Time Nick Message 00:41 rubenwardy change log seems outdated. Missing my cdb security fix at least 00:45 rubenwardy also looks like we have a few last minute security hardenings/fixes which are not in the changelog and do not have advisories 00:53 rubenwardy android done 02:34 red-001 rubenwardy: it seems the changelog groups all the security fixes under one item at the end 02:36 red-001 https://github.com/luanti-org/luanti/security/advisories/GHSA-5cp9-g6f5-xv8p 02:36 red-001 you can move that to draft if you want a separate advisory 06:26 red-001 vcpkg caching is broken 06:27 red-001 what horrible software it is 06:28 red-001 https://learn.microsoft.com/en-us/vcpkg/reference/binarycaching#abi-hash horrible 06:28 red-001 they hash on the version of powershell used 06:28 red-001 why 06:54 red-001 rubenwardy: I think you complained about vcpkg before, what do you think of conan, etc? 07:17 sfan5 conan ships ancient luajit, not a good look 07:17 sfan5 (see irc logs the other day) 07:33 red-001 seems no search engine really works with the irc logs 07:53 red-001 #17481 hotfix for vcpkg stuff in CI 07:54 ShadowBot https://github.com/luanti-org/luanti/issues/17481 -- [CI] Make VCPKG ABI hash more stable by red-001 07:54 red-001 if that passes CI it would be great if someone could merge it in and delete the old caches 07:55 red-001 getting the actual ABI hash for vcpkg seems to rely on very carefully greping it out of user facing output, but I suppose should be using that for the cache key 09:10 sfan5 since we have the android builds ready I'll go and update the website 10:04 red-001 I feel like version could be a template parameter in that template 12:28 rubenwardy Krock: can you explain what "Improved collision handling" actually means for the user 12:28 rubenwardy is it just this PR? https://github.com/luanti-org/luanti/pull/17127 12:30 rubenwardy looks like a bug fix 12:30 rubenwardy "Collision handling accuracy has been improved, reducing jittery movement caused by floating-point errors." 12:32 pgimeno red-001: I think it starts here: https://irc.luanti.org/luanti-dev/2026-08-15#i_6336673 12:40 red-001 that's suboptimal 13:40 rubenwardy > I wanted to let you know: 13:40 rubenwardy When I launched Luanti today, it offered me a new version. After downloading version 5.17.0 from https://www.luanti.org/de/downloads/, 13:40 rubenwardy the Windows virus scanner quarantined the file, flagging it as a "Trojan:Win32/Wacatac.H!ml" Trojan. 13:42 rubenwardy rubenwardy: I think you complained about vcpkg before, what do you think of conan, etc? 13:42 rubenwardy well, personally, my solution was to not use C++ 13:43 rubenwardy I've not used conan 13:43 Sheriff_U3 They are probably trying to push people to MineCraft... 13:44 rubenwardy https://www.virustotal.com/gui/file/048c132b8b4e637741144d1b551ff64f6c562a286617d4d8ddbf17e974fb746a 13:48 sfan5 IIRC nsis stuff gets misdetected quite often 13:49 sfan5 the engine itself comes up all green https://www.virustotal.com/gui/file/1fb30fba4a244482031e2cb8b71e0d1dbdba469fdc942238c5498e30783dc74d/details 14:12 Krock rubenwardy: I mentioned it in case there's glitch reports, which I hope won't occur. Didn't go into detail because players do not find that interesting - at least usually. 14:13 Krock will merge #17401 #17411 #17449 in 15 minutes 14:13 ShadowBot https://github.com/luanti-org/luanti/issues/17401 -- Add textcolor to label docs by jordan4ibanez 14:13 ShadowBot https://github.com/luanti-org/luanti/issues/17411 -- Don't use getNodeNoEx for mapblock meshgen by cx384 14:13 ShadowBot https://github.com/luanti-org/luanti/issues/17449 -- [no squash] Misc fixes by SmallJoker 14:15 sfan5 feel free to add #17464, #17442, #17360, #17384 too 14:15 ShadowBot https://github.com/luanti-org/luanti/issues/17464 -- [no squash] Move some CI workflows to newer Ubuntu + more stuff by sfan5 14:15 ShadowBot https://github.com/luanti-org/luanti/issues/17442 -- [no squash] Misc fixes and filesystem stuff by sfan5 14:15 ShadowBot https://github.com/luanti-org/luanti/issues/17360 -- Improve volumetric lighting performance by ubuntuyou 14:15 ShadowBot https://github.com/luanti-org/luanti/issues/17384 -- Optimize image transform modifier by sfan5 14:20 Krock sure 14:30 Krock done 15:03 rubenwardy any final reviews for https://github.com/luanti-org/blog/pull/178 15:13 Krock rubenwardy hands leak?? 15:14 sfan5 needs more hands 15:15 Krock did a quick proofread. seems fine. 15:22 rubenwardy wish my nails were still painted, had to remove the varnish yesterday as chipped 15:22 [MatrxMT] what 16:36 MTDiscord rubenwardy's fingers are properties of rubenwardy, all rights reserved 16:57 MTDiscord <.zenonseth> let me post here instead: I mentioned this in engine discord a while ago > with the new collision fixes, pressure plates in InsideTheBox are kinda broken It's still an issue, just tested with 5.17 client on the Inside The Box server. The reason seems to be, iirc, that for nodes below y=0 the rounded player y position now rounds to the node below the player's feet, rather than at the player's feet. This is seemingly not a server issue 16:57 MTDiscord - ITB server is running 5.14, because just logging in with a 5.17 client breaks the pressure plates for me. The same logic may affect other games/mods, idk who may use it 16:59 Krock https://github.com/minetest-mods/mesecons/issues/726 17:00 Krock the previous imprecisions might've made it magically work 17:00 Krock now that it's clear that there's an issue - best fix up the affected mod(s) in this case. Well... as long the player coordinates are actually accurate. 17:01 MTDiscord <.zenonseth> luatic made that after i mentioned the issue, but afaik, it doesn't affect mesecons, tested locally with their pressure plate <0 works fine. ITB has its own logic, but I don't know how unique the code is 17:02 MTDiscord <.zenonseth> also i've already made a suggested PR to fix it on ITB for sofar to review, but i haven't tested all mods to know so 🤷 17:07 Krock if it's somehow possible to reproduce this in singleplayer and a logic error in the mod is unlikely, then please let me know of the steps to reproduce. 17:10 MTDiscord <.zenonseth> i can send you a tiny minimal mod to reproduce it but you can decide whether that's a mod error or not - it is a different behavior from 5.16 for sure 17:13 Krock I sure hope the behaviour is different. Otherwise my change wouldn't fix some of the ceiling collisions ;) 17:45 MTDiscord <.zenonseth> here you go Krock: https://gist.github.com/ZenonSeth/0151a99ba3f37e89a416cc7e9ee9b3c9 - tiny mod which works for walk-over pressure plate detection in 5.16 but not in 5.17 (just walking over it doesn't, jumping will temporarily trigger pressure plate, but un-trigger when you land again) 17:49 MTDiscord so #16809 then 17:49 ShadowBot https://github.com/luanti-org/luanti/issues/16809 -- Inconsistent rounding with position at negative y levels (y < 0) 17:54 MTDiscord <.zenonseth> maybe related, but issue is from Jan 5th -and when i talked about it in engine channel here, I tested that this behavior change was due to this commit : * commit fa956d13e7c1173d82763ebcedfc226fad6aacf4 | Date: Mon Apr 20 19:15:52 2026 +0200 17:57 MTDiscord that issue basically asserts that the problem is not new, it complains about vector.round not giving you the node you want for -x.5 5.17.0 may have made it worse. 17:57 MTDiscord <.zenonseth> also i specifically am not calling this an engine bug :) idk if it should be considered - just letting you know that it may affect other mods/games too 17:59 MTDiscord <.zenonseth> > 5.17.0 may have made it worse. maybe but anecdotally, i spent quite a bit of time on Inside The Box and never ran into pressure plate problems before i pulled that fix 18:02 Sheriff_U3 Looks like I'll need to check circuits then. 19:29 red-001 https://github.com/luanti-org/docs.luanti.org/commit/6f18f9b90716d8f0f198bd3da3707af97bd4aaad#diff-34f4c2353fdbdef50d563c94b71881957783d2eab025c920b3ebe9c6e69b6884R70 I think it's kinda weird how there's a separate entry for the security fixes at the bottom and for the general security and stability fixes at the top, maybe just have one entry saying security and stability? 19:30 red-001 (if it's updated again) 19:31 rubenwardy agreed, "multiple security fixes" is also too vague 19:32 red-001 yes that's true but they do have the GHSAs for two of them 19:37 red-001 got like all the mesh and client fixes which are maybe a little more on the improving security without it being clear someone could have abused those bugs, the TSAN bug that JosiahWI found if the entry is about that also falls under that probably?, you never know what evil things a thread safety bug could be used for in theory, then that some sandbox escapes got fixed, one of them critical, then another entry for the RCE fix for custom CDB 19:37 red-001 the zip thing, not sure where that falls exactly 21:44 pgimeno hm, regarding #16809, the repro mod would probably be more useful if it printed string.format("%.17g", pos.y) instead of (or in addition to) tostring(pos.y) 21:44 ShadowBot https://github.com/luanti-org/luanti/issues/16809 -- Inconsistent rounding with position at negative y levels (y < 0) 21:48 pgimeno that would reveal hidden imprecisions if there are any