Time Nick Message 07:59 SFENCE_arch Btw: Bad project description and web link here: https://github.com/luanti-org/luanti-org.github.io, @celeron55 09:09 Krock where are the locale files cached in CMake? The incremental build fails due to the removal of ia/LC_MESSAGES/luanti.mo 09:25 Krock okay. the error is caused by "minetest.po~" files inside the "po/" directory. Apparently the locale scan by gettext is imperfect. Optimally it should skip over "empty" directories. 09:26 Krock they're remainders from the engine rename procedure, I suppose. The old files weren't removed. 09:40 celeron55 SFENCE_arch: fixed 09:49 MTDiscord @luatic possible mesh code bug? https://git.minetest.land/VoxeLibre/VoxeLibre/issues/4930 10:35 Krock @herowl Difficult to say. What's the mesh format? .obj? https://git.minetest.land/VoxeLibre/VoxeLibre/src/branch/master/mods/PLAYER/mcl_skins/models 10:35 Krock I've seen 180° rotation bugs before. There is/was a duct-tape fix for that in Luanti 10:37 Krock this part: https://github.com/luanti-org/luanti/blob/5.10.0/src/client/content_cao.cpp#L1507-L1546 10:39 MTDiscord B3d 10:41 Krock https://github.com/luanti-org/luanti/commit/8719a816e7 10:41 Krock you could try to revert this commit and see whether the issue disppears 10:42 MTDiscord https://git.minetest.land/VoxeLibre/VoxeLibre/src/branch/master/mods/ITEMS/mcl_armor/models 10:42 MTDiscord Krock: above is the correct link 11:02 sfan5 merging #15736, #15739, #15408, #15784 in 15m 11:02 ShadowBot https://github.com/luanti-org/luanti/issues/15736 -- Tweak main menu server list behavior by siliconsniffer 11:02 ShadowBot https://github.com/luanti-org/luanti/issues/15739 -- [no sq] Enable ipv6_server by default by sfan5 11:02 ShadowBot https://github.com/luanti-org/luanti/issues/15408 -- Collision: more accurate computation with acceleration and long dtime by kno10 11:02 ShadowBot https://github.com/luanti-org/luanti/issues/15784 -- Correct keycode URL in settingtypes.txt/minetest.conf.example by MiguelPL4 11:13 sfan5 also pushing a back to sdl2 commit after that 11:13 sfan5 or before actually 11:16 Krock If you'd like, please also merge #15733. 11:16 ShadowBot https://github.com/luanti-org/luanti/issues/15733 -- Clean up Irrlicht matrices a bit more by appgurueu 11:17 sfan5 I will 11:17 Krock thanks :) 11:18 Krock #15719 would be another candidate. In case you're concerned about stability - I'd argue that we're in early -dev phase 11:18 ShadowBot https://github.com/luanti-org/luanti/issues/15719 -- Optimize add_area_node_boxes in collision code by sfan5 11:19 sfan5 ah sure 11:23 sfan5 ok all done 12:56 MTDiscord Herowl: not a bug: https://github.com/luanti-org/luanti/issues/15692 13:44 MTDiscord So does that mean if I include scale (1,1,1) it'll work? 13:45 sfan5 someone still needs to create a forum topic for the release 13:53 Krock hijack the rc1 post :3 13:54 Krock nah I'll create a new topic 14:01 Krock https://forum.luanti.org/viewtopic.php?t=31369 14:07 sfan5 stickied it 14:08 Krock thx 14:49 MTDiscord Herowl: as I write in that issue, it depends. that is one workaround, but it also requires you to fix the rotation, and it'll only work for luanti client versions that support bone overrides with scale. 14:50 MTDiscord Fix rotations... how? And what will be the simplest solution? 14:52 MTDiscord by composing with the 180° rotation that would otherwise live in the scale 14:53 MTDiscord the simplest solution, which also works best across older clients, probably is to re-implement the broken decomposition 14:53 MTDiscord that is, if you have a 180° rotation along a single axis for a bone in the original model, turn that into negative scaling and pass that to set_bone_override 14:53 MTDiscord forcing new clients to exhibit the fixed bug so that the wrong rotations work again 14:54 MTDiscord i can write a workaround mod that does this 14:55 MTDiscord with modlib it's quite straightforward: read the b3d, turn the rotations into matrices, check if the sum of absolute values on the diagonal is close to 3 (that is, all diagonal entries have an abs value of ~1), if so turn the diagonal into the scale, hook set_bone_override to apply this 16:32 MTDiscord Do you think VL should just include modlib? 16:34 MTDiscord Anyway, I don't like reimplementing broken behavior... broken linear algebra even less so. 16:35 MTDiscord I'd be down for checking protocol versions or server versions or whatever and branching though. 16:37 MTDiscord problem is that the branching is nasty, because there are no per-player bone overrides. you could try something with per-player entities but that'd be nasty too. 16:37 MTDiscord No, that'd be dumb 16:38 MTDiscord Hmm 16:39 MTDiscord So would rotating by like 1⁰ every bone that exhibits the behavior fix it? 16:39 MTDiscord if you also fix the rotations you set via bone overrides, yes 16:40 MTDiscord How do I find out the correct new values? 16:40 MTDiscord precompose with the the 180° rotation that would previously have lived in the scale 16:40 MTDiscord Man I'd like to be able to just pass a transform matrix to the bone override and be done with that... 16:40 MTDiscord 😬 no 16:41 MTDiscord Why? 16:41 MTDiscord doesn't interact well with a number of things 16:42 MTDiscord I had to do mad calculations to add a new animation like two months ago... 16:43 MTDiscord for absolute bone overrides specifically, a matrix could be considered, but there are still a bunch of problems and things you shouldn't be able to do (but arguably probably are able to do already) 16:44 MTDiscord We also had hellish bugs about player swimming and flying turned on the back. 16:45 MTDiscord Actual heisenbugs mind you, where changing mod load order changed globalstep register order, which changed their execution order, which changed bone override order, which messed things up in a way that made it hard to track down. 16:46 MTDiscord And imagine my horror when bisect consistently led me to a commit that changed only mod.conf files... 16:46 MTDiscord oof 16:48 MTDiscord Tbh I'd just treat the whole player animation code in VL with fire, but I don't have a better way to determine all the angles and stuff, so it comes down to a bunch of overrides with magic numbers. Comments make it readable, but changing it in any way is insanely hard. 16:49 MTDiscord i mean part of what makes it hard certainly was the bug which 5.11 just fixed 16:49 MTDiscord What's the axis order then? 16:50 MTDiscord can't tell you off the top of my head but it's XYZ or ZYX and it should be possible to infer it by reading the source 16:51 MTDiscord fatal flashbacks 16:51 MTDiscord no i mean i have literally documented and unit tested it for some matrix methods 16:51 MTDiscord it's also not that hard to test 16:52 MTDiscord but beware: it's very well possible that the object and the bones use different conventions :juanchi_face: 16:52 MTDiscord :juanchi_face: indeed... :facebook: 17:07 Desour merging #15784 and #15774 in 10 units 17:07 ShadowBot https://github.com/luanti-org/luanti/issues/15784 -- Correct keycode URL in settingtypes.txt/minetest.conf.example by MiguelPL4 17:07 ShadowBot https://github.com/luanti-org/luanti/issues/15774 -- [nosq] Don't use fps_max_unfocused for the pause menu, and a little more by Desour 17:21 Desour merged. don't question my units 17:22 MTDiscord as long as you test your units, i'm happy ;)