Time  Nick       Message
07:18 nrz        Hello, i just published a new irrlicht build cleanup here: https://github.com/minetest/irrlicht/pull/251. Note: windows failing part are always failing on current master branch, then don't just trust the build status, read logs 🙂 (no windows part is touched in this PR too).
07:19 nrz        2 less build flags which are always enabled, more pragma once (it should be now everywhere, it seems i miss some on previous pass, weeks ago), 110 lines of crap removed 🙂
10:40 ROllerozxa fwiw the irrlicht android asset reader is unused, it would be useful but once we've switched to SDL2 and SDL2's file read functions are used it can read directly from the asset folder as a fallback removing the need for extracting an assets zip
12:47 sfan5      not really
12:47 sfan5      you also need to teach Lua/LuaJIT in addition to all of our internal filesystem code about those ZIP files to make that work
12:48 sfan5      not impossible but definitely not worth it
12:52 celeron55  GNU/Minetest
12:53 celeron55  using the zip file system
12:53 celeron55  why? well it just happened
12:55 [MTMatrix] <localhost> zip -9
13:04 erle       fwiw irrlicht already has WAD support, are those not like zip files?
13:04 erle       (i think they are, but i might be wrong)
13:06 erle       so here https://irrlicht.sourceforge.io/docu/example002.html it says to just use     device->getFileSystem()->addFileArchive("../../media/map-20kdm2.pk3");
13:06 ROllerozxa ...yes, irrlicht has WAD support, they work like any file archive irrlicht supports
13:06 erle       > So we add the .pk3 file to our irr::io::IFileSystem. After it was added, we are able to read from the files in that archive as if they are directly stored on the disk.
13:06 erle       it was .pk3 files, not WAD files i think
13:06 erle       pk3 files are zip files
13:07 ROllerozxa pretty sure irrlicht supported WAD archives too, but irrlichtmt only has ZIP and the android asset archive left
13:07 rubenwardy Is there a way to mount a zip file such that standard file APIs work?
13:07 rubenwardy Probably, but is likely to be unreliable
13:08 ROllerozxa anyways I'm aware that it would be hard to make everything be loaded from assets, but even if fonts wouldn't have to be extracted that would be quite an improvement in initial loading times on android
13:08 erle       rubenwardy do you mean device->getFileSystem()->addFileArchive() only works for reading and not writing?
13:08 rubenwardy That only works for Irrlicht
13:08 rubenwardy A lot of our code doesn't using Irrlicht's filesystem utils
13:08 erle       oh i see
13:12 ROllerozxa it's possible to make a fopen wrapper for android assets like described here: http://www.50ply.com/blog/2013/01/19/loading-compressed-android-assets-with-file-pointer/
13:15 nrz        ROllerozxa thanks for the notice, i just kept the build mode iso, i didn't removed any code, except noop conditions 😄
13:17 erle       apart from there being no test cases for this PR, is there anything holding it up? https://github.com/minetest/irrlicht/pull/243
13:18 erle       (nothing requires test cases)
13:18 erle       (i mean: there is no policy to require test cases)
13:35 nrz        merging #13871
13:35 ShadowBot  https://github.com/minetest/minetest/issues/13871 -- build: Allow disabling documentation build + print more build flags by nerzhul
14:10 [MTMatrix] <localhost> minetest.iso? 🤔
14:24 [MTMatrix] <localhost> imagine DVD-version
15:00 sfan5      merging #13560, #13876, #13870, #13509 in 10m
15:00 ShadowBot  https://github.com/minetest/minetest/issues/13560 -- ActiveObjectMgr fixes by Desour
15:00 ShadowBot  https://github.com/minetest/minetest/issues/13876 -- Don't print ASCII art when using ncurses by sfan5
15:00 ShadowBot  https://github.com/minetest/minetest/issues/13870 -- Prevent early respawns caused by up/down button in the death screen by srifqi
15:00 ShadowBot  https://github.com/minetest/minetest/issues/13509 -- Optimize PNG files by Bituvo
16:13 [MTMatrix] <localhost> 122 files changed, 1207 insertions(+), 238 deletions(-)
16:13 [MTMatrix] <localhost> wow
20:59 MTDiscord  <luatic> Hey folks. Just got yet another instance of the insanely common mistake of modifying the inventory and having it overwritten by a callback which expects an item stack as return value. I think we should definitely document this better. We could also detect this kind of mistake in the engine, as in, save the wielditem, if the wield item isn't what it ought to be after the callback was ran, warn. Would this be worthwhile?
21:00 erle       luatic wdym with “if the wield item isn't what it ought to be after the callback was ran, warn”
21:00 erle       like what is it supposed to be?
21:02 MTDiscord  <luatic> erle: It should not be modified while the callback is running, as any such modifications will be overwritten when the callback returns.
21:03 MTDiscord  <luatic> In this specific case, a semgrep pattern of on_rightclick = function(...) ...:set_wield_item(...) end would have sufficed to catch it.
21:03 MTDiscord  <luatic> set_wielded_item*
21:03 erle       sounds like you could make a repo with linter rules for mods?
21:59 [MTMatrix] <Zughy> Hello. Is there any Windows user who can help lhofhansl by suggesting them how to profile/debug Minetest on Windows? Both on #13545 and #13518
21:59 ShadowBot  https://github.com/minetest/minetest/issues/13545 -- 5.7.0 is sometimes extremely slow on Windows (11) and crashes
21:59 ShadowBot  https://github.com/minetest/minetest/issues/13518 -- Shutdown takes a long time on Windows