Time Nick Message 12:15 sfan5 merging #15636, #15631 12:15 ShadowBot https://github.com/minetest/minetest/issues/15636 -- Fix show_debug setting causing inconsistency between debug control and shown debug info by Desour 12:15 ShadowBot https://github.com/minetest/minetest/issues/15631 -- Rename getGameMinetestConfig to getGameConfig by wrrrzr 12:47 Krock FYI: I'll be mostly absent for a few weeks starting from tomorrow. 15:05 sfan5 merging #15634 after I have turned on the laundry 15:06 ShadowBot https://github.com/minetest/minetest/issues/15634 -- [no sq] Use polygon offset to fix z-fighting for overlay tiles by sfan5 15:07 sfan5 and #15555 too 15:07 ShadowBot https://github.com/minetest/minetest/issues/15555 -- Add unit test for raycasts falsely skipping nodes by appgurueu 15:08 sfan5 Krock: enjoy your vacation 15:18 Krock sfan5: thank you! :) 16:31 MTDiscord The new raycast unittest failed: https://github.com/grorp/minetest/actions/runs/12621061165/job/35167475123 16:31 MTDiscord [FAIL] test_raycast_noskip - 27ms 16:31 MTDiscord ...test/bin/../games/devtest/mods/unittests/raycast.lua:73: assertion failed! 16:43 sfan5 looks like it's flaky 16:53 sfan5 works locally for me 17:02 Krock already the 2nd "works on my machine but randomly fails on CI" bug. That's not nice. 17:07 sfan5 well this one uses math.random so it can be explained to some degree 17:21 MTDiscord Hurm.... Sounds like I/we need some stronger offline testing, besides the CI 17:21 MTDiscord Something that runs a lot more, on a lot more platforms (qemu?) 17:22 sfan5 how would that have helped? 17:24 MTDiscord Anything where rand is involved typically takes a ton of test cases to catch edge cases 17:34 Krock seed the rand beforehand so that it's a guaranteed sequence 17:47 Krock Meeting in 13 minutes, if there's anything to discuss. 18:00 MTDiscord https://github.com/minetest/minetest/pull/12142 18:28 [MatrxMT] Thanks for the meeting haha 18:34 MTDiscord Maybe I could gauge interest: Should I consider adding a way for the engine to automatically turn .gltf into .b3d (when possible)? Would someone be willing to review that? 18:36 MTDiscord And/or conversely a way to turn .b3d into .gltf (needs just a .gltf writer for SkinnedMesh essentially). 18:37 MTDiscord seems like something that would become legacy debt to support, and potentially be a source of issues 18:38 MTDiscord for the first one 18:39 MTDiscord Definitely push people away from .b3d, the format itself is a source of crashes in Luanti for large meshes 18:39 MTDiscord the latter seems like a good idea for an external tool, since you cant import (with animations) b3d into blender 20:01 Krock @luatic I could imagine having a conversion tool baked into Luanti by CLI. That could lower the entry barrier. However, how many releases do have sufficient gltf support? Players first need to update. 20:03 Krock Servers need to be recent too. I don't think there's any errors raised by unknown media formats. Clients simply wouldn't receive the models. 20:03 Krock that's however a concern for modders, or guides respectively. 20:52 rubenwardy Converting gltf to b3d could allow more modders to adopt gltf. I feel like it may turn out to not be so easy in practice 20:55 Desour @luatic: I'd review (after at most 1 month). gltf->b3d would be good to make adoption of gltf easier for modders. b3d/x->gltf would be useful to drop b3d,x from the protocol 21:33 Desour I've been able to reproduce the failing devtest unittest locally. but unittests should really print their randomness seed 21:58 Desour and it also depends on pos. works with `pos = vector.new(6, 32, -3); math.randomseed(1596190898)`, if someone wants to test. (sanitizers gave me nothing) 22:00 MTDiscord I'll look into fixing it 22:01 MTDiscord I'll also make devtest unittests seed random and print the seed while I'm at it. 23:22 MTDiscord It looks like the unit test is correct and there is an edge case bug with our raycasts