Time Nick Message 00:00 nephele if you can freeze the state the players are in you could also make severall screenshots from different positions, and then catch up on player movement or so 00:00 rubenwardy Mostly fixed maps 00:00 appguru TBH one should just record actions to be able to make screenshots later on when replaying 00:00 rubenwardy Requires intervention 00:00 rubenwardy Less fun 00:01 rubenwardy Having a system to take screenshots and then rate the interesting-ness of them is a cool problem 00:01 rubenwardy But can end up very computer-visiony 00:01 nephele stir the pile of data till it starts loooking right ;) 00:05 appguru data science in a nutshell 00:11 appguru nephele: are you https://github.com/nephele ? 00:12 nephele no 00:12 nephele I am not on github 01:36 tf2ftw majochup, apologies 01:43 tf2ftw how does the gold key work? Can I lock a regular chest? 02:25 freelikegnu howdy all 02:46 Copenhagen_Bram hai 07:08 cheapie Gotta love it when functions don't do what the docs say they do... 07:08 cheapie The docs state that minetest.forceload_free_block() "stops forceloading the position" 07:09 cheapie But... nope, it just decrements the reference count (that minetest.forceload_block() had been incrementing) and then only actually does what it says *if* the reference count becomes zero. Grr. 07:10 * cheapie just had to manually free a bunch of stuff since a mod had been using it the way it says it's supposed to work, and as a result wound up with hundreds of mapblocks forceloaded that shouldn't have been, some of them tens of millions of times over. 09:46 theenemystando[m is there a way to get the name of the game for the current world? 09:47 theenemystando[m i'm new to minetest's API, apologies if it's an obvious problem 09:49 sfan5 there is certainly no obvious way 09:50 theenemystando[m how would mods implement cross-game compatibility then? 09:50 theenemystando[m or is it not implemented yet? 09:51 sfan5 check if an item, node or API specific to a game exists 09:51 theenemystando[m that sounds unassuring - what if a game is just a really close fork of the original? 09:52 theenemystando[m can't expect a mod to check everything then... 09:52 sfan5 no, why would it? 09:53 sfan5 if the fork is close enough for the mod's purpose, existing support code will work fine 09:53 theenemystando[m guess so. I'm not writing a cross-game compt mod though so it'll be fine. 10:00 jas_ it feels to me like 'game' is a collection of mods, and if a server adds a single mod to worldmods/ directory then is that `game` still `game`? it's an interesting question i'd never considered nor heard asked -- but it made me think of minetest server list which does show the game name 11:18 nephele theenemystando[m, checking for globals checking for node availability is unfortunately the only way currently 11:18 nephele there is no get current gamename or anything 11:19 nephele cheapie, oof, that sounds really annoying... i was planning to use that call soon, but just forceload while something is going on... guess i'll have to do something more complicated 11:20 sfan5 annoying? 11:20 sfan5 "forceloading honors the reference count" isn't a bug, it's a feature 11:34 nephele sfan5, if it doesn't match the doc it is definetely a bug 11:34 nephele either in the code or the doc 11:34 nephele For my usecase reference counting is more a detriment if anythign, adding it ontop of an api is easy... but reversing it is hard :) 11:35 sfan5 the documentation should explain the reference counting, yes 11:36 sfan5 however what is your situation you have where reference counting is a hindrance? 11:39 nephele Well, i have mostly mechanics, i would likely forceload each block in a net from a central position, if it's a noop on calls after that i can just do that, if its not a noop i need to do carefull locking to not lock them severall times 11:40 sfan5 hm 11:41 sfan5 if it was possible to attach a string key to a forceload, your situation would be easier 11:41 sfan5 but that means more overhead for the engine 11:43 theodixon hi 11:43 Krock hi 11:44 theodixon To help with the popularity issue of minetest, I have promoted it on my website. https://prospective-technology.co.uk/game_reviews/minetest/ 11:46 theodixon The website is not fully finished. When it is i hope it gets popular. Minetest was the first article I wrote on it 11:46 nephele sfan5, yeah, or just one global forceload function that doesn't do counting 11:47 Krock nephele: that sucks in terms of mod-interaction 11:52 tf2ftw why is it so much easier to find silver than steel?? Where should I be looking for steel? 11:55 Krock there's no steel ore 11:55 Krock you need iron 11:56 Krock https://github.com/minetest/minetest_game/blob/master/mods/default/mapgen.lua#L710 11:56 Krock small clusters can be found from -255 to -128 11:56 theodixon i think like bronze is made from copper and tin, steel should be made from iron and coal 11:56 theodixon because iron +carbon = steel 11:57 Krock large clusters, but more distributed can be found flow -256 11:57 Krock theodixon: in the furnace that works just fine. you burn iron with coal to get steel 11:57 Krock also, this was already discussed a few times. check out the MTG issues on GitHub 11:58 theodixon oh ok 11:58 theodixon maybe its a mod i can make, with more alloys 11:58 Krock see also: 11:58 Krock !mod technic 11:58 MinetestBot Krock: Technic [technic] by RealBadAngel - https://forum.minetest.net/viewtopic.php?t=2538 - https://github.com/minetest-mods/technic 11:58 theodixon forgot about that 12:35 tf2ftw thanks for the info on the steel 12:36 tf2ftw are there any "sea level" indicators? 12:37 nephele <0 if you mean mtg 12:55 jas_ some ambience: https://www.youtube.com/watch?v=CvTd-MY64s0 12:56 jas_ !server jastest 12:56 MinetestBot jas_: jastest | jastest.duckdns.org | Clients: 0/15, 0/1 | Version: 5.3.0-dev / minetest | Ping: 133ms 14:54 oil_boi texmex, do you think I should add snow blocks? 15:19 texmex oil_boi: Yes, but what mechanism do they bring? 15:19 oil_boi Spleefing 15:21 texmex Sure. 15:22 oil_boi Or snowmen? 15:22 oil_boi Or even ice cubes :T 15:23 texmex Love the way you focus on core mechanisms though, rather than lighter biome-related functionality… 15:23 texmex It’s easy to get lost in biomes, in my experience 15:28 oil_boi Thanks texmex, the only thing I've gotten lost in so far is how to tell https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L4357 not to hurt the player if they fell into some water 15:28 oil_boi I don't like having biomes do specific functions besides creating certain nodes or it quickly becomes a mess :L 15:31 oil_boi It doesn't look like it's possible because returning true or false in register_on_player_hpchange doesn't do anything and resetting the hp on one server step after can bring up the death screen with full health 15:33 oil_boi Also even return({modifier=false,hp_change=0}) is non functional, I wish this explained how to tell the engine the new damage 15:38 sfan5 oil_boi: http://sprunge.us/c7Mye8?lua 15:42 oil_boi Thanks sfan5 15:43 nephele Is there any way to specify /how/ a specific biome is supposed to look in regards to elevation and stuff 15:43 oil_boi I'll have to make a note on the explanation style that certain api references like that use 15:43 nephele like... plains should have relative low elevation across the board, increasing one decreasing one block per 20-30 blocks or so at most 15:44 sfan5 I don't believe the mapgen offers this flexibility, several noises decide the terrain and biomes decide what nodes to place 15:44 nephele .-. 15:45 nephele Guess i'll write my mapgen in lua then, without that it's really unusable for me 15:46 sfan5 this seems very fundamental so I have a hard time believing that Minetest can't do this 15:46 oil_boi nephele, you could set the biome to air and then have a custom lua gen in place worse comes to worse but it'll look like there's chunk errors slightly 15:47 sfan5 or set it to some special filler node and then post-process the terrain in on_generated 15:48 nephele sfan5, well, i have found nothing in the lua docs for biomes that suggest i could tie biomes to features 15:48 nephele I have seen some mapgen already that would look fine as a plain, if it actually did generate blocks for my plains biome there 15:49 nephele currently i just get it randomly /assuming/ the height is correct... but really absolute height too seems like a terrible way to specify terrain, it always seems to take all or nothing 15:49 sfan5 nephele: sure but it might be possible to control this indirectly by placing plains only where the elevation is smooth 15:50 oil_boi Ooooo, that's a great idea sfan5, you could use a node as a "grass" layer with the rest set to 0 and then do lua mapgen using the node as a noise reference point and apply amplification to it to create a mountainous region!!! 15:50 nephele And how would i acomplish that? the biome definition let me tie in only in heat, humidity and absolute heights 15:50 sfan5 I'm not saying it is possible, just speculating 15:51 calcul0n_ biome defs have a y_min/max and min/max_poq params, i guess they are done for that 15:51 calcul0n_ you could register different biomes for different levels 15:51 calcul0n_ *min/max_pos 15:52 nephele sfan5, ah okay 15:52 nephele Well, I would like to be able to specify a median size, with maximum and minimum absolute sizes aswell for biomes 15:52 nephele In any case, minetests insistence that sea level is ==0 is really wierd imo 15:52 sfan5 https://github.com/minetest/minetest/blob/master/src/mapgen/mapgen_v7.cpp#L304-L318 15:53 nephele See level should be variable too somewhat, or tinier sees higher or so 15:53 sfan5 well as evident here, terrain is always the same and biomes just control the nodes 15:53 nephele Yes, i know 15:53 nephele i don't like that, it's not what i would want my biomes to /mean/ :P 15:54 calcul0n_ note that a mapgen in lua would probably be terribly slow 15:55 sfan5 I think nephele is willing to accept that in exchange for being able to do biomes "properly" 15:55 calcul0n_ yep, i just mean be prepared to suffer :p 15:57 rschulman Am I right in assuming that once a chunk is generated, there's no way to gen new ores (after adding moreores) in that chunk? 15:57 FreeFull That does sound right 15:58 sfan5 "no way" isn't correct, but it's certainly not easy 15:58 Krock rschulman: use VManip. https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L3755-L3757 15:58 Krock https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L4764-L4771 16:01 sfan5 and unless you want double the ores you should turn existing ores back to stone before calling the method Krock suggested 16:01 rschulman Oh, I meant from within the game. 16:01 sfan5 yeah there isn't, you'll need a mod (and not only need but also need to program) 16:03 rschulman Ok cool, that's what I figured. Thanks. 16:04 nephele calcul0n_, no it wouldn't 16:04 oil_boi sfan5, I'm sorry to change the topic but the latest commit that fixed bones well 16:05 oil_boi https://i.imgur.com/anHyluu.png this is third person in a fresh install of minetest with a few seconds old git clone of minetest gmae 16:05 oil_boi game* 16:06 oil_boi I just git fetched and recompiled minetest engine and I was messing around, I went into 3rd person on my game, and I was like, oh no, I broke something, but the same thing happens in MTG 16:07 sfan5 huh which commit? 16:08 oil_boi Oh, I thought the bone fixes were pushed, that uhhhhh, makes me even more confused 16:10 oil_boi sfan5, I cleared the cache and it's all better?? What the 16:10 nephele you should start making wild acusations in confusion :3 16:11 oil_boi nephele, YOU DID THIS 16:11 oil_boi There 16:11 nephele I am honored, beyond honored, to be important enough to be a probably cause for compile errors on your machine 16:12 oil_boi But in all seriousness this is why I get super nervous when I find stuff like this, I still remember when entities would sometimes keep old functions stored in their metadata even though you completely redefined them and cleared their serialize and deserialize functions completely, that was a very long time ago 16:13 oil_boi Thank you nephele for being such a kind and caring glitch gremlin 16:13 nephele That's a new one 16:14 nephele Hmm... I'll add "re-implement mapgen in lua" to my todo list... right after "make a new crafting api" 16:14 oil_boi oh yeah, I didn't even bother bug reporting it because after 1-2 server restarts in singleplayer at least it would fix itself, but it was fixed around 2014 or so but good lord it was annoying 16:21 oil_boi nephele, https://github.com/paramat/lvm_example 16:22 nephele ? 16:26 texmex oil_boi: I wonder, since you’ve added Aether-like features, are you keen on some other MC mods that feel like they should’ve been in vanilla? Related to adding biomes, I’m interested in Tough As Nails functionality myself, as it makes biomes really behave different. 16:34 oil_boi texmex, I'm watching a video on it right now and taking notes 16:35 texmex Awesome ? 16:37 texmex This video is the best one I’ve found for explaining its mechanics. https://m.youtube.com/watch?v=kgNWUWjpKwk 16:40 texmex Sorry, I think I meant this one https://m.youtube.com/watch?v=5SjTtFcYMuU 16:42 oil_boi texmex, in that last one you sent, he has made an EXTREMELY good point, we should have modding teams 16:43 oil_boi Err, wasn't really a point, he just mentioned modding teams 16:44 oil_boi Before I even finish watching all 3 of these videos, time to add ice 16:45 orbea If a mod calls 'set_hp(-1, "foo")' then 'reason.type' for 'minetest.register_on_player_hpchange()' should be "foo" and not "set_hp", right? 16:46 texmex oil_boi: Agreed. It was touched upon in this thread: https://forum.minetest.net/viewtopic.php?p=371343 16:47 oil_boi Linuxdirk: Minetest development is super slow and the devs are uber scared of changes so it is just the easiest and simplest way to code things you need by yourself. 16:47 texmex But the idea should be expanded upon and tried. From what I can tell it was quite common in the MC community. 16:48 oil_boi >if you can make the thing run as fast as it would in the engine in LUA then there is no need to add it into the engine because 97% of anyone will not agree that it should be added into the engine anyways 16:48 texmex More like this post, specifically https://forum.minetest.net/viewtopic.php?p=371343#p371343 16:48 orbea oh, im doing it wrong now that I look at other examples... 16:50 oil_boi texmex, your idea is good, I looked over that thread and was inspired by your post before adding https://github.com/oilboi/Crafter/blob/master/mods/new_functions/init.lua 16:50 orbea Now: 2020-04-26 09:49:56: ERROR[Server]: Bad type given! *sigh* 16:51 oil_boi Touch based damage, damage inside node, suffocation inside regular drawtype nodes, all based in lua, all extremely fast 16:51 orbea how do you give it a custom damage type? 16:51 oil_boi orbea, can I see your code? 16:52 texmex oil_boi: Yeah, especially when you’ve client-side mod at your disposal \o/ 16:52 oil_boi texmex, Oh my man, if I added that into clientside, I think I'd hover above my chair and do a flip in excitement 16:53 orbea oil_boi: not my code, but I am trying to modify hunger_ng on this line: https://gitlab.com/4w/hunger_ng/-/blob/master/system/hunger_functions.lua#L187 16:53 orbea it now has: player:set_hp(new_health, { type = reason }) 16:54 oil_boi Clientside api basically creates the possibility to go from singlethreaded to threads = (however many players are in the game) + server thread 16:55 orbea both hunger_ng and 3d_armor report the type correctly as "starving", maybe the error is bogus? 16:55 oil_boi orbea, see https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L4363 16:56 texmex oil_boi: Yes, it's quite exciting and I think you're doing the right thing by starting using it right away even before server-sent csm is possible. 16:56 oil_boi orbea, then https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L5869 You must give reason a custom string to work with it 16:56 orbea oil_boi: i have read it many times, maybe I dont understand what it means? I undestood that if it gives a reason as indicated it should use that instead? https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L5869 16:57 oil_boi so set_hp(10, "you_are_cool") or set_hp(-10, "you_dun_goofed") 16:57 orbea i did that at first and it just reports set_hp then 16:58 orbea (in 3d_armor) 16:58 oil_boi orbea, Oh man, I dunno then, you might have to PM Linuxdirk on the forum 17:00 oil_boi texmex, once the server does CSM automatically all of my fears of people using CSM to hack like crazy in the Crafter, or any game, will be completely relieved!! 17:01 texmex Build now, worry never :) 17:01 oil_boi That's very true 17:05 orbea oil_boi: not sure its specific for hunger_ng, this is just what I was trying it with. See https://github.com/minetest/minetest/blob/master/src/server/player_sao.h#L252 and https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_object.cpp#L242 17:05 orbea it seems the code assumes custom reasons are not possible? 17:05 orbea even though the docs indicate it is? 17:06 orbea and despite the error message it continues to use the custom type... 17:07 oil_boi orbea, I think you need a moderator or engine developer to help you at this point because I've never tried what you're trying so I don't know hot to resolve it 17:07 oil_boi !mod 17:07 MinetestBot oil_boi: Sound Reactions 1.0 [soundreaction] by Glory! - https://forum.minetest.net/viewtopic.php?t=13185 17:08 oil_boi Oh.. 17:08 orbea ah, thanks for trying anyways :) 17:17 rubenwardy orbea: I wrote that code, you can provide custom reasons but the type must be "custom" 17:18 rubenwardy Well, or an existing valid type 17:20 orbea rubenwardy: so how would I determine between two different custom types, say "foo" and "bar"? 17:21 orbea doesn't work? player:set_hp(new_health, { type = "custom" } 17:23 oil_boi texmex, oh my god the models for entities and players are randomly not loading in, I'm about to throw this motherboard in the garbage 17:23 rubenwardy { type = "custom", some_other_field = "foo" } 17:23 rubenwardy perhaps `type` should allow arbitrary types that start with `_` or something like that 17:25 orbea even with type = "custom" it still prints an error 17:25 orbea also dont see where it checks "custom" in the code 17:26 rubenwardy hmmm 17:26 rubenwardy try not specifying the type? 17:26 rubenwardy this is broken 17:26 rubenwardy oh, looks like I made up the custom type 17:27 rubenwardy it's not documented 17:27 rubenwardy the `set_hp` type should probably be used 17:28 rubenwardy set_hp`: A mod or the engine called `set_hp` without giving a type - use this for custom damage types. 17:28 orbea adding { hunger = reason } works, I guess I can do that 17:28 orbea i dont think disabled armor damage for "set_hp" is a good idea, that would break armor damage from tnt and probably other things I think 17:28 orbea *disabling 17:29 rubenwardy armor doesn't make sense with set_hp 17:29 rubenwardy armor is strictly a tool caps thing 17:29 rubenwardy so you should use punch 17:29 rubenwardy you can specfiy a nil puncher, you use the player as the puncher 17:29 rubenwardy or an entity 17:29 rubenwardy actually, citation needed 17:30 rubenwardy yes, confirmed 17:30 orbea fwiw 3d_armor handles armor damage here https://github.com/mt-mods/3d_armor/blob/master/3d_armor/init.lua#L403 17:31 orbea it should also be damaged by things like falling or lava 17:31 rubenwardy oh right, I misunderstood what you meant by armor damage 17:31 rubenwardy I see the problem - you don't want hunger to affect armor 17:31 orbea so my idea was to define what types it should not be damaged by, in my use case that mostly means drowning, eating things that damages the player and from starving 17:31 oil_boi sfan5, I found the bug: If I turn shaders off no models load?? 17:32 orbea drowing is a defined type so that was easy 17:32 orbea *drowning 17:33 oil_boi sfan5, this has to be a bug with manjaro or arch 17:33 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Fix mapgen settings in minetest.conf being ignored (#9737) 13eca6ee9 https://git.io/Jftd5 (152020-04-26T17:32:29Z) 17:33 MinetestBot 02[git] 04theviper121 -> 03minetest/minetest: Fix UpdateBonePosition() breaking animations (#9577) 13e1fc72c https://git.io/Jftdd (152020-04-26T17:32:04Z) 17:33 sfan5 oil_boi: works for me here 17:34 * Lone_Wolf claps 17:35 rubenwardy works her for me too (Manjaro) 17:36 rubenwardy I don't turn shaders off though 17:54 oil_boi rubenwardy, sfan5 https://youtu.be/27-JEmC3FDQ This is something 17:54 oil_boi !title 17:54 MinetestBot oil_boi: What even is this bug - YouTube 17:57 sfan5 flat clouds is a new feature 17:57 sfan5 or rather varying cloud "densities" 17:59 Krock oil_boi: your menu clouds are really weird 17:59 oil_boi sfan5, oh okay, I'm more concerned about the vertices outlines showing up on everything around 5:36 17:59 sfan5 either way here's an existing bug report https://github.com/minetest/minetest/issues/9072 17:59 Krock #9072 17:59 ShadowBot https://github.com/minetest/minetest/issues/9072 -- Faces show as if in wireframe mode 17:59 Krock these ninjas again 18:00 oil_boi So then the models not showing up at all is the new bug then? 18:00 sfan5 yea looks like it 18:01 sfan5 so to confirm the wireframe bug happens with antialias x2 regardless of whether shaders are enabled? 18:03 oil_boi shaders don't matter, x2 worse, x4 is slightly less worse, if formspec that's translucent in front it's even worse 18:03 oil_boi if shaders on, entity and player model loads, shaders off, entity and player model doesn't load 18:04 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Remove unused lookup table from noise.cpp 1368f45fc https://git.io/JftFB (152020-04-26T17:52:09Z) 18:06 sfan5 entities not loading is a new bug, possibly related to https://github.com/minetest/minetest/pull/9374 18:06 sfan5 would be good to know if it works with a build before Feb 16 18:06 oil_boi sfan5, how to get build before feb 16 18:07 Krock reset your shaders 18:07 sfan5 https://minetest.kitsunemimi.pw/builds/ there's one from 2020-02-10 18:07 sfan5 if you want to compile from source you can just use git to reset to any commit you want 18:08 Krock (nvm @ shaders) 18:09 Krock might be related to the new use_texture_alpha distinction? https://github.com/minetest/minetest/pull/9374/files#diff-90993ca3bb550ffc1c77a1578f3545d3R591 18:09 sfan5 that was there before 18:09 oil_boi sfan5, tested in wine, the problem still happens 18:10 * Krock opens his eyes 18:10 sfan5 oil_boi: hm good to know 18:10 nephele Krock, your eyes were closed all this time? 18:10 oil_boi Damn I hate opening these huge problems 18:10 rschulman How would folks go about creating a lua node that prevents mobs (say from mobs_redo) from passing above it or coming w/in a certain number of blocks of it? Is there a way to influence `minetest.find_path`? 18:10 sfan5 if this is broken with 5.1, 5.0, ... too then it must really be due to some change in Mesa/AMD's code 18:10 Krock how about a 0.4.x build? if it happens there it's likely a driver issue 18:10 oil_boi sfan5, LETS FIND OUT 18:11 sfan5 https://github.com/minetest/minetest/releases/download/5.1.1/minetest-5.1.1-win64.zip 18:11 sfan5 https://github.com/minetest/minetest/releases/download/5.0.1/minetest-5.0.1-win64.zip 18:11 sfan5 ^ for your convenience 18:11 orbea oil_boi: any chance you tried the mesa git master? 18:11 Krock rschulman: only manual influence by changing the code 18:11 sfan5 rschulman: there are some hacks possible e.g. overwriting minetest.find_path from a mod, but such a thing should really be implemented in mobs_redo itself 18:13 oil_boi 5.1 wine doesn't render player either way, same with 5.0 18:13 oil_boi orbea, let's give it a try cause I'm about to throw this RX580 in the toilet 18:14 rschulman Ok, makes sense, thanks! 18:15 orbea i have amd too and I find its best to follow the git for libdrm/mesa and the most recent llvm stable release 18:15 orbea if all else fails apitrace might help 18:16 orbea (would help track down the bad GL calls) 18:16 Lone_Wolf IIRC if you disable mipmap the entity missing bug no longer occurs 18:16 oil_boi Lone_Wolf, the mipmapping was disabled in all tests 18:17 Lone_Wolf Or something else was up. Been a while since I had that problem 18:18 oil_boi I had this problem on windows and I thought it was just me being a dumb with the visualstudio stuff 18:19 sfan5 wait wait, the clouds turned wireframe-y on windows too? 18:20 Lone_Wolf OH. I fixed the bug by removing the directional_lighting PR from my client 18:20 oil_boi sfan5, yeah 18:20 sfan5 oh no 18:20 Lone_Wolf I could pop over to Windows and test for the wireframes if wanted 18:20 oil_boi I cannot get into the windows 10 install because the new windows update completely killed the bootloader 18:20 sfan5 if this was a bug in just Mesa it would be somewhat okay to fix/diagnose 18:21 sfan5 but on Windows the situation is entirely different (unless AMD ships Mesa as OpenGL implementation there?) 18:21 orbea might indicate that minetest is doing something wrong and it just happens to work on some drivers 18:21 Lone_Wolf Ping me if you need some testing 18:21 orbea or maybe they just have the same bug by chance (the video drivers) 18:22 oil_boi One second, you have to remember that the drivers on windows, if shipped with the same implementation of opengl could be using the stable branch, I'm compiling the git version of mesa AMD drivers right now to check 18:23 Lone_Wolf Lemme know if you succeed. I couldn't figure out how to get the custom-compiled mesa to override my current one 18:23 oil_boi AMD couldn't get their rx5800XT running right for months I don't doubt they couldn't figure out how to do OpenGL 18:29 oil_boi sfan5, no need to worry, you and orbea's assumptions were correct and it was the shit AMD drivers 18:29 sfan5 so it's fixed in Mesa master? 18:32 oil_boi On Mesa 20.1.0-devel (git-bf3c9d2770) latest dev working, with wine 5.0 is working, with wine 5.1 is working (shaders on and off for all 3) 18:33 oil_boi But, what this means is, anyone who has AMD GPUs and is on windows or linux with the stable OpenGL implementation is going to have this bug and it's not anyone's fault but AMD 18:33 oil_boi I would make a forum sticky about this for the time being :T 18:35 oil_boi I can't test this on W10 with unstable OpenGL implementation because 1, I don't work at AMD and 2, I don't want W10 to wipe my grub again 18:50 oil_boi_ sfan5, I'm sorry to keep piling stuff up but, it seems like AA 2 and 4 are the only one's that do anything bad is because they're the only ones doing anything https://imgur.com/a/s5E24Cv 18:51 sfan5 useful to know 18:51 oil_boi_ Alright I'm gonna stop bug reporting because I'm being an ultra downer rn 18:51 sfan5 so in conclusion: AMD recently broke anti aliasing 18:51 oil_boi_ When doesn't AMD break literally anything 18:52 sfan5 their CPUs are pretty good 18:52 oil_boi_ Yes, yes they are, but their GPUs get the big oof from me 18:53 oil_boi_ I have a full AMD system so it's like heaven and hell wrapped into one 18:54 tf2ftw speaking of specs, what type of cpu/ram are needed to host ~100 users? 18:54 Lone_Wolf speaking of 100 users how are you planning to get that many players? :thinkeyes: 18:55 tf2ftw ha yeah 18:55 oil_boi_ tf2ftw, you're gonna need a singlecore that does 7GHZ 18:55 tf2ftw I live in a neighborhood with a lot of kids ;) Maybe my son can make it go viral at school 18:55 sfan5 before you run out of RAM the poor cpu utilization of Minetest will bring you down 18:56 sfan5 so grab the CPU with the best single-core performance and hope for the best 18:56 tf2ftw single core at 7ghz? is MT not multithreaded? 18:56 oil_boi_ tf2ftw, oh man, I'm sorry to pull that bandaid off for you 18:56 sfan5 to some extent 18:56 sfan5 but everything that (really) matters happens in the server loop in a single thread 18:57 tf2ftw well.. hm 18:57 sfan5 of course how responsive this is also depends on what the mods do 18:57 tf2ftw ok true. im not complaining. 18:57 tf2ftw good to know though. 18:58 oil_boi_ one day we will have multiple thread capabilities! I hope! I am just too dumb to put ideas into code through C++ 18:58 Krock GHz is not the only factor. If the RAM latency were lower and instructions took less cycles you'd get the same performance for lower clock rates 18:59 tf2ftw Well for now, MT is perfect for my 5yo sold and me to play with a few of his friends :) I was just wondering what the hit was per player 18:59 Krock basically the goals of RISC. only a few instructions, but those must be somewhat fast 18:59 tf2ftw * son 18:59 oil_boi_ the only way I can see doing this in pure LUA is https://www.lua.org/pil/9.4.html and .after statements but they will have to funnel back into a single thread right now 19:00 tf2ftw Go has very nice thread management but I digress 19:01 oil_boi_ tf2ftw, if you have a multicore you can set the emerge_threads in the conf and it will generate the map much faster which is probably what you're more worried about with default MTG 19:01 sfan5 the best case for threading optimization I see is running object's on_step outside the server loop 19:01 sfan5 but you can't "just" put that on another thread 19:02 oil_boi_ sfan5, oo that's a good idea :T I mean you could but it wouldn't communicate properly with the terrain and you'd have to have the objects work with the server instead of the server work with the objects 19:02 Krock it would need an entire map modification pipeline 19:02 oil_boi_ Exactly 19:02 oil_boi_ The objects would have to work from 1 memory pool and calculate using that 19:03 sfan5 the most likely way I can see this working is that mods can submit work to run in the background 19:03 sfan5 those functions then run in a different Lua environment with limited/low-priority/read-only(?) access to the map and stuff 19:05 oil_boi_ That would make memory management so ultra extremely efficient, the closest thing we have now is having objects offload data to the closest player's client using modchannels and calculate using that essentially distributing processing threads to as many people that join the game 19:05 tf2ftw Why not at least separate thread to render things beyond the players view - or maybe that is already happening and the real issue is the mods? 19:06 sfan5 "render"? 19:07 tf2ftw fair. prepare to render? 19:07 oil_boi_ I think tf2ftw is talking about caching chunks that are not currently stored in the clients mapdata memory pool 19:08 tf2ftw Thanks. Yes 19:09 tf2ftw I'm sure we're talking about re-arch no matter the direction though. right? 19:09 oil_boi_ That does technically happen tf2ftw, when you hit unlimited viewing range that's why your client fps tanks if you've walked very far 19:09 tf2ftw i see 19:09 oil_boi_ But I assume you're talking about extending the distance of viewable chunks? 19:09 tf2ftw Na I'm just saying to offload work beyond the players view 19:10 tf2ftw basically limit the player view so other things can happen at a "slower" pace 19:10 sfan5 are we talking about the server now or client? 19:10 tf2ftw ah right. good call out. I guess that would be the client 19:11 tf2ftw Didn't mean to take us down a rabbit hole. Thanks for the info 19:12 sfan5 the server already puts objects to sleep if there is no client within 48 (by default) nodes 19:14 oil_boi_ Offload all object ABM LBM etc work seems to happen already, in minetest.conf you have max_block_send_distance which limits the amount of mapblocks you can see, then there's active_object_send_range_blocks which seems to set how far objects can stay alive from the player, and then there's active_block_range which controls how far from the player ABMs can run 19:15 oil_boi_ Doing this per client would cause some serious problems 19:15 sfan5 active_block_range also applies to objects 19:15 sfan5 but when active_object_send_range_blocks < active_block_range the objects are kept alive but you don't see them 19:16 oil_boi_ Holy moly so any forceloaded block can still process entities? That's extremely useful 19:16 MinetestBot 02[git] 04Panquesito7 -> 03minetest/minetest: Add new command /revokeme (#9584) 132fe4641 https://git.io/JftNE (152020-04-26T19:15:05Z) 19:16 sfan5 they do, yes 19:20 oil_boi_ *Brain has quit (Quit: Turned to charcoal) 19:39 oil_boi_ You know what we really need: A fishing game 19:40 Lone_Wolf With MC-like fishing rods? 19:41 oil_boi_ Lone_Wolf, Nah like https://youtu.be/xcZhg77asYU 19:41 oil_boi_ !title 19:41 MinetestBot oil_boi_: Action Bass ... (PS1) - YouTube 20:13 texmex Watch out for viral diseases on your favorite servers https://gitlab.com/sztest/epidemic 20:13 sfan5 !title 20:13 MinetestBot sfan5: sztest / epidemic · GitLab 20:17 texmex oil_boi_: Nay to synchronous, individual rod fishing. Yay to asynchronous, coop net fishing! 20:27 tf2ftw texmex, nice 20:48 oil_boi_ texmex, that's crazy 20:48 oil_boi_ And awesome! 20:50 oil_boi_ Also texmex I already added the prototype of it but it acts weird :T https://youtu.be/e1Uauh0szgU 20:50 oil_boi_ Maybe just simple fishing for now and make a complete fishing game later :T :T :T 20:52 oil_boi_ Ahh man, Action Wrass in Minetest is gonna go on the backburner 20:55 oil_boi_ All the elements of making a fishing game are already in the engine, you can do a singlenode worldgen and literally load maps in by clearing the area using vm and then loading in a stored schematic, hell you can create doom right now with all the stuff that's in the engine 21:49 jas_ https://www.youtube.com/watch?v=CTNostYetvE 21:49 jas_ rocket jump on jastest 21:49 jas_ shooter shoots players but not mobs :/