmmmm because i was writing to nodes that were in the buffer zone 05:25 hmmmm i didn't realize how intricate the VoxelArea thingy was 05:52 hmmmm hrmm 05:54 hmmmm the segfaults that happen when "too much is going on" seems to be due to updateLighting() adding too many nodes to light_sources 05:55 hmmmm i'll leave it unfixed (because an actual fix would require a lot of work) since that lighting code may be ripped out soon anyway or not used much anymore 05:58 RealBadAngel hey, dont stop 05:59 RealBadAngel make it work, please 05:59 hmmmm it seems like this is an irrlicht bug 06:00 hmmmm well 06:00 hmmmm actually a libsupc++.so.1 bug 06:00 RealBadAngel have you checked 1.8 code for it? 06:00 hmmmm what....? 06:00 hmmmm that has nothing to do with what i'm talking about 06:01 RealBadAngel you told irrlicht bug 06:01 hmmmm yeah it is an irrlicht bug, if it's smashing the heap like i think it is 06:01 RealBadAngel we are using outdated version of it 06:01 hmmmm that's okay. 06:01 RealBadAngel like 1.7.1 06:02 hmmmm we aren't, maybe you are. i'm personally using 1.7.3 06:02 RealBadAngel why not current stable? 06:03 hmmmm because it's too new? 06:03 hmmmm that won't fix the problem anyway 06:03 RealBadAngel since they have stable released each 6 months 06:03 RealBadAngel stable means well tested 06:03 RealBadAngel read their forums 06:04 RealBadAngel before release the stable they have a month of freeze 06:05 hmmmm yes, but that doesn't change that most operating systems don't have 1.8 available 06:05 RealBadAngel thats the other thing 06:06 RealBadAngel but if it appears that the problem is in just upgrading to new libs... 06:07 hmmmm it's not a problem at all, because we're not upgrading to new versions of libraries for no reason 06:07 RealBadAngel im dont know, just a thoight 06:08 RealBadAngel Jeija was doing something and said movin to it is necesary 06:08 RealBadAngel but i dont know the reason now 06:08 hmmmm probably because of the texture mipmapping bug 06:09 hmmmm it's fine if you use 1.8 in windows builds or anything where the library is redistributed, but you can NOT rely on 1.8's behavior 06:10 Calinou since they have stable released each 6 months 06:10 Calinou lol nope 06:10 hmmmm if you really must do so, you need to include a preprocessor check of which irrlicht version it's being built with 06:10 Calinou every two years for debian rusty... 06:10 Calinou 8 months before the release: debian stale is frozen 06:10 hmmmm and use the appropriate non-breaking behavior for older ones 06:10 RealBadAngel Calinou, i was talkin bout Irrlicht engine 06:10 Calinou oh 06:12 * Calinou should ctrl+f more often 06:13 RealBadAngel Calinou, have you tested UI? 06:14 Calinou my inventory flashes when i click on something in creative inv 06:14 RealBadAngel delete default creative mod 06:15 hmmmm okay this just turned into a real problem 06:15 RealBadAngel need to find a way to disable it without deleting tho 06:15 Calinou the creative inv you made is pretty good 06:15 Calinou search is case-sensitive, though 06:15 Calinou shouldn't ;) 06:15 RealBadAngel thats why im askin folks to test 06:16 RealBadAngel to gather info what should be improved 06:16 RealBadAngel btw enter case is solved already, but needs another patch to be merged 06:17 Calinou bag textures look 2D in hand 06:17 Calinou you probably didn't threshold alpha ;) 06:17 RealBadAngel the code is cornernote's 06:17 RealBadAngel i left it almost intact 06:18 RealBadAngel they propably could be better 06:19 RealBadAngel with the latest patch search is even more comfortable 06:19 RealBadAngel search button is not needed at all 06:20 RealBadAngel im workin now on updating the search results realtime 06:20 RealBadAngel when user is typing the query 06:21 Calinou that'll make the server lag since the search is server-side.. 06:22 RealBadAngel will have to be tested then 06:22 RealBadAngel if it will really be producing lags i will drop the idea 06:24 RealBadAngel offtopic: http://faildesk.net/2012/02/21/finally-a-true-purpose-for-internet-explorer/ 06:27 Calinou lol 06:32 RealBadAngel http://faildesk.net/2012/11/30/ie-10-sucks-less-and-microsoft-wants-you-admit-it/ 06:32 RealBadAngel thats good :) 08:42 celeron55 hmmmm: the voxel lighting will not be ripped out anytime soon, if ever 08:42 celeron55 if it's up to me, not in years 08:43 hmmmm mm yes, nevermind that. 08:43 hmmmm it was my fault, i was corrupting the the heap a little 08:43 hmmmm it really looked like irrlicht's problem though. 08:43 celeron55 it does a good job of making the game look different to everything generic opengl stuff out there 08:44 celeron55 (and also it has the gameplay utility of making it easy for stuff to look up where it is dark and where not) 08:45 hmmmm then perhaps it could use an overhaul 08:45 celeron55 why? 08:45 Calinou change its default colors to look more like shaders, and make the top face of blocks brighter? 08:45 Calinou minecraft uses voxel lights, sides of blocks are darker, so this is possible 08:45 celeron55 Calinou: ...the shaders branch uses voxel lighting 08:46 hmmmm at the very minimum, the part where it allocates about a half million nodes can be changed 08:46 hmmmm 3/4ths of a million actually 08:46 celeron55 if you want to optimize it, go ahead 08:46 celeron55 you'll have hard time though 08:46 hmmmm i'm confident i can do something about it 08:46 hmmmm but, that'll come later 08:46 celeron55 do you mean the VoxelManipulator part? 08:47 celeron55 it's pretty important, because it makes a linear array of the whole problem 08:47 celeron55 the regular minetest map abstraction is too slow for huge bulk operations, and it cannot be used in threads 08:47 hmmmm mm no, unlight_from and light_sources 08:47 hmmmm those are the offenders 08:47 celeron55 ok 08:48 hmmmm and it takes up the vast majority of the mapgen time 08:48 celeron55 i assume you can figure out what they do 08:48 hmmmm no i know exactly what they do 08:48 hmmmm i spent way too much time on the lighting problem 08:48 hmmmm i'm pretty familiar with all that code 08:50 RealBadAngel hi celeron55 08:51 RealBadAngel if you want to try whats possible thanks to the changes to formspec: http://realbadangel.pl/UI.zip 08:52 RealBadAngel i made complete overhaul of inventory 08:53 celeron55 RealBadAngel: how about a github repo and screenshots; much handier 08:53 RealBadAngel its WIP 08:54 hmmmm having to unzip someone's file is so cumbersome though 08:54 RealBadAngel screenshots wont tell you how it works on the other hand 08:54 RealBadAngel finding an item is now damm fast and easy 08:54 RealBadAngel even within thousands of aviable ones 08:56 hmmmm oh 08:57 RealBadAngel i used to have 60+ pages in creative, now i do have less then 20 08:57 hmmmm completely different subject, but what do you say about eventually removing farmesh entirely? 08:57 celeron55 it will be rewritten in a different way... some day 08:57 hmmmm with a super fast mapgen like this and much faster rendering, i am not so sure if it's necessary 08:57 celeron55 could be removed currently though, it's useless as-is 08:58 PilzAdam (working) farmesh would be a nice feature 08:58 hmmmm would you like to make it happen!? :) 08:58 RealBadAngel hehe 08:58 RealBadAngel PilzAdam you have been choseen as a volunteer ;) 08:58 hmmmm no seriously don't 08:59 hmmmm we need to come up with a new idea for farmesh first (if we even actually need it by the time we get around to it) 08:59 celeron55 i have plans to make it work using "mapblock impostors" 09:02 celeron55 there are much details of implementation to find out and figure out, but i am quite positive it will work, somehow 09:03 PilzAdam I guess the new_style_water can be removed since we got shaders 09:03 hmmmm wouldn't that still kinda take up a lot of memory 09:04 celeron55 PilzAdam: no need to hurry for that 09:04 celeron55 hmmmm: dunno; it's all in the details 09:05 celeron55 the close quality of them can be screwed up by a huge lot without making them useless 09:06 celeron55 on modern computers, we're talking about things further away than like 160 nodes 09:06 celeron55 on slower ones, people will be happy to see even something 09:07 hmmmm this is just me, but i'm using an 8400gs and i don't typically have a problem with full range view enabled 09:07 hmmmm especially after that texture switch reordering 09:08 hmmmm oh yes this is good 09:08 hmmmm with my new biome setup, you can assign an average color to each 09:08 hmmmm for example, the average color of a desert biome would be orange 09:09 celeron55 lol... why would you do that 09:09 celeron55 that's not even close to how real biomes are formed 09:09 hmmmm huh? 09:10 RealBadAngel seen red deserts in aegypt 09:10 hmmmm you mean the biome code i committed to my own repository? 09:10 hmmmm it probably isn't, i don't know how real biomes are formed, that's just what came to me 09:11 celeron55 how does it handle two biomes that have the same average color but completely different "functionality" 09:11 hmmmm oh no no you misunderstood 09:11 celeron55 you're quite unclear in what you say 09:11 hmmmm with the fake mapblocks you're going to have reduced detail 09:11 RealBadAngel celeron55, https://github.com/RealBadAngel/unifiedinventory 09:12 hmmmm and instead of using the color of each node or whatever 09:12 celeron55 oh yes 09:12 hmmmm if that fake mapblock is in a certain biome you can just make a sub-node group of say 4x4 with that average color 09:13 celeron55 if your mapgen turns out good and has the preview generation abilities you've been talking about, it will be useful for creating the data for far rendering 09:13 hmmmm yea 09:13 SitDog offtopic: some feedback on code quality & style. http://ideone.com/6gh1J7 09:13 celeron55 but the colors have to be found out on the client side from the textures it is using 09:13 celeron55 the server can't know how what it is sending is supposed to look 09:13 hmmmm oh gosh that's true 09:14 hmmmm i need to get to sleep, it's 4:15 am 09:14 hmmmm that's a pretty big problem... i'll think of something 09:14 celeron55 you'll have to use node ids instead of colors 09:14 celeron55 no other option 8) 09:17 dannydark I was thinking about adding noclip a priv and splitting it from fly, so if someone had the noclip priv they could still fly through the ground etc but without it they would still be able to fly (with the fly priv of course) just not through the ground as my brother wants to give some people on his server fly privileges but doesn't want them noclipping through the ground to cheat at getting ores etc. Would you be happy to accept this change 09:17 celeron55 maybe change the equivalent of what you currently transfer as a color to two node ids (to be able to average from them; better than one but not much worse than 3), and the client will figure out which side of them it will use based on context 09:18 celeron55 dannydark: got cut at "this change u" 09:18 dannydark accept this change upstream? as I feel it could be useful for other servers to well...maybe ^_^* 09:18 dannydark sorry it said it had sent fine on mine :S 09:19 celeron55 irc cuts too long messages; you should get a client that will split them 8) 09:19 dannydark haha yeah, I'm on my windows box so using trillian 09:20 celeron55 anyway; that would indeed be useful, maybe someone could look into it 09:22 dannydark Ok, I will take a look at the code in a sec and get working on it 09:23 celeron55 it's probably quite simple 09:24 celeron55 just add a new description of a privilege and modify localplayer.cpp 09:24 dannydark yeah I did it in 0.3 ages ago for his server but as hes running latest master now he wants it to be added again haha 09:24 RealBadAngel dannydark: +1 for the idea 09:24 RealBadAngel that could make the jetpack idea usable 09:25 dannydark ^_^ oh yeah think I saw that jetpack mod, so it needed fly priv to work? 09:25 celeron55 not really; a jetpack requires taking away direct control from the player but keeping the collision detection going 09:26 celeron55 and be independent of privileges 09:26 dannydark yeah was going to say if it did use fly priv that it wouldn't be the ideal way to do it 09:26 RealBadAngel separation of fly and noclip is needed in the first place 09:27 RealBadAngel i thought of simple tool when weared and charged could let player fly 09:28 celeron55 well, yeah, a hacky implementation could indeed be done with a fly privilege without noclip 09:28 RealBadAngel like with fly priv, but noclip makes it nonsense 09:28 PilzAdam https://gist.github.com/4187946 09:29 dannydark yeah, ideally though It would be nice to have it work without fly priv, especially on survival only servers where admins don't want people to be able to fly about (at least without the jetpack...which would also be limited flying with some sort of fuel consumption maybe) 09:30 RealBadAngel i got the code for it almost done 09:31 RealBadAngel just need correct privs 09:31 PilzAdam should the client be able to swtich noclip if he has the noclip priv? 09:31 RealBadAngel its a power tool 09:31 RealBadAngel you have to craft it, charge and then you can use it 09:31 dannydark PilzAdam: nice, wouldn't that stop people from using fly though (if noclip priv is not allowed)? 09:32 dannydark PilzAdam: I don't think so, I think if someone has noclip and fly then let them noclip while flying (otherwise it means adding more controls etc) 09:32 PilzAdam in this case the patch is ready 09:33 RealBadAngel that problem wont rise on most of the servers 09:33 dannydark RealBadAngel: ah nice, will be nice to see it in action when its ready 09:33 RealBadAngel need just a way to switch on/off nocliped fly 09:33 Calinou i'd say: make fly priv allow noclip by default. "disallow_fly_noclip = true" in minetest.conf disables noclip 09:34 Calinou people with "server" privilege can always noclip, though 09:34 celeron55 PilzAdam's patch is not ready 09:34 dannydark yeah won't that stop people using standard fly without noclip? 09:35 RealBadAngel when using the tool fly priv will have nothin to do with it 09:35 celeron55 it won't let a player fly if the player doesn't have fly AND noclip, and there is no fly mode with collisions 09:35 dannydark yeah 09:35 RealBadAngel if your jetpack runs out of energy you gonna fall to the ground and dye 09:35 RealBadAngel die 09:36 RealBadAngel thats the idea 09:36 dannydark another thing that might need changing is when using fly with collision, if you fly into blocks it will kill you....where it shouldn't unless your falling... 09:36 dannydark at least that problem happened in 0.3 09:36 PilzAdam celeron55, with my patch you can fly without noclip 09:37 RealBadAngel so its ready or not? :) 09:37 PilzAdam you die if you fly down at the ground 09:37 PilzAdam with superspeed 09:38 dannydark yeah I don't think it should do that 09:38 dannydark unless your falling 09:38 RealBadAngel i see it other way 09:38 RealBadAngel if it is a priv 09:38 dannydark because you will also die if you fly into walls 09:38 RealBadAngel jetpack runnin out of energy will just disable it 09:39 dannydark RealBadAngel: yeah I think thats right, because then you would be falling...so if your high your going to take damage ^_^ haha 09:39 RealBadAngel and player will find himself in walkin mode over the clouds 09:39 celeron55 "oh dog" 09:40 RealBadAngel hi the lake, hi grass 09:40 RealBadAngel *splash* 09:40 * dannydark misses lake.... *splat* 09:41 RealBadAngel hehe 09:42 PilzAdam were is the falling damage handled? 09:42 RealBadAngel on the ground? ;) 09:43 dannydark localplayer.cpp 09:44 dannydark line 296 (Report collisions) 09:44 RealBadAngel so ground level is line #296 ;) 09:44 dannydark should just need an && !free_move in the if statement 09:44 dannydark or something like that 09:45 dannydark PilzAdam: well line 296 with your patch 09:45 celeron55 uhm 09:45 celeron55 what are you doing 09:45 PilzAdam it works 09:45 PilzAdam Im messing arround with noclip 09:46 PilzAdam ok, it doesnt work 09:46 celeron55 hmm, disabled collision damage for flying without noclip? well i guess that is reasonable 09:47 dannydark PilzAdam: you still getting damaged? 09:48 PilzAdam its a bit strange; the damage is delayed 09:48 dannydark I'll run a compile now and have a look 09:49 PilzAdam at the first collison the hearts dissapear and at the second one i die 09:52 dannydark yeah same here will have a look at the code, I'm sure that in 0.3 I only had to amend that collision if statement although I guess much has changed since then ^_^ hehe 09:53 RealBadAngel celeron55, http://minetest.net/forum/viewtopic.php?pid=54671#p54671 together with some screenshots 09:54 dannydark oh wait 09:54 dannydark its because free_move is false 09:54 dannydark as we don't have the noclip priv 09:55 PilzAdam so g_settings->getBool("free_move") 09:55 dannydark yeah that should work 09:56 PilzAdam yep, it works 09:56 dannydark hehe >_< 09:56 dannydark yeah so I guess thats the noclip/fly split patch done 09:57 RealBadAngel is it able to set on/off from lua? 09:57 RealBadAngel hi Taoki 09:57 PilzAdam with minetest.setting_set() 09:58 PilzAdam updated patch: https://gist.github.com/4187946 09:59 celeron55 minetest.setting_set() changes server settings 10:00 PilzAdam RealBadAngel, why do you need that? 10:01 RealBadAngel for jetpack 10:01 PilzAdam why dont you use entities for that? 10:01 RealBadAngel need to be able to turn it on when user wears jetpack 10:01 RealBadAngel and off when puts it away or energy runs out 10:02 PilzAdam attach the player to a Lua entity 10:02 celeron55 hmm, indeed, attach the player to an entity 10:02 RealBadAngel no way 10:02 dannydark RealBadAngel: maybe this is useful: http://minetest.net/forum/viewtopic.php?id=3915 10:02 RealBadAngel theyre lagged as hell 10:02 celeron55 it's the right way 10:02 dannydark I've not seen the code so not sure how he's done his jetpack but yeah might be useful 10:02 RealBadAngel thanks to the entities nothin based on them is not useful 10:03 RealBadAngel lasers shoots 20s after fireing 10:03 PilzAdam depends on the server connection 10:03 RealBadAngel local 10:03 celeron55 if you could set client settings from the server, it would just annoy the player if he for example wants to have free move set for all the time and then switches through a jetpack in the inventory or so 10:04 RealBadAngel jetpack is meant for surrival servers 10:04 RealBadAngel its not a kind of priviledge 10:04 RealBadAngel like high end tool aviable after long days of playing and gathering resources 10:04 dannydark will it have some sort of force pushing the player? erm...mmm not to sure how to explain it 10:05 RealBadAngel hmmm with entities used i think it could be possible 10:05 celeron55 free_move should be made to be sent from the server and saved to there though and not be a client setting 10:05 RealBadAngel celeron55, of course 10:05 celeron55 but i don't think there is any proper framework for such things 10:05 dannydark basically so it feels more realistic with some sort of gravity rules applied to it rather than just feeling like free_move? because that would be awesome 10:06 RealBadAngel celeron55, im using prived stuff to turn on/off already 10:06 celeron55 it's wrong and wouldn't work in that case anyway 10:06 RealBadAngel see set daytime buttons in Unified Inventory 10:06 RealBadAngel its working 10:07 celeron55 ... 10:07 celeron55 i am sorry but i have to request you to not be an idiot once again 10:07 RealBadAngel im setting it on the server side 10:07 RealBadAngel not the client 10:08 RealBadAngel so please stop 10:08 RealBadAngel im using the commands like chat does 10:18 PilzAdam celeron55, bug in water shaders: http://www.zimg.eu/i/1782686636 -> http://www.zimg.eu/i/4257623532 10:21 celeron55 known 10:29 RealBadAngel hmmm, stupid question, can entity emit light? 10:29 PilzAdam currently not 10:29 RealBadAngel :( 12:06 Taoki PilzAdam: Can you commit a small change for me please? I tested anaglyph stereo again, and 0.1 is too strong of a default. Can you please change anaglyph_strength = 0.1 to anaglyph_strength = 0.05 in minetest.conf.example? 12:06 Taoki I also think there might be a default in the code that needs to be changed too 12:06 PilzAdam yea, in defaultsettings.cpp 12:06 Taoki right 12:06 Taoki 0.05 is good 12:07 PilzAdam k, will do 12:07 Taoki tx 12:08 thexyz i guess that's too personal setting, other people were claiming 0.05 is too low 12:08 Taoki PilzAdam: Also, can you merge this one commit from my branch? https://github.com/MirceaKitsune/minetest/commit/14409e115eced2f76670e7cc8f069a25d67c347f NOT the whole branch, I spoke with c55 and the server-side already prevents sending the same position, but the client doesn't 12:09 Taoki thexyz: I tried it again and it's too eye-straining to work 12:10 PilzAdam people have to set this for themselves 12:10 PilzAdam so maybe its not a good idea to change it every day in upstream 12:10 thexyz http://minetest.net/forum/viewtopic.php?pid=11162#p11162 — your post 12:11 PilzAdam just leave it as it is 12:12 Taoki The current default won't work for many, it's too eye straining. But ok 12:12 thexyz yep 12:12 thexyz Taoki: we'd better ask people instead of making those conclusions 12:12 Taoki makes sense 12:13 PilzAdam Taoki, dunno about the commit you mentioned 12:13 Taoki Also: Can someone confirm that for both players and lua entities, position is only updated from server to clients only when it changed? So if the object is still it doesn't waste andwidth still sending updates? 12:13 Taoki PilzAdam: I tested it and it does no harm, or bugs I can think of. Just keeps the local player from wasting bandwidth for no reason 12:13 Taoki Always a helpful thing 12:14 Taoki Well, feel free to test it, but that commit is certainly fine. Would be surprised if any harm could come from that 12:16 PilzAdam thexyz, what do you think? 12:17 Taoki That branch also contained the same thing for client_sao, but c55 said that the server already knows to avoid sending the same position for still players. I'd like someone else to confirm though, since I can't find the code that does that except for ItemSAO 12:17 Taoki Anyway, local player doesn't, and what's what the commit I linked is for 12:17 Taoki **that's 12:17 Taoki BRB, but feel free to answer as I'm only afk 12:20 thexyz PilzAdam: have you tested it? 12:20 thexyz seems fine as for me 12:24 PilzAdam 13:24:02: ERROR[main]: ClientEnvironment::addActiveObject(): id=1 type=e: SerializationError in initialize(): unsupported ObjectProperties version: init_data="\u0001[...] 12:24 PilzAdam when i connect to minetest.ru:12345 12:25 PilzAdam a randomly choosen 0.4.3 seems to work 12:25 thexyz https://github.com/celeron55/minetest/pull/326 12:25 thexyz PilzAdam: it's not up-to-date 12:26 PilzAdam do you know any up-to -date 0. 12:26 PilzAdam 4.4-dev servers? 12:26 thexyz there were some commits that fixed version incompatibility, i'll rebuild right now 12:26 PilzAdam without Taoki's patch? 12:27 thexyz why should i include it? it only affects client 12:27 thexyz done 12:28 PilzAdam I just want to test if it breaks something 12:28 thexyz try to connect now 12:28 PilzAdam already done 12:28 PilzAdam no errors 12:28 celeron55 Taoki: about your commit: you should store the last values in LocalPlayer, not Player 12:29 celeron55 Taoki: and you should initialize them in the constructor 12:30 thexyz whoa! bug 12:31 celeron55 i wonder if i should try cramming in the meta_set_nodedef branch too... 12:31 thexyz http://i.imgur.com/DRnxl.png → http://i.imgur.com/r2o9W.png 12:32 celeron55 it's not a bug, it's a shortcut! 12:32 PilzAdam celeron55, what would you do if you dont make it? 12:33 PilzAdam bbl 12:39 Taoki back. Will try to make those changes and re-create the branch 12:48 celeron55 oh dog 12:48 celeron55 i just started looking through the meta_set_nodedef branch 12:48 celeron55 this is really funky stuff 12:48 celeron55 didn't even remember what this contained 12:49 celeron55 this for example creates a fake full game environment for... ehm... something 12:54 celeron55 this pushes a fake node definition manager in a wrapped game definition to the mesh generator and creates fake node ids for the dynamic nodes 12:57 thexyz celeron55: https://github.com/celeron55/minetest/pull/326 13:08 celeron55 database.h should be designed so that it doesn't depend on stuff like sqlite3.h 13:08 celeron55 (like headers generally) 13:09 celeron55 (yes, old code in minetest is baaaaaad) 13:10 celeron55 thexyz: see the sound stuff for an example of the current recommended way of making abstractions 13:11 celeron55 sound.{h,cpp}, sound_openal.{h,cpp} 13:11 thexyz ok 13:24 Taoki celeron55: Did your changes. https://github.com/MirceaKitsune/minetest/tree/localplayer_check_position 13:25 Taoki https://github.com/MirceaKitsune/minetest/commit/ef3b53bac6722e0a2a132189d3e68e829fb9c049 Compared to how I did it the first time 13:25 Taoki That's a new branch so no rebase needed 13:26 Taoki Someone should test this on a server with >= 2 clients connected, just to be sure each sees position being updated, though it should be perfectly good 13:28 PilzAdam Taoki, building; I will join minetest.ru:12345 13:28 Taoki will try to connect there too 13:29 thexyz celeron55: but pythonmodulo depends on sqlite3_int64 13:30 thexyz hm, seems there's no difference between sqlite3_int64 and long long 13:31 celeron55 ehm 13:31 celeron55 what is pythonmodulo 13:31 thexyz // modulo of a negative number does not work consistently in C 13:31 thexyz static long long pythonmodulo(long long i, long long mod) 13:31 thexyz it's used by Database::getIntegerAsBlock 13:32 celeron55 hmm 13:33 celeron55 well, it should be just a signed 64 bit number 13:34 celeron55 does irrlicht always define that 13:35 celeron55 apparently no 13:39 Taoki PilzAdam: Forget about what I just said. It also sends pressed keys so it needs to be send for attachments too. So yes, feel free to merge that branch as is now 13:39 Taoki https://github.com/MirceaKitsune/minetest/tree/localplayer_check_position 13:40 thexyz well, they use this https://gist.github.com/4188724 in newer irrlicht version 13:41 celeron55 improvised from sqlite and existing stuff in minetest: https://gist.github.com/4188733 13:42 celeron55 should support MSVC and GCC, which suffices for us well 13:42 Taoki celeron55, PilzAdam: So can that change go upstream? Just to be sure I don't need to do any other improvements 13:42 celeron55 and is a smallish change 13:43 PilzAdam Taoki, as it seems to work fine its ok to me 13:44 Taoki ok, switching back to master then :P 13:46 celeron55 there is a small change i want to make but i'll make it when i merge those 13:47 Taoki I'm curious: Can shader sun allow fully smooth lighting transitions now? So when it's sunrise / sunset the light doesn't update itself into various steps. It would sound like this should make that possible 13:48 Taoki Since shaders should allow each frame to check sun position and calculate brightness at that moment 13:50 celeron55 it's actually somewhat possible even without shaders, but once again it's like it is for sparing the crappiest hardware... i guess i'll do that too 13:52 Taoki ok. In the code it would probably be expensive. But for shaders I'd assume you can do something like light = sun_position / 1000 each frame (assuming sun position ranges between 0 and 1000 and maximum light between 0 and 1) 13:52 Taoki Since it doesn't need to update the voxel lightmaps I believe 13:53 celeron55 yeah it's basically equal to doing nothing 13:53 celeron55 the classic code updates the vertex colors based on a cached store of day/night brightness pairs for each vertex 13:53 celeron55 it's quite lightweight too, actually 13:54 Taoki right, that's why it would be expensive. But shader light can do a tiny formula each frame that even the oldest GPU wouldn't be influenced by 14:11 thexyz celeron55: i've updated pull request https://github.com/celeron55/minetest/pull/326 14:20 Taoki Regarding what I last asked: I just tested if m_position_not_sent is set properly based on what the server gets from the client, and with my change m_position_not_sent doesn't become true if a player stands perfectly still and presses no keys. So this should have double the effect on bandsitch saving :) 14:20 Taoki **bandwidth 14:21 Taoki Well perfectly still players will use no bandwidth at all now, so yeah 14:21 Taoki Hard to test for lua entities. For fropped items I already noticed a code which checks if position is greater than a given amount before sending... so I assume this is all that's needed 14:22 celeron55 thexyz: indentation seems to have problems; i'll look at it later 14:22 celeron55 too much to do already 14:23 Taoki Although, I do have one curiosity: Does content_sao send the position of a player also to the player which sent the position to it? Or only to all clients which are not our specific player? 14:23 Taoki Cuz then we still have a useless ping-pong there 14:24 celeron55 it sends but that doesn't matter; it's useful for debuggin too 14:24 celeron55 +g 14:24 Taoki ok. If it does have an use then that's good. 14:24 Taoki https://github.com/celeron55/minetest/pull/327 14:25 * PilzAdam points to his player controls branch: https://github.com/PilzAdam/minetest/commits/controls 14:25 PilzAdam just if someone is interested 14:26 Taoki If we would optimize bandwidth to the max, I would be tempted to try fixing this as well, and forget about such debugging. But eh, perhaps another time if anything 14:33 Taoki Actually, I wanna see if I can fix this... need to understand how ActiveObjectMessage works though, so I can add an ID filter or something 14:39 Taoki Eh, some other time... the ActiveObjectMessage system makes this too hard 14:39 Taoki Since you add a message to that it's hard to filter by ID who it's sent to and persist the initial player to know to avoid him 14:40 Taoki So my last change on that branch is good to go as is :P 15:11 Taoki Hmmmmm... an idea for the more distant future (perhaps) just popped to mind. What if we could make a server setting that could allow us to use custom block grids? When set to 1, you place blocks like you do currently, all at distance of one block. When 0.5 however, you can place blocks in between other blocks 15:12 Taoki But this might require a major re-write of the engine. No idea there 15:12 Taoki 0.25 would allow you to place blocks 1/4 in any direction on other blocks, so on 15:12 Taoki With values like 0.5 however, some servers could make this interesting 15:13 thexyz celeron55: sorry for being totally unrelated, but http://minetest.net/forum/viewtopic.php?pid=54697#p54697 15:37 Taoki Can https://github.com/celeron55/minetest/pull/327 be taken care of if everything is ok with it? 15:41 Calinou looks good 15:41 Calinou did you test if new player position is sent if player is teleported with a command? 15:41 Calinou just to be sure :P 15:41 Calinou should be, but we never know 15:41 Calinou (use /teleport player1 player2) 15:46 Taoki Calinou: It certainly is, but we can test that more if PilzAdam still has the server up and can login now 15:46 Taoki I fall in a cave on his server either way, can't get out :P 15:47 PilzAdam its not my server 15:47 Taoki ok. Well if it has the latest code 15:47 Taoki I'm in now 15:48 Calinou now code player yaw smoothing, i tried doing it this morning and I failed again :D 16:03 celeron55 i herd u liek commits: https://github.com/celeron55/minetest/commit/804a7875a2018f3d8cb3fbe917c11ecc1666d25a 16:04 PilzAdam :D 16:04 Taoki :D 16:04 Taoki will switch over and test that soon 16:09 VanessaE testing that now. 16:09 PilzAdam epic. 16:09 Taoki updated and cimpiled, testing now 16:09 Taoki indeed 16:10 PilzAdam time_speed = 96000 -> disco 16:10 VanessaE oh that's niiiice 16:10 VanessaE night->day is nice and smooth now 16:11 VanessaE mostly :-) 16:11 Taoki yes, very nice! 16:11 celeron55 http://ompldr.org/vZ2phOQ 16:11 celeron55 my intense SHADER DEVELOPMENT THING. 16:11 VanessaE light-emitting nodes look right again also 16:12 VanessaE whoa, seriously intense :D 16:12 celeron55 it has multiple inches of screen space 16:12 celeron55 (less than my shader-incapable laptop at the right) 16:12 PilzAdam time_speed = 9600 -> ~4 sec. day/night 16:13 Taoki if it works, that's the important part 16:14 celeron55 well, it usually works 16:14 darkrose now it just needs shadows that follow the sun 16:14 celeron55 the computer itself is a year old i3 but the display is a tad older 16:14 PilzAdam dannydark, +water relections +wind in grass = MC mod 16:14 Taoki torches look nice too :) 16:15 Taoki If we keep going with commits like this, MineTest will soon be the best voxel-based space out there 16:15 VanessaE PilzAdam: that begs the question, how *would* you make plantlike nodes wave in a breeze anyway? 16:15 darkrose _animated 16:16 VanessaE darkrose: no, I mean how would you make the model itself warp around like they do in MC 16:16 celeron55 that is probably laughably simple to do with a shader 16:16 Taoki Hmmmmmm. Now I wonder if shaders can be used to do marching cubes as well :] 16:16 VanessaE celeron55: I figured it would be. maybe a group setting in a node that (assuming the node is plantlike) allows that shader to distort the node 16:16 VanessaE groups = {bendy=1} or something. 16:17 PilzAdam does this water 0.2 nodes lower shader apply to all liquids? or just to water and lava? 16:17 Calinou all liquids, normally 16:18 celeron55 now it applies to all liquids 16:18 Calinou night->day is nice and smooth now 16:18 Calinou it was always smooth :P 16:18 Calinou it's less smooth without shaders but it can be made more smooth with source code change 16:18 VanessaE Calinou: if you count 14 steps of daylight as smooth :P 16:18 VanessaE or however many there are 16:18 Calinou just change the smoothing time to 10 seconds (or enough to make the transition look 100% smooth) > done 16:18 VanessaE nevertheless, this is a nice improvement. 16:19 Calinou shaders are very slow, I get below 60FPS if I start two instances of the game 16:19 Calinou without I can start like 4-5 and have 60FPS 16:19 celeron55 buy more GPUs 16:19 Calinou the light/water improvements could be backported to shader-less mode, no? 16:19 Calinou (new_style_water worked before 0.4 ;)) 16:19 celeron55 too difficult, and you can't brighten things further with vertex colors 16:20 Taoki Calinou: I never seen a performance decrease with shaders 16:20 thexyz https://github.com/celeron55/minetest/pull/330 16:20 * PilzAdam neither 16:21 VanessaE here's an interesting one, I don't know how old this is because I never noticed it before: lava disappears behind water now, but objects behind that lava are still visible. 16:21 RealBadAngel thexyz: what if it drops more items of the kind? 16:21 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/screenshot_1550605553.png 16:22 thexyz RealBadAngel: umm? 16:22 VanessaE behind the cursor is three or four lava columns and those glass blocks. 16:22 thexyz it was returning list of strings in one case and list of ItemStack's in other 16:22 thexyz and api.txt says it returns list of strings 16:22 RealBadAngel ah ok 16:53 thexyz PilzAdam: https://github.com/celeron55/minetest/pull/330 16:56 PilzAdam looks ok to me 16:57 VanessaE wtf!? http://minetest.net/forum/viewtopic.php?pid=54680#p54680 16:57 VanessaE oops, wrong channel. 17:11 celeron55 thexyz: to spare my effort, could you tell how the database wrapper and leveldb code has been tested so far? 17:12 celeron55 i'm too tired to actually look at it 8) 17:12 thexyz OldCoder was using leveldb on his server for long time, no problems were found 17:12 thexyz well, the only problem is that it doesn't compile on windows (that's why there is ENABLE_LEVELDB setting) 17:13 celeron55 how does the design work for possible future improvements and modifications 17:13 celeron55 like starting to use sqlite to store more columns for faster access of some things without loading/saving everything 17:15 thexyz backend only defines beginSave, endSave, saveBlock, loadBlock and listAllLoadableBlocks 17:15 RealBadAngel leveldb is used on ALL oldcoders servers 17:16 RealBadAngel and its because sqlite corrupted all the worlds 17:16 celeron55 so if i add saveTimestamp and loadTimestamp, who is going to implement them for leveldb? 17:16 RealBadAngel so the migration was forced 17:17 RealBadAngel leveldb is tested on oldcoders server for more than a month 17:17 RealBadAngel no problems reported 17:17 celeron55 dual backend at least doubles work, it might not be viable for a project of this size 17:18 RealBadAngel in fact it allows to play without server corruption 17:18 celeron55 i have never had any corruption on my servers that i have ran some months ago for months 17:18 RealBadAngel all bigger servers were affected with it, when worlds started to regenerate 17:18 thexyz well, leveldb backend may exist as non-upstream patch too 17:18 VanessaE celeron55: did your maps exceed 4GB? 17:19 celeron55 nope 17:19 RealBadAngel you wont get the problem then 17:19 celeron55 but that shouldn't be any kind of a limit to sqlite 17:19 RealBadAngel there is 17:19 celeron55 there are limits but they are thousands of times more 17:19 RealBadAngel and worlds starts to regenerate 17:19 RealBadAngel destroyin all players made stuff 17:20 RealBadAngel thats why leveldb migration was forced 17:20 thexyz then how about just implementing that backend thing without actually adding leveldb to main repo? 17:20 celeron55 thexyz: i could take in the db abstraction and leave leveldb for later 17:20 thexyz that'd be fine 17:21 RealBadAngel pity, folks wont be able to play on their servers with latest features 17:21 celeron55 how is the backend selected? 17:21 VanessaE celeron55: some people have reported 4GB limits in sqlite outside of the MT community. From the SQLite home page, "The policy was that if it would fit in memory and you could count it with a 32-bit integer, then it should work." 17:21 thexyz celeron55: i've added setting to world.mt 17:21 thexyz then there's simple if statement which chooses backend depending on its name 17:22 celeron55 so you start a leveldb world by creating the directory and adding world.mt with the backend, and then telling minetest to use the world? 17:22 thexyz or by just migrating sqlite one 17:22 celeron55 how does that work 17:23 RealBadAngel we already do have HUGE leveldb worlds 17:23 RealBadAngel on like 20 servers 17:23 thexyz it lists all blocks using current backend (like sqlite) and writes them to another one (leveldb in that case) 17:23 thexyz command line is ./bin/minetestserver --worldname world --migrate leveldb 17:24 dannydark dannydark, +water relections +wind in grass = MC mod - huh? 17:25 PilzAdam https://www.youtube.com/watch?v=PHHxUzp29HA 17:26 RealBadAngel ive seen a video of another voxel game 17:26 RealBadAngel with way better shaders 17:26 dannydark Oh yeah I've seen that, just didn't get why I was highlighted? haha 17:27 RealBadAngel tryin to recall its name, it is still WIP 17:27 PilzAdam oh, it was darkrose 17:27 PilzAdam sry 17:27 dannydark hehe no worries ^_^ 17:28 PilzAdam btw: BF3 FTW 17:28 RealBadAngel got the name its Terasology 17:29 dannydark Yeah ^_^ can't get enough of it recently...just waiting for Tuesday for the new dlc :D 17:29 thexyz guys, can you discuss that somewhere else? 17:30 OldCoder Hello. I was invited by RealBadAngel. 17:31 RealBadAngel celeron55, oldcoder can tell you bout problems with sqlite on his servers 17:31 RealBadAngel and leveldb migration 17:31 celeron55 i don't need to hear more 17:32 celeron55 i'm going to ask thexyz more questions and you two better be quiet 17:32 OldCoder One question only if I may... 17:32 hmmmm http://ompldr.org/vZ2piOQ/screenshot_1554513921.png 17:32 hmmmm progress lel 17:32 Calinou dat minimal game 17:32 OldCoder As I have just arrived. Have decisions about upstream merges been delegated to three people? If so what do they say? 17:33 celeron55 i take care of the big ones 17:33 OldCoder All right 17:33 OldCoder It's good to see you're involved 17:33 OldCoder I take it there's some controversy. Very well. 17:34 celeron55 ...i've been coding minetest 24/7 (minus sleep) for the whole week 17:34 celeron55 (and minus work) 17:34 OldCoder Rest 17:34 OldCoder Why do this? 17:34 celeron55 why not 17:34 OldCoder Then it is your decision 17:35 celeron55 i know how to not burn out; i was just pointing out your wrongness in my non-involvement 17:35 OldCoder Who is "your" ? Debate should be technical. celeron55 is involved. 17:35 celeron55 anyway, let's stop this crap 17:35 celeron55 --- 17:35 hmmmm it's certainly not engine development discussion 17:36 celeron55 thexyz: has somebody used that database backend wrapper with sqlite for extended lengths of time? 17:36 hmmmm he just said so 17:36 celeron55 also, has it been tested with super-ancient worlds (the sectors format) or was it dropped? 17:37 celeron55 i see some related code but code doesn't say much about that, it's likely to not work if it hasn't been tested 17:38 thexyz don't know about sqlite, i personally wasn't using it with that patch; it shouldn't break anything though as it's just redesign without changing sqlite-related code 17:38 thexyz and i haven't tested sectors at all 17:39 thexyz because i don't have any worlds which use sectors 17:39 celeron55 the sectors support should probably be dropped, as there is the conversion tool for that 17:39 celeron55 and it works badly anyway (at least in current c55/master) 17:40 celeron55 mainly because there are some miscellaneous other incompatibilities with that old data 17:40 celeron55 there shouldn't be, but many have slipped in 17:41 hmmmm indeed, there are many opportunities for cleaning up minetest... with farmesh removed, old map support, cleaned up lighting et al. i bet the code base can be shrunken by 2kloc 17:44 celeron55 thexyz: i'm leaving this on it's whole out from 0.4.4, but it will be handled in the near future by doing testing and then merging the abstraction part upstream 17:45 celeron55 we need to be *very* sure the sqlite part isn't going to do anything bad 17:45 thexyz ok 17:46 OldCoder Could the SQLite3 part be dropped except for import/export? 17:46 celeron55 and an another thing is to structure the code of handling multiple backends so that it's unlikely to cause merge conflicts when server admins apply leveldb support to any version 17:47 celeron55 OldCoder: at least not anytime soon 17:47 OldCoder rubenwardy, Hi 17:48 rubenwardy hi 17:50 celeron55 sqlite has the huge benefit of being widely available for external tools made in any language 17:51 RealBadAngel what about the fact it also corrupt worlds? 17:52 celeron55 singleplayer worlds never grow so large and admins of huge servers know what they are doing 17:53 Calinou admins of huge servers don't put world limiting mods, rarely (and anyway, cheating lets you go through these walls) 17:53 Calinou ask redcrab/ghostshell 17:54 celeron55 i am not saying leveldb would never be used upstream, but i am not confident in doing that now 17:54 RealBadAngel redcrab also suffered that issue 17:54 celeron55 i haven't even searched any other alternatives yet 17:57 OldCoder Decision seems to be made. I'll note for future though (a) 4GB limit is gone (b) Faster storage (c) Supports import and export (d) Less corruption. As an operator of 18 worlds including some approaching 0.5GB I am pleased. 17:58 OldCoder thexyz, thank you again for this marvelous work.\ 17:58 OldCoder The \ was a typo 17:59 OldCoder celeron55, another technical question. Simple one. Could not this feature be present as optional? Then everybody might be satisfied. 18:00 celeron55 19:18:24 < celeron55> dual backend at least doubles work, it might not be viable for a project of this size 18:01 celeron55 i don't want to have anything upstream that won't be always up to date 18:01 OldCoder Very well 18:01 celeron55 i am committed to keeping sqlite up and running until i decide otherwise, and i don't want to double my work 18:02 OldCoder All right. Good luck. thexyz branch will remain my standard. There can be no going back to SQLite3. We will feed patches upstream when possible. And merge some downstream too. The software is hereby forked. 18:03 celeron55 well, i'm going to make sure it's easy to apply the leveldb branch to any version (after the abstraction layer is merged) 18:03 OldCoder Good. celeron55 call on me if you need assistance. I trust you are enjoying the work despite exhaustion. 18:04 celeron55 has anyone been testing the db layer with this? https://github.com/celeron55/minetest-worldtest/ 18:04 OldCoder R 18:04 celeron55 that could possibly be extended to support testing of backend migrations 18:05 celeron55 try running it 8) 18:05 celeron55 it's fun to watch 18:06 celeron55 it'll test compatibility between various versions from 0.3.1 to 0.4.3 18:06 OldCoder Later perhaps. Time has barely permitted maintaining the existing worlds. They are an interesting test in and of themselves. I have Calinou World, Zegaton, Harbor, and even LM Creative. 18:06 OldCoder All seem to work perfectly relative to database format. 18:06 OldCoder I asked Ghostshell for Gameboom world. He did not answer and now that world is gone again :-( 18:07 OldCoder I presume the discussion here is done 18:07 celeron55 the current sqlite support was coded by some guy related to gameboom because the sectors format didn't scale up to their server needs 8) 18:07 OldCoder Heh. They have hit the limits. And then some. 18:08 celeron55 that's a long ago... more than a year 18:08 OldCoder In Internet time it is millenia 18:08 celeron55 more like year and a half, actually 18:08 OldCoder Billenia then :p 18:09 Calinou september 2011, 1.25 year :p 18:09 hmmmm arhg 18:09 hmmmm all of my mapgen's problems were either directly or indirectly related to the damn VoxelArea's buffer zone 18:10 celeron55 wut? 18:10 hmmmm things are working much better 18:10 hmmmm now it's time to work on the lua stuff 18:29 Taoki Any chance for minecarts to make it in 0.4.4? 18:29 PilzAdam they are too buggy 18:29 Taoki I tested them on many servers. The basic idea would be that currently, they work well, only that at some corners they suddenly stop and throw you out for no reason. It also seems that if the server is gagging you get stuck in them until you re-log yourself 18:29 Taoki ok 18:30 Taoki **lagging 18:34 Taoki Damn. Since today the client keeps closing itself on any server I go after some time 18:35 Taoki Hope it's not one of those segmentation things again. Happened all day today >_> 18:35 Taoki and now again 18:48 celeron55 http://pastebin.com/4WyUFK0k 18:48 celeron55 hmm... 18:49 PilzAdam farwasteland? 18:49 darkrose someone has a branch fetish 18:50 celeron55 i leave them hanging around just in case, and then forget which of them are deleteable and which are not 18:51 celeron55 PilzAdam: it was a test to generate some kind of wasteland to far places 18:52 Taoki I wonder if torches and light-emmittnig blocks can specify the intensity / distance of lights. Though I assume not... 18:52 celeron55 it was actually quite cool, but dunno; people wouldn't generally like it 18:53 dannydark any screenshots? ^_^ 18:53 celeron55 Taoki: they can, but torches are almost as bright as it allows 18:53 Taoki oh, ok. Good to know 18:54 Taoki BTW. If anyone wants you can join VanessaE's server. I'm curious to see what the improvement on bandwidth is since that change with idle players 18:54 Taoki At an average 18:54 OldCoder Is there an idle players patch I should merge? 18:54 Taoki OldCoder: No, it's been merged 18:55 Taoki OldCoder: I made a patch so players who stand still don't send any position updatee packets to the server, and in turn the server doesn't sent position updates to other clients 18:55 celeron55 dannydark: i seem to have two: http://c55.me/random/2012-04/screenshot_2198089407.png http://c55.me/random/2012-04/screenshot_2116280199.png 18:56 OldCoder Taoki, I don't have upstream. Could you PM me the patch link? 18:56 VanessaE celeron55: I'm curious if the performance is better for you on my server also 18:56 celeron55 can't test now 18:56 celeron55 crappy connection that doesn't even pass udp through 18:56 Taoki OldCoder: It's a modified GIT branch. Sadly it's hard if you aren't keeping up with latest upstream 18:56 Taoki ok 18:57 OldCoder All right ty 18:57 VanessaE ok. Whenever you are ready to, it will be up. 24.181.193.82 port 30000. 18:57 celeron55 (yes, my adsl is once again broken... it has a tendency to break on friday evenings) 18:57 dannydark celeron55: don't seem to be loading for me at the moment :( just timing out will try to have another look later ^_^ 18:58 celeron55 oh 18:58 celeron55 you're right 18:58 celeron55 it's on this side of the broken adsl 18:58 celeron55 well whatever... 18:59 dannydark ah makes sense, well I've pinned them so will try again at some other time. just intrested to see what it looked like as I quite like the idea of wasteland type area 18:59 celeron55 i guess i'll upload somewhere then 19:00 dannydark don't worry too much about it I don't mind looking again later when you're adsl is playin nice 19:00 dannydark +g 19:01 celeron55 http://misc.minetest.net/farwasteland_1.png 19:05 Taoki looks nice 19:05 Taoki BTW... what about farming mod? That's beenr eady for sooooooo long too 19:06 PilzAdam Im waiting for RealBadAngel to finish the textures 19:06 Taoki oh, righty then 19:07 dannydark nice ^_^ the lava looks a tad out of place but I reckon thats just a texture thing, it would be nice to see this make it back into the mapgen at some point ^_^ 19:12 celeron55 dunno; i guess we'll get something like that as a mod if and hopefully when it gets some modding support 19:12 * celeron55 points at hmmmm 19:12 celeron55 ehm 19:12 celeron55 s/it/mapgen/ 19:12 Taoki BTW, a problem I wanted to point out for some days (not sure if it's really a problem). Why are we still using 2 nodes for doors? I saw doors in custom nodes which are made of only one node, only that the nodebox / collision size extends to 2 blocks. Only one entity, one collosion box, etc 19:12 Taoki I'd find it better in all ways for all doors to be like that 19:13 PilzAdam it causes some textures bugs 19:14 Taoki ahh, ok 19:14 Taoki VanessaE: poke. It's 4 players on the server now... what's average bandwidth? 19:23 Taoki celeron55: BTW. The same way top surfaces of blocks are brightened, shouldn't bottom surfaces be darkened? Would look even nicer for eg: Cave ceilings 19:24 Taoki So sides = unmodifies, top = brighter (like shaders do now), but bottom = darker (by the same amount top is brighter) 19:45 VanessaE Taoki: about 15 kB/sec 19:45 VanessaE but people are moving around doing stuff 19:45 Taoki VanessaE: Ok. Weird since it shouldn't be using that much 19:45 hmmmm farwasteland? 19:45 Taoki More like 1kbps from what I know 19:46 VanessaE I'm not sure how much of that is the server and how much is coming from routine network access on my box and on abe's 19:46 hmmmm what's that, screwed up map past coordinates 4096? 19:46 VanessaE brb 19:47 VanessaE ok. 19:49 Taoki VanessaE: Ah... if it's your total machine bandwidth it's likely other things too 19:49 VanessaE ok, got nethogs running. 19:50 Taoki you'd need to know how much minetestserver is using specificallyt 19:51 VanessaE heh, it's so low it isn't even showing up in the list. 19:53 Taoki might even be under 1kbps 19:55 VanessaE bah, nethogs only shows tcp, not udp. 20:00 VanessaE there we go, let's see how iptraf works. 20:03 VanessaE Taoki: sign off, clear your cache, sign back on 20:04 VanessaE I wanna make sure this tool is actually reporting on UDP:30000 20:13 VanessaE strange.. is the server using any port other than 30000? 20:50 Taoki VanessaE: back. Still need me to do that? 20:51 VanessaE no 20:51 VanessaE RBA did but I didn't see anything in iptraf that looked like it was coming from MT 20:51 VanessaE (yet he says it transferred at a reasonable speed this time). 20:52 VanessaE does MT use any other port besides 30000 (or whatever)? 21:05 Taoki Just had another idea: Could clients store a permanent cache (up to a certain size limit) of the world on each server? So that when they connect to a server again they don't have to receive all of the chunks? Maybe just the changes after that... not sure if this would work out at all 21:16 jin_xi hm so when shaders are used item images of nodeboxes look alright, but in-game they look flat 21:31 Taoki VanessaE: Suddenly, your server is extremely slow to login to again 21:32 VanessaE weird, I'm not even on it right now, but I am doing some testing in another MT instance, trying to debug something unrelated. 21:33 VanessaE (and my tests are hogging CPU pretty badly, but I expected it) 21:39 celeron55 oh interesting... i happened to get the editbox segfault while using a debug build of irrlicht with core dumps enabled 21:43 celeron55 https://raw.github.com/gist/4191188/1fc5b2ae84412a9ae449d6d0dcf9412486c428d4/gistfile1.txt 21:43 celeron55 looks useless to me, but in case somebody is interested 21:53 Taoki https://github.com/celeron55/minetest/pull/333 21:59 celeron55 dim bottom suits less lighting situations than just a bright top 22:00 celeron55 also you're checking three too many variables, you could just check for y < -0.5 22:02 Taoki celeron55: Just pulled latest master and noticed you already changed some things there, so I guess that would no longer make sense now. If you think darker bottom works too you can probably add it in... I think it looks a lot better but that depends 22:02 celeron55 i guess i'll try it some day 22:03 Taoki Well it's just one line of code, so it can be tried in 2 minutes :P 22:03 celeron55 i modified that part a bit to make it do the multiplication in linear-ish color space; it is the "right" thing to do 22:03 Taoki yeah, that sounds correct 22:04 celeron55 otherwise it's sRGB the gamma of which makes multiplied colors screwed 22:04 celeron55 you see that way screwed colors everywhere but it doesn't make it less wrong 22:05 celeron55 even in minetest 8) 22:05 Taoki I noticed some blocks appeared overly whitened 22:05 celeron55 it's the lazy thing to do 22:06 celeron55 they do that probably because i added an exposure function in the shader; it allows bright things to still have their detail visible 22:06 celeron55 it kind of lessens colors while doing it at the bright end 22:06 Taoki Ahh... I forgot to take any screenshots >.< But I can say bottomside darkness looks a lot better too. 22:06 Taoki Also, 22:07 Taoki I'd suggest making water 1 or 2 pixels lower. 3 seems a bit too much to me 22:08 celeron55 i like it as is 22:08 celeron55 if you want to change it, you'll have to collect votes 8) 22:08 Taoki ok 22:08 celeron55 i'll probably reimplement it in C++ when i'm less lazy though; it doesn't make much sense in a shader 22:09 Taoki Yeah... this one is simple enough to do there 22:10 Taoki I wanna see if I can find some free post-processing code and add it in. But I'm not sure yet how you call shaders properly from the MineTest code 22:10 Taoki I should be able to adapt such shaders if I find any though, and make a setting for each 22:10 Taoki My main focuses are bloom, motion blur, depth of field, reflective / refractive water. I know some won't like this, but some will... so each should have a menu setting and everyone is happy then :P 22:11 Taoki Just wonder if I can find any free code for Irrlicht shaders of this type