Time Nick Message 10:11 Zughy[m] 7 PRs to <100 PRs, nice 10:13 MTDiscord yeah 11:51 appguru minetest/src/network/serverpackethandler.cpp: In member function ‘void Server::handleCommand_Init(NetworkPacket*)’: 11:51 appguru minetest/src/network/serverpackethandler.cpp:111:20: warning: comparison is always false due to limited range of data type [-Wtype-limits] 11:51 appguru 111 | if (depl_serial_v < SER_FMT_VER_LOWEST_READ) 12:18 appguru Using a fresh 5.6.0-dev shift sometimes gets "stuck" despite me not pressing it; were there any control-related changes? 12:26 sfan5 there werent 12:55 MTDiscord mhm it (almost) works 12:55 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/979729677138677800/Bildschirmfoto_von_2022-05-27_14-55-21.png 12:56 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/979729706960175144/Bildschirmfoto_von_2022-05-27_14-55-39.png 12:56 MTDiscord just need to clear the rotation lol 12:58 MTDiscord sfan5: this also allows me to get rid of the dirty mesh cloning because I properly set the transform 12:59 MTDiscord furthermore the translation is now done on the GPU 12:59 MTDiscord (not that it would matter) 13:24 erle luatic is this rotated entities or rotated nodeboxes? 13:26 appguru erle: rotated entities 13:51 appguru why is rotation a property of UnitSAOs rather than SAOs 13:55 appguru would anyone have a problem with me moving rotation to SAOs? 13:55 sfan5 what difference does it make 13:56 appguru I need the rotation for raycasts on the server to be accurate 13:57 sfan5 you can dynamic_cast<> a SAO to a child class if you need one of its methods 13:57 appguru So all SAOs are guaranteed to be UnitSAOs? 13:58 sfan5 yes but that's beside the point 13:58 appguru I'll just cast then 13:58 sfan5 if the cast fails then you just pretend there is no rotation 13:59 sfan5 (of course this is not proper OOP and all but eh) 14:02 appguru fine fine 14:06 rubenwardy gross 14:06 rubenwardy if the position is on SAO, then the rotation should be too imo 14:06 rubenwardy I also don't get the point of unitsao vs sao 14:06 rubenwardy also why isn't position and rotation on the activeobject 14:07 sfan5 that stuff could use a big cleanup 14:07 rubenwardy Just seems a bit complicated 14:08 rubenwardy In my game, I just have a single Entity (=object) class for both players and luaentities, no inheritence. It handles things like position, rotation, scale, and attachments. The behaviour is done externally 14:09 rubenwardy well, I have a lot less features 14:18 appguru I kinda agree with ruben here 14:18 appguru I guess I'll have to downcast CAOs to GenericCAOs :/ 14:20 Krock why does CONTENTFEATURES_VERSION exist? 14:27 Krock or rather: would it be okay to drop this concept in favour of PROTOCOL_VERSION? 14:35 sfan5 since there's no way for the server and client to agree on a contentfeatures version there's not much point in the extra version number 14:35 sfan5 but we have a lot of these 14:35 sfan5 maybe celeron55 knows what the original purpose was 14:36 erle https://en.wikipedia.org/wiki/Wikipedia:Chesterton's_fence 14:36 Krock plus IItemDefManager is basically useless. this could be implemented using "const" functions and return values 14:37 Krock probably inspired by Irrlicht style 14:45 Krock @ contentfeatures. I assume this was made to avoid a protocol version bump for every change. However, there's still >65000 bumps possible so I consider this as unnecessary worry 14:47 rubenwardy plus we're doing protobumps for each version anyway 14:48 rubenwardy huh, only used for sounds 14:49 rubenwardy oh nevermind, it's read into a variable in nodedef 15:14 Krock rubenwardy: but it's entirely unused 15:15 celeron55 whatever reason there was for CONTENTFEATURES_VERSION it's probably not relevant today, it doesn't seem worth it to even try to remember 15:18 Krock alright thanks 15:19 celeron55 MT's object stuff is much more entity / component flavoured these days than what it originally was in the C++ only versions 15:19 celeron55 while it's still built on the same base classes 15:20 celeron55 it's basically what you would expect given the history 15:20 celeron55 the current state, that is 15:21 celeron55 you indeed probably could just flatten it by not just one but a couple of layers 15:23 Krock hmm. sound handles are never removed server-side unless you manually stop them 15:29 appguru Workflows are dying for obscure reasons: "Error: Failed installing dependency: https://luarocks.org/say-1.3-1.rockspec - Error fetching file: Failed downloading https://github.com/Olivine-Labs/say/archive/v1.3-1.tar.gz - v1.3-1.tar.gz" 15:29 appguru But the file is there 15:29 appguru See https://github.com/minetest/minetest/runs/6627908360?check_suite_focus=true 15:30 appguru Anyways, allow me to introduce #12379 15:30 ShadowBot https://github.com/minetest/minetest/issues/12379 -- Allow rotating entity selectionboxes by appgurueu 15:50 Krock and what's the purpose of the Futurama quote? 15:55 appguru Krock: To aid in explaining the code obviously 15:55 appguru (I transform the ray rather than the object) 15:55 appguru also fun :P 15:57 celeron55 the workflow is running on github, and it's failing to fetch stuff from github? 16:01 appguru celeron55: figured it out 16:01 appguru the luarocks is outdated 16:01 appguru it needs to be bumped to 2.4.4 16:01 appguru current is 2.4.2 16:02 appguru but the ubuntu 20.04 repos are probably outdated 16:02 appguru so I'd suggest just bumping that to 22.04 16:03 appguru NVM, 22.04 doesn't seem to have a newer luarocks in its repos 16:04 appguru NVM I'm spouting BS 16:05 appguru 22.04 ships with 3.8 which is way newer 16:12 appguru upgrading to 22.04 isn't as easy though :/ 16:19 appguru alright, I found a solution: #12381 16:19 ShadowBot https://github.com/minetest/minetest/issues/12381 -- Lua workflow: Use Leafo's Luarocks action by appgurueu 16:54 x2048 I've build Minetest 5.6.0-dev for RPI, and when running I get GLXBadFBConfig. Any pointers where to look or what settings would work? 16:56 sfan5 make sure glxgears works 16:56 sfan5 first 16:57 x2048 Yes, it works 16:57 x2048 5.3.0 from flatpak also works 16:58 sfan5 at this point I'd check glxinfo for anything that look off 16:58 sfan5 but are you sure you don't want ogles support anyway? 16:59 x2048 I compiled irrlicht with ogles2 and selected the driver in the config 17:00 sfan5 that goes via EGL so if you get a GLX error then something went wrong there 17:00 x2048 Are there compile-time flags in Minetest? 17:00 sfan5 yes -DENABLE_GLES=1 17:02 x2048 Trying this ^ 17:03 x2048 Yes, Irrlicht's cmake complained it "could not find EGL (name mismatch)", but the properties for EGL are set 17:04 sfan5 which did you set? 17:06 x2048 I did not set anything specific for EGL 17:07 x2048 only installed the libraries and set OGLES2 in IrrCompileConfig.h 17:07 x2048 Checking CMakeLists.txt, EGL is checked/required only when OGLES1 is enabled, but not OGLES2 17:08 x2048 source/irrlicht/CMakeLists.txt is the file 17:09 sfan5 gles2 uses egl too it's just in another file 17:09 sfan5 my current guess is that you don't have the right mesa package installed but for more the cmake log would be useful 17:12 x2048 My guess is that /usr/lib/arm-linux-gnueabihf/libEGL.so confuses cmake scripts a bit, because it's not /usr/lib, but the path to the library it finds is correct. 17:12 x2048 My raspbian also has EGL_mesa in the same folder 17:16 sfan5 like I said, cmake log 17:17 x2048 -DENABLE_GLES=1 fixed it :) thank you, sfan5 17:17 sfan5 ? 17:44 sfan5 x2048: so how well does it run? ;) 18:39 x2048 without shadows I get 30-60 FPS in devtest with a range of 110 18:39 x2048 sfan5: ^ 18:48 Pexin what model of pi? 18:51 x2048 PI4 model B, 4GB RAM, not the latest one 18:52 x2048 Don't really remember whether "Model B" applies to RPI4 18:57 * Pexin ponders pi zero2 w 18:58 x2048 :D 19:01 x2048 I have a couple of pi zeroes. They could probably run a small Minetest server, if not the I/O speed. Have they fixed it in Zero2? 19:02 Pexin literally all I know about zero2 is it has 4 cores 19:02 rubenwardy pi zero2 w has massively better specs than the zero 19:02 Pexin I wish I got more zeros back before covid when they were still cheap 19:02 rubenwardy I think it's similar to pi 2 or 3, I forget 19:03 rubenwardy > Blending the form factor of the original Raspberry Pi Zero with the processing power of the Raspberry Pi 3, the new Zero 2 W is an upgrade on almost all fronts 19:04 rubenwardy from a medium post so your mileage may vary 19:04 Pexin it's probably for the best I can't get a zero2. if I had one I'd probably waste my life playing katamari on a playstation 2 emu. or at least waste my life trying to get a ps2 emu working on it 19:05 rubenwardy another source: https://hackaday.com/2021/11/01/the-pi-zero-2-w-is-the-most-efficient-pi/ 19:15 sfan5 rpi zero has the same slow cpu as the original first gen rpi 19:21 rubenwardy I preordered a rpi zero and I've pretty much never used it 19:21 rubenwardy I should use it to make a handheld emulator or smth 19:22 rubenwardy offtopic zone 19:28 Krock will merge #12304 and #12377 in 10 minutes 19:28 ShadowBot https://github.com/minetest/minetest/issues/12304 -- DevTest: Add nodes and items for testing overlays by Wuzzy2 19:28 ShadowBot https://github.com/minetest/minetest/issues/12377 -- Optimize misc helpers by appgurueu 19:39 Krock merging 19:40 Krock done