Time Nick Message 00:07 est31 RealBadAngel, what was your idea again to turn off shaders when compilation fails? 00:08 est31 can you make a pr with it 00:09 RealBadAngel ive coded that in fact 00:10 RealBadAngel an i do curse that code when im developing shaders 00:10 est31 ? 00:10 RealBadAngel before there was fail, edit and next try 00:11 RealBadAngel now i have to turn on shaders again ;) 00:11 RealBadAngel real pain in the ass 00:11 est31 we can make it only happen on release builds 00:12 RealBadAngel tommorow i will push fixes for non tileable textures 00:13 RealBadAngel then i will male a pr with shaders fallback, ok? 00:13 est31 ok if you fix what I've said in the github comments 00:13 est31 ok 00:13 RealBadAngel i ve coded by now vertical case 00:13 RealBadAngel need to take care of horizontal too 00:14 RealBadAngel not a problem tho, just swap the axes 00:15 RealBadAngel what was cut on y, shal be cut on x the very same way 02:10 paramat hmmmm, #2905 seems ready to go 02:10 ShadowBot https://github.com/minetest/minetest/issues/2905 -- Mgv6/treegen: (Re)Add fallback nodes for compatibility with subgames by paramat 02:15 paramat now pushing game#558 02:15 ShadowBot https://github.com/minetest/minetest_game/issues/558 -- Add new texture for default:leaves and style simple by RealBadAngel 02:19 paramat complete 04:46 hmmmm paramat: looks really good 05:21 est31 is it possible that a shadow mapgen bug still exists? 05:21 hmmmm of course 05:21 est31 since VanessaE rebuilt her server with debug mode, I've got two shadow areas 05:22 est31 before none 05:22 VanessaE oh yeah I guess I should have mentioned that :) 05:22 hmmmm the shadow mapgen bug is not going away any time soon 05:22 hmmmm i know what needs to be done with it 05:22 VanessaE I built it in Debug mode and have it running through gdb, and set up so that when it exits, it'll throw a full backtrace into the debug.txt 05:23 hmmmm meh don't bother 05:23 VanessaE (mostly I just wanted to catch these random segfaults I get from time to time() 05:23 hmmmm i know the reason why it happens and i have a rough idea of what i would need to implement to fix it 05:23 VanessaE -( 05:24 VanessaE and yeah, that old shadow bug exists. not too big a deal, worldedit works fine to fix them. 05:29 hmmmm fixing the network and data serialization is much higher priority 05:29 hmmmm god this code should not be allowed to exist 05:29 hmmmm who the fuck wrote this nonsense 05:29 VanessaE eh...well... 05:30 VanessaE I decline to say who most likely did, as it'll just start a shitstorm :P 08:10 hmmmm https://github.com/kwolekr/minetest/commit/08663b05b60f4ef1e72c76cdf05fb0e2d6d65168 08:10 hmmmm night 08:10 * hmmmm & 09:08 est31 hmmmm, looks good. 09:08 est31 except one if( 09:08 est31 ctrl + f for it :) 09:08 est31 and i cant push it because it bases on another commit 09:08 est31 nore, any judgement yet on my pr? 09:10 RealBadAngel hi est31 09:10 est31 hi 09:10 RealBadAngel im preparing now shaders pr 09:10 RealBadAngel will push that asap 09:11 RealBadAngel the one with grass solution i mean 09:12 est31 rubenwardy, can you read through the doc changes in #2898 09:12 ShadowBot https://github.com/minetest/minetest/issues/2898 -- Add AreaStore data structure by est31 09:12 est31 checking for grammar errors 09:12 est31 and so on 09:12 rubenwardy sure 09:12 nore est31: areastore.cpp, L253: if accept_overlap is true, you don't need to check AST_CONTAINS_AREA 09:13 nore so you can use a ternary instead 09:14 est31 nore, yea 09:15 rubenwardy https://github.com/minetest/minetest/pull/2898/files#diff-1305560bd8befb32862f0feeefabd02eR2602 If `accept_overlap` is true, areas are also returned which have nodes in common with `area` 09:16 rubenwardy so accept_overlap means that areas found don't have to be fully contained, but can extend over the boundaries of the queried area? 09:16 est31 yes 09:17 nore hm... is the commented code normal in the lua api files? 09:18 est31 yea 09:18 est31 dunno whether its ok 09:18 est31 serialisation should be added later 09:18 est31 i had no time to think about it 09:18 est31 libspatial has its own serialisation 09:18 nore hm... 09:18 est31 the one i coded was generalized 09:19 nore serialiazation should be the same regardless of the way data is stored I think 09:19 est31 libspatial has a bulk load mode 09:19 nore (so you can move a world to/from a server with/without libspatial) 09:19 est31 which speeds up loading by 10x or so 09:20 est31 but i think it can only be done if its serialisation is used 09:20 nore I agree it may slow loading 09:20 nore but creating a R*-tree is not that expensive 09:20 est31 agreed 09:20 nore and there will never be more than 10^6 areas in a tree 09:21 est31 also there is a problem in cpp because the size is written directly as size_t 09:21 est31 which is platform dependent 09:21 nore yeah, so a serialiazation format should be decided 09:21 rubenwardy maybe: If `accept_overlap` is false, areas will only be returned if they are fully contained by 'area'. 09:22 rubenwardy but I guess that loses detail 09:22 nore I think it wouldn't be too difficult, given what we already have 09:22 rubenwardy apart from that, it's all good 09:23 est31 nore: https://github.com/libspatialindex/libspatialindex/blob/master/include/spatialindex/RTree.h#L88 09:23 est31 those are the three methods 09:23 est31 I'm using the first 09:23 est31 line 79 09:23 est31 also the problem is how to implement forEach 09:23 est31 for libspatial 09:24 est31 perhaps pass an area with HUGE_NUM or how its called 09:24 est31 then make a forEachVisitor 09:56 RealBadAngel est31, updated, waitin just for travis 09:58 RealBadAngel for the very first time i am removing 2 files from the project :) 10:16 Amaz RealBadAngel, for some reason, with #2897 everything goes red (the normal shaders not working thingy), but before, I've always been able to use shaders without a problem... 10:16 ShadowBot https://github.com/minetest/minetest/issues/2897 -- Fix relief mapping issues by RealBadAngel 10:30 Amaz I'm not sure if it would help, but this is the error it gives: http://pastie.org/10289619 10:33 rubenwardy Amaz, what is your opengl version? 10:33 VanessaE OpenGL version "old as shit" 10:33 VanessaE :) 10:34 Amaz 3.0 10:34 VanessaE that's old. 10:34 rubenwardy How would I update OpenGL? 10:34 Amaz ^ 10:34 VanessaE however if that's all you got, then the shader code needs patched. 10:34 rubenwardy xubuntu 10:38 Krock that's old. 10:38 Krock heh 10:38 Krock I can just laught about this with my 2.1.x version here 10:39 VanessaE heh 11:03 VanessaE RealBadAngel: some players are reporting a "double images" effect on my server now. Probably from parallax maps that don't work for them. On my screen, they work fine but some of them seem...blurry 11:03 VanessaE (I just git pull'ed from minetest_game before the restart) 11:03 VanessaE specifically, ores 11:04 VanessaE and I note that there only seem to be a small number of such images. default grass lacks for example 11:04 VanessaE I guess you meant to use these with the "generate normalmaps" feature 11:09 Amaz I just switched to a propitiatory Ndivia driver, which gave me opengl 3.3, which works :) Sorry to cause more trouble! 11:09 VanessaE actually it matters 11:10 VanessaE because you described a regression 11:10 Amaz Ah, okay :) Glad I could be a help in that case! 11:14 VanessaE RealBadAngel: turning on "bumpmapping" is what's leading to that "blurry" effect, for me. 11:15 VanessaE I am not yet using the "fix relief mapping issues" patch, btw. 11:21 VanessaE anyways, more later. for now, off to bed. 13:05 RealBadAngel texture2Lod was the problem 13:05 RealBadAngel i will change it back to regular sampler 13:29 RealBadAngel Amaz, here? 13:36 Amaz RealBadAngel, yep 13:36 RealBadAngel Amaz, could you switch drivers back and check the pr again? 13:37 Amaz Okay 13:41 Amaz I need to restart to change graphics driver, right? 13:43 * Amaz goes ahead and restarts. brb 13:48 Amaz Great, that works brilliantly RealBadAngel! 13:48 RealBadAngel thx 13:49 Amaz np 15:04 rubenwardy I'd like help with this: #2908 15:04 ShadowBot https://github.com/minetest/minetest/issues/2908 -- Add map limiting (WIP) by rubenwardy 15:05 rubenwardy When I set the map_generation_limit to 80, the map goes from 30,30 to 45,45 roughly 15:05 rubenwardy paramat, hmmmm :P 15:06 nore rubenwardy: the map is generated in 80x80x80 chunks 15:06 rubenwardy yeah 15:06 nore so setting it to 80 will only create the chunks that fit inside 15:06 rubenwardy I'll measure again 15:07 rubenwardy Goes from (-32,-32) to (47,47) - 32+47 = 80 15:07 rubenwardy so that does make sence 15:07 rubenwardy * 32+47 = 79 15:08 rubenwardy So it's 80x80x80, but also offsetted 15:08 nore yes, it is always offset like that 15:09 crazyR could someone point me in the direction for where the list formspec is parsed in minetest core. ive been searching for a while. not had much look. to many references to the word formspec and list lol 15:10 rubenwardy https://github.com/minetest/minetest/blob/master/src/guiFormSpecMenu.cpp#L300 15:11 rubenwardy or https://github.com/minetest/minetest/blob/master/src/guiFormSpecMenu.cpp#L777 15:11 rubenwardy depending on what you mean 15:14 crazyR im trying to find where the event data gets prepared when a list formspec is parsed. i may have over thought it a little thought. as maybe its actually in the "on_player_receive_fields" section 15:14 rubenwardy Updated #2908 15:14 ShadowBot https://github.com/minetest/minetest/issues/2908 -- Add map limiting by rubenwardy 15:34 RealBadAngel nore, it was you that coded camera offset? 15:35 nore yes 15:35 RealBadAngel is there any chance to achieve the same effect without the offset? 15:36 RealBadAngel offset thingy is blocking (making it slower in any case) the VBO code 15:36 nore hm... I'm not sure there is a way 15:37 nore except forcing irrlicht to have more precision 15:37 RealBadAngel you know where it should be applied? 15:40 nore I'm not sure it is even possible :/ 15:41 RealBadAngel either the camera offset, or need to resend the meshes to HW buffers on its change 15:42 RealBadAngel if we didnt have the offset VBO could be way more efficient 15:42 nore yeah, you could try without offset at first :) 15:43 jin_xi well, its wrong selection box without, anything else? 15:43 jin_xi i only did quick look 15:43 nore jin_xi: without offset, you get visual problems at large coordinates 15:43 nore flickering nodes, etc 15:44 RealBadAngel worse 15:44 RealBadAngel if we do ignore offset all world meshes are fucked up 15:44 nore what do you mean? 15:45 RealBadAngel if you go more than 200 in each direction, meshes are displaced 15:45 nore that's because offset is not completely disabled 15:45 jin_xi yes, thats the camera offset step, i set it to 1000000000000 or so to try without 15:45 nore ^ that should work better 15:45 nore don't set it too high though, it shouldn't be too large 15:45 RealBadAngel problem is that offset modifies vertices coords 15:46 nore I guess 15000 is safe for testing 15:46 RealBadAngel if those are changed, mesh have to be re-sent to the buffers 15:47 RealBadAngel and thats slow, if you move just by one node further and have to wait for all the meshes to be resent 15:47 jin_xi so many weird offset and BS factors to mt 15:47 RealBadAngel btw, proller started to remove BS thing 15:48 jin_xi try helicopter mod, due to limitations of attachement system and coding around those its not save to travel across camera offset steps with it 15:48 nore RBA: offset is not changed enough 15:48 RealBadAngel if this was meant only to make the coords different as c55 said, proller is right 15:48 nore you need to travel 200 nodes to get it changed 15:48 RealBadAngel 201 and 199 border i meant 15:49 nore yes, if you go at 200, then offset is set at 200 15:49 RealBadAngel here you will meet the switch in offset value 15:49 nore and then, you need to either go to 0 or 400 so it gets changed again 15:49 RealBadAngel but imagine just walkin at the border line 15:49 RealBadAngel resending triggered all the time, an overkill 15:50 nore no, if you walk past the border line, the borders get changed 15:50 nore and you need to walk 200 nodes again to cross a border 15:50 RealBadAngel ah i see 15:50 RealBadAngel so its a bit better 15:51 RealBadAngel nore, what do you think about that BS case? 15:51 paramat we need to remember how far out the problems start, i'll dig up the commit 15:51 RealBadAngel we are doing *10 , /10 all the time just for sake coords not to be mistaken 15:52 nore well, I don't know what it is for :) 15:52 RealBadAngel i asked c55 bout that already 15:52 RealBadAngel he said its just for this purpose 15:53 paramat offset every 200 is not needed, perhaps a few thousand is good 15:53 RealBadAngel i wonder how much time we spent on doing those conversions... 15:53 nore paramat: I started noticing problems even with ~1000 y coordinate... 15:53 jin_xi should have been 1 not 10 from the start, but it has leaked into many other places and calculations so its not easy to change 15:53 RealBadAngel paramat, nore, where the problem with the camera starts to happen? 15:54 RealBadAngel 30k? 15:54 nore RBA: IIRC, there are problem near 500 - 1000 that are already noticeable 15:54 jin_xi 1k nodes and visual glitches due to precision sounds fishy 15:54 RealBadAngel i noticed that only with selection lines 15:55 RealBadAngel if they were not so thin that would be propably not noticeable 15:56 RealBadAngel btw. i just spent a few hours lookin for a bug that didnt exists 15:56 RealBadAngel im talkin bout my last pr 15:57 RealBadAngel i just forgot to include networkprotocol.h with raised protocol number 15:57 RealBadAngel so all my code was inactive lmao 16:01 RealBadAngel paramat, i will clean the pr again and push it, will need then changes to mt game 16:01 RealBadAngel all the dirt+something node defs will have to be changed 16:02 RealBadAngel i will change the ones that come in minimal game 16:02 RealBadAngel http://pastie.org/10290216 16:08 RealBadAngel paramat, also i have figured out how to tint the grass 16:08 RealBadAngel interested? 16:09 paramat nope hehe 16:11 RealBadAngel why not? 16:11 RealBadAngel we could have grass tinted depending on biome 16:11 paramat tinting is never as good as individual textures, changing the colour usually also needs careful tuning of saturation, brightness, contrast too 16:11 RealBadAngel have you saw tone maps used in mc for that feature? 16:11 paramat also only a very few biomes would use it, currently 2 16:12 RealBadAngel i dont think so 16:12 paramat yes i've seen the MC version, don't like it 16:12 RealBadAngel make it depending on humidity or temperature 16:12 RealBadAngel more wet env you get more juicy green 16:12 RealBadAngel hotter it gets, more brown 16:13 RealBadAngel and imho transitions could be nicer thx to it 16:14 RealBadAngel not to mention grass being single nodedef 16:14 paramat cool idea but i don't like smooth variation, hmmmmm may be more interested for v8 and his new biome system 16:17 RealBadAngel also i cannot wait till we have swamps in :) 16:17 paramat perhaps then the camera offset could be around 512? 16:17 paramat yeah all rainforest will dip into water to create swamps 16:18 RealBadAngel paramat, i want the need of resending be as rare as possible 16:18 RealBadAngel so greater the number the better 16:19 RealBadAngel i just dont know what kind of glitches were there apart from selection box lines 16:20 RealBadAngel so if it was happening at 30k, half the value just 16:20 paramat i remember problems at only 10000, if nore says 500-1000 then we should trust that 16:21 nore (well, when I wrote that code, I used the smallest value for which I didn't notice problems) 16:22 nore but anyway: you can write the code, and then we can test how expensive a resending is 16:22 RealBadAngel code for that is trivial 16:22 nore and change offset step accordingly 16:22 RealBadAngel and you can just feel the border 16:23 RealBadAngel fps drop is noticeable 16:23 nore how much time does it cost? 50ms? 100? 16:23 RealBadAngel basically VBO code doubles the fps 16:24 RealBadAngel when you are at the border fps drops for a while to level as it could be without vbo 16:24 nore except when crossing a border? 16:24 nore ah, so it is no real problem :) 16:24 RealBadAngel its especially noticeable when moving fast 16:24 nore hm... 16:25 RealBadAngel you have a feeling that you have crashed on an ivisible barrier 16:25 Miner_59 Hi, I'm here to ask about my pull request made 8 days ago which I think fixes jumping on bouncy blocks. Link: https://github.com/minetest/minetest/pull/2886 16:25 nore well, you could maybe try to raise offset step to 500 or so 16:25 nore but I wouldn't raise it more if I were you 16:25 paramat yeah 500 is good 16:25 RealBadAngel nore, we can get the correct value with some testing 16:26 Calinou Miner_59, it breaks jumping on all bouncy nodes, even if disable_jump is not 1 16:26 nore also: I would advise to make it a number close to a power of 2, but below 16:26 Miner_59 ??? but why, I tested over a week and it clearly works always for me 16:26 nore (not too close, though) 16:28 paramat so 500 is better than 512? 16:28 nore yes, 500 would be a good choice I think 16:29 Miner_59 Calinou, do you have time to test? I have a server for with jumping mod running on ip 80.128.163.64 Port 30.000 16:29 Calinou ok, cloning and compiling your fork 16:29 nore but there would need to be some tests 16:29 Miner_59 ok cool thanks 16:29 paramat okay 16:30 RealBadAngel if choosing between 500 and 512 i would rather pick the second 16:30 Miner_59 if it is too much work you can connect with unmodified client too and watch me jumping 16:30 RealBadAngel pot numbers are always better in this job ;) 16:30 nore no, too close can have problems (the glitch would be easier to trigger I think) 16:31 paramat ah 16:31 nore (because you could place a node past 512 without going there yourself) 16:31 paramat of course because it's precision error 16:32 RealBadAngel what is creative range for placement? 16:32 nore something like 10? 16:32 Calinou 10 16:32 RealBadAngel what about technic laser? 16:32 RealBadAngel lets make it like 512 - 64 16:32 RealBadAngel just to be sure 16:32 nore (the problem will be there anyway) 16:33 nore (because you can place the node, go back to 0, 0, 0, restart the game and go where the node is) 16:33 nore it's just that it would be a bit less easy to see 16:34 nore and all that because of irrlicht :/ 16:35 RealBadAngel nore, mc is not using irr and had same problem ;) 16:35 nore yes, but they can decide to use better precision, we can't :| 16:36 RealBadAngel can irrlicht be forced to use more precise format? 16:36 nore idk 16:37 RealBadAngel we shall investigate it then 16:37 RealBadAngel we are not the only voxel game made on top of irr 16:38 RealBadAngel propably others encountered it too 16:45 Calinou hmmmm, RealBadAngel, paramat, Miner_59 actually fixed bouncing, 16:45 Calinou his patch is probably ready to merge now 16:45 Calinou jumping on trampolines works as expected 16:45 paramat cool 16:45 RealBadAngel so lets merge it 16:46 RealBadAngel this is bouncing around for too long already ;) 16:56 Miner_59 Ok thanks again for testing with me Calinou. 17:10 hmmmm Calinou, you keep saying that but I simply do not see how it's fixed 17:11 hmmmm disabling jumping on a node is not the same as fixing it 17:11 Calinou hmmmm, OKAY 17:11 Calinou connect on Miner_59's server 17:11 Calinou with his client 17:12 Calinou and see how it works. 17:12 Calinou it works perfect 17:12 Calinou you seem to have misunderstood what I said 17:12 hmmmm alright will look 17:12 Miner_59 hi, hmmmm it doesn't disable jumping 17:13 Miner_59 ok, join my server ip 80.128.163.64 Port 30.000 17:17 Miner_59 why removing the code which I did should disable jumping? This was just additional to the normal jumping and it gave the player way too much speed 17:21 hmmmm erm 17:21 hmmmm Assertion failed: (mediaReceived()), function afterContentReceived, file /usr/home/ryan/minetest/src/client.cpp, line 1719. 17:21 hmmmm well this can't be good 17:22 Miner_59 what does this mean? 17:22 hmmmm i think i might be having trouble connecting to your server 17:23 rubenwardy C1ffisme really annoys me :( 17:23 hmmmm trying again 17:23 Miner_59 a player with empty name tried to connect from 130.0.11.193 17:23 hmmmm that's not me, my ip is in the 72 range 17:24 Miner_59 and this:19:21:17: ERROR[CurlFetchThread]: servers.minetest.net/announce not found (Couldn't connect to server) (response code 0) 17:24 Miner_59 then can you try with jeija's version of jumping mod, the original one? 17:25 hmmmm nevermind fixing bugs, let's add some new features! 17:25 hmmmm Miner_59, I haven't given up yet 17:25 hmmmm it's still on the Media... part 17:26 Miner_59 oh, I just restarted server now 17:27 Miner_59 here, mod to test with:https://github.com/Jeija/minetest-mod-jumping 17:30 hmmmm eh 17:30 hmmmm too late i got on the server 17:30 hmmmm alrighty where am I going to now? 17:33 hmmmm did you take the server down again? 17:33 Miner_59 it crashed 17:33 Miner_59 sorry 17:34 Miner_59 19:33:20: ACTION[ServerThread]: Server: A mismatched client tried to connect from 72.79.173.196 17:34 Miner_59 19:33:20: ERROR[ServerThread]: ERROR: An unhandled exception occurred: basic_string::_S_construct null not valid 17:34 Miner_59 up again 17:34 hmmmm shoot 17:35 hmmmm somebody somewhere tried to make a std::string with NULL it seems 17:36 hmmmm did it happen again?? 17:36 Miner_59 yes 17:37 Miner_59 I get it sometimes, but I havent latest minetest version 17:37 hmmmm i don't recall anything like that getting fixed since 0.4.12 17:39 hmmmm did you get a backtrace after that unhandled exception by any chance? 17:40 Miner_59 no, but I will try to get one maybe 17:40 hmmmm thank you 17:40 Miner_59 can it be because of modifications I did in mapgen source code? 17:40 hmmmm no 17:42 hmmmm i'm going to guess the error exists somewhere within the lua api 17:45 Miner_59 I'm not sure if it is real not caused by modifications because why then nobody reported it already. I got it first I think a month ago 18:07 Calinou paramat, https://github.com/minetest/minetest_game/pull/565#issuecomment-120997369 18:07 Calinou please drop the "Mesh ALL the things!" fad 18:07 Calinou mushrooms look better as plantlike than they would with a mesh 18:07 hmmmm drink all the booze! hack all the things! 18:08 hmmmm zero through three, we're in every single ring! 18:09 Miner_59 hmmmm: success I have reproduced the crash with server thread attached in gdb 18:09 hmmmm great 18:09 hmmmm let's see the `bt full` 18:10 Miner_59 ok 18:12 Miner_59 I think i need another thread but here is bt: http://pastebin.ubuntu.com/11873585/ 18:13 hmmmm huh 18:13 Miner_59 Here, threads and bt full: http://pastebin.ubuntu.com/11873625/ 18:13 hmmmm i don't see any crash 18:14 hmmmm it looks like you just pressed ctrl+C 18:14 Miner_59 oh can be, but after error server doesn't respond anymore 18:14 hmmmm oh so it still *lives*? 18:14 Miner_59 yes 18:14 hmmmm but it doesn't respond?? 18:15 Miner_59 can you help me what i need to do? 18:15 hmmmm there are like 3 problems here all at the same time 18:15 Miner_59 hmm ok 18:15 hmmmm 1). the exception handling is messed up 18:15 hmmmm 2). there's a std::string(NULL) problem 18:15 hmmmm 3). there's a version mismatch when there shouldn't be any to begin with 18:16 hmmmm but let's ignore all that and work on new features!!! yay!! 18:18 hmmmm wait, is that minetest server you're using compiled as debug or release? 18:19 Miner_59 oh, i think as release 18:19 hmmmm yeah hrmm, i dunno, it could be some strange interaction between gdb and whatnot 18:20 Miner_59 hmm ok 18:20 hmmmm can you restart the server with gdb already attached? 18:20 Miner_59 yeah I had to press Ctrl+C because i cant type anything in gdb 18:21 hmmmm you should be able to catch the assert(0) 18:21 Miner_59 how can I restart it? I think it doesnt respond 18:25 hmmmm try ctrl+D? 18:25 hmmmm if that doesn't work, switch over to a different console and kill -9 `pgrep gdb` 18:33 Miner_59 i think i need to attach another thread 18:34 hmmmm if you start it with gdb you won't need to attach anything 18:34 Miner_59 ah ok i will try 18:40 paramat now pushing #2905 18:40 ShadowBot https://github.com/minetest/minetest/issues/2905 -- Mgv6/treegen: (Re)Add fallback nodes for compatibility with subgames by paramat 18:40 RealBadAngel im about to push #2897 and close issue 2815 18:40 ShadowBot https://github.com/minetest/minetest/issues/2897 -- Fix relief mapping issues by RealBadAngel 18:40 hmmmm why not wait for your build to finish? 18:41 RealBadAngel i am waiting 18:41 RealBadAngel anyway, building here ended a few minutes ago 18:41 hmmmm RealBadAngel, are you aware that the code style is to not have any spaces after the function name, right? 18:41 hmmmm it's supposed to be like 18:41 hmmmm void getTextureFlags() 18:42 RealBadAngel are you aware that its not c++? (jokin, will change that) 18:43 paramat RBA im about to push, is this okay? 18:44 RealBadAngel im not touchin the same files, its ok 18:44 paramat okay thanks 18:45 paramat complete 18:47 hmmmm RealBadAngel: out of all that code, is this line + return intensity(texture2D(baseTexture, clamp(uv, 0.0, 0.999)).rgb); the one that actually fixes the texture problem? 18:48 RealBadAngel are you jokin? 18:48 hmmmm if that line wasn't there, the texture would wrap around 18:48 RealBadAngel you think ive changed the protocol version because i used clamp somwhere? 18:48 paramat hmmmmm next i will reverse what i did to minimal game, a while back i added some nodes to solve the 'areas of ignore in snow biomes' problem 18:49 hmmmm RealBadAngel: who did you get approval from with that PR? 18:49 RealBadAngel read the pr 18:50 hmmmm nerzhul +1ed it? 18:50 hmmmm :| 18:50 paramat lol 18:50 RealBadAngel https://github.com/minetest/minetest/pull/2897/files#diff-82a77dc39dc952bc7ea07b74b73fc2a5R114 18:50 hmmmm he's practically a yes man at this point 18:50 RealBadAngel this is the core change 18:50 hmmmm don't make me crank the requirement up to 3 approvals 18:50 hmmmm nerzhul +1s everything on the github PR tracker 18:50 RealBadAngel but to get this working i had to get those bools 18:51 RealBadAngel and also i wont be doing anti repetitive tricks inside shaders, forget it 18:52 RealBadAngel code is too sensitive for that 18:52 RealBadAngel you may require that in core, but stop demanding that in shaders please 18:52 hmmmm you just make it into a function 18:52 RealBadAngel no i wont 18:52 hmmmm repetitive code is not just ugly 18:52 hmmmm it's CONFUSING 18:52 RealBadAngel it have to be fast not nice 18:53 RealBadAngel hands off please ;) 18:53 hmmmm other people work on this you realize 18:53 hmmmm right now it just so happens that you're the one modifying the shaders the most 18:53 hmmmm you keep doing this with your code in the core too 18:53 hmmmm it's copy pasted from somewhere else, and it usually has a copy/paste error or two in it 18:53 hmmmm not only that 18:54 RealBadAngel thats copy pasted? 18:54 hmmmm but at some later point in time, you modify one variant of the code ever so slightly but not the other one 18:54 RealBadAngel wtf are you talking about? 18:54 RealBadAngel youre claimin thats not my code? 18:54 hmmmm no... 18:55 hmmmm that is your code, but you copied one instance and pasted it into another portion of the code 18:55 hmmmm sorry, I can't get behind that argument that this is somehow safer and somehow better 18:55 hmmmm when code is sloppy, there are usually bugs 18:56 hmmmm if it smells like shit, there's probably shit somewhere nearby 18:56 RealBadAngel you dont even understand the code yet have to comment it 18:56 RealBadAngel your question bout clamp showed that 18:56 hmmmm I'm still reading the code 18:58 RealBadAngel ok 18:58 RealBadAngel waitin for the comments then 19:05 hmmmm just to make sure i understand this 19:05 hmmmm ds is the distance between the eyes 19:05 hmmmm uv is the left eye, and uv2 is the right eye 19:05 hmmmm ? 19:05 RealBadAngel no 19:06 RealBadAngel ds is displacement, based on eye ray 19:06 hmmmm isn't there supposed to be two eye rays in parallax occlusion 19:06 RealBadAngel uv are texture coords, modified by the displacement 19:07 RealBadAngel uv2 is the modified one, by displacement. if texture is seamless it will become current texture uv 19:07 RealBadAngel when texture is not seamless, we have to check the border cases 19:08 RealBadAngel http://www.cescg.org/CESCG-2006/papers/TUBudapest-Premecz-Matyas.pdf 19:08 RealBadAngel please do read some theory on the effects used 19:10 RealBadAngel both displacement methods we are using are described here 19:13 hmmmm and the displacement is the distance between each intersection probe 19:14 RealBadAngel displacement depends on heightmap 19:14 hmmmm you said ds == displacement 19:14 RealBadAngel when you march a ray, and meet point that should be higher 19:14 hmmmm ds is the step each probe takes to find an intersection 19:15 RealBadAngel then difference beetween an original point and found one will be ds 19:15 Miner_59 Ok, I'm back. Here's a backtrace of all threads of minetest after it displays mismatched client error: http://pastebin.ubuntu.com/11873884/ 19:15 hmmmm so i don't quite understand this part 19:15 RealBadAngel displacement is not constant 19:15 hmmmm you have uv -= ds 19:16 RealBadAngel yes 19:16 hmmmm but then you modify uv to be += dist * ds 19:16 RealBadAngel ive modified original method a lot 19:16 hmmmm so doesn't that make all the displacements minus 1? 19:16 hmmmm shouldn't it be (dist + 1) * ds? 19:16 RealBadAngel i do shift texture by maximum possible ds value, thats an offset 19:17 RealBadAngel kinda like i was using inverted heightmaps 19:17 RealBadAngel then add found actual displacement value 19:17 hmmmm Miner_59: interesting.. the serverthread is not there at all 19:17 hmmmm it keeps dying without breaking into the debugger it seems 19:18 hmmmm this should not happen 19:18 RealBadAngel substracting max value helps to accomodate texture shifts 19:18 RealBadAngel because it mainly depends on view angle 19:20 hmmmm hmm 19:20 RealBadAngel i suggest you to see grass texture before and after 19:20 hmmmm i think i understand what your code does well enough, but what does that have to do with leaving part of it repetitive 19:21 RealBadAngel it may look nearly the same but its not the same 19:21 RealBadAngel vertical changes y values, horizontal x 19:21 RealBadAngel the second section is dynamically compiled 19:22 hmmmm I am looking at it and I legtiimately can't see a difference 19:22 hmmmm actually nevermind there is one difference 19:22 hmmmm you compare to 2.0 instead of 1.999 in the second version 19:22 hmmmm i don't think this is intentional though.. 19:23 RealBadAngel it is 19:23 hmmmm why 19:23 RealBadAngel values are chosen after heavy tests 19:24 Miner_59 hmmm I think it happens only when the client which tries to connect is build with a rather recent (1-2 month??) minetest version. I can't reproduce it myself when connecting with clients with older minetest version 19:24 RealBadAngel autogen maps are totally different 19:24 hmmmm did you really test the difference between " >= 2.0" and " > 1.999"? 19:24 RealBadAngel ofc, its visible 19:24 hmmmm if the difference is less than 0.0001 then it shouldn't make any logical difference at all 19:25 RealBadAngel not when it comes to normal maps 19:25 hmmmm well, then, don't you think you should add a comment about that? 19:25 hmmmm that seems like a very important detail 19:25 RealBadAngel its black magic anyway 19:25 hmmmm it's not black magic 19:25 hmmmm code is not black magic 19:25 RealBadAngel ofc it is 19:25 hmmmm it needs to be understandable 19:26 RealBadAngel displacement mapping doesnt have such code 19:26 RealBadAngel what we are doing here is a workaround for non existant bug 19:26 hmmmm i don't get it 19:27 RealBadAngel the method assumes that texture has to be seamless 19:27 RealBadAngel got it now? 19:27 hmmmm you're refusing to write a comment about this very minute difference of less than 0.0001 because "it's black magic anyway" and it "solves a non-existent bug"? 19:27 RealBadAngel im not refusing 19:27 hmmmm well 19:27 hmmmm i can't speak for you but 19:27 RealBadAngel im just saying that theres no theory for it, no equations 19:27 hmmmm i want everybody to understand how the shader works 19:28 RealBadAngel values were chosen for textures to look right 19:28 hmmmm this is the reason why I don't just rubber stamp your commit 19:28 hmmmm i am reading about parallax occlusion and actually trying to understand why you coded what you coded 19:29 RealBadAngel i will try make the story short 19:30 RealBadAngel main thing about parallax is that whats gone on one edge appears on the opposite one 19:30 RealBadAngel so displaced texture is continuos 19:30 RealBadAngel in case of grass it caused dirt to be visible on top 19:31 RealBadAngel i just had to stop it while having displacement working 19:32 paramat PR to make minimal minimal again #2909 19:32 ShadowBot https://github.com/minetest/minetest/issues/2909 -- Minimal: Remove recently added unnecessary nodes by paramat 19:35 RealBadAngel hmmmm, have you tried grass before and after? 19:35 hmmmm no 19:36 RealBadAngel how the fuck you want to understand the gfx code without seeing what it does? lol 19:36 RealBadAngel i should hand you my asm demos code and wish you good luck reading ;) 19:37 hmmmm this is why people inveted comments 19:37 RealBadAngel propably with unrolled loops i could hear you cryin even here in poland ;) 19:43 hmmmm okay there's a problem here 19:44 hmmmm I don't know how this happened but the protocol version is not the same as the contentfeatures serialization version 19:44 hmmmm take a look at ContentFeatures::deSerialize() 19:44 hmmmm you get the version from the first byte of a serialized ContentFeatures 19:45 hmmmm I am sort of upset 19:46 hmmmm this isn't your own problem, but the whole concept behind having different serialization versions is all screwed up 19:46 hmmmm really wish we could nuke it and start over again 19:48 hmmmm alright, I finished commenting 19:50 RealBadAngel with dropping the texture i lost it 19:50 RealBadAngel textures are deleted on shutdown, cant do that runtime 19:51 hmmmm okay, that's good to know 19:51 RealBadAngel i could delete them, but i do need them 19:52 RealBadAngel about asserting image creation 19:52 RealBadAngel are you sure that i should check if there were 4 bytes aviable for its creation? 19:52 hmmmm if it's a function from some third party library, you cannot speculate on whether or not it will always succeed 19:53 hmmmm this is how you write robust code 19:53 hmmmm minetest needs more robust code 19:53 RealBadAngel i can add that but i will have to stop the engine then 19:53 RealBadAngel code cannot work without that texture 19:54 hmmmm I dunno if you realized it or not, but lately there's been a lot of buggy behavior caused by fragile code because people don't check all the error cases or some other equally simple thing 19:54 RealBadAngel i see 19:54 RealBadAngel will add check for it 19:54 hmmmm it needs to be absolutely rock solid, not something that'll shatter in a million pieces if I look at it the wrong way 19:54 hmmmm save that for your personal projects maybe 19:55 VanessaE RealBadAngel: did you see the comments I made earlier? about the "double image" effect, and the "blurriness" ? 19:55 RealBadAngel i saw, but i dont get it :) 19:55 RealBadAngel can you explain? 19:56 RealBadAngel whats the "double image" effect? 19:56 VanessaE http://irc.minetest.ru/minetest-dev/2015-07-13#i_4320666 19:56 VanessaE better yet, moment 19:56 RealBadAngel a screnshot maybe? 19:57 RealBadAngel parallax not workin. on which node? whats doubled there? 19:57 VanessaE this is what the guy sees, with just parallax enabled: https://dl.dropboxusercontent.com/u/183644134/O_o/screenshot_20150713_124748.png 19:57 VanessaE and if he turns on bumpmappint too: https://dl.dropboxusercontent.com/u/183644134/O_o/1.png 19:58 RealBadAngel he has modified client propably 19:58 VanessaE I advised him to just turn those features off for now. he wasn't reporting a problem before because default textures didn't have included normalmaps in the past. 19:58 RealBadAngel i can see fsaa applied there 19:58 RealBadAngel do you see the hand? 19:58 VanessaE that's not fsaa. 19:58 VanessaE that's some wield mesh glitch 19:59 RealBadAngel caused by fsaa 19:59 VanessaE nope, pretty sure it isn't. others have had the same problem 19:59 RealBadAngel we havent changed lately anything related to wieldmesh 19:59 VanessaE it's something related to kahrl's wield mesh rework ages ago 20:00 VanessaE so that's an olllllld glitch 20:00 RealBadAngel yet the node textures and the wield mesh seems to be affected the very same way 20:00 VanessaE it's not related at all to the current discussion 20:00 RealBadAngel lemme turn the fsaa on 20:01 VanessaE the guy who reported the issue is on another network now, trying to get him to come here 20:02 hmmmm RealBadAngel, can you explain to me how insertSourceImage and getTexture works? 20:02 hmmmm :) 20:03 RealBadAngel its not my code 20:03 RealBadAngel it just works ;) 20:03 hmmmm you're using something you don't understand? 20:04 hmmmm okay, I have not written much client-side graphics code at all so I could be totally missing something here 20:04 RealBadAngel i know what it does, how it is doing such primitive is not my concern 20:05 hmmmm insertSourceImage, from what it seems, inserts an *image* into the texture source cache 20:05 RealBadAngel thats not possible to understand everything in such big project 20:05 hmmmm so let's just follow this logic for a while 20:05 hmmmm the flags image is created 20:05 hmmmm some pixels get set 20:06 hmmmm you insert this IImage into the source cache 20:06 RealBadAngel yes 20:06 hmmmm then you use this image as the source for a new ITexture 20:06 hmmmm this ITexture is never added into the texture cache 20:06 hmmmm is it? 20:07 hmmmm I am asking you 20:07 RealBadAngel driver handles the textures for me 20:08 RealBadAngel why should i dig the irrlicht engine when im using a simple texture? 20:09 hmmmm because I want to make sure there's no memory leak here 20:09 RealBadAngel texture cache is flushed on shutdown 20:09 hmmmm shutdown of what 20:09 RealBadAngel image used is in cache (original dropped) 20:09 RealBadAngel the engine 20:10 hmmmm the point is 20:10 hmmmm can the TextureSource be destroyed and recreated hundreds of times without recreating the same textures 20:12 RealBadAngel is the TextureSource code point of my PR or what? 20:12 RealBadAngel im just using it 20:12 hmmmm and you might be possibly using it incorrectly 20:12 hmmmm if you use it incorrectly, that makes a memory leak 20:13 RealBadAngel 12 bytes.... 20:13 hmmmm i don't give a shit how many bytes 20:13 hmmmm this must be done CORRECTLY 20:13 hmmmm otherwise your code is shit 20:13 hmmmm i am not going to allow memory leaking code to get into the engine 20:13 RealBadAngel show me the proof its shit 20:13 RealBadAngel or something is incorrect here 20:14 hmmmm that's why i'm asking you to go over this with me so that I understand why this code is solid 20:14 RealBadAngel otherewise it looks like youre just WANT to find problems here 20:14 hmmmm look 20:14 hmmmm irrlicht might delete the textures on shutdown 20:14 hmmmm but that's not the point 20:14 hmmmm any program with memory leaks will free the memory on shutdown 20:15 hmmmm if you're making such an assertion that you don't need to worry about allocated memory, you need to understand 20:15 RealBadAngel so where is the leak in your opinion? 20:15 hmmmm - the lifetime of TextureSource 20:15 hmmmm - the ramnifications of irrlicht reference counting 20:15 hmmmm - whether or not your texture or image is being tracked 20:15 hmmmm I am not saying there is a leak 20:16 hmmmm I'm saying the code looks suspicious 20:16 hmmmm and what's even more suspicious is that you don't want to help me understand why it's correct 20:16 hmmmm you want me to gloss over it instead and just give you a rubber stamp 20:16 RealBadAngel check its destructor 20:16 RealBadAngel driver->removeTexture(iter->texture) 20:16 RealBadAngel flushed? 20:16 RealBadAngel end of story 20:17 hmmmm and where is that texture coming from? 20:17 hmmmm you never added it to the texture cache 20:17 RealBadAngel // Insert an image into the cache without touching the filesystem. 20:17 RealBadAngel // Shall be called from the main thread. 20:17 RealBadAngel void insertSourceImage(const std::string &name, video::IImage *img); 20:18 hmmmm an image is not the same as a texture 20:18 hmmmm look at what insertSourceImage does 20:19 RealBadAngel it inserts image to cache 20:19 hmmmm which cache 20:19 RealBadAngel get texture grabs the image from cache and return the pointer 20:19 hmmmm does it really 20:21 RealBadAngel i dont care how it does it, waiting for santa's visit with box full of pointers or whatever, but my code gets the stored pointer to the image 20:21 RealBadAngel i havent wrote that code 20:22 RealBadAngel theres no point for me to read whole engine code, be it our own, or irrlicht when using primitives 20:23 hmmmm but you have to understand how to properly use those interfaces 20:23 hmmmm if you don't use it properly, there could be memory leaks or other assorted bugs 20:23 hmmmm you make them, the rest of us fix them 20:25 RealBadAngel who else the fuck you are asking to read and understand the whole engine? 20:27 RealBadAngel also comments like "wtf is this string" "i could wrote that line in different manner" are just plain stupid 20:27 RealBadAngel since some time youre just jumping on my code 20:28 RealBadAngel also modyfing it without knowledge what it does 20:30 UjE . 20:30 RealBadAngel if youre so smart, please write your own code 20:30 hmmmm RealBadAngel, code has to at least look nice too 20:30 hmmmm but seriously 20:30 hmmmm wtf IS the point of strue and sfalse? 20:31 VanessaE ah there he is 20:31 hmmmm I am not asking you to change it necessarily, I'm just asking why did you decide that you needed to do that 20:31 UjE here 20:31 RealBadAngel i just defined a string that im using in many places 20:31 VanessaE RealBadAngel: UjE is the guy with the double-image problem 20:31 RealBadAngel what else could it be? 20:31 RealBadAngel hi UjE 20:31 UjE hi 20:31 UjE im UjEdwin in the forum 20:31 RealBadAngel UjE, can you pastebin your config in the first place? 20:32 hmmmm you can also do the same with "1" and "0" 20:32 hmmmm have you ever heard of string pooling? 20:32 UjE after load a world or into the server? 20:32 RealBadAngel your game config 20:32 UjE k 20:32 hmmmm in any case, RealBadAngel, the point I wanted to make is that if you had read the code you'd notice that you create twice as many images 20:33 hmmmm one that does get tracked by TextureSource and one that does not 20:33 hmmmm twice as many textures, I mean 20:33 UjE server_unload_unused_data_timeout = 1500 20:33 UjE server_name =Floating Forest O_o 20:33 UjE server_description = Temporary Tiny Server 20:33 UjE motd = hi, say: help to teleport to spawn. 20:33 UjE disable_fire = true 20:33 UjE enable_damage = true 20:33 UjE enable_tnt = false 20:33 UjE default_privs = 20:33 UjE --interact, shout, home, protect 20:33 UjE viewing_range_nodes_min = 0 20:33 UjE viewing_range_nodes_max = 80 20:33 RealBadAngel oops ;) 20:33 UjE server_dedicated = false 20:33 UjE remote_port = 30001 20:33 UjE port = 30000 20:33 UjE rc.server = 2.68.32.20 20:33 UjE irc.channel = 3231 20:33 UjE irc.nick = O_o Ser 20:33 UjE irc.password = un1cyc132 20:33 UjE irc.sasl.pass = un1cyc132 20:33 UjE max_users = 8 20:33 UjE static_spawnpoint = -138, 12, 224 20:33 RealBadAngel pastebin.com or pastie.org for example ;) 20:33 UjE disallow_empty_password = false 20:33 UjE creative_mode = false 20:33 UjE enable_pvp = true 20:33 UjE server_address = 20:33 UjE server_announce = true 20:33 UjE ahux1_descends = false 20:34 hmmmm UjE, you sure that was a smart idea to reveal your password to the entire world? 20:34 UjE serverlist_url = servers.minetest.net 20:34 UjE creative_mode = false 20:34 UjE enable_shaders = true 20:34 UjE maintab_LAST = settings 20:34 UjE menu_last_game = minetest 20:34 UjE mg_name = singlenode 20:34 UjE name = O_o 20:34 UjE noclip = true 20:34 UjE doubletap_jump = false 20:34 UjE mainmenu_last_selected_world = 2 20:34 UjE selected_world_path = C:\Users\Edwin\Desktop\minetest\bin\..\worlds\FP 20:34 UjE public_serverlist = false 20:34 UjE sound_volume = 1 20:34 UjE fast_move = true 20:34 UjE fixed_map_seed = 20:34 UjE free_move = false 20:34 UjE enable_rollback_recording = false 20:34 UjE areas.self_protection_max_areas = 3 20:34 UjE areas.self_protection_max_size = 32,32,32 20:34 UjE areas.self_protection_privilege = protect 20:34 hmmmm gahh.. 20:34 UjE areas.self_protection = true 20:34 UjE address = breballlove.no-ip.org 20:34 UjE keymap_backward = KEY_KEY_S 20:34 UjE keymap_chat = KEY_KEY_T 20:34 UjE keymap_cmd = / 20:34 UjE keymap_console = KEY_F10 20:34 UjE keymap_drop = KEY_KEY_G 20:34 UjE keymap_fastmove = KEY_KEY_F 20:35 UjE keymap_forward = KEY_KEY_W 20:35 UjE keymap_freemove = KEY_KEY_Q 20:35 UjE keymap_inventory = KEY_KEY_E 20:35 VanessaE EEK! 20:35 UjE keymap_jump = KEY_SPACE 20:35 UjE keymap_left = KEY_KEY_A 20:35 UjE keymap_noclip = KEY_KEY_C 20:35 UjE keymap_print_debug_stacks = KEY_KEY_P 20:35 UjE keymap_rangeselect = KEY_KEY_R 20:35 RealBadAngel omfg :) 20:35 UjE keymap_right = KEY_KEY_D 20:35 UjE keymap_sneak = KEY_LSHIFT 20:35 UjE keymap_special1 = KEY_LCONTROL 20:35 UjE wieldview_update_time = 2 20:35 hmmmm usually we'd kick people who do this 20:35 UjE aux1_descends = false 20:35 UjE wieldview_node_tiles = false 20:35 UjE gui_scaling = 1.00 20:35 UjE client_unload_unused_data_timeout = 600 20:35 UjE world_config_selected_mod = 17 20:35 UjE world_config_hide_gamemods = true 20:35 hmmmm i'm sure it'll end soon though... hopefully 20:35 UjE world_config_hide_modpackcontents = true 20:35 UjE unlimited_player_transfer_distance = true 20:35 UjE cinematic = false 20:35 UjE enable_clouds = true 20:35 UjE anisotropic_filter = false 20:35 UjE bilinear_filter = false 20:35 UjE connected_glass = true 20:35 UjE enable_node_highlighting = true 20:35 UjE generate_normalmaps = false 20:35 VanessaE G*d damn, UjE... 20:35 UjE mip_map = false 20:36 UjE video_driver = opengl 20:36 UjE enable_bumpmapping = true 20:36 UjE enable_parallax_occlusion = true 20:36 UjE trilinear_filter = false 20:36 UjE ill use dropbox 20:36 UjE ? 20:36 UjE how? 20:36 UjE https://dl.dropboxusercontent.com/u/183644134/O_o/minetest.conf 20:36 UjE ohh, that was not that pass 20:36 UjE its dont using that pass somewhere 20:36 VanessaE UjE: you pasted the entire file right into the channel 20:36 UjE i did not really understand how that worked, so i just made one 20:36 VanessaE RealBadAngel: well that confirms that he's not using fsaa. 20:37 VanessaE UjE: it's okay. next time just do the dropbox thing instead :) 20:37 UjE kk 20:38 RealBadAngel wieldview_node_tiles, whats that? 20:40 RealBadAngel we dont have such setting 20:41 UjE i have no ide 20:41 VanessaE that's probably for the 3d armor modpack 20:41 RealBadAngel can you zip whole game folder? with mods and everythin? 20:41 UjE cash too? 20:42 UjE its about 170mb 20:42 RealBadAngel without cache 20:42 UjE k 20:43 RealBadAngel VanessaE, what was about blurry? 20:43 VanessaE RealBadAngel: moment. I'll make a screenshot. 20:45 UjE the upload will take a while 20:45 UjE 40mg 20:45 UjE mb 20:46 VanessaE RealBadAngel: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/random/Screenshot_2015-07-13_16-45-54.png 20:46 VanessaE I distinctly remember normalmaps looking a LOT sharper than this. 20:46 RealBadAngel yes 20:46 UjE the sand stonebrick looks really weird too 20:47 RealBadAngel autogen cannot make softer normalmaps 20:47 UjE https://dl.dropboxusercontent.com/u/183644134/O_o/screenshot_20150713_133746.png 20:47 RealBadAngel i just made them softer 20:47 VanessaE RealBadAngel: then the "soft" normalmaps you provided for minetest_game have got to be redone. they look absolutely horrible. 20:47 RealBadAngel horrible? 20:47 VanessaE I'm serious, it feels like the camera is...out of focus 20:48 RealBadAngel when theyre softer? 20:48 RealBadAngel folks were voting bout them 20:48 VanessaE this is a 16px texture you're making them for. they have to be sharp to match it 20:48 RealBadAngel there were no single voice again the style i was makin them 20:49 RealBadAngel *against 20:49 VanessaE I don't think anyone had time to comment on them 20:49 VanessaE I never even saw that discussion 20:49 RealBadAngel on the other hand, when i made them sharp, folks didnt liked them 20:49 VanessaE wat 20:49 VanessaE they were perfect 20:49 RealBadAngel hey, there were no maps before :P 20:49 VanessaE if you use soft normalmaps with a hard 16ps texture they look horrible 20:50 VanessaE 16px* 20:50 RealBadAngel for the very first time we have some 20:50 RealBadAngel what are you comparing then? 20:50 UjE Done: https://dl.dropboxusercontent.com/u/183644134/O_o/MT.zip 20:50 RealBadAngel UjE, thx, downloading it 20:51 VanessaE RealBadAngel: I'm comparing your hand-made normals to the "sharp" effect you get with the auto-generator. The sharpness MUST match that or it looks awful. 20:51 UjE its win8.1 64 20:52 RealBadAngel VanessaE, we can open a thread on it again 20:52 RealBadAngel so folks can decide 20:52 VanessaE what's to decide? it looks horrible 20:52 RealBadAngel because its soft? 20:53 * cheapie agrees that it looks horrible 20:53 cheapie MT is supposed to be blocky. These aren't. 20:53 VanessaE RealBadAngel: it's not "soft", it's out of focus. 20:53 VanessaE like you applied a blur filter or something 20:53 RealBadAngel damn it. when there was a pr with everybody liked them :P 20:54 RealBadAngel now you say "its horrible" 20:54 RealBadAngel you have saw the screenshots before, didnt you? 20:54 VanessaE hell, look at your iron normal and compare it to say, the diamond one 20:54 VanessaE https://raw.githubusercontent.com/minetest/minetest_game/166b1c623add33a0ad4ba469da30035df5293bca/mods/default/textures/default_mineral_iron_normal.png 20:54 RealBadAngel so? 20:54 VanessaE https://raw.githubusercontent.com/minetest/minetest_game/166b1c623add33a0ad4ba469da30035df5293bca/mods/default/textures/default_mineral_diamond_normal.png 20:54 cheapie The stone, especially, looks like somebody made the map in MS Paint. 20:55 VanessaE you can see the difference in sharpness between these two alone 20:56 RealBadAngel i made a diamond a bit sharper just 20:56 VanessaE and then compare it to the normal for stone: 20:56 VanessaE https://raw.githubusercontent.com/minetest/minetest_game/166b1c623add33a0ad4ba469da30035df5293bca/mods/default/textures/default_stone_normal.png 20:56 crecca oh god, now I'm affraid to look at minetest code knowing RealBadAngel opinions :) 20:56 VanessaE it looks like you actually baked the lighting into the texture here 20:56 RealBadAngel for them not to be lookin exactly the same 20:57 RealBadAngel ok, i will make sharper ones and will upload them to review 20:57 VanessaE I'm sorry RBA but if I had the authority, I'd have blocked your PR here :( 20:57 RealBadAngel but honestly, there was PR with lotsa screenshots hanging there 20:57 cheapie RealBadAngel: OK. I'm looking forward to testing that. 20:57 RealBadAngel why didnt you said anything before? 20:57 RealBadAngel both of you :P 20:57 VanessaE RealBadAngel: because you weren't showing stone. 20:58 VanessaE you were focusing on grass and stuff 20:58 VanessaE at least in the screenshots I DID see. 20:59 RealBadAngel https://github.com/minetest/minetest_game/pull/559 20:59 RealBadAngel "Looks good, although the brick could be a bit more smooth IMO" 20:59 VanessaE these need to be remade to be at least as sharp as what the auto-gen-normals code would have produced given the same amount of info about the image 20:59 RealBadAngel folks even asked to make some more smooth :) 20:59 VanessaE I think by "smooth" he meant the surface of the brick faces. 20:59 VanessaE not the edges 21:00 VanessaE i.e. less displacement among neighboring pixels 21:00 RealBadAngel https://cloud.githubusercontent.com/assets/2177790/8579768/c9150a68-25b6-11e5-8f75-7cd85cc81bdf.png 21:00 RealBadAngel anyway, can we see stone here? 21:00 cheapie The mese looks horrible in that picture... 21:00 VanessaE yeah, that's the same effect I get. the stone is terrible. 21:00 VanessaE the mese is okay-ish 21:00 RealBadAngel rotfl 21:01 VanessaE but that sand. G*D no. 21:01 RealBadAngel mese ok, mese bad, decide you two ;) 21:02 VanessaE I said "okay-ish" :) 21:02 RealBadAngel sand came out to be as you see after several different versions 21:02 cheapie Well, I'm still of the opinion that mese should be a yellow block with "ME SE" hastily scribbled on it... 21:03 RealBadAngel what the mese block has in common with mineral mese texture? 21:03 VanessaE cheapie: well there IS still a mese block and it does generate underground, it just doesn't look hastily drawn anymore; pictured here is mese crystals 21:03 cheapie VanessaE: I know... I've been grumbling about the whole mese crystal thing more or less ever since it got put in. 21:04 RealBadAngel and Vanessa, autogen produces such sharp images because i couldnt apply there blur 21:04 RealBadAngel i made such code but it was damn to slow 21:04 VanessaE RealBadAngel: and thank G*d it can't! 21:04 VanessaE in any case, RealBadAngel you have GOT to revert this commit. 21:04 cheapie RealBadAngel: The sharp images look nice. That's the whole point of this discussion. 21:04 VanessaE "Sharp is safe, dull is dangerous.: 21:04 RealBadAngel what about folks that think softer are nice? 21:05 VanessaE let them make their own normalmaps. 21:05 VanessaE it takes all of 1 minute to apply a guassian filter over an image. 21:05 VanessaE but you can't UN blur the image if it's already done that. 21:05 VanessaE that way* 21:05 RealBadAngel theyre not made with simple gausian filter 21:05 VanessaE you get my point 21:05 sofar bit of a personal opinion, but I like how the sand looks :^( 21:05 VanessaE don't split hairs 21:06 RealBadAngel that is what autogen is doing 21:06 VanessaE well it looks like you did varying amounts of Gaussian blur over an otherwise sharp image 21:06 UjE i like that sand too 21:07 VanessaE but my point is you can blur the texture if it's too sharp, but you can't UN-blur a fuzzy one to any reasonably good result. 21:07 VanessaE so err on the side of caution and supply sharp textures. 21:13 RealBadAngel UjE, do you have a link for the build youre using? 21:13 UjE this? https://github.com/minetest/minetest/releases/download/0.4.12/minetest-0.4.12-win64-msvc.zip 21:13 RealBadAngel ah i see 21:14 UjE i using win0.1 64 21:14 UjE win8.1 21:14 RealBadAngel for the heightmaps and relief mapping you should use more up to date client 21:14 RealBadAngel youre build is too old 21:14 UjE i have no ide 21:15 RealBadAngel lemme find another build for you 21:15 UjE i looking for new buildes next to everyday, but no newer realesed 21:15 UjE NO, im dump 21:15 UjE here is it 21:15 UjE https://forum.minetest.net/viewtopic.php?f=42&t=6745&sid=218dc6eb53e1537791d982de4926ac36 21:16 UjE i using dev version 21:16 UjE i for got that 21:17 RealBadAngel i can see your version of shaders 21:17 RealBadAngel theyre totally outdated 21:17 UjE Latest dev build, 0.4.12 version with leveldb support, commit f16ebbfecf, from 26.05.2015. 21:17 RealBadAngel you dont have minimap even 21:17 hmmmm hey when are you going to finish minimap btw 21:18 RealBadAngel https://github.com/minetest/minetest/commit/8a85e5e58decf77df27d114574026ff8032b8a45 21:18 RealBadAngel you need a build that with this commit at least 21:18 RealBadAngel hmmmm, whats there to finish? 21:19 RealBadAngel you mean api for it? 21:19 hmmmm no 21:19 UjE should u not make a mpa for a full texture, not only parts? 21:19 hmmmm there's a github issue full of things todo for the minimap 21:19 RealBadAngel UjE, point is that your build doesnt have new relief mapping code 21:20 UjE ohh 21:20 VanessaE RealBadAngel: meanwhile, please revert your normalmaps commit until you can make up better images 21:21 UjE yeah its from 26.05.2015 21:21 RealBadAngel hmmmm, A better scanning strategy would be more performant. Do you have any better idea? i dont have 21:21 UjE make full pics instand of parts 21:21 RealBadAngel Surface scanning should be done in the minimap thread, not the mesh update thread - completely false 21:21 hmmmm then explain why, in the issue 21:21 RealBadAngel Only the left shift key can be used to change the - whats wrong with left shift? 21:22 UjE i made many nurmal maps for mod to UT3, taht they using nurmalmaps for everything 21:22 RealBadAngel There is an older bug in minetest @VanessaE calls main mesh thread lag, which greatly affects the responsiveness of the minimap. Perhaps it should be fixed. - not belonging here 21:22 VanessaE UjE: that doesn't matter, the maps can combine just fine. it's the fact that the individual maps are fuzzy is what's so ugly. 21:22 RealBadAngel The buffers have constant size - thats only reasonable 21:22 UjE k 21:23 hmmmm so basically you're saying that you're not going to finish what you started 21:23 hmmmm this is the RealBadAngel way 21:23 hmmmm add a feature 21:23 RealBadAngel i just said im not agreeing with all 21:23 hmmmm make sure it's full of edge-case bugs, overly verbose code, memory leaks, etc. 21:23 hmmmm never finish it 21:24 hmmmm leave that part for other people to worry about 21:24 hmmmm let's just worry about new features 21:24 RealBadAngel in case you havent noticed im hardly trying to fix parallax mapping issue 21:24 hmmmm the thing that you call "not a bug" 21:24 RealBadAngel yet i have to fight with you is using a string "0" and "1" nice or not 21:25 RealBadAngel ofc its not a bug. it just looks bad 21:25 hmmmm the enitre point of your graphical effects is to look nice 21:25 RealBadAngel if it was a bug i could propably fix that in no time 21:26 hmmmm so 21:26 hmmmm i didn't see a reply to #2844 21:26 ShadowBot https://github.com/minetest/minetest/issues/2844 -- Minimap code and behaviour problems 21:26 hmmmm were you planning on replying to it at all 21:26 RealBadAngel please let me finish what i do have open, ok? 21:27 RealBadAngel i cant work on a few issues at one time 21:27 hmmmm k i just wanted to know if you had ever planned to fix it in the first place 21:27 RealBadAngel last time i was checking that minimap thread there were just 6 "issues" 21:28 RealBadAngel btw, A better scanning strategy would be more performant - please do compare what we have to what was done in gsmapper 21:29 RealBadAngel it took me a few weeks to make it being so fast as it is now 21:29 RealBadAngel yet you have to make an issue on it :P 21:29 RealBadAngel dont you feel thats just not right? 21:30 hmmmm i'm trying hard to direct minetest development to have a higher level of quality 21:30 hmmmm maybe this new direction isn't what some old devs want 21:31 hmmmm maybe they want to continue shitting out half-assed code that handles no error cases and uses the most naive algorithm possible 21:31 hmmmm with duplicated work 21:31 hmmmm i think the userbase is fed up enough with all this though 21:31 RealBadAngel IF you will have better algorithm for scanning, just let me know 21:31 hmmmm yes 21:32 RealBadAngel atm i managed it to work almost real time 21:32 hmmmm don't scan downward, store the result in a map, and then scan downward a second time 21:32 hmmmm it shouldn't be struggling to do real time 21:33 hmmmm and btw the minimap cache is never evicted, so over time it'll get really large and bogged down 21:33 hmmmm dude let's just put it this way 21:33 RealBadAngel cache is dynamic... 21:33 hmmmm the bar has been raised 21:33 hmmmm will you rise to meet the challenge? 21:34 RealBadAngel you again showed that you dont know the code youre commenting 21:34 RealBadAngel unloading the block is freeing the cache 21:34 hmmmm and how does that work 21:36 RealBadAngel sending pos with empty data deletes block from cache 21:36 RealBadAngel ofc if you didnt changed that 21:36 celeron55 i wish RBA would just make code of reasonably quality in the first place... 21:37 celeron55 this is painful to watch 21:37 celeron55 and a waste of everyone's time 21:38 hmmmm RealBadAngel: so maybe you should have added a note about that instead of mislead people 21:38 hmmmm are you absolutely sure that unloading a block produces an mesh update? 21:38 RealBadAngel hmmmm, actually i was explaining this piece of code to you before 21:38 hmmmm or are you going to say, "I don't need to know how all the rest of the code works" 21:39 hmmmm and then berate me for commenting on code where I might be wrong on some particular detail about it 21:39 hmmmm if I am wrong on some detail about your code that I'm commenting on, there's an opportunity for me to learn more about the code, which is a good thing. everybody's knowledge increases 21:40 hmmmm scrutinizing code helps a lot. who cares if we're stuck on some mundane detail about reference dropping for 20 minutes, we all come out more knowledgable after that 21:40 hmmmm and it's not personal 21:43 RealBadAngel im ok with discussing actual improvements. but things like with those strings are just a matter of taste, not even a style 21:43 hmmmm you dismissed every improvement suggested so far 21:43 RealBadAngel and are completely meaningless. yet you flood me with similar things 21:43 RealBadAngel no 21:44 RealBadAngel i am investigating texturesource thing atm 21:44 hmmmm if you really want to define a variable for a constant you should at least define it with the const-qualifier to show your intentions 21:44 hmmmm in regard to the string thing 21:44 hmmmm you really seem hung up on that suggestion 21:44 RealBadAngel it was an example. already changed that 21:44 hmmmm lol you'd never last at my job 21:45 RealBadAngel tname += normalmap_present ? "1" : "0"; 21:45 RealBadAngel tname += tiledef->tileable_horizontal ? "1" : "0"; 21:45 RealBadAngel tname += tiledef->tileable_vertical ? "1" : "0"; 21:45 hmmmm before I started my current job and worked on a real software development team with real development methodologies, I thought that having a backlog of bugs and memory leaks and all this was common for large projects and par for the course 21:45 hmmmm but it's not 21:45 hmmmm this is not normal at all 21:45 hmmmm minetest really sucks 21:46 sloantothebone https://www.bountysource.com/teams/minetest 21:48 sloantothebone I've been banned from #minetest for bad connection again :( 21:48 sloantothebone Anybody want to earn some money by contributing to minetest? 21:49 sloantothebone https://www.bountysource.com/teams/minetest 21:52 hmmmm I suppose I can do the text-on-signs one, but any bounties must be donated to the website hosting 21:56 RealBadAngel hmmmm, what makes you think that i do have two copies of generated images? 21:58 hmmmm because I read the code 21:58 hmmmm that code that was too unimportant for you to read 21:59 RealBadAngel instead playing the smartest person in the world, you could just answer 22:00 hmmmm .. I already answered this before 22:01 hmmmm so getShaderFlagsTexture() 22:01 hmmmm on first for that flags texture, you create the image 22:01 RealBadAngel no you havent answered it 22:01 hmmmm write the image 22:01 hmmmm insert the source image into the IMAGE cache 22:01 RealBadAngel yes 22:01 hmmmm then create the texture directly from irrlicht 22:01 hmmmm this is texture instance #1 22:02 hmmmm next time that function gets called, isKnownSourceImage() returns true 22:02 hmmmm so you execute getTexture() 22:02 hmmmm getTexture calls getTextureId() 22:02 hmmmm if the texture ID is not known but the image is, it'll call generateTexture() and then call driver->addTexture 22:02 hmmmm that's copy #2 22:03 hmmmm the difference between these copies are... the one generated in getTexture() is tracked by the TextureSource, and gets deleted on TextureSource destruction 22:03 hmmmm TextureSource has a lifespan of a Game 22:03 hmmmm you can have multiple Games per instance 22:03 RealBadAngel ok, fixed that 22:04 hmmmm so in theory, if you were to keep entering and exiting games, you'll have lots of leaked textures 22:04 hmmmm before you said, "what, it's only 12 bytes!" 22:04 hmmmm oh it's okay guys, I'm only leaking 12 bytes 22:04 hmmmm don't worry 22:04 crazyR would someone be able to add a "warning" log level if they have the time. my C++ skills are virtually non existant or i would add and merge request it myself 22:04 hmmmm crazyR: that exists in ShadowNinja's log refactoring 22:05 RealBadAngel hmmmm, sorry, you were right. already fixed it 22:05 hmmmm i.e. coming soon 22:05 hmmmm RealBadAngel: it's okay, I'm only leaking out 4 cc of piss per hour, it's okay 22:05 hmmmm my pants are totally not looking wet 22:06 crazyR hmmmm thanks didnt see that before :D 22:08 Calinou heh http://paulmillr.com/posts/github-pull-request-stats/ 22:08 Calinou C++ projects are untolerating of pull requests 22:31 celeron55 that's not very surprising considering how much easier it is to shoot oneself in the foot with C++ 23:00 crecca hmmmm: are you kwolekr on github? 23:05 RealBadAngel hmmmm, updated my pr 23:33 VanessaE est31: you remember that in-engine database thing for region storage i.e. Areas mod? um... we need it worse thant we thought. my server peaked at 47 users today so far, and that Areas HUD is getting very laggy at that count. 23:43 est31 hmmmm, http://irc.minetest.ru/minetest-dev/2015-07-13#i_4321308 23:43 est31 whats the problem here? 23:43 VanessaE ah, he DOES read the logs :) 23:49 luizrpgluiz it is possible to access the server remotely minetest? 23:49 est31 luizrpgluiz, how do you mean that 23:50 est31 you can start the client, and then you "access the server" :) 23:51 VanessaE I think he means via a remote terminal, ssh or something 23:51 luizrpgluiz yes 23:52 luizrpgluiz because it has a line in minetest.conf called remote_acess 23:54 luizrpgluiz because I was curious if it is possible to access the remote server without being graphically the game 23:54 est31 does it? I can't find it in code 23:54 est31 not yet 23:54 est31 i think hunterz is writing a console, I'm not sure 23:55 est31 it is possible to some extent already, with the irc mod, and the irc_commands 23:56 luizrpgluiz :)