Time Nick Message 01:40 AdamS So I downloaded minetest off pacman on arch. When I started it, it asked me to update it, so I guess my question is, can it be updated safely in linux using the built in update feature or do I need to just wait for my distro to update the package? 01:40 AdamS I know it's probably a stupid question :D 01:42 AdamS Sorry, I mean Luanti 01:42 AdamS (still called minetest in arch's repo) 01:45 specing Every distro I know of disables update checks particularly as you're supposed to install/update via package manager 01:45 specing so, my recommendation: find a better distro and let the package manager handle it 01:46 AdamS or just compile it myself off github and build a package myself ^_^ 01:46 MTDiscord looks like arch has minetest at 5.9.0, unless you need a bug fix in 5.9.1, just ignore it for now 01:46 MTDiscord or better yet make a pr to the arch repo 01:47 AdamS I was trying to download VoxelLibre twice and it failed, so I suspected that it was a version issue or something. 01:47 AdamS Backrooms test worked so, who knows. 01:48 MTDiscord they support the last two major versions of minetest iirc 01:49 specing AdamS: you can probably take the packaging script and change 5.9.0 into 5.9.1 and run it 01:50 AdamS ooh, I found a aur package for latest version :) 01:52 AdamS or not, it's also 5.9.0 :/ 01:54 AdamS github time :D 02:26 AdamS Really good build instructions :) it worked well 02:27 erle AdamS does mineclonia work? similar game, similar size 02:28 erle like is it just big packages? 02:28 erle AdamS otherwise, you can always use web browser to download whatever from https://content.minetest.net and put it in mods/games directories 02:28 erle good night 02:32 AdamS night erle 02:52 AdamS erle: with the compiled version, it downloads the packages fine :) 11:44 MinetestBot 02[git] 04sfan5 -> 03minetest/serverlist: Some documentation adjustments 135d19189 https://github.com/minetest/serverlist/commit/5d191896f3221cd1194643cf4152c0d221ffb3b3 (152024-11-03T11:38:50Z) 11:44 MinetestBot 02[git] 04sfan5 -> 03minetest/serverlist: Set short max_age for list.json 134d320bb https://github.com/minetest/serverlist/commit/4d320bbcf552e8e565ae34e77566b76643e9628e (152024-11-03T11:42:37Z) 14:11 MinetestBot 02[git] 04appgurueu -> 03minetest/minetest: Apply "and" to server list & content search terms (#15365) 130e06590 https://github.com/minetest/minetest/commit/0e06590ffda8e701c30b731f09466e55ab4ecad4 (152024-11-03T14:09:47Z) 14:11 MinetestBot 02[git] 04appgurueu -> 03minetest/minetest: Fix model[] being lit wrongly if shaders are disabled (#15364) 133064f3c https://github.com/minetest/minetest/commit/3064f3ccb769df03ef693bf8ce464481a627512a (152024-11-03T14:10:21Z) 14:11 MinetestBot 02[git] 04SmallJoker -> 03minetest/minetest: JsonCPP: restore '1.0.0+' compatibility (#15368) 13c884e71 https://github.com/minetest/minetest/commit/c884e7181fd6cc0e19e9c5abb4dce102f1485233 (152024-11-03T14:10:39Z) 14:11 MinetestBot 02[git] 04veprogames -> 03minetest/minetest: Replace occurences of 'forum.minetest.net' with 'forum.luanti.org' (#… 139982c56 https://github.com/minetest/minetest/commit/9982c563730f294d404119a2b8ffe26073134884 (152024-11-03T14:10:58Z) 16:48 erle https://github.com/minetest/minetest/issues/15370 16:48 erle > Note: With menu clouds relying on shaders (and shaders obviously being on by default) in 5.10.0, we will be able to tell quite well from user complaints if our assumptions are wrong. 16:49 erle yeah and then you can say “maybe just ignore that person” 16:49 erle from my experience, about 1 in 50 to 100 people complains loud enough that devs notice, unless it is a very small user base 16:51 erle past exhibit: mineclone2 breaking because of sync → async transition. very few people filed issues about it and the one that was filed was closed almost immediately. 16:52 erle but the race condition affected every single player of mineclonia, mineclone2, mineclone5 who ever made a map. 16:53 erle anyways, i think as long as the no-shader-codepath *works* (which it does perfectly now, AFAIK) it should not be gated behind a refucktored cloud menu. 16:55 erle again: i volunteer to keep the fixed pipeline working. it does not necessarily need feature parity with all cool new effects. it already does not have a lot of stuff. 17:08 erle i wish people saying “let's change/remove something and see if someone complains” would handle it more like linux does – last time i saw someone proposing some old drivers, it was “if even one person complains, we are going to put it back, no discussions about it” 17:08 erle otherwise you get “this person/hardware/usecase is not important” and support tends to more-or-less-track what the devs have (meanwhile, you can never become a dev if your hardware is not already supported) 17:14 cheapie I'm not sure what the devs have, but it must be fast - I went out and bought a 5950X and RX 6600 XT a while back and even that is *still* not enough to consistently get 60 FPS with the default view range if you're in an area with stuff built in it. 17:16 Krock cheapie: hmm. is that with full-blown shaders enabled? Because you should be getting 500+ FPS with this powerful hardware. 17:17 sfan5 it depends very much on what the "stuff" is 17:17 cheapie Krock: That was with the settings turned nearly all of the way down (I think the only eye-candy feature I had on was a low-ish AA level). If I'm out in the middle of nowhere I can get performance like that (more or less). 17:17 Krock To be fair, transparency sorting can heavily impact rendering performance. See https://github.com/minetest/minetest/pull/15076#pullrequestreview-2276014490 17:18 cheapie I had to turn transparency sorting off, that was destroying performance to insane levels. 17:18 erle sfan5 btw what *are* the advantages of making the main menu require shaders except to annoy me in particular (and everyone who has “garbage hardware”)? 17:18 sfan5 empty MTG terrain is fast to render, nodeboxes and mesh nodes not so much, and semi-transparent stuff not at all 17:19 erle semi-transparent stuff is the worst in terms of performance on any hardware i have seen. guess as long as you do not have order-independent transparency it will stay that way. 17:19 cheapie The glass panes seem to kill performance more than anything else, which is particularly annoying since basically /any/ building that isn't some tiny house is going to have lots of those. 17:19 Krock I find it rather interesting that transparency sorting is not handled by OpenGL's depth buffer 17:19 Krock or rather depth sorting 17:20 erle cheapie try glass walls mixed with watercubes on oysterity anarchy. 5fps on the “garbage hardware” that gets 30fps usually (i recently found out that is basically what i limited the fps to, to prevent useless busywork) 17:23 erle cheapie what i can guess about the devs' setups: 1. they probably do not have intel integrated GPUs, otherwise they would have noticed some things they did not notice. 2. at least the people involved in the menu redesign probably all have 16:9 or wider screens, as on 4:3 it has both text escaping its box and some free vertical space. 3. *probably* mostly using QWERTY/QWERTZ or similar, as evidenced by various key handling issues. 17:24 sfan5 cheapie: I can't promise a fix but it would help if you "exported" such a scene so it can be viewed standalone and opened a bug report 17:24 Krock Intel HD (3000) is in fact capable of rendering dynamic shadows, as well as any other shaders. It just happens to be slow as hell. 17:29 sfan5 if this turns out to be caused by mods blindly using use_texture_alpha = "blend" to get rid of the warnings it'll be funny 17:30 Krock for minetest-mods projects there's a nonzero chance that I might've introduced a blooper 17:31 MTDiscord there is/was a game that set all nodes to blend by default 17:34 rubenwardy I have an intel integrated gpu 17:38 rubenwardy also the CDB thing is unrelated to 4:3 /16:9 screens - https://forum.minetest.net/viewtopic.php?p=440386#p440386 17:40 erle well, on a 16:9 screen it looks nice and has no empty space. 17:41 rubenwardy it's fairly unpredictable based on the screen resolution and dpi 17:41 erle ok thx 17:44 erle Krock i have not tested this with the current release candidate, but generally with intel integrated graphics i have made the experience that regardless of what hardware advertises, the older it is, the more likely the fixed pipeline is the one that runs best/correct (or, in the case of several older thinkpads, it is the only one that works). meanwhile, on newer arm laptops that have a weak GPU, the fixed pipeline performs worse (but playable), than shader-based 17:44 erle setups. 17:45 MTDiscord sfan5: "if this turns out to be caused by mods blindly using use_texture_alpha = "blend" to get rid of the warnings it'll be funny" - i'm afraid it partially is, at least i've seen this with citysim. 17:45 erle grab some thinkpad that still says “IBM” on it and you can check this for yourself 17:46 erle luatic what's the correct thing to do there? 17:46 MTDiscord erle: most of the time you probably want use_texture_alpha = "clip", sometimes use_texture_alpha = "opaque" to prevent cheating for solid blocks 17:46 MTDiscord I've added a note to the docs that use_texture_alpha = "blend" is NOT zero cost 17:50 MTDiscord erle, cheapie: As for the inquiries regarding my hardware, I have a ca. 5 year old Lenovo laptop with a Ryzen 5 3500U with integrated Radeon RX Vega and 8 GB of RAM. I believe this is pretty medium hardware in today's day and age. 17:50 erle luatic by standards of “people who only buy refurbished hardware” this is a dedicated gaming machine 17:50 MTDiscord I am considering upgrading my RAM for a better developer experience, toolbelts can get heavy and I like to have things open at the same time. I may eventually upgrade my CPU too for better compile times. At that point I might upgrade my laptop altogether. 17:50 erle what *dev tools* have issues with 8G of RAM? 17:50 rubenwardy android studio 17:51 erle browsers fill it nicely, but they fill any amount of RAM 17:51 erle oh okay 17:51 erle luatic did some error-message suggest to use blend mode? 17:51 rubenwardy my next laptop will probably be 32GB ram 17:51 MTDiscord 🤤 17:51 rubenwardy then finally I'll be able to run android studio 17:52 MTDiscord 😄 I can't upgrade this one beyond 20 GB, 4 GB are soldered on, and I can only swap the other 4 GB for a 16 GB one at most according to hardware limitations 17:52 rubenwardy i exaggerate, 16GB is enough for AS you just can't run many electron apps at the same time 17:54 erle well, each electron app is another unpatched chrome version 17:55 erle and yeah, i'd bet money on any random electron app grabbing 500M to 1G of memory for itself minimum 17:55 MTDiscord erle: I don't think the error message did that but I'm too lazy to look it up right now 17:55 MTDiscord erle: Yeah, that's why I sometimes have to close Discord when compiling :P 17:56 Desour > empty MTG terrain is fast to render, nodeboxes and mesh nodes not so much, and semi-transparent stuff not at all 17:57 Desour it's not about the nodeboxes or meshes. it's rather the number of different tiles 17:57 Desour because that's what causes the many meshbuffers and drawcalls 17:58 MTDiscord I think bindless textures will provide a decent way to fix this for people on moderate hardware 17:58 MTDiscord (not erle, I'm sorry) 17:58 erle LOL 17:58 erle luatic you are probably the only dev for who i believe that 17:58 MTDiscord (Note that I don't think we should drop support for people without bindless texture support) 17:59 MTDiscord would the texture atlas pr help? 17:59 MTDiscord or completely unrelated here 17:59 MTDiscord The texture atlas is an alternative approach 17:59 erle would it help to draw them in different order? like, try to draw as much with each material as possible? or is that already done? 17:59 MTDiscord I'm inclined to believe that on hardware which supports bindless textures, it is pretty much strictly worse 18:00 erle luatic even being able to run discord sets a minimum on the hardware. i have some doubt about even owning a computer that can run discord (and i have 3 old thinkpads and 2 newer arm laptops). i have access to 2 computers that probably can, but both are owned by my employer. 18:00 erle “newer” as in, an arm laptop from 2018 (reform 1) and from 2021 (reform 2) 18:01 MTDiscord erle, what are the specs of your current machine that it can't run Discord? 18:01 ROllerozxa mark.wiemer: it's a thinkpad R60 18:01 erle mark.wiemer i do not know what exactly prevents discord, because other stuff runs fine. i even used matrix with it (but it used a lot of resources) 18:01 erle it can do voice and video calls 18:02 erle but i have 2G RAM and Genuine Intel(R) CPU T2400 @ 1.83GHz 18:02 MTDiscord weird, yeah 2GB ram should be enough but I guess Discord, like most apps, doesn't optimize RAM usage, surprise surprise 18:02 erle perfectly fine for low-spec stuff like 3D RTS or minetest/luanti, but discord makes the computer hang 18:03 erle well, not entirely hang, but stuff-is-very-choppy-until-i-quit-the-signup-dialog 18:03 MTDiscord Yeah not worth it. I'm new to IRC, can I ask how you connect? 18:04 sfan5 Desour: then cheapie could also test the wip texture atlas PR that's open 18:04 erle you can do this: /CTCP erle VERSION 18:04 erle oh wait you are not on IRC 18:04 Desour @luatic: AFAIK bindless textures aren't *that* common. i.e. according to opengl wiki, intel often doesn't have it https://www.khronos.org/opengl/wiki/Bindless_Texture#Extension_implementation 18:04 erle mark-wiemer i use hexchat. why? 18:05 MTDiscord erle, just curious, still learning 🙂 18:05 MTDiscord Desour: I've heard about that, I don't think it's terribly problematic, but we'll have to see 18:05 MTDiscord mark: https://libera.chat/guides/clients 18:05 ROllerozxa mark.wiemer: this channel is #minetest over at libera.chat. install an IRC client, connect and register an account: https://libera.chat/guides/ 18:05 erle mark.wiemer, about *any* IRC client will be less resource desiring than discord if that is your question 18:05 ROllerozxa (if you want to get deep you can set up a bouncer like ZNC, but if you're on discord you get a free bouncer through the bridge) 18:06 MTDiscord Thanks guys, I'll look at libera.chat 🙂 18:06 erle the most software projects i have seen that *only* use discord are some that have absurdly high hardware requirements (e.g. gaming) 18:07 erle mark.wiemer https://hexchat.github.io/ is the client i use, but there are a TON of irc clients. just make sure you register your name with libera.chat so no one else can log in as you while you are away. 18:07 Desour @luatic: the thing is, we'd likely need to keep a fallback path for systems that don't support bindless textures, which makes the code more complicated 18:07 MTDiscord I know 18:08 ROllerozxa I wonder if anyone has stepped up to fork hexchat now that it has officially been abandoned 18:08 rubenwardy I use quassel as it has good sync between devices 18:09 rubenwardy znc but better 18:09 Guest71 (Mark here testing out web.libera.chat) 18:09 MTDiscord (OK I won't spam anymore, sorry) 18:10 SwissalpS hi mark, pong 18:10 MTDiscord 🙂 18:10 erle ROllerozxa why and when has it been abandoned? any info? 18:10 ROllerozxa erle: last news article, https://hexchat.github.io/news/2.16.2.html 18:11 ROllerozxa not that an IRC client needs much changing nowadays but it's unfortunate it never saw the jump to at least GTK3 18:11 erle oh well 18:11 erle i don't see the issue with not making such a jump if it can keep working 18:11 erle i mean i don't miss gtk4 in XFCE either 18:11 erle all it would achieve is … losing theming? 18:11 MTDiscord hexchat is no more, everyone switch to base64chat 18:13 ROllerozxa I was considering switching to one of those CLI IRC clients because *cool*, but I instead just switched to another GUI client, quassel 18:13 ROllerozxa if I frequented IRC more I would probably have tried to learn some CLI client but alas, it is not 2016 anymore 18:15 MTDiscord personally use thelounge 18:15 sfan5 same 18:16 erle ROllerozxa have you checked ii(1)? :D 18:16 MTDiscord last i knew warr used it as well 18:17 Desour ROllerozxa: I've seen that you maintain a bunch of mod and subgame packages in the AUR. out of interest, do you spend much time updating them? also, does it even make sense to have them if we have in-game contentdb gui? 18:18 ROllerozxa erle: "ii is a minimalist FIFO and filesystem-based IRC client." haha oh gosh, I was more thinking something like weechat 18:19 MTDiscord Desour: we don't use the s word anymore ;) 18:19 MTDiscord well our code unfortunately does, but we don't 18:19 Desour I do 18:20 Desour it's just so more unambiguous ;) 18:20 erle the s word 18:20 erle sense 18:20 MTDiscord lol 18:20 Desour erle: ii=-1, not 1, btw 18:20 MTDiscord subgame. it is indeed clearer than game. 18:20 ROllerozxa Desour: I still sorta maintain them, update them whenever people mark them as outdated. I was planning to make creating and updating packages as automatic as possible for me through the CDB API but never got around to doing it so now they're just sitting there. they probably will not survive the rename, most likely 18:21 MTDiscord Desour: well, they were just multiplying i² with 1 so I think it's fine ;) 18:21 Desour ROllerozxa: :eyes: 18:21 erle Desour no idea how it is for AUR, but a) system wide installs exist b) for distributions that do not always have the newest engine, packaging definitely makes sense. 18:21 [ erle: GTK2 has the major issue of not having native wayland support 18:22 Desour @luatic: mhm, i^2*1=-1*1=1, yes 18:50 MinetestBot 02[git] 04sfan5 -> 03minetest/minetest: Fix ScriptApiSecurity::checkPath mangling non-existent paths 13294a30e https://github.com/minetest/minetest/commit/294a30e445ec9cd94d66efc8f94048606716ad35 (152024-11-03T18:27:08Z) 19:10 cheapie [12:04:00] Desour: then cheapie could also test the wip texture atlas PR that's open 19:10 cheapie Does it actually /work/ this time around? 19:11 cheapie I seem to remember 0.4.3ish doing some sort of atlas stuff, and all it accomplished back then was an error message every time it started. 19:28 erle when is the next game jam going to be? 19:30 ROllerozxa there is one like, right now 20:24 erle what i don't get btw, is why “make game crash at startup if user has no shaders” is not handled like every other change of support. even the BMP removal required a first-deprecate it, then remove it in the next version. 20:24 erle Krock btw, unless fixed, that *is* something that belongs into the changelog 20:26 erle luatic can you help me catch the shader exception? i found where it is thrown, but do not know what causes it. 20:26 erle alternatively, i wonder if just using the code that causes the exception to set the shader setting would be viable. at that point the game does know whether shaders are supported or not. 20:27 erle but i worry it is too deep in one code path 20:27 erle do you use exceptions as control flow in C++? in python it is pretty common 20:31 MTDiscord erle: yes, we use exceptions as control flow as reasonable 20:32 erle luatic well, is catching the exception from shader.cpp (or so) in main.cpp and then setting some value based on that *reasonable* or not? 20:32 erle i do not know idiomatic cpp 20:33 MTDiscord erle: seems okay-ish to me. it's not the cleanest imo, generally i like if there aren't too many layers between throwing and catching, but if the exception is specific enough it should be ok. 20:34 erle luatic well the other method could be to query the driver if it can do that and then set the value right at the start, but that would make the setting useless (e.g. for those GPUs that can do shaders, just badly, there would be no way to disable it) 20:34 erle the “shaders are recommended” settings thing should have checked if they *work* though 21:34 * cheapie looks through old screenshots, hmms at some road tunnels that more or less haven't changed at all since they were put in back in 2015 21:36 cheapie Both on North Street on VE-Creative, specifically the Robert D. Toukmond tunnel and Thomas E. Stock tunnel. Nobody ever gets the references on those...