Time Nick Message 00:45 * Sokomine tries to pacify Fixers doom bosses 01:03 Fixer Sokomine: fail, brutality intensifies* (*tom(m)or(r)ow) 01:04 Fixer did not think doom can be that much fun in brutal doom 05:04 Megaf https://xkcd.com/1692/ 05:10 Megaf Im impressed https://www.youtube.com/watch?v=bl7WljhLa7Y 05:10 Megaf !title 05:10 MinetestBot Megaf: Einstein the bird - YouTube 06:28 benrob0329 (off) Space cadet keyboard! https://www.youtube.com/watch?v=oDozftThFMw 12:54 Krock "template < parameter-list > requires constraint function-declaration (2) (since C++20)" 12:54 Krock > since C++20 12:55 Krock duh.. if they already have that ready for the next standard, the better :) 14:44 Megaf Morning 14:45 twoelk o/ 14:47 Megaf ok, I just updated my Debian Jessie for the first time in 3 years 14:47 Megaf reboot time 14:47 Megaf hopefully it will boot 14:47 Megaf That Jessie instalation is so old that it was a Jessie/Sid 14:49 Megaf megaf@BigSam:~$ uname -a 14:49 Megaf Linux BigSam 3.16.0-4-amd64 #1 SMP Debian 3.16.51-3 (2017-12-13) x86_64 GNU/Linux 14:49 Megaf quite recent build of the kernel 16:25 Krock Calinou, smooth yaw: http://krock-works.16mb.com/u/rotation-2017-12-23_17.21.58.mp4 16:25 Krock (not done with a low server step - using the SmoothTranslator) 16:36 Megaf Random people poping in and out of my servers channel... Some saying topic doesnt contain enoug information... 16:36 Megaf How dare them criticize my topic?! 16:36 Megaf (That was irony) 16:54 Krock Megaf [...] Some saying topic doesnt contain enoug information... 16:54 Krock perhaps only one. me :3 16:54 Megaf :) 16:57 Calinou Krock: yay for 60 FPS video 16:57 Calinou (what hardware do you have now?) 16:57 Calinou anyway, this is pretty nice 16:58 Calinou it could be even better if linear/angular velocity was computed and sent, so you could make use of Hermite interpolation (instead of linear) 16:58 Krock Quadro FX 4600 / Core2 Quad CPU Q6700 running with low-res texture pack and no shaders 16:58 Calinou but that's much more complex 16:58 Calinou so linear is fine for now 16:58 Krock yes, everything can be overengineered :P 16:59 Krock the reason for the 30 FPS videos most of the time is basically because my internet isn't too fast and the benefits are only there if it's about eyecandy 17:02 Krock next step will be extending set_look_horizontal and the camera with cinematic mode so it doesn't rotate from 359° all the way down to 0° 17:02 Krock these wrapped numbers really cause more work than they should -.- 17:04 Megaf I like when devs use old hardware 17:04 Megaf it teaches them how to code 17:04 rubenwardy lol 17:04 Megaf too easy coding for newer hardware, you just dont care 17:04 rubenwardy * teaches them to learn non-framework things 17:05 Megaf a you want to get me impressed. Show me MT running on Windows 2000 on an AMD K6-2 around 300MHz with 128 MB of RAM 17:06 Megaf that would show some skill 17:06 Megaf and it is very doable because irrlicht, it just demands skill 17:06 Krock the main reason for my old/low-end hardware is re-using from other people it as long it works. Production and recycling takes way more energy than use afterwards 17:06 Krock it don't run on 2000 due zlib 1.2.8 17:07 Megaf so just dont use that lib 17:07 Megaf or use older version 17:07 Megaf for example, no need for freetype and gettext 17:07 Megaf theres libpng and libjpg, drop support for one of the formats 17:07 Megaf Id say drop png and use just jpg 17:08 Megaf again, find solutions, show me your skill and cleverness 17:08 Megaf if you think development has became boring is because you are not after challenges 17:12 Megaf and for the K6, I said Windows 2000 because I don't think you can run Linux there 17:12 IhrFussel Last-minute Christmas shopping, see ya later^^ 17:12 * twoelk runs mt on vista - sometimes 17:12 Megaf maybe if you have a AGP motehrboard and use a Radeon 9xxx or nvidia 6xxx 17:13 Megaf 3 years ago I was running MT on Windows XP with integrated Intel GPU and Pentium 4 in the office 17:14 Calinou Id say drop png and use just jpg 17:14 Calinou JPEG for pixel art is a recipe for disaster :P 17:15 Calinou by the way, the proprietary NVIDIA driver is dropping 32-bit support next year 17:15 Calinou (on Windows and Linux, starting from 390.xx) 17:15 Megaf Calinou: why you say that? 17:15 Megaf about jpg 17:15 Megaf jpg files are smaller 17:15 Megaf maybe we should just use jpeg2000 17:16 twoelk being small is the only interesting feature of jpg 17:16 twoelk else they are crap 17:20 Megaf lets just use uncompressed bitmap files 17:21 Krock Megaf, you can run Linux pretty much everywhere. Just the 32-bit builds are slowly disappearing 17:24 rubenwardy loooool 17:25 Megaf "Now, the VRAM consumed by your images will roughly be width * height * bitdepth for each image loaded in VRAM. There are a couple of things to notice here:" 17:25 Megaf "The width and height in which your images are stored in VRAM will not necessarily match the ones of your original image. Some GPUs can only handle textures with sizes in powers of 2, so your 320x240 image may actually be stored in a 512x256 space in VRAM, with the unused memory effectively wasted. Some times you're not even allowed to load textures with sizes that are not powers of 2 (like in GLES 1.1)." 17:25 Megaf That ^ 17:25 Megaf -> https://gamedev.stackexchange.com/questions/48304/which-image-format-is-more-memory-efficient-png-jpeg-or-gif 17:25 Megaf !title 17:25 MinetestBot Megaf: graphics - Which image format is more memory-efficient: PNG, JPEG, or GIF? - Game Development Stack Exchange 17:25 rubenwardy err 17:26 Calinou Megaf: old GPUs don't support NPOT textures indeed 17:26 Calinou (< 2007 or so on desktops) 17:26 rubenwardy because JPEG is lossful, and you don't want to losing quality on single pixels is bad 17:26 rubenwardy also 17:27 rubenwardy there's a difference between an over reliance on VMs like JVM and the Node ones 17:27 rubenwardy and supporting ridiculously old hardware 17:27 rubenwardy which Minetest shouldn't do 17:30 Megaf so, should we drop support to JPG? 17:31 Megaf Perhaps we should make PNG, 32x32, 16 bit the default and enforce mods to use that 17:31 Megaf also 17:31 Megaf "But no matter what you do, the most important factor when considering the amount of VRAM your game uses, is the amount of images you have in VRAM at a given time. This is the number you most likely want to keep as low as possible if you want a good performing game. Loading and unloading textures into VRAM is expensive, so you can't just load each image whenever you're going to use it. You must find a balance between preloading the images you will 17:31 Megaf most likely use, and unload them when you're sure you're not going to use them anymore. Doing this right is not trivial, and you have to think of your own strategy for your particular game." 17:31 Calinou not gonna happen 17:33 Megaf Cool. Now I will write a list of changes I believe would be good to Minetest and for sure will be ignored. 17:38 Megaf I wonder if Writer supports md 17:41 Megaf will give a try to retext and formiko 17:45 Krock rubenwardy, by supporting old hardware and trying to make it fast as possible opens the possibility to add more eyecandy for those who want it 17:45 Megaf ReText is quite cool for writing markdown 17:46 rubenwardy Krock, so slow hardware 17:46 rubenwardy not old hardware 17:48 Krock oh I see 17:48 Megaf http://git.megaf.info/Megaf/TODO/src/master/TODO.md 17:52 rubenwardy "Drop support to anything but sqlite3" 17:52 rubenwardy loool 17:53 Megaf Ive done it in two forks I made in the past already, totally worth it 17:54 Megaf Formiko editor is the best in GTK and ReText in Qt 17:55 Krock Megaf, a few server owners wouldn't like that at all 17:56 rubenwardy yeah, that's not being merged XD 17:56 rubenwardy no real benefits at all 17:56 Megaf That's what I said possible fork. It's mostly for my server really 17:56 rubenwardy just don't compile the files using a CMake option 17:56 Megaf rubenwardy: what's the point in maintaining code and checking for thins you dont use? 17:57 Megaf And whats the point in just leaving dead code behind? 17:57 rubenwardy just don't compile the files using a CMake option 17:57 Krock "Gogs Version: 0.11.34.1122 Page: 0ms Template: 0ms" Hats off! That's very fast :) 17:57 rubenwardy and the code is used 17:57 Megaf Go language is very fast :) 17:58 Megaf Gogs is written in Go. No credits to me on that. Just using pre built version from Gogs and Go from Debian repos 17:58 Megaf Tho I started using Gogs in my Raspberry Pi first gen, then everyting built by me an it was just as fast 17:59 Megaf Thou here... "© 2017 Gogs Version: 0.11.34.1122 Page: 34ms Template: 4ms" 18:00 Megaf Krock: thats a good browser benchmark :P 18:01 Megaf Chromium -> Page: 22ms Template: 1ms 18:02 * Megaf sips his wine 18:06 Megaf do we support wav files? 18:07 rubenwardy yes 18:07 rubenwardy also, to unsupport a file format completely you'd need to fork Irrlicht 18:08 Megaf not messing with that 18:08 Megaf what about compressed files? What do we use zlib and bz2 for? 18:08 twoelk map blobs? 18:09 twoelk zlib that is 18:09 Megaf bz2 is a requirement too 18:09 Megaf what for? 18:11 Megaf http://git.megaf.info/Megaf/TODO/src/master/TODO.md 18:11 Megaf no reference to audio files 18:11 Megaf We could drop OGG support, but I dont thing there would be any benefit at all 18:11 Megaf well, we would have less dependencies 18:12 rubenwardy Irrlicht again 18:12 Megaf Does irrlicht depends on OGG? 18:13 twoelk why drop ogg? 18:14 twoelk don't most mods use it? 18:14 Megaf Actually I thinking more in shipping ogg libs 18:15 Megaf I think we could built in the most elementary libs, we already do with jsoncpp, gmp and lua 18:15 Megaf I think sqite3, zip and ogg should be there too 18:15 twoelk i'd rather bloat it a bit more and add a nice midi support :P 18:15 Megaf +1 18:16 Megaf It would be awesome a FPS subgame with weapons runnning in CSM 18:16 Megaf and a nice DOOM/Duke3D inspired midi soundtrack 18:16 Megaf ^^** Super idea for subgame ** ^^ 18:17 * twoelk admires the tree that stands straight at last 18:18 twoelk now where did I leave the lights? 18:20 Megaf define lights 18:21 Megaf I wonder if I should change to 16x16 pixels textures 18:21 Megaf that would be very good for Android phones 18:21 twoelk putting up the xmas tree :-D 18:21 Megaf we should make Minetest with a 32 bit CPU at 800 MHz, 512 MB of RAM and 32 MB for video in mind. That should be our target. 18:22 Megaf And that's totally fine for something based on Irrlicht 18:25 benrob0329 Megaf: then it would look worse, because the shader renderer would still be unoptimized 18:26 Megaf I'm not using the current shaders either 18:26 Megaf hate them 18:26 Megaf I mean 18:26 Megaf I believe they are not ideal 18:26 benrob0329 Waving water sucks 18:27 rubenwardy I'd rather have a larger draw distance 18:27 benrob0329 Waving leaves is fine (mostly) and waving plants 18:27 rubenwardy which could require more caching so more memory 18:27 benrob0329 Shaders != slow 18:27 rubenwardy shaders should be faster than doing everything in CPU 18:27 benrob0329 ^^ 18:27 rubenwardy but our shaders suck 18:27 benrob0329 Exactly 18:28 Krock Megaf, look at other software - many of them have more dependencies than Minetest 18:28 benrob0329 IMO we should drop the fixed renderer and fix the shader renderer 18:28 rubenwardy Do LVMs no longer generate the map? 18:29 * twoelk wonders where to get a shaders expert 18:30 ThomasMonroe LMD on the forums is pretty good with them 18:30 benrob0329 Godot dev team :P 18:30 benrob0329 Calinou: who can we hire? :D 18:30 Megaf Krock: I know, I think MT is amazing regarding that already. That doesn't mean we can't make it even better 18:31 ThomasMonroe Blender dev team is pretty good with shaders too 18:31 benrob0329 I'd honestly put money towards fixing one renderer, and dropping the other. Along with adding some useful effects 18:33 Megaf we just need ONE mapgen, ONE db, ONE renderer 18:33 Megaf why have many broken things instead of ONE working thing? 18:33 Megaf v5 was broken, then made v6, it was broken, no mountains, they made v7 18:34 Megaf That's alright! Just drop the old ones! 18:34 Megaf and just call the mapgen, wait for it... mapgen! 18:35 Krock Megaf, and one setting with one node per world 18:36 Krock v7 wasn't good enough, thus valleys and carpathian 18:36 Megaf IMHO v6 would be the perfect mapgen if it had mountains 18:36 Megaf the other mapgens you can clearly see the noise functions 18:36 Megaf and sine waves and stuff 18:36 Megaf its depressing 18:37 Megaf v6 is the most organic 18:37 benrob0329 I quite like Valleys 18:37 Megaf valleys is too mathematical 18:37 benrob0329 The river system makes boat travel possible 18:37 Megaf benrob0329: well, in my server I created/dug several miles of canals and rivers interconnecting things 18:38 Megaf but I agree, rivers are god 18:38 Megaf good 18:39 Megaf so we should just add rivers and mountains to v6 18:39 benrob0329 Add mountains, cliffs, plateaus, and rivers to v6 18:40 benrob0329 And the extra books from v5 and v7 18:40 benrob0329 *bioms 18:41 Krock sine waves? are you talking about the "eased" cosine interpolation? 18:41 Krock raw perlin noise doesn't have any sine waves, Megaf 18:42 Megaf Krock: I think v5 has sine waves 18:42 Megaf anyway, I can see waves 18:42 Megaf and math functions on all mapgens but v6 18:42 Krock yeah, most likely due that interpolation 18:43 Krock it makes the terrain smoother 18:43 rubenwardy v5 isn't broken 18:43 rubenwardy well 18:44 rubenwardy the version of v5 we have isn't 18:44 rubenwardy it's more CPU intensive due to use of 3d noise 18:44 Krock eeks. Apparently I've now got a greenish name on the forums. Thanks, I guess. 18:44 rubenwardy adding more renderers doesn't change anything 18:44 rubenwardy err 18:44 rubenwardy the additional renderers doesn't change anything 18:44 rubenwardy Irrlicht handles that 18:45 rubenwardy different world formats are needed for different reasons 18:50 Megaf yep, I'm sure irrlicht already have a nice rendered that we are just not using 19:01 benrob0329 Does anyone know how I could generate a dithered pattern? Like draw every other pixel type thing 19:27 IhrFussel Was someone else able to confirm that loading an *unloaded area that has been generated before* takes up to 1+ secs? Or is my dedi HDD defective? 19:28 paramat by teleporting? and do you mean the server stalls for that time? 19:28 IhrFussel By area I mean: Player teleports to the middle of an unloaded area and it takes up to 1.5 secs for the chunks to display correctly 19:29 paramat ah, chunks displaying is not server lag 19:29 paramat yes it takes a while for all the meshes to be generated 19:29 IhrFussel It actually does lag the server as long as not all mapblocks are displayed client side 19:30 paramat hmm 19:30 IhrFussel Only generation is its own thread, loading/unloading happens in the main thread no? 19:30 paramat that surprises me, i would have thought meshgen was separate or at least done in small batches 19:31 paramat someone else will know 19:32 IhrFussel My server has 30+ cities that players can *freely* teleport to at any given time... and I get the feeling that the server lags wayy more when more player hop between cities 19:34 IhrFussel Another problem is that more advanced nodes take more resources to load compared to default ones or water 19:37 paramat 'feeling' isn't proof, you need to test to be sure 19:38 IhrFussel paramat, I tested it in 0.5.0-dev SP with mapblock_limit set to 50, generated some small area, then tp far away, generated another 50 blocks to make sure the first area gets unloaded from RAM and tp back to that first area and it took 1.2 secs 19:39 IhrFussel And /status showed max_lag=1.2 19:40 IhrFussel Since the default mapblock_limit is 5,000 I think it's not really possible to test it wothout setting that to something low 19:41 IhrFussel In SP* 19:48 IhrFussel No wait, mapblock_limit applies on servers as well I think so it should be the same as SP 19:57 IhrFussel But I want others to test it under the exact same conditions too and report their results... I find it highly unlikely that my dedi's HDD AND my local PC HDD have the same fault but nonetheless I'd like to see other people test it 19:58 paramat it needs to be tested in MT with no mods, to rule those out 19:58 IhrFussel I oonly have MTG on my local PC ... do you think MTG can cause 1+ secs lag? 19:59 paramat you can test it with mapblock limit 5000 by just exiting and re-entering a world 20:01 IhrFussel paramat, going back to the main menu will not unload the mapblocks from RAM AFAIK 20:02 * benrob0329 just decides to write a script using love2d for generating the textures 20:02 IhrFussel Only re-opening the game will clear the RAM ... cause when I first load my SP world with just default nodes max_lag is 1.0-1.4 secs, after exiting to main menu and re-loaing the world without closing the client it only takes 0.4 secs 20:04 IhrFussel But yes I can still reproduce it that way...and you have to imagine that's with DEFAULT MTG nodes...now think about what kind of lag could happen with custom nodes that may even contain meta 20:07 IhrFussel Or like in my case a city to be loaded... a city has quite a bit more nodes inside the chunks compared to regular terrain 20:18 paramat i might be wrong but exiting a world should clear world data from the server 20:19 paramat all chunks have equal number of nodes 20:22 IhrFussel paramat, you misunderstood....a city has wayy more nodes placed in the chunks compared to regular terrain... a city has lots of custom buildings PLUS the regular terrain surrounding it 20:23 IhrFussel Another way to say it is: City map areas have less air nodes 20:28 IhrFussel paramat, RAM usage stays the same when I exit to main menu 20:44 paramat yes i know =) i was being pedantic, air is a node 20:44 paramat maybe ram is only cleared when entering a new world 20:49 IhrFussel Nope, only part of the RAM gets unloaded, I switched between 2 worlds a few times and RAM usage is now higher than when I only open 1 world 20:50 IhrFussel Looks like MT keeps mapblocks of both maps in memory 21:07 Megaf hi paramat 21:08 Megaf so, v6.1 it is? 21:08 Krock are you sure it's the mapblocks and not meshes? 21:09 Megaf IhrFussel: loading mapblocks were always painfuly slow 21:09 Megaf is that a bug? 21:12 IhrFussel Krock, does MTG load any meshes in normal terrain? If no then not 21:12 Krock IhrFussel, the player. and then there are also the textures 21:13 IhrFussel Megaf, yes and now I'm pretty sure the slow loading of mapblocks and thus hanging the server thread is the main issue on my server and explains why it still lags even with no custom mods 21:14 Megaf IhrFussel: I experience that since the begining 21:14 IhrFussel My server seems to have more teleport options compared to 99% of other servers..hence why other server owners barely notice it 21:14 Megaf when mapblocks are loading the whole server hangs 21:14 Megaf no matter the hardware settings 21:15 Megaf tried on Raspberry Pis, ODroid, XEONs wih SSDs, Phenom II x4 at 3,2 GHz and fast HDDs and my current dedicated 21:15 Megaf all show the same thing. All hangs when unloaded areas are loading 21:16 IhrFussel Megaf, so you can definitely confirm it with unloaded but generated areas too? Then it must be an engine issue 21:16 Megaf IhrFussel: confirmed 21:17 Megaf IhrFussel: we can test on my server right now 21:17 Megaf !up 5.196.88.227 30003 21:17 MinetestBot 5.196.88.227:30003 is up (10ms) 21:18 IhrFussel Of course you will NEVER see that lag if your server doesn't have travelnets/teleport potions/pads/other random teleport commands ... the server doesn't lag when you just walk with regular walking speed of 4 in 1 direction cause it only needs to load mapblocks for a few rows 21:19 IhrFussel But if the server has to load several rows 360° around the player it's overkill 21:23 paramat hm i can understand it may be seen as the priority. MT was probably designed not thinking about teleportation 21:24 IhrFussel Megaf, I just joined...did your server spike to 1.9 secs cause of that? 21:27 IhrFussel Megaf, maybe it helps your server to set the block_send_distance lower? 21:28 Megaf it is set to 4 already 21:28 Megaf and 21:28 Megaf Floating point exception 21:28 Megaf reglnx@Aluminium:~$ 21:28 Megaf I just got that in the client ^ 21:28 Megaf and client eats up 2.1+ GB of RAM 21:29 Megaf I miss the time when Zeno was hunting memory leaks 21:29 Megaf now a days people just want to add more stuff and leave the leaks behind 21:34 IhrFussel Well I am thinking about how I can make my server "servicable" again... players right now use teleportation methods extensively ... I could try block_send_distance of 2 21:34 Megaf say sorry and tell them to have patience, thats how Minetest it is 21:34 IhrFussel Not sure how much it would help though when the whole area around the player is unloaded 21:37 VanessaE IhrFussel: add maptools if you haven't already and put admin world anchors next to all of the most common destinations 21:37 Megaf I set my server to keep items loaded for 24 hours 21:49 Megaf 2017-12-23 21:49:28: [Server] Item "darkage:sandstone_cobble" not defined at position (-106,30,-38) 21:50 Megaf I will never ever update my server again once I get it fixed 21:50 Megaf everything breaks 21:50 Megaf ALWAYS 21:52 Megaf wand we need to make boats turn faster 21:52 Megaf they accelerate way quicker than they turn 21:52 Calinou turn it up to 11… 21:52 Calinou …degrees per second 21:58 twoelk how much meta info does normal terrain contain? I would imagine all the nodes with extra info in params would make settlements somewhat heavier than normal terrain 22:18 paramat fresh terrain has no metadata 22:21 IhrFussel VanessaE, how is that supposed to help? 22:22 VanessaE IhrFussel: I assumed much of your lag problem is from the map taking forever to loadd 22:23 VanessaE anchors would ensure the surrounding land is kept loaded at all times 22:30 IhrFussel VanessaE, that could work for the larger cities yeah...but there are lots of travelnet stations per player and I can't mark all of those areas :/ 22:31 IhrFussel Maybe I should just set the unload_unused_data_timeout to something very high so that the chance is greater that a player visits a standby-area from RAM 22:33 IhrFussel I set it to 1800 which is 30 minutes ... maybe that improves the loading 22:44 IhrFussel Maybe I need to set max_simultaneous_block_sends_per_client higher too? 23:34 rubenwardy <_< https://i.rubenwardy.com/FjCLcdLmMI.png 23:45 IhrFussel max_simultaneous_block_sends_per_client << does "simultaneous" in this context mean "per step" ?