Time Nick Message 00:02 RealBadAngel rotfl 00:02 RealBadAngel replaced air with water 00:03 RealBadAngel no wonder mgv6 is still so slow 00:03 Jordach kahrl, i useb MGV7 as default now 00:03 RealBadAngel http://i.imgur.com/ZU3F75t.png 00:04 RealBadAngel its just doing the caves all the time 00:04 RealBadAngel without any height limits 00:05 RealBadAngel i made once patch for it but it was rejected, it was like more than yr ago 00:14 VanessaE I think it's time to revisit the idea of patching that out. 00:14 VanessaE there's zero reason to be generating caves above roughly 30m or so. 00:21 RealBadAngel mgv7 is not generating caves in the air 00:26 RealBadAngel ok, i know the reason 00:26 RealBadAngel if (!ndef->get(c).is_ground_content) 00:26 RealBadAngel continue; 00:26 RealBadAngel content air for mapgen is ground content 00:31 RealBadAngel if (!ndef->get(c).is_ground_content || c == CONTENT_AIR) thats enough to dont let it carve shit above ground 00:33 RealBadAngel https://github.com/minetest/minetest/blob/master/src/cavegen.cpp#L251 00:33 RealBadAngel any objections i will push the fix directly? 00:35 kahrl air is ground content? I guess someone took their biology class too literally :P 00:35 RealBadAngel mgv7 does the same check 00:36 RealBadAngel https://github.com/minetest/minetest/blob/master/src/cavegen.cpp#L559 00:39 RealBadAngel but i was wrong, the check alone is not enough for mgv6 00:39 RealBadAngel it is still doing the caves 00:46 RealBadAngel checking for air simply doesnt work, below theres also another check 00:46 RealBadAngel not working too, funny 00:50 nore_ Rba: air -> not ground content in builtin would be easier? 00:50 nore_ kahrl: that was when is_ground_content was defaulted to true... 01:01 RealBadAngel the check here simply doesnt work for some reason 01:02 RealBadAngel when i set it to if (1) caves are disabled just 01:02 RealBadAngel but when i do if (c == CONTENT_AIR) it makes the caves everywhere 01:03 RealBadAngel to be more funny i added there msg about skipping, it prints the message and does the caves 01:03 RealBadAngel wtf? 01:04 RealBadAngel if (c == CONTENT_AIR){ 01:04 RealBadAngel dstream<<"Skipping\n"; 01:04 RealBadAngel continue; 01:04 RealBadAngel } 01:27 RealBadAngel holy shit, i replaced cave air with stone and guess what i got :) 01:28 nore_ No caves? :) 01:28 RealBadAngel http://i.imgur.com/upjvf46.png 01:28 RealBadAngel rotfl 01:29 VanessaE turned caves into dirt eh? 01:30 nore_ Floatlands! 01:31 RealBadAngel btw, why the fuck ores are generated in the stone above the clouds? 01:31 RealBadAngel wasnt there a max height limit? 01:32 RealBadAngel http://i.imgur.com/avam3Km.png 01:33 VanessaE um, that shouldn't happen. 01:33 nore_ RBA: absheight flag 01:33 nore_ Mese will generate above 256 for example 01:33 nore_ s/256/128/ 01:34 RealBadAngel ok 01:41 Sokomine the cavegen and mudflow is still giving me trouble; is_ground_content is only a partial solution as the houses are built out of diffrent materials (which may be stone as well); only way to get around the cavegen seems to spawn multiple times? 01:41 RealBadAngel anyway, but why the heck the check even if workin (i mean finds content air) carves the caves as usual? 01:42 RealBadAngel while setting the condition to 1 disables it completely? 01:55 RealBadAngel lol, found the reason 01:55 RealBadAngel that dumb code was makin caves in CONTENT_IGNORE 01:59 nore_ lol 01:59 RealBadAngel but adding such check is not a solution, it effects with far less caves in underground 02:00 nore_ Wait, it makes caves *before* the rest is generated?? 02:00 RealBadAngel it looks like 02:00 RealBadAngel before even block is loaded 02:01 RealBadAngel it looks like it doesnt care if the block is loaded or not 02:01 RealBadAngel it just happily creates caves everywhere 02:01 nore_ Strange 02:02 RealBadAngel yup, with check on CONTENT_IGNORE it makes caves only in loaded blocks 02:02 RealBadAngel and i can see just a few caves underground 02:02 RealBadAngel and none above the ground 02:03 RealBadAngel i think the only solution is to limit the height 02:22 paramat RBA actually i disagree with the more saturated colours, sorry if i wasn't clear. also i noticed you've made EW faces brighter than NS faces? fairly sure the management want it kept unchanged, i did a comparison and now think EW darker is better :) 02:23 * VanessaE throttles paramat 02:24 paramat lol 02:25 paramat i'll test the pull request soon, not sure i have time right now, but within a day 02:25 VanessaE test it now :P 02:37 RealBadAngel RBA,colors are not more saturated 02:37 RealBadAngel actually levels in pull are lower than before 02:38 RealBadAngel top face is 1.0 instead of sqrt(1.3) 02:38 RealBadAngel why i wrote to myself? rotfl 02:39 VanessaE haha 02:39 RealBadAngel paramat, that was addressed to you :) 02:39 Exio go to sleep RealBadAngel! 02:39 RealBadAngel and about switching faces i dont like sides facing the sun being darker 02:40 RealBadAngel its against any logic 02:40 RealBadAngel btw, i have fixed those damn caves 02:41 RealBadAngel now just makin sure that they will be able to carve holes in the mountains or cliffs 02:41 paramat okay will test the pull to check colours 02:42 RealBadAngel later on i will make also overall brightness and contrast settings possible 02:42 RealBadAngel because i can clearly see no matter what i will code there will be always some1 to get the opposite 02:42 paramat i made comparison screenshots, with the sun in view EW darker actually looks better because you expect shadows towards you ... i'll get those images ... 02:43 RealBadAngel this is not really about shadows 02:43 RealBadAngel its about what face is being better lit 02:43 RealBadAngel and thats whole difference 02:44 VanessaE paramat: besides which, sooner or later there will be real shadows being cast by the sun and light sources. 02:44 VanessaE (in shaders anyway) 02:44 RealBadAngel not only 02:44 VanessaE you think you can do it without? 02:44 RealBadAngel it will work with and without shaders 02:44 VanessaE even better. 02:45 RealBadAngel lighting? ofc 02:45 RealBadAngel engine will do that for us 02:45 RealBadAngel http://i.imgur.com/T8wf2tp.png 02:45 RealBadAngel this is without shaders 02:46 RealBadAngel and faces shading too, btw 02:49 RealBadAngel paramat, and i havent saw a thing IRL that could cast a shadow onto itself ;) 02:50 RealBadAngel especially any cube ;) 02:50 paramat current EW darker > http://i.imgur.com/1enABtU.png ... NS darker > http://i.imgur.com/E4Rlrro.png ... both look okay but i perhaps prefer EW darker 02:52 VanessaE I can see what he's saying, RealBadAngel. 02:52 VanessaE this can only be solved by making the face lighting actually follow the sun. 02:53 RealBadAngel i can live with partially wrong solution better than with completely wrong one 02:53 VanessaE however, having EW faces be darker during midday is definitely wrong for sure 02:54 RealBadAngel like they said better to have at least one hand than none 02:55 VanessaE paramat: can we agree that EW faces being brighter than NS is not, itself, a reason to stop otherwise correct figures from going in? 02:56 * VanessaE especially wants the patch to go in because it finally fixes light sources and liquids once and for all 03:00 RealBadAngel i can revert faces switch if there will be more votes to do so 03:00 paramat its not all or nothing, RBA could just swap X face values for Z face values 03:01 RealBadAngel thats just what i did 03:01 RealBadAngel and you thought how thats done? 03:02 paramat i'm just keeping you out of trouble ;) i think celeron wants EW darker as is tradition for Minetest 03:02 RealBadAngel tradition of mt is having things buggy and upside down :P 03:02 RealBadAngel like those damn caves 03:03 paramat okay tested pull, faces shading looks okay, colours look perhaps okay ... 03:03 VanessaE "perhaps"? 03:04 RealBadAngel i dont fucking belive that yrs old issue is still here and eating CPU like nothing happened 03:05 RealBadAngel folks are blaming our mods to slow down the mapgen while it happily puts shitload of caves above the ground 03:07 RealBadAngel paramat, check out lightsources and liquids 03:07 paramat yeah colours are fine actually 03:08 RealBadAngel and btw, that pull makes it again faster 03:08 RealBadAngel first of all, all the top nodes are not boosted anymore so 1/6th calls went off 03:09 RealBadAngel 2nd, since color.b is the lightsource value theres no reason to multiply it 03:09 RealBadAngel so the call multiplies only r and g 03:09 paramat ah i see. i appreciate the more lightweight code 03:12 RealBadAngel and finally you can see real texture in the game 03:14 RealBadAngel it sounds like a shitload of features for such small commit ;) 03:19 RealBadAngel anyway, i will wait for more devs to check it and vote 03:19 paramat if you make EW faces the darker pair im happy with this pull (but i'm hardly an authority here) 03:20 RealBadAngel since you were the one to open issue about it i can do that right now to avoid argues with others 03:20 RealBadAngel but then you will have to close it ;) 03:26 paramat #1462 is closed, or do you mean #1546 03:31 paramat toodlepip! 04:14 RealBadAngel ok, here comes cavegen fix: https://github.com/minetest/minetest/pull/1554 04:14 RealBadAngel its still being able to carve caves in cliffs and mountains 04:15 RealBadAngel but cannot generate caves in the sky 04:22 RealBadAngel https://github.com/minetest/minetest/pull/1555 04:22 RealBadAngel this one adds main menu music 07:39 Zeno` I have no idea why I did this, but I suspect I just don't like unnecessary globals don't make sense (and the comments associated with them don't either... they've been there since MT was first uploaded to github): http://codepad.org/6McXOdQp 07:41 Zeno` All of the tests remain the same, *except* "Testing floating-point vector speed" which takes about twice as much time (which I guess invalidates the comment that they were global to minimise optimisations) 07:43 Zeno` err, from main.cpp btw 08:54 Zeno` RealBadAngel, ping 10:07 RealBadAngel Zeno`, yes? 10:09 Zeno` RealBadAngel, just wondering if I should squash https://github.com/minetest/minetest/pull/1543 10:09 Zeno` I was unsure because it, really, addresses two issues :3 10:10 RealBadAngel so 2 pulls? 10:11 RealBadAngel but then you will have to rebase before the 2nd 10:11 RealBadAngel just squash it 10:11 RealBadAngel it touches the very same code 10:14 Zeno` yep ok 10:14 Zeno` thanks 10:22 PenguinDad RealBadAngel: #1553 seems to only fix the liquid shading when shaders are disabled https://mediacru.sh/e8fbdbfee631 10:23 Zeno` that didn't work out so well, lol 10:25 Calinou I have the same problem, PenguinDad 10:27 RealBadAngel hmmz, thats pretty strange 10:30 Zeno` can't rebase now I've stuffed up 10:30 RealBadAngel does the water and lava work ok? 10:30 Zeno` oh well, close the merge requests I guess 10:31 RealBadAngel PenguinDad, ? 10:31 Zeno` I'll write them again in a few weeks I guess 10:31 PenguinDad Water and lava look ok :/ 10:32 RealBadAngel so whats that yellow thingy? 10:33 PenguinDad That green liquid is acid from carbone 10:33 RealBadAngel i see, i will check out whats wrong with it 10:33 RealBadAngel propably mod issue not the code 10:37 Zeno` or maybe I can clone 10:37 * Zeno` thinks 10:42 RealBadAngel i have to go now, bbl 10:42 Zeno` wait! 10:42 Zeno` lol 10:42 Zeno` oh, nvm... if it doesn't work it doesn't work 10:43 Zeno` done 10:44 Zeno` looks right to me anyway 10:48 Zeno` :( 17:50 luizrpgluiz hi devs and guys 18:54 luizrpgluiz devs I had an idea of a minecraft world generator that could work in minetest 18:55 luizrpgluiz because guys do not create the generator has the world Amplified minetest? 18:55 sfan5 tell us your idea 18:56 luizrpgluiz because guys create the generator has the world Amplified minetest? 18:57 sfan5 what? 19:03 luizrpgluiz the idea would be to put the generator amplified world of minecraft in minetest 19:04 sapier "put"? 19:04 sapier luizrpgluiz: you're aware that minecraft uses java while minetest is c++/lua? 19:07 luizrpgluiz I mean for example create a world generator amplified as in minecraft 19:08 sapier luizrpgluiz: please first define the term "world generator amplified" ;-) 19:09 luizrpgluiz seen the generator worlds in minecraft? 19:09 Amaz No. 19:10 sfan5 what exactly do you mean? 19:10 sapier No too 19:11 luizrpgluiz amplified means to have more mountains, huge caves, plains least 19:11 sapier ok coroutines and minetest are a no-go 19:12 PenguinDad why? 19:12 sapier coroutines implicit create their own lua stack therefore you can't use minetest api from a coroutine ... at least not all of it 19:14 Amaz luizrpgluiz, picture of what you are talking about please. And have you seen watershed? https://forum.minetest.net/viewtopic.php?f=11&t=8609 Might have some of the things you want in... 19:17 PenguinDad It's a pity that the full api isn't available in coroutines :/ 19:22 luizrpgluiz http://cdn6.portalprogramasnet.com/milbits/imagenes/Modo_amplificado.jpg 19:26 luizrpgluiz :) 19:31 luizrpgluiz this image is an example of biome amplified 19:33 sfan5 oh 19:33 sfan5 that 19:33 sfan5 you could achieve that by changing the perlin noises for mgv6 19:35 luizrpgluiz but how do I change it? 19:37 sfan5 look in map_meta.txt 19:39 celeron55 that won't give satisfactory results 19:39 celeron55 if you want that much vertical stuff 19:40 celeron55 v7 might be able to do it (not sure); a lua generator is the best bet 19:41 luizrpgluiz I loved the update can render the game in direct3d and software 19:41 sapier wow coroutines are quite heavy stuff ... even if we can fix the api issues they might not help for lots of small tasks 19:43 luizrpgluiz rendering in Direct3D already left with more fps here minetest 19:46 luizrpgluiz bad in Direct3D 8 does not open, but he usually opens at 9 20:46 VanessaE sapier: by coroutines, in this context it almost sounds like you mean C++ mapgen hooks being used outside of the engine core? 20:46 VanessaE e.g. by another C++ extension? 21:07 sapier no lua coroutines 21:08 VanessaE ah 21:09 sapier coroutines use voluntary preemption thus they don't use multicore cpu's but if they'd been lightweight enough they'd be usable for small things done pseudo parallel ... but seems like they're way to heavy 21:10 VanessaE eh 21:10 VanessaE I don't think that's all that useful to us 21:10 sapier true 21:10 VanessaE we already have ABMs and callbacks 21:10 VanessaE which in practice look more or less the same anyway if they're coded well 21:11 VanessaE they're just not realtime, as such 21:11 sapier the only way to use them is create a coroutine and push jobs there ... but that's almost as heavy as full async using multicore cpus 21:12 sapier the only benefit would be as coroutines are not parallel fixing api to be available to coroutines would be possible ... while it's way to much work for async things ... at least at start 21:17 VanessaE I think at this point there are more important things to focus on, honestly 21:17 sapier true but spawning requires a way to do things async as it can take a lot of time 21:17 VanessaE this is true 21:18 sapier usually it doesn't but one spike is enough to cause major lag 21:18 VanessaE don't I know that. 21:18 VanessaE I've been seeing major lag spikes lately from sign-ons, and it ain't from media downloads (these users already have it) 21:19 sapier guess some mods do things on join 21:19 sapier I was thinking about adding some sort of mod profiling ... maybe it's time to do it now 21:20 VanessaE I have some code xyz gave me for that purpose 21:20 VanessaE want it? 21:20 VanessaE it's pretty thorough 21:20 VanessaE but sorta barebones as far as how it's actually used. 21:20 VanessaE could use a little "fleshing-out"\ 21:21 sapier I have a profiler too but I thought about some built in mechanism telling you mod x using y% of cpu per step xxx ms at max 21:22 VanessaE yes 21:22 VanessaE that's precisely the sort of thing we need 21:22 VanessaE it would be useful if that can be broken down a little further, e.g. by function name, let's say 21:22 VanessaE (or by some other slightly more fine-grained measure anyway) 21:23 sapier well function name is hard as generic feature as it's gonna be quite slow 21:23 VanessaE true but one need not always have the profiler turned on 21:23 sapier but e.g. on_step on_generate on_join might be possible without to much overhead 21:23 VanessaE yeah, that's what I mean 21:23 VanessaE just pass callbacks through it 21:23 VanessaE stuff you can intercept easily 21:24 VanessaE *just* enough to give the coder some idea where the bottleneck is, if there is one to be sorted out 21:24 sapier yet I need same thing as needed for cpu quota ... us precise time ... so we need to extend api as lua usually doesn't provide that exact timings 21:25 VanessaE "plants_lib / register_on_generated : 15% CPU over 1:23:45, 25% in last 5 mins" 21:25 VanessaE etc 21:26 sapier try installing mobf enter "mobf" on console enable statistics restart and see on info tab ... that's about what I mean 21:28 VanessaE fix your screenshots in your github page :) 21:28 VanessaE https://github.com/sapier/animals_modpack/wiki/Menu <-- google shows a relevant image, now I see what you were talking about, but the image file actually missing from there. 21:29 sapier ah another lost image ... yes 21:29 sapier thanks I need to update 21:29 VanessaE all of the images there seem to be lost 21:29 VanessaE anyways, I see what you're getting at 21:30 VanessaE same idea as I suggested, just worded differently 21:30 sapier yes my old webspace thought it's not ok to provide android apk's from their servers 21:30 VanessaE something to allow a modder to quickly identify a bottleneck at any rate 21:31 VanessaE in fact, this sort of profiler is needed right now in minetest_game. there's a CPU usage issue there that I have not identified yet. Might be in the game's mods, or one of the extras used by my server. 22:05 fu-fu minetest.forceload_block(pos), Is that a 16x16x16 block or..? 22:09 Zefram_Fysh the block that is forceloaded is a 16x16x16 mapblock 22:13 fu-fu Zefram_Fysh: Thank you.