Time Nick Message 03:31 MTDiscord Well, it's not using type restriction on its generics lol 05:00 MTDiscord Neko, I greatly appreciate your distaste for the raw pointers. But the output streams are so deeply ingrained into every facet of the engine, it would take a truly monstrous pr to replace them 05:01 MTDiscord I'm talking, 9/10 files you come across with an implementation 05:18 MTDiscord are you referring to me complaining about our log streaming? 05:19 MTDiscord looool, i havent figured out exactly how our logging system works well enough, so i've just been using raw printf calls or iostreams 05:21 MTDiscord I was already looking to do the, once thought to be impossible, soon... swiping out a lot of Irrlicht's builtin utilities. Might put on some music and zone out, just replacing shit bit by bit 05:38 MTDiscord The only interface you need to logging is to use one of the streams to output (errorstream, warningstream, actionstream, infostream, verbosestream, tracestream) corresponding to the log level you want 05:38 MTDiscord additionally, rawstream outputs to everything 05:39 MTDiscord should probably not ever use that except for debug prints that you intend to remove 05:40 MTDiscord The complexity is due to wanting it to be somewhat fast (each thread has its own local char buffer, so that each operator<< does not require a separate mutex lock and fwrite) 05:40 MTDiscord and also so that verbosestream is (almost) zero-cost when verbose logging is disabled 05:41 MTDiscord or tracestream more importantly 07:56 sfan5 nekobit: what would be the practical advantages? 07:58 sfan5 supporting opengl and opengl es is a solid basis and if apple actually drops opengl they will have to live with the consequences of drastically reduced application availability, because few people have the resources to maintain a special graphics backend just for one platform (nor will we) 07:59 sfan5 what I don't know is how opengl looks on consoles if we ever want to expand to that in the future 07:59 sfan5 also note that with the unified opengl3 / gles2 driver we have in irrlicht the "get rid of awful legacy gl code" reason starts to fade 08:00 sfan5 lastly a question: if we were to use bgfx, what happens to our glsl shaders? we can't afford to maintain multiple variants 08:00 nekobit opengl support is already ass on Windows, and directX comes out with new versions each time i forget what year it is 08:01 nekobit macOS? i think thats quite the audience to ignore, esp considering some of our devs 08:01 nekobit what about Vulkan? 08:01 nekobit are we maintaining multiple variants? 08:02 nekobit https://bkaradzic.github.io/bgfx/tools.html#shader-compiler-shaderc 08:02 nekobit but it sems like some things would need to be ported 08:03 nekobit its essentially vanilla GLSL tough with some nice changes 08:03 sfan5 "nice" is relative but that answers that 08:03 nekobit i really cant objectify why we'd want to keep irrlicht's gpu shit which is probably full of little issues 08:03 nekobit tons of games are backing and battletesting BGFX from what i see 08:04 nekobit irrlicht? supertux. Take it or leave it. 08:04 nekobit *kart 08:04 nekobit i dont trust a game that makes disney sound effects to justify an outdated graphics library and-then-some 08:05 nekobit nice isnt relative considering its done to be portable 08:06 nekobit if you can objectify against bgfx over anything else, than be my guest 08:06 nekobit but ive dug around it all night, especially with EFL as a stub library, and i genuinely cant see why we couldn't consider it 08:06 sfan5 not really. I have worked with projects that use shaderc directly without a GLSL dialect and it works fine. 08:07 sfan5 either way "irrlicht's code is probably bad" is not a good reason. have you read the code of the unified driver? 08:07 nekobit ive read enough irrlicht code and ive already gotten sick of it mate 08:08 sfan5 that's a no then 08:08 nekobit its a no 08:08 nekobit i mean ive even though of throwing bgfx as an added driver to irrlicht for now 08:09 nekobit shaders might be a hack for that 08:09 nekobit but if its stable, and people seem to like it... then we decide further, i suppose, not sure what point i was making? 08:10 sfan5 "opengl support is already ass on Windows" citation needed btw. vendor drivers provide the implementation and it's just as important for gaming as other apis (vulkan) 08:11 nekobit for amd gpus 08:11 nekobit all i can say 08:11 sfan5 no, for all gpus 08:11 nekobit pardn me, opengl spport is already ass on amd gpus 08:11 nekobit sorry, just tired 08:11 nekobit but if we are already considering axing out irrlicht's gui shit and its mediocre libraries, then.. hard to justify keeping its 08:12 nekobit im gonna sleep 08:12 sfan5 oh that. news to me 08:12 sfan5 as for macOS we have been lucky if we barely have even a single person to look at macOS issues so it's not like (effectively) the platform is super important to us 08:12 nekobit ruben would like to chat? 08:12 nekobit and probably many players 08:12 nekobit its as gross as keeping irrlicht than to justify not caring about macOS users 08:13 nekobit not that i care about them either 08:13 nekobit just sayin 08:15 sfan5 an open source project lives from its contributions and we've barely had a macos dev in the past. if apple now decides that everyone supporting macOS has to invest 40 hours writing a new 3d backend then we will simply not do that. not because we don't like macos or because someone decided not to support them but simply because we CAN'T. 08:15 sfan5 anyway 08:15 sfan5 in conclusion: I can very well believe that bgfx is generally better than whatever irrlicht (or we in the meantime) came up with but personally I'd like some concrete arguments for it 08:16 nekobit with bgfx we wont have to 08:16 nekobit i dont think anyone wants to 08:16 nekobit only way to keep them afloat in the case of opengl being dropped (very likely, even if they "lose users", dont forget that x32 was dropped and essentially killed steam), all we would have is ANGLE for that 08:16 nekobit and ive never used nor can i vouch for ANGLE other than what is written on the box 08:17 sfan5 there is also a proprietary implementation of GL on Metal https://moltengl.com/ 08:17 nekobit we talked in 'cord a tiny bit, i want a graphics dev to really vouch for it though 08:18 nekobit but as i said, we are obviously planning on axing out a lot of irrlicht, its been mentioned far too often, so it all boils down to if we should keep the most mediocre thing that irrlicht does well if we strip it of all its nutrients 08:19 nekobit gute nacht 08:20 sfan5 if you want an opinion from other people than me feel free to open an issue 08:21 MTDiscord yah will in the morning maybe but let me sleep before i die 08:21 MTDiscord If you want to use bgfx that's going to need to literally be an entire development 0.Y.0 because that's going to break so many commits 08:21 MTDiscord which is my fault, not urs 08:21 MTDiscord its gonna break shit fo sho 08:22 MTDiscord but we already seem to have a policy on axing old commits 08:24 MTDiscord I can only imagine the hell it will be to remove the irrlicht pointers 08:24 sfan5 here's the code I mean by the way https://github.com/minetest/minetest/blob/master/irr/src/OpenGL/Driver.cpp 08:24 sfan5 this is rather new, and different from the mess the OpenGL 1 driver is 08:24 MTDiscord Let me 👃 08:26 MTDiscord https://github.com/minetest/minetest/blob/master/irr/src/OpenGL/Driver.cpp#L1684 08:26 MTDiscord ? 08:27 MTDiscord i mean dont think youd ever need that many but just seemed funny hah 08:28 MTDiscord What is an scolor 08:29 MTDiscord Scholar 08:32 MTDiscord You're not supposed to use cassert anymore 08:33 MTDiscord Maybe I should just start firing up prs but it might send me into a void of never ending insanity 08:42 MTDiscord There are pointers getting set to 0 instead of null, no docs, uses non standard types, I have been defeated by words 08:53 MTDiscord I think the non standard types, possibly an easy fix. Just tedious 10:04 MTDiscord Please anybody re-review the ambient light PR with the new fixes included recently? 10:04 MTDiscord I would like to try to get it merged in 5.10 10:23 [MatrxMT] Reminder: 5 days to merge features, then it's feature freeze 11:08 MTDiscord I hope the feature freeze isn't as long as last time 11:20 [MatrxMT] The goal is to end it in two weeks, as agreed a few months ago, following the new release cycle 11:23 MTDiscord it'd be great if someone reviews the animated gltf pr so we can ship animated gltf in 5.10 (and not just static gltf). i can offer to review a pr in return. 11:26 [MatrxMT] I guess we won't be able to enable SDL by default with 5.10, too many things missing 11:27 [MatrxMT] #14964 and #15010 11:27 ShadowBot https://github.com/minetest/minetest/issues/14964 -- SDL: Use scancodes for keybindings by y5nw 11:27 ShadowBot https://github.com/minetest/minetest/issues/15010 -- OS doesn't receive keys captured by Minetest [SDL] 11:28 rubenwardy :'( maybe december 11:28 rubenwardy SDL2 for Christmas 11:28 Desour which december? 11:41 [MatrxMT] Next one is end of January 16:49 MTDiscord Rockin' around the SDL. Have a fancy minetest scene 16:55 sfan5 speaking of feature freeze: #14659 16:55 ShadowBot https://github.com/minetest/minetest/issues/14659 -- [no sq] Generic IPC mechanism between Lua envs by sfan5 17:52 MTDiscord i think i can look at that tomorrow 18:38 MTDiscord ill try to roll my scrollbar fixes tonight maybe 18:38 MTDiscord before feature freeze 18:42 MTDiscord it will improve trackpad scrolling dramatically 18:42 MTDiscord and in general 19:29 Krock will merge #15244 and #14623 in 15 minutes 19:29 ShadowBot https://github.com/minetest/minetest/issues/15244 -- Document performance cost of use_texture_alpha=blend by appgurueu 19:30 ShadowBot https://github.com/minetest/minetest/issues/14623 -- scrollcontainer: Add automatic scrollbar calculation by SmallJoker 19:44 Krock merging 19:45 Krock done 22:52 MTDiscord sfan5: seems https://github.com/minetest/minetest/pull/15257 has the same pr linked twice as like conflicts