Time Nick Message 00:15 lebruhgamer[m] "being poor is more a mentality..." <- this is such an entitled bigoted loaded BS I've ever seen from someone on the internet, seemed like an asumption that people can be not poor by doing anything to get money -- are you suggesting people should be doing illegal stuff just to not be poor?? for example, by selling something to get money when selling stuff even in small amount require license in some country but you need 00:15 lebruhgamer[m] to pay to get the license to begin with; or by robbing the bank; or by trying to get job by lying because you're not qualified to get that job; or by smuggling out of the country to get job in other country because you can't apply for passport due to government discrimination; etc?? there's always circumstances for people being poor and not just a mentality, if you think otherwise then YOU are part of the problem that caused 00:15 lebruhgamer[m] poverty, put your foot down and see actual people around you, stop being so ignorant 00:59 schwarzwald[m] What causes poverty is relevant to development decisions how? 01:03 lebruhgamer[m] idk, my chat earlier was just replying to them 08:32 MTDiscord lebruhgamer: please be aware that this is MT-dev 08:33 MTDiscord there is a decision to be made with how old hardware we want to support 08:34 celeron55 kilbith's opinions don't have anything to do with that though, that's just offtopic 08:34 MTDiscord yes 08:34 MTDiscord we should actually figure out how many people are running on how old hardware though 08:35 MTDiscord I suppose if you can afford Internet you can also afford 5 year old hardware 08:35 MTDiscord (just a guess though) 08:35 celeron55 well that's definitely not true 08:35 celeron55 but let's say 12 year old hardware 08:35 MTDiscord which hardware lasts 12 years nowadays ;-) 08:36 celeron55 many students and such get internet for free, but have to pay for hardware 08:36 celeron55 so internet is top notch but hardware is utter crap 08:36 celeron55 a lot of hardware lasts 12 years easily. not the consumer stuff, but business stuff 08:37 celeron55 that's for PCs of course 08:37 celeron55 when talking about android, it's a shitshow 08:37 celeron55 they sell phones new that are basically useless for any apps 08:37 MTDiscord so we draw the line at the top-notch hardware when Minetest was written? 08:38 celeron55 well, i think that could happen to be a reasonable limit 08:38 rubenwardy should be a sliding limit, top end hardware 12 years ago 08:41 celeron55 i can imagine what eg. erle would say to this. something along the lines of "everything should be supported if possible" 08:42 celeron55 but MT can't really afford to lose the top end hardware users of today by being held back by extremely old hardware 08:42 celeron55 that's the general problem i think 08:43 celeron55 top end hardware 12 years ago kind of matches the 200€ laptops you can get new today (that nobody should buy) 08:46 celeron55 the only way the latter is better is it that it consumes about 5% of the power of the former 08:46 celeron55 -it 08:46 MTDiscord oi, looks like you can get a thinkpad with 8gb RAM and a core i5 for 250 bucks nowadays 08:46 celeron55 new? 08:47 MTDiscord arguably refurbished 08:47 celeron55 i guess i'm getting old. where are the atoms and other slow processors? 08:48 celeron55 i'm not interested in used prices. you can get great hardware for free if you're not in a hurry 08:49 rubenwardy another issue is that current cheap hardware tends to support newer standards than old topend hardware 08:49 rubenwardy also iGPUs are much much better these days 08:50 rubenwardy although top end hardware wouldn't be iGPUs 08:51 celeron55 12 years old integrated gpus are pretty much obsolete today 08:51 celeron55 10 years old could still make sense 08:53 celeron55 the intel HD series was launched 2010 and in a year or two from that it got into normal hardware 08:54 celeron55 any intel HD is perfectly adequate for MT 08:54 celeron55 anything before those, i think it obsolete 08:54 celeron55 is* 08:55 celeron55 and here, erle would got mad with his gma900 series 09:01 nrz i started to dev MT on a radeon HD 4890, which is very obsolete for today, but always good for MT, as we don't added fancy graphics features ? 09:01 celeron55 anyway that's pretty much my view on this. 12 year old with an external GPU and anything with intel HD should still run MT 09:01 celeron55 12 year old top spec, i mean 09:03 rubenwardy anyway, there's a difference between old and cheap. Eliteism, bigotry, and such ignorance is not acceptable 09:04 celeron55 basically like... the official system requirements of windows 10 09:04 celeron55 except that MT actually runs on this hardware, i wouldn't say so about windows 10 09:33 sfan5 on that topic we should figure out how many users can't run opengl with shaders 09:34 celeron55 it's basically the users running pre-intel-HD iGPUs 09:34 celeron55 that i just proposed to shove under the bus 09:35 celeron55 that's precisely where the pain point has been for years now 09:35 celeron55 i agree getting some numbers wouldn't hurt 09:35 sfan5 hopefully no "well it can technically run shaders but not at acceptable performance" issues crop up with iHD then 09:36 celeron55 no they won't. it's so much faster 09:36 celeron55 of course depends on the shaders 09:38 celeron55 in 2010 it was considered ridiculous MT didn't require or use shaders. 12 years have passed since that. if i happened to have a computer with an external gpu back then, MT would have never supported the fixed pipeline 09:38 celeron55 it's quite the dice roll 11:16 kilbith "Eliteism, bigotry, and such ignorance is not acceptable" -> I didn't add up to the topic to respect the channel rules but do not fuel the flames please 11:16 kilbith and also fuck off, you have a personal issue with me 11:37 MTDiscord funny you're the one saying that 13:15 MTDiscord Please don't feed the trolls 16:41 MTDiscord I’d just have a rolling 10 year window with each release for a minimum supported Intel CPU generation 16:42 MTDiscord If the iGPU doesn’t support OpenGL 3x then drop it 16:47 MTDiscord A 600 nvidia GPU or later should run Minetest under any CPU generation just terribly 17:16 celeron55 well i'd hope we'd require opengl 2 first 17:29 kilbith GL 2.1 is 20 years old 17:31 Desour wikipedia says 16 years 17:31 kilbith GL 3.x is a more reasonnable guess, especially that we can do things more efficiently with the glMultiDraw functions that are vacant from OpenGL 2.x 17:32 kilbith any refurbished 200€ laptop today can support OpenGL 3.x 17:33 kilbith and anyone in the Western world is able to spend 200€. go work at McDonald and boom, you're not poor anymore! 18:01 MTDiscord hell haswell CPUs that are close to turning 10 years old support 4.2 18:02 MTDiscord which are still capable of excellent game performance and even run a current day GPU without much issue 18:02 MTDiscord it's just the thinkpad cult who believes they should be catered to 18:04 MTDiscord even if it's the detriment of everyone else 18:05 MTDiscord oh my old haswell from 2014 supported 4.3 18:06 celeron55 anyway does someone not agree intel HD is a reasonable minimum requirement? i think those do support gl 3 18:06 celeron55 or more like these, i have one also... like millions of people 18:07 MTDiscord yeah intel HD is the reasonable bareminimum since they're the same iGPU on mobile 18:07 Desour will there be a separate minimum requirement for android? 18:07 MTDiscord "if it runs it runs" 18:07 Desour (or other devices that currently use the gles stuff) 18:07 MTDiscord nobody can guarantee manufacturers to not do stupid shit with their ES drivers 18:08 celeron55 the main platform for MT is PC, requirements on other platforms have to be derived from that in some reasonable way 18:09 MTDiscord i'd treat android on a per manufacturer basis for how good are their drivers scenario 18:09 MTDiscord rather than pure performance 18:09 rubenwardy The minimum requirements for Android should be higher by age but lower by power 18:09 rubenwardy Like, I wouldn't support older than 5 years 18:10 MTDiscord that's a reasonable minimum given the pace of the mobile market 18:10 celeron55 well, if we choose to drop GL<3.0 and GLES<3.0, then that defines the devices 18:10 celeron55 not that any manufacturer advertises anything based on what GL version it supports 18:10 celeron55 it's going to have to be a community maintained list 18:11 MTDiscord similar to the old cyanogenmod compatibility list 18:11 Desour everything supports opengl 4.6, with the mesa software thing 18:11 MTDiscord (so does my M1 mac under linux) 18:12 Desour btw. is it planned to have separate implementations for a bit older and very recent gpus? 18:12 MTDiscord that would be bloat 18:13 MTDiscord and requiring two lots of maintenance 18:13 Desour but it would be sad if we can't use (for optimization) any features that exceed the new minimum 18:25 celeron55 there probably isn't enough resources to do that 18:25 celeron55 it would require new engine internal apis to be designed and implemented 18:26 MTDiscord just list the required GL[SL] requirements when you check it's box 18:26 schwarzwald[m] It's not like Minetest is out of stuff to improve and needs more. 18:29 MTDiscord if people want the graphics fancies turned on they should have the requesite hardware, but not locking out the base game out either 18:29 MTDiscord having them as pure optional with (X OpenGL requirement) is fine 19:22 celeron55 that more than doubles the workload though 19:22 celeron55 with those requirements i personally aren't expecting anyone to want to develop gl 3 stuff 19:23 celeron55 that's what i talk about when i say old systems are holding MT back 19:24 celeron55 anyway personally i don't mind, it's just that it may not be in the best interest of the project 19:25 MTDiscord for every 100 users the fancy graphics bring in, 1 is likely to end up writing engine code 19:28 celeron55 every year opengl 1 and 2 become more obsolete. one of these years must be the year that they're dropped 19:29 MTDiscord macOS can run upto OpenGL 4.2 via the usage of OpenGL contexts# 19:29 MTDiscord which is apparently the proper way to enable/disable certain OpenGL features 19:35 sfan5 do we have any concrete features from 3.0 we want or why is this being proposed? 19:36 Desour it would be very useful to have array textures 19:36 Desour to reduce the number of drawcalls per mapblock 19:36 rubenwardy New opengl versions have features that allow you to improve performance by communicating with the gpu less. Fewer draw calls 19:38 MTDiscord we should probably have MT 6 when we break a lot of things by chopping out even more irrlicht and introducing a hard requirement for OpenGL 3.x 19:39 celeron55 the problem is, i don't know enough modern opengl to propose any actual features to use 19:39 celeron55 if you search the logs for what hecks has been blabbing about, that'll give an actual introduction to the subject 19:40 celeron55 altough i wouldn't rely on him doing any of the work 19:40 MTDiscord i treat MT 6's "roadmap" or "milestone" as exploratory and experimental 19:40 MTDiscord if you don't know what to do with it, start experimenting 19:40 MTDiscord let mad science produce results and learn from them 19:40 sfan5 I don't remember what hecks mentioned but I do remember that he proposed writing our own renderer that'd essentially use the common subset of OpenGL 2.0 / GLES 2.0 19:41 sfan5 Desour: sounds like binding multiple textures would be more flexible there unless we run into limits there 19:41 celeron55 yes 2.0 was more of the intention 19:42 celeron55 altough hecks had plans for newer versions also 19:42 Desour btw. with modern gl, we could try to do client-side mapblock occlusion culling by doing occlusion querries (or something similar with atomic bools, because afaik there's just one occlusion querry at once allowed) of mapblock bounding boxed using the depth buffer of last frame 19:42 sfan5 irrlicht can do that IIRC 19:42 sfan5 for example I bet you could merge the buffers of the two most common nodes in mapblocks if it's stuff like stone & dirt 19:44 sfan5 I've been thinking that particles would be a good first target for an entire rewrite with shaders in mind because it's fairly small and self-contained (you don't get into the million options nodes have) 19:45 Desour sfan5: what I meant was that you then pass an index for the texture in the vertex buffer. AFAIK, this does not work if you use multiple bound textures, because `if (idx==1) col=texture(tex1, uvpos) else if (idx==2) col=texture(tex2,uvpos)` has to sample both textures for all framgents 19:46 sfan5 hmm, not sure 19:47 Desour bindless textures would also work, but they are just an extension (AFAIK many intel gpus still don't support them) 19:47 rubenwardy particles are especially well-suited for GPUs as well 19:47 rubenwardy assuming you don't want oncollide callbacks 19:47 Desour we can also do animation in shaders then btw 19:48 sfan5 looks like you can bind multiple textures into a sampler2D array 19:51 Desour ^ that does also not work. see https://www.khronos.org/opengl/wiki/Data_Type_(GLSL)#Opaque_arrays 19:51 Desour (at least if the sampler2Ds are bound textures and not bindless texture handle things) 19:52 celeron55 what i don't see is leaving out obsolete gl versions to have a large effect in comparison to just implementing more aggressive batching on the cpu side, basically just bigger meshes 19:53 celeron55 i'm trying to get to working on some of that 19:53 MTDiscord My pc is from 2009 and only supports 2.1 btw...maybe thew version is 15+ years old but the industry doesn't stop releasing older versions with newer hardware right away 19:54 MTDiscord Electronic stores here also still sell Android 9 devices 19:54 rubenwardy 2009 would be outside of the proposed support Window 19:54 Desour we could also do gl 2, but with some extensions 19:54 rubenwardy Will OpenGL ES work on old devices that support GL2? 19:55 rubenwardy if we were to support two renderers, say using bgfx, could we have a OGL ES 2.0 one and a OpenGL3/4/Vulkan one 19:56 rubenwardy this all feels quite hypothetical though, do we even have anyone that would want and have time to rewrite the renderer to use newer features 19:56 sfan5 anyway my entire point was that we should aim for opengl 2.0 and if whoever does the work notices that savings could be made using some newer feature we can talk about it 19:57 rubenwardy yeah that sounds like a solid approach 19:57 MTDiscord Would this OpenGL3+ only version be 6.0 then? 19:57 rubenwardy perhaps, but doesn't need to be 19:58 celeron55 what i'm saying is if someone with actual knowledge looks into it and propose to use features that aren't available in old gl versions, it will be considered 19:58 celeron55 proposes* 19:59 celeron55 _without_ requiring them to implement a separate mode for supporting obsolete hardware 20:01 celeron55 how this relates to getting a more modern look for the rendered result... who knows. depends on the artistic skills of the graphics person i guess 20:03 rubenwardy I think if you can sprinkle some fancy effects whilst modernising and improving performance, you get both a faster game and pretty screenshots for release posts 20:06 celeron55 but you basically have to first respect them and only then expect to get something 20:06 celeron55 saying "but it has to run on fixed pipeline" to a graphics programmer is an insult 20:35 MTDiscord @IhrFussel I suppose if you don't mind, could you elaborate on why you're still using a 12 year old laptop? I get not needing to upgrade, but eventually software is gonna need newer stuff 20:35 MTDiscord Unfortunately with something like OpenGL, it's very much tied to the hardware 20:35 MTDiscord Sorry, 13 year old 20:36 celeron55 it's probably not a laptop 20:37 MTDiscord I thought I read laptop, same difference for the discussion I suppose 20:37 celeron55 a laptop of that age is practically useless, a top spec desktop can be useful 20:38 celeron55 my 8 year old top spec laptop is perfectly usable for anything i do for work or hobbies 20:38 celeron55 and it better be, it was very expensive and i better get my money's worth 20:38 MTDiscord I don't have much money so I simply looked for an used older pc 20:39 MTDiscord And until this one dies I don't think I'll need another 20:39 MTDiscord My mid-tier 6-year-old laptop does what I need to most things, but it doesn't game well. MT works alright on it though, and I've done IKEA dev on it 20:41 MTDiscord @IhrFussel I mean, at what point is older hardware not worth the dev time for us to support? If there are benefits to be had for using more modern tech, when should we be able to pull the plug on old hardware? 20:41 celeron55 IhrFussel: what CPU does it have? 20:43 MTDiscord Intel Pentium Dual-Core 20:44 celeron55 so the predecessor to i3 basically 20:45 celeron55 or well, basically the same thing, different name 20:45 celeron55 oh wait 20:46 celeron55 no, it's probably a 65 nm processor, Allendale or something 20:46 celeron55 similar to core 2 duo 20:48 MTDiscord Nope, 45 nm 20:49 celeron55 ok, that's basically the fastest processor you can possibly get without getting an intel HD iGPU with it 20:49 celeron55 it will still have the integrated GPU separate from the CPU so the GPU will be much slower than intel HD 20:52 kilbith https://www.amazon.fr/Lenovo-ThinkPad-X230-8Go-128Go-Reconditionn%C3%A9/dp/B00CBWH4YU/ref=sr_1_1?qid=1660855877&refinements=p_36%3A9733299031%2Cp_n_feature_twenty-three_browse-bin%3A27388180031%2Cp_n_feature_fifteen_browse-bin%3A8323130031%7C8323131031%7C8323132031&rnid=8323110031&s=computers&sr=1-1 20:52 celeron55 i guess one of those has enough single thread performance to make it still useful 20:52 kilbith Intel HD Graphics 4000 for only 180€ 20:54 celeron55 that's like 2.5x faster than the pentium dual-core and almost infinitely faster graphics 20:54 MTDiscord Btw I didn't say that MT absolutely needs to support my Open GL version in the future...it would be cool if there would be no breakage between server & client so that I could still play even with 5.6 or 5.7 (assuming 5.8 would be the newer graphics) 20:57 celeron55 i'd say whatever iGPU the pentium dual-core comes with will have the fate of pretty much a dice roll depending on the exact extensions it supports 20:57 celeron55 in the upcoming years 20:57 celeron55 it suffers from being a high end processor from a retiring cpu generation 20:58 celeron55 as you already said i think 20:58 MTDiscord Would it be possible to update the client's graphics while still allowing older clients to connect to newer servers? Or is that too much work? 21:00 celeron55 can't predict such details 21:01 celeron55 but it's true it would give a bit more time for old systems if the graphics updates wouldn't be done in an otherwise breaking release 21:02 celeron55 even this seems too much detail to plan now though. i don't think we have anyone wanting to do even GL 2 stuff atm 21:03 celeron55 i just wanted to bring this up so that nobody following development won't be surprised when it happens 21:04 rubenwardy updated #libresprite 21:04 rubenwardy heh 21:04 rubenwardy updated #12542 21:04 ShadowBot https://github.com/minetest/minetest/issues/12542 -- Prevent loading a world with unresolved dependencies by rubenwardy 21:04 rubenwardy need to check what's on my clipboard before pasting into IRC 23:01 diceLibrarian what did you do?