Time Nick Message 00:37 MinetestBot 02[Git] 04ShadowNinja -> 03minetest/minetest_game: Fix farming unloaded node crash 1344dc611 http://git.io/mdk0gA (152014-09-06T20:36:40-04:00) 03:02 RealBadAngel https://www.youtube.com/watch?v=CWOW_kihBIM&feature=youtu.be 03:05 RealBadAngel https://www.youtube.com/watch?v=TMlAcMiW8ww 03:31 zat OT, which can be better for sitting? 03:31 zat http://www.lightinthebox.com/yoga-anti-explosion-fat-reducing-ball-65cm_p461556.html 03:31 zat http://www.lightinthebox.com/yoga-anti-explosion-fitness-thicken-ball-65cm_p461545.html 03:55 TheBeast Hey Who is on that remembers me 03:56 TheBeast Vanessa Or Vazon you there 03:56 VanessaE yes. 03:56 TheBeast Do you remember me 03:57 TheBeast How it been 03:58 Cylus Your name seems familiar ... 03:58 zat I know many beasts 03:58 Vazon yes 03:58 TheBeast I Am TheBeast Beast Boss 03:59 Vazon yes im aware who you are 03:59 TheBeast Vazon How your server going 03:59 Vazon TheBeast: good, im working on a second server right now 04:00 Vazon other wise i have no time to do anything between school and work 04:00 TheBeast Could i help while i am back 04:02 TheBeast What has changed out minetest 04:30 jordan4ibanez http://youtu.be/M-wvQ00llEo 04:30 jordan4ibanez I think I'll make this like a 2.5d minetest/haven and hearth thing. 05:22 blaise !server MidnightSystems 05:22 MinetestBot blaise: MidnightSystems | midnightsystems.net | Clients: 1/5, 1/3 | Version: 0.4.10 / minetest | Ping: 642ms 05:23 blaise what causes ghosts? 05:28 jordan4ibanez ghost blocks? 05:31 blaise ghost clients 05:31 blaise I really want to run this mod.. :( 05:31 blaise https://gitlab.com/CiaranG/railcarts/blob/master/README.md 05:31 blaise I'm afraid it will cause way to much lag though 05:31 blaise :( 05:32 blaise and it says it won't work on vanilla minetest! 05:32 blaise :((((( 05:33 VanessaE heh, good luck with that 05:33 blaise I think I'm just going to try it and see what happends.. :G 05:35 blaise who knows 05:35 blaise maybe a miracle 05:39 blaise damn 05:39 blaise VanessaE: it has some really good features! 05:40 blaise VanessaE: would it be even possible to make it compatable with vanilla minetest? 05:40 VanessaE idk 05:40 VanessaE ask CiaranG 05:41 blaise :`( 06:47 e1z0 are now minetest can be compiled with ms visual studio ? 06:48 VanessaE MSVC and MinGW 06:48 VanessaE and of course on Linux with gcc et al. 06:50 e1z0 what version of MSVC is supported ? 06:51 VanessaE dunno, I use Linux for this. 06:53 VanessaE I guess just make sure it's recent 06:53 VanessaE doesn't say in the README what version is needed 06:53 JTE_ Minetest's engine leaves much to be desired. :< I don't even understand how it manages to make chest inventories lag when you add a bajillion blocks to the game, even in single computer mode..? 06:53 VanessaE wait 06:53 VanessaE "Visual Studio 10" 06:53 VanessaE JTE_: formspecs. 06:54 VanessaE JTE_: the engine is quite capable. it's just a matter of learning to use it 06:54 VanessaE oh, lag. 06:54 VanessaE missed that 06:54 VanessaE it's because the client actually asks the server first if it can open the chest 06:55 VanessaE rather than just relying the info the server already sent to the client about the chest's formspec 06:55 JTE_ You click the item you want, and it jumps to your cursor (But it's still really sitting right where it was before), then you click where you want to put it and it appears to jump right back to where it started ... and then five seconds later, it suddenly swaps the items around. 06:55 VanessaE yeah, too much communication between the server and client for operations like that, not enough prediction 06:55 VanessaE that's because the server is doing all of the work to manage the chest inventory 06:56 JTE_ If you try to interact with it in the meantime, it just does random undesirable things like giving you half a stack of the un-swapped items and then picking up the item you just tried to put down again :s 06:56 VanessaE the devs could use some help in making it better 06:56 JTE_ How can it have so much communication problems on a single machine with an internal server? I cannot even fathom why it is like this, much less how to fix it... >_<; 06:58 * JTE_ goes to get CMake 06:59 VanessaE I don't exactly know. however, singleplayer mode still operates as a server-client setup, 06:59 VanessaE it just hides it 06:59 VanessaE I guess file an issue report on github :-/ 07:00 JTE_ What, does it connect to 127.0.0.1 instead of just shooting messages back and forth entirely in-memory..? :P 07:00 VanessaE something like that 07:00 VanessaE it launches a server and connects to it 07:01 VanessaE however in singleplayer normally it's pretty quick 07:01 VanessaE are you sure the lag isn't from the map generator slowing things down? 07:01 VanessaE it's not multi-threaded 07:01 JTE_ Yeah. It's just after you start adding decorative mods that add a bajillion new node types that it starts to have trouble. ... for some reason. 07:02 JTE_ I sure hope one of them isn't invisibly replacing the default chest with an older version or something dumb like that. 07:02 Krock meow 07:03 * JTE_ purrs. =o.o= 07:05 VanessaE Krock: -dev. now. 07:06 Krock VanessaE, I'm there. 07:06 VanessaE your input re: #1607 is needed 07:06 Krock Already viewing it lol 07:06 VanessaE short version: update to current HEAD and try to reproduce it. 07:06 Krock clicked right before you told me to do 07:06 VanessaE heh ok 07:08 Krock wait. HEAD = master? 07:08 VanessaE master branch, latest commit. 07:08 Krock ah k 07:08 VanessaE 62ea342ad1b725ac44a787c60388cbc8a1c383c5 in this case. 07:09 Krock updated from 8948907 to 62ea342ad 07:09 Krock compiling.... 07:13 * Krock likes his 5-click minetest building script 07:13 VanessaE nice 07:13 * sfan5 meows at Krock 07:14 * Krock meows at sfan5 07:16 sfan5 blaise: if you manage to rework the formspec not to require this change https://github.com/CiaranG/minetest/commit/45dc321b3c48f449bac5f9985db587847490d3e2 you can run the carts mod 07:16 sfan5 the other change is entirely server side 07:32 Krock hmm I'm getting many compilation warnings 07:38 * Krock meows at PenguinDad 07:38 Krock good meowing 07:43 * PenguinDad squawks at Krock 07:44 VanessaE bbl. bed 07:44 Krock bai VanessaE 07:46 * gorilla throws a pillow at VanessaE.... sleep well 07:54 PenguinDad <3 xyz bitching 'bout macros 07:56 * gorilla is missing something 07:57 * Krock gives gorilla a metal detector 07:59 gorilla Krock: Hehe... a metla detector for Minetest. just doesn't tell you what metal it's detecting. 07:59 Krock gorilla, there's only iron and copper by default... 07:59 Krock where one of them has a bigger use than the other 07:59 gorilla Krock: okay.. that 07:59 gorilla s a valid point 08:00 * Krock wonders if his freetype addition works 08:02 PenguinDad Krock: there's also gold and bronze 08:02 Krock PenguinDad, bronze isn't an ore 08:02 Krock (not generated by nature) 08:03 PenguinDad s/nature/mapgen/ 08:04 Krock pseudo-nature 08:04 PenguinDad PNMG :D 08:06 gorilla PenguinDad: Write a book of Computer Acronymns according to Minetest :-) 08:08 * Krock points at minetest wiki 08:11 gorilla Krock: Just a reflection on how different industries and fields tend to reuse acronymns. 08:14 PenguinDad gorilla: PNMG == Pseudo-Nature Map Generator 08:14 gorilla PenguinDad: I gathered that much 08:15 Krock PNBMG, biome map generator.. 09:23 * sfan5 meows at john_minetest 09:25 sfan5 !tw https://twitter.com/asiekierka/status/508542322423103488 09:25 MinetestBot So, let's see. ~30 Bukkit developers quit, Mojang doesn't actually care, and I fear Sponge will die in all the inter-developer conflicts. (@asiekierka) 09:25 sfan5 apparently the drama doesn't stop 09:26 asie sfan5: i see you follow me 09:26 sfan5 asie: i see you follow me too 09:26 asie sfan5: i see you follow me following you too 09:27 sfan5 asie: i see you follow me following you following me too 09:27 sfan5 was that correct? 09:27 sfan5 :D 09:27 asie yes 09:28 PenguinDad Followception \o/ 09:29 sfan5 I think I should fix !tw 09:29 gorilla PenguinDad: When does following become stalking??.... love social media. :-) 09:29 sfan5 !tw 508542322423103488 09:29 MinetestBot Sorry, couldn't get a tweet from https://twitter.com/twitter/status/508542322423103488 09:29 sfan5 yes I should 09:51 asie i'm still considering writing a Minecraft mod which makes Minecraft support Minetest mods 10:02 PenguinDad Minetest doesn't compile with -std=c++11 :'( 10:46 Jordach this is not a trick https://cdn.mediacru.sh/19IRE-YksDtV.jpg 11:23 PenguinDad Jordach: using vmware? 11:23 Jordach there may be a second tower 11:27 * PenguinDad throws a kitten at 0.9c at Krock 11:27 * Krock wonders about 0.9c 11:30 PenguinDad Krock: You don't know what c is? O_o 11:30 Krock PenguinDad, a lower-case letter 11:30 PenguinDad ·_· 11:30 Krock °*° 11:31 Krock \_/ 11:32 * Krock now uses freetype instead of the png fonts 11:32 SylvieLorxu https://www.youtube.com/watch?v=MOXQo7nURs0 11:33 PenguinDad Krock: c as in the speed of light 11:34 * Krock throws a kitten with 2c at PenguinDad 11:34 PenguinDad Time traveling kittens \o/ 11:36 Krock best question in school ever: "I don't get this maths sheet. How do I convert 4.5 kg to 0.00002s?" 11:38 * PenguinDad didn't know that maths and physics are combined at school :D 11:47 JTE_ huh... so I can't actually influence how the map generator produces the landmass, only what blocks are placed along the surface there as a result of its wacky random terrain hijinks. .-.; 12:11 Krock http://awesomegifs.com/wp-content/uploads/thumbs-up.gif 12:13 Krock I only knew Chuck norris :/ 12:18 * Krock gives sfan5 http://awesomegifs.com/wp-content/uploads/cat-loves-to-nuzzle.gif 12:18 PenguinDad You don't know the cat? D: 12:18 sfan5 !cat 12:18 MinetestBot http://i.imgur.com/ERebecg.jpg 12:20 PenguinDad !doge 12:20 MinetestBot DOGE is at 0.00000040 BTC 13:06 Krock !doge 13:06 MinetestBot http://i.imgur.com/606BPbS.png 13:27 reactor :( 13:28 reactor 16 bytes? 13:31 reactor double? 13:31 reactor and 32 is quad? 13:31 reactor Ah, long int, not double 13:42 reactor Aha. 13:43 ecutruin And.. https://dl.dropboxusercontent.com/u/53977587/Stone.gif - https://dl.dropboxusercontent.com/u/53977587/Texture%20Test%202D.png - End result of this morning's art romp. ^-^ I created my own stone and dirt textures. XD 13:43 * Krock knows how to use the noise function in PDN and could do a similar image 13:44 * Krock has no idea how the rotating node got created 13:44 ecutruin It's not a node, and the textures were created from real-life textures, resized and touched up. 13:44 ecutruin Just stuff I'm working on for my own art project blog. 13:46 ecutruin This stuff is a lot more detailed - https://www.dropbox.com/sh/cxi9vzdzmk6ywpa/6NDGwDrWBl#/ However, its based off MC textures.. so I'm remaking my own textures slowly so its no longer derivitive. 13:58 reactor OldCoder: eh 14:56 reactor That's alright. 14:57 PenguinDad john_minetest: He's another friendly komputer user ;D 14:58 Krock >komputer 14:58 * PenguinDad throws a citten at Krock :p 14:59 jentron What's the trick to powered tracks? 14:59 * Krock throws grahhhhhmah at PenguinDad 14:59 Krock jentron, "trick"? 15:00 jentron If I put two mese blocks down with a switch I can get the cart to start, but it quickly slows down and stops, even with more powered track 15:00 Krock you need to activate all powered tracks 15:01 jentron How do I activate them? 15:01 Krock with meseons, connect them on one side 15:05 jentron Thanks, that works... 15:05 Krock fine 15:06 PenguinDad "You have nyaned for 20000 seconds!" \o/ 15:09 T4im hmm is there a way to determine the chunksize from within lua? 15:11 Krock T4im, vector.subtract(maxp, minp) 15:11 T4im i mean outside of the on_generate... it seems unnecessary to calculate it over and over again 15:11 T4im and get everything that depends on it over and over again 15:12 T4im its not like it will change constantly, will it? 15:12 T4im at least not during runtime 15:12 Krock minetest.register_on_mapgen_init(function(mgparams) 15:12 T4im ah, nice 15:12 T4im thanks 15:12 Krock mgparams.chunksize 15:12 Krock should solve it 15:12 T4im yes, thanks :) 15:14 Krock np 15:17 T4im ah, and blocksize seems constant.. nice :) 15:18 * Krock uses blocksize 4 15:18 PenguinDad s/blocksize/chunksize/ 15:19 T4im blocksize seems defined within constant.h to 16 and being multiplied with a configurable chunksize then 15:20 T4im its not really a think among servers to change that, is it? O_o 15:24 PenguinDad T4im: changing chunksize is quite unproblematic 15:24 PenguinDad But I'm not too sure what happens if you change MAP_BLOCKSIZE 15:24 T4im well.. thats what i was asking.. its not common for people to change that one? 15:24 T4im blocksize that is 15:25 T4im recompile their engine with different defines and all that :D 15:26 PenguinDad I don't think anybody ever changed this constant in a "production" environment 15:26 T4im good :) 15:58 CreativeBlob hi 16:04 CreativeBlob hi 16:26 Creativeblob KLFNLK:N#KLNHRJKNF:LKHJF 16:26 Creativeblob PKNTONE:G# 16:26 Creativeblob GL:LKE#TN:KJLTN$E:JHNOBTKN$ 16:26 Creativeblob T 16:26 Creativeblob EROPERJOJHNOE:FNHLKJN 16:28 sfan5 what? 16:28 PenguinDad ^ He created entropy \o/ 16:29 Creativeblob hmmmm 16:32 Creativeblob ok 16:34 VanessaE Creativeblob: some reason for all that gibberish 16:34 VanessaE ? 16:34 Creativeblob ? 16:34 VanessaE [09-07 12:27] KLFNLK:N#KLNHRJKNF:LKHJF 16:34 VanessaE [09-07 12:27] PKNTONE:G# 16:35 VanessaE etc. 16:35 Creativeblob oh 16:35 Creativeblob i didnt know which one to use 16:38 VanessaE didn't know which key to use? :) 16:38 Creativeblob i was trying to talk on irc to see if iot would do CreativeBlob@irc hia 16:39 Creativeblob i mean CreativeBlob_irc 16:39 VanessaE next time try just talking :) 16:39 Creativeblob :) 16:39 Creativeblob ok 16:40 Creativeblob :( 16:40 Creativeblob XD 16:40 sfan5 o.O 16:40 PenguinDad 05H04e07r08e 03a09r10e 11s02o12m06e 13c05o04l07o08r03s 09f10o11r 02y12o06u 13:05D 16:40 MinetestBot 4Coo7ooo8olo3ooo12ooo6rs! 16:42 * PenguinDad squawks at MinetestBot 16:42 * Creativeblob 16:45 Creativeblob XD 16:46 Creativeblob whos on this channel 16:46 VanessaE Creativeblob: just do /names 16:47 VanessaE you can see all of the usernames who are connected. whoever is actually here will speak when they feel the need to 16:54 * CreativeBlob is afk 16:55 sfan5 CreativeBlob: use /away, don't tell us when you go afk 16:57 CreativeBlob ok 17:04 Krock http://i.imgur.com/BWwyvtu.jpg 17:06 VanessaE Krock: that issue with Yappy and current HEAD, are you certain it isn't a bug in your mapgen? 17:07 VanessaE does it happen in singleplayer also? 17:08 Krock VanessaE, It happens on singleplayer and multiplayer. I found a possible way to solve this bug - simply don't skip generation 17:08 VanessaE don't skip? 17:08 VanessaE can you elaborate please? 17:08 Krock https://github.com/SmallJoker/yappy/commit/c68d39d2c399927710969aa69f8516c2d44cfa6c 17:09 Krock when skip_overgen is set, the mapgen skips some areas 17:09 VanessaE I ask because I couldn't reproduce this issue with the MG mod and new-enough git, which is vmanip based, and which I use on one of my servers. 17:09 Krock but those were already generated one server step earlier 17:09 VanessaE you should mention this in the bug report 17:09 VanessaE it may be relevant 17:09 Krock yes but I wasn't sure 17:09 Krock that kept me off doing it 17:10 Krock until now, I found no new corruptions when I disable it 18:09 Krock omg this talk. 18:10 hmmmm Krock, what is skip overgen...? 18:10 hmmmm I'd love to help you get your voxelmanipulator stuff working 18:10 Krock hmmmm, this is when an area gets generated 2 times in a row 18:11 hmmmm what do you mean?? 18:11 hmmmm two times in a row 18:11 hmmmm how can an area get generated two times in a row? 18:11 Krock it double generates. idk how to say. 18:11 Krock but when I remove this filter, it works fine and the bug disappears 18:12 hmmmm perhaps they're not duplicates... 18:12 hmmmm and this skip_overgen worked fine before those three commits? 18:12 Krock yes 18:12 Krock I'm looking for the maxp and minp difference. maybe the mapblock size is diffrent 18:13 Krock *chunk 18:13 hmmmm can you set skip_overgen to true in your mod, and then try your mod after compiling minetest after doing 'git revert 3fa4f782d90dac0d800251a9ab0f0afb9d32560c' ? 18:13 hmmmm looking for the maxp and minp difference? 18:13 hmmmm what do you mean by that 18:14 Krock nvm 18:14 hmmmm if previous_maxp == maxp and previous_minp == minp then exit? 18:14 Krock this is what I meant: http://pastebin.com/yda90PiD 18:14 Krock line 5 and 6 18:14 Krock and so on 18:14 hmmmm hrm 18:15 Krock and yes 'if previous_maxp == maxp and previous_minp == minp then' -> skip 18:15 Krock when I let it skip, then it causes those bugs 18:15 hmmmm could you look at your console window and tell me if you see any block saving errors? 18:16 VanessaE hmmmm: what are you thinking here? 18:16 hmmmm i have no idea 18:16 VanessaE (rather than guess at it, this time I'll just ask :P ) 18:16 hmmmm let's logically reason this here 18:16 Krock hmmmm, no single error 18:17 Krock trying with log level 4 18:17 hmmmm don't bother.... 18:17 hmmmm it goes in the errorstream 18:17 hmmmm exactly two months ago i committed this piece of art https://github.com/minetest/minetest/commit/a2e1b0fc7fa9a470f645e130115ce14a89aa2e02 18:18 hmmmm because we were experiencing that android "buildings getting splattered with grass" fiasco 18:18 Krock okay. I instantly see the change when skipping the double-generation 18:18 Krock disabled it and no corruptions happen 18:18 hmmmm krock: yes, but how about after reverting that patch I posted earlier? 18:20 hmmmm you see, the whole issue is that the emergethread would get block emerge requests queued up in the same chunk 18:20 Krock aagh. I won't build that stuff again, the problem has been solved 18:20 hmmmm so you get initBlockMake called and it determines the bounding chunk in the imaginary 'grid' that the mapblock being requested belongs in 18:21 hmmmm if it determines that it's not in the database and needs generation 18:21 hmmmm so makeChunk generates that, runs the mods on it, sets it as modified so that clients will get it sent on next step 18:21 Krock hmmmm, last time I reverted to 3fa4f78 and compiled it, it had the same bug as HEAD 18:22 hmmmm but in the queue also exists another mapblock with the same exact bounding chunk 18:22 hmmmm and it gets a duplicate chunk area 18:22 hmmmm makeChunk runs through that 18:22 Krock oh 18:22 hmmmm because none of it is CONTENT_IGNORE, it won't bother placing any nodes though 18:23 hmmmm but through some combination of unforseen logical conditions it splatters the area with dirt nodes and then grows grass, because that doesn't check for content_ignore 18:23 hmmmm so to counter this i added in a little optional thing in voxelmanipulator to not blit back a chunk if it had already been generated 18:24 hmmmm the only blitback this was used in was in the initial blitback for after map generation 18:24 hmmmm butttt 18:25 Krock s 18:25 VanessaE but? 18:25 hmmmm you see, stupid me never considered that i would still need to blit back to already generated blocks (bordering ones specifically) for trees and stuff 18:25 hmmmm so last week I start up minetest and I see trees getting cut off and everything 18:25 hmmmm and i figured out through some commit bisecting that I actually caused that bug 18:25 hmmmm whoops 18:26 Krock yes but wouldn't it be a problem if the bordering chunks contain cut-off stuff, too? 18:26 hmmmm so I just disabled the part where I don't blit back already generated blocks by removing that little ", false)" from emergethread 18:26 hmmmm at this point, the code is logically the same as it had originally been before 2 months ago 18:26 hmmmm krock, what? 18:27 Krock sorry. Only understood the half of this all. therefore maybe an unlogical question 18:27 Krock I'll take some time to read it again 18:28 hmmmm so when you get two duplicate minp/maxps in your mapgen, it really has already been generated 18:29 hmmmm but 18:29 Krock yes. I also can see this in some cases in the client, when trees get removed and new appear 18:29 hmmmm you're using singlenode 18:29 hmmmm aha 18:29 hmmmm singlenode doesn't check for CONTENT_IGNORE before placing the single node 18:29 hmmmm thanks, celeron... 18:29 Krock lol 18:29 hmmmm he's the one who made singlenode btw 18:30 Krock well, singlenode isn't that complicated to make :3 18:30 VanessaE hmmmm: I believe the MG mode uses singlenode also... 18:30 VanessaE mod* 18:31 hmmmm hmm no that can't be it 18:31 hmmmm singlenode does wisely only overwrite if content ignore 18:31 Krock bingo. some chunks are 64x64x64 and some 96x96x96 at chunksize = 4 18:32 hmmmm oh god 18:32 hmmmm you set a custom chunksize and you expected no problems? 18:32 hmmmm that wasn't notable enough to mention? 18:32 Krock never had problems 18:32 hmmmm are you kidding me 18:32 * hmmmm falls out of chair 18:32 * Krock catches hmmmm and puts him back on the seat 18:33 hmmmm so there are math problems with calculating the chunk area 18:33 hmmmm that's celeron's original code btw 18:33 hmmmm i didn't modify that 18:33 hmmmm i always found his division-type code suspect 18:33 hmmmm fitting things into other things 18:34 Krock chunksize = 5 + skip_overgen causes the same problems as all other sizes. 18:35 hmmmm odd 18:35 RealBadAngel btw, offtopic, it seems when one modifies a mesh for a mapblock, in any way, a few copies of it are still displayed at the same time 18:35 RealBadAngel and slowly get deleted over time 18:35 hmmmm hmm 18:36 Krock what? copies when modify? 18:36 RealBadAngel place a node, it will cause mesh to regenarate 18:36 RealBadAngel for whole block 18:36 RealBadAngel then engine displays two meshes for one block for a moment 18:37 blaise hrmm 18:37 blaise I feel like maybe a mod named tesla would be good 18:37 blaise XD 18:37 RealBadAngel ive added node highlighting and with fast movement i can see up to four copies of the mesh 18:38 RealBadAngel i can see that because each mesh hold own copy of highlighted node position (praise the idea of making global variables be passed as locals) 18:39 RealBadAngel so i see trails of highlighted nodes 18:40 Krock RealBadAngel, but that's not the case for the selection box 18:40 hmmmm Krock: just to verify, you ARE using current HEAD right? 18:40 Krock hmmmm, yes. 18:40 Krock hmmmm, compiled from scratch 18:40 RealBadAngel we are not talkin bout selection boxes now, we are talkin bout meshes 18:41 Krock but, mm.. I thought too far 18:41 RealBadAngel if engine is really displaying multiple copies of meshes for the very same block i say we have a serious problem 18:41 hmmmm we already have serious problems 18:42 RealBadAngel that can be main reason for fps hitting the ground for built up worlds, like spawn areas on VE servers 18:42 RealBadAngel hmmmm, so i just found another :) 18:42 hmmmm wait a minute wait a minute wait a minute 18:42 Krock so 3 minutes in total? 18:43 hmmmm ah nevermind 18:43 Krock RealBadAngel, if I've read this correctly, then there's a big potential to speed up minetest 18:44 RealBadAngel it looks like 18:44 RealBadAngel first of all, if something is modified, new mesh is created, no matter nature of modification 18:45 RealBadAngel i guess thx to threading kind ask to delete old one goes somwhere 18:45 Krock john_minetest, stahp. first fix all the bugs :/ 18:45 RealBadAngel and it takes up to a few seconds to delete old mesh 18:46 hmmmm krock, there are multiple people working on minetest you know 18:46 RealBadAngel within those few seconds 3-4 copies of the mesh are being displayed at once 18:46 hmmmm and people are not equally proficient in every area of it 18:47 Krock *nods* 18:47 RealBadAngel that means not only mesh as itself, it means lighting calcs etc 18:47 RealBadAngel thats why digging is propably causing so heavy spikes 18:47 meldrian hi john_minetest 18:47 Krock RealBadAngel, means, one mesh per crack-step? 18:48 RealBadAngel yes, it works like that, each update is causing new mesh to generate 18:49 RealBadAngel but since crack is made as animation and not continuos you wont be able to see that before 18:49 * Krock starts wonder how much gets calculated double or twice for a single animation 18:49 * Krock hides 18:49 kahrl RealBadAngel: but why doesn't this cause nodes to linger after you've dug them? 18:50 RealBadAngel i have no idea 18:51 RealBadAngel i only see what happens with my own code, each instance of a mesh holds own position of highlighted node 18:51 RealBadAngel and after update has a few seconds time to live (not always, but with a great chance) 18:52 RealBadAngel its impossible for my code to highlight a few nodes at once, so copies used at the same time are the only answer 18:55 RealBadAngel also the fact the copies (old positions) dissapear after some time also confirm that 18:59 RealBadAngel huh, the effect is more visible when i record a vid 19:01 RealBadAngel when CPU has more to do 19:01 RealBadAngel that is threading related bug 19:04 kahrl RealBadAngel: is the node highlighting code online so I can try to reproduce this bug? 19:04 RealBadAngel i will make a repo with that 19:04 RealBadAngel gimme 15 mins or so 19:04 OldCoder Yo 19:05 RealBadAngel and i think i know whats responsible for that already 19:05 RealBadAngel void Client::addUpdateMeshTask(v3s16 p, bool ack_to_server, bool urgent) 19:05 OldCoder Who rang? 19:05 RealBadAngel this is called with node position v3s16 p 19:06 RealBadAngel and later on added and compared by p value 19:06 RealBadAngel so calls to the very same block are treated as separate ones 19:07 kahrl RealBadAngel: nah, p is a blockpos 19:07 RealBadAngel rly? so crack is abusing that 19:07 kahrl where? 19:08 RealBadAngel http://pastebin.com/xVmGgpuZ 19:08 kahrl that's addUpdateMeshTaskForNode, not addUpdateMeshTask 19:09 RealBadAngel ah right 19:10 RealBadAngel https://www.youtube.com/watch?v=qrOPvnLSdQo&feature=youtu.be 19:10 RealBadAngel ^^ vid on the effect 19:13 RealBadAngel kahrl, gonna upload it now, brb 19:14 kahrl alright 19:20 VanessaE RealBadAngel: weird effect - and the color continues to cycle even after the mesh loses "focus" too. very weird. 19:20 RealBadAngel mesh animation is still working 19:20 RealBadAngel so there are copies of it working for sure 19:23 RealBadAngel kahrl, lemme just compile the rebased one to make sure if its complete 19:24 Wings Aye peeps 19:24 kahrl RealBadAngel: from the coordinates in the video, I suspect these are actually different mapblocks 19:24 Wings Oops wrong one 19:24 Wings XD 19:24 Wings I was supposed to be in inchra 19:24 Wings Not this one 19:25 Wings XD lolz oops 19:25 Wings Bye 19:25 kahrl RealBadAngel: you can see that the effect only happens when you cross a certain boundary (I suspect the mapblock boundary) 19:26 RealBadAngel no because i was able to get up to 4-5 highlighted nodes at once 19:26 RealBadAngel not just two of them 19:27 kahrl yeah, if you're at the corner of a mapblock (ignoring the Y axis) then there are 4 mapblocks next to each other 19:27 RealBadAngel i will check that shortly, compiling before making a branch with the code 19:28 kahrl (and if not ignoring the Y axis, there can be 8 mapblocks touching a single vertex) 19:29 VanessaE kahrl: I saw in that video where some nodes were highlighted in a checker pattern 19:29 VanessaE as in, 19:29 VanessaE - X - 19:29 VanessaE X - X 19:30 VanessaE later the interrim blocks were being highlighted. that would seem to rule out mapblock boundaries? 19:30 VanessaE interim* 19:30 kahrl VanessaE: timecode? 19:31 VanessaE kahrl: about 30 seconds in, or I thought I saw it before 19:31 VanessaE now I don't :P 19:31 RealBadAngel http://pastebin.com/UZhfUns4 19:31 VanessaE however I do see multiple nodes highlighting in a pattern at that time point that would seem to rule out a mapblock boundary anyway 19:32 RealBadAngel i just placed 3 nodes, it happens when i do highlight 1) and 2) 19:32 RealBadAngel never happens when trying to highlight 2) and 3) 19:32 kahrl ah, so indeed mapblock boundaries 19:32 VanessaE ok I'm wrong then :P 19:33 kahrl RealBadAngel: did you add the update mesh task with the "urgent" flag set? 19:33 RealBadAngel yes 19:33 kahrl also for removing the highlight? 19:34 RealBadAngel http://pastebin.com/Fz3UAJPY 19:34 RealBadAngel checking if new pos is same as old one was removed here 19:34 kahrl looks good. hmm... 19:35 RealBadAngel i made it very same way as sending crack pos is done 19:36 RealBadAngel hold on a few minutes, compiling is almost done so you will have whole code 19:38 RealBadAngel https://github.com/RealBadAngel/minetest/commit/2f258d444bac20f81882e0b97dea403a2628764b 19:38 RealBadAngel this comes together with tile switching, i had no time to separate the code, it is done in the very same places 19:45 kahrl lol git fetch is stuck downloading the main menu music :P 19:46 kahrl ok there we go 19:51 RealBadAngel kahrl. got the code? 19:51 kahrl yes 19:52 kahrl I can reproduce the bug 19:53 RealBadAngel go to the positions i pasted before if you can 19:53 RealBadAngel http://pastebin.com/UZhfUns4 19:54 kahrl well I'm at x=z=0 but that shouldn't matter 19:54 RealBadAngel i noticed lag happens when you move from node 2 to 1 19:54 RealBadAngel but never when 1 to 2 19:55 RealBadAngel thats kinda weird 19:59 kahrl that could be because addUpdateMeshTaskForNode adds the neighbor mesh in certain cases 20:00 RealBadAngel so we do have two blocks, a) for node 1), b) for node 2) 20:01 RealBadAngel why 2->1 triggers that and 1->2 dont? 20:01 kahrl I have an idea but let me test first 20:01 RealBadAngel ok 20:02 RealBadAngel btw, how do you like the effect? you can see it ingame now 20:10 kahrl well it's bit hard to say with the bug :P 20:11 kahrl I think it's a bit cheaty because you can use it to map out dark caves 20:12 VanessaE kahrl: given how easy it is to make a torch, that's not much of a cheat :P 20:12 kahrl not in a challenge map with no trees ;) 20:12 VanessaE heh 20:13 kahrl but yeah for building I could definitely see myself using it 20:15 kahrl the effect still needs to be tweaked a bit of course (less tint IMO, darker on dark nodes, no "jump" in the animation) 20:16 RealBadAngel ofc 20:16 RealBadAngel instead of n multipiers n*PI have to be used 20:17 kahrl aha! RealBadAngel: try https://gist.github.com/kahrl/4ee8911c341f7158343c 20:18 VanessaE ahhh 20:18 VanessaE even *I* can read THAT 20:18 VanessaE :) 20:19 kahrl heh 20:23 RealBadAngel damn :) 20:24 RealBadAngel it works 20:28 RealBadAngel kahrl, but about "darker on darker" nodes, no go with the approach, it is not aware of the textures used 20:29 kahrl true 20:29 kahrl I guess looking at the vertex colors it the best compromise 20:30 kahrl so textures won't affect it but light will 20:30 kahrl is the* 20:43 MinetestBot john_minetest: zeg9 was last seen at 2014-06-11 18:45:28 UTC on #minetest 20:51 ShadowNinja ^ And it's very slow. 20:52 ShadowNinja john_minetest: I think Zeg9 made one with a modified Minetest client. 23:37 CreativeBlob HI GUYS