Time Nick Message 00:00 paramat phew 00:00 VanessaE (I still get to argue with it, but for a different reason now.) 02:30 paramat sofar could you look at game#1678 ? i get the feeling doors was broken by a recent commit 02:30 ShadowBot https://github.com/minetest/minetest_game/issues/1678 -- Doors do not appear in the creative inventory. 02:49 paramat nevermind, found the cause 12:04 glut32 I would like to know which version of irrlicht do you use on minetest mac ? 12:12 nrzkt glut32: on travis you mean ? 12:13 glut32 https://github.com/krondor-game/minetest/releases 12:13 glut32 on this build 12:13 glut32 I contacted the guy 12:14 nrzkt i don't know, it's a minetest interesting user but not official core member 17:17 Krock Need MTG testers for this patch? https://github.com/minetest/minetest_game/issues/1678#issuecomment-290773342 17:17 Krock Table references seem to have manipulated the groups, thus this misbehaviour 17:18 Krock s/?/:/ 17:33 VanessaE I've had that happen before. 17:33 VanessaE (in a different circumstance) 17:35 VanessaE (I had to resort to a deep copy to get around it, too) 18:04 Krock VanessaE, I expected register_item would copy the contents but apparently it doesn't :/ 18:13 Thomas-S May I ask, why, when "sudo make install"ing minetest, new files are installed but deleted ones aren't deleted in /usr/local/share/minetest ? 18:27 Fixer https://github.com/minetest/minetest_game/issues/1680 18:34 sfan5 Thomas-S: cmake only knows about the files to install, not about the files previously installed 18:34 sfan5 this is why you should use a package manager 18:34 Thomas-S oh, ok. Thanks. I solved my problem by deleting /usr/local/share/minetest. I assume this has no negative impacts? 18:35 sfan5 no 18:35 Thomas-S Thanks! 18:44 nerzhul sfan5, will fix points you mention. For the path requirement it's required by cmake itself because add_subdirectory only allow, without this argument, subdirectories, and move them outside of src make this rule invalid. What i suggest is to move the detection path in CMakeLists.txt in lib/ folder and do add_subdirectory of this part in root/CmakeLists.txt before the add_subdirectory src/, are you okay ? 18:45 sfan5 sounds good 18:45 nerzhul nice :) i'm experimenting this better approach now 18:45 nerzhul can you review #5487 ? 18:46 ShadowBot https://github.com/minetest/minetest/issues/5487 -- [CSM] Add function and chat command to disconnect from server. by red-001 19:36 nerzhul sfan5, Krock i updated #5473 and fixed the android build :) 19:36 ShadowBot https://github.com/minetest/minetest/issues/5473 -- Update embedded jsoncpp from unk version to 0.10.6 + move libs to lib/ instead of src/ by nerzhul 19:39 nerzhul can you review and accept it for a merge ? 19:41 Krock at least travis should be done before reviewing 19:41 nerzhul Krock, exact :) 19:42 nerzhul travis is not happy hmm 19:42 nerzhul interesting... 19:42 nerzhul my cmake version doesn't have this error 19:44 nerzhul Krock, travis is happy on 5487, the change is simple, but required for our second CSM roadmap 19:56 nerzhul merging #5490 thanks sfan5 for reopening & others to look at this 19:57 ShadowBot https://github.com/minetest/minetest/issues/5490 -- Wieldmesh: Natural orientation depending on pitch by sfan5 19:57 nerzhul and thanks to kil 19:57 nerzhul and thanks to kilbith 19:59 nerzhul oh 19:59 nerzhul github changed commiter 19:59 nerzhul i need to fix this sorry :) don't pull 20:00 nerzhul done 20:23 nerzhul what is minetest codestyle column limit ? because clang-format seems to use 80 but it's very tiny, 100 seems more reasonable no ? 20:25 celeron55 we do have this http://dev.minetest.net/Code_style_guidelines 20:25 celeron55 which is mostly written by hmmmm, but mostly valid 20:25 celeron55 it says 90 is a strict limit 20:25 nerzhul yeah i saw it and it doesn't talk about cols, just refering to linux kernel which says 80 cols except sometimes for readability 20:26 celeron55 actually it talks 20:26 nerzhul oh sorry 20:26 nerzhul misread 20:26 nerzhul okay, i'm pushing a trivial fix on clang-format to take account of that limit 20:28 nerzhul and apply a style fix on l_client.cpp 20:28 nerzhul (which is not conflicting) 20:28 nerzhul and l_client.h 20:28 nerzhul you won't have to fix it red-002 then 20:43 Fixer This is finished it seems, please consider reviewing https://github.com/minetest/minetest/pull/5416 20:43 Fixer Windows: Skip cmd for release builds #5416 20:43 ShadowBot https://github.com/minetest/minetest/issues/5416 -- Windows: Skip cmd for release builds by adrido 20:45 nerzhul Fixer, you have my approval, just require a second approval, from Krock maybe 20:46 Fixer nerzhul: you was too fast, also not my PR, but i'm interested to get rid of cmd window :) 20:46 Fixer i tested it 20:46 nerzhul it's more proper to not having it in releases build 20:46 Fixer on mingw crosscompile, originally it was for msvc 20:47 Fixer nerzhul: with it you will have -DENABLE_CMD_WINDOW = 1/0 for cmake to choose 20:48 nerzhul i hope it's 0 for release :) 20:49 rubenwardy is it possible to opt-in to a cmd window, either with a setting or a different launcher? (ie: launch using bash) 20:49 sfan5 possibly 20:49 rubenwardy s/bash/batch/g 20:55 Fixer rubenwardy: why not just tail debug.txt ? 20:55 rubenwardy er - windows? 20:55 rubenwardy also, convenience 20:56 Fixer rubenwardy: in windows I found mtail, very useful if console is disabled 20:56 Fixer rubenwardy: http://ophilipp.free.fr/op_tail.htm 21:02 Fixer nerzhul: for msvc - yes, for mingw cross-compile... it will be hidden for any build by default 21:07 Fixer it seems, will check 21:10 red-002 updated #5487 21:10 ShadowBot https://github.com/minetest/minetest/issues/5487 -- [CSM] Add function and chat command to disconnect from server. by red-001 21:52 VanessaE well, that's that. I just evicted LuaJIT from my servers. 21:52 VanessaE too many OOM crashes. 21:53 sfan5 you probably "just" need to wait until gc64 is stable 21:53 sofar surprised you hang on this long, tbh 21:53 VanessaE I already had it compiled with gc64 enabled. 21:53 sfan5 are you sure? that mode crashed for me very often 21:53 VanessaE yup. 21:54 VanessaE doesn't matter. 21:54 sofar better lua code probably helps you more 21:54 sfan5 (crash meaning segfault) 21:55 VanessaE sofar: I suppose it would, but if I have a machine with 16GB of RAM, by G*d I should be able to use it ;) 21:55 sofar yes but all that memory is slow if you thrash your caches with bad lua 21:55 VanessaE and besides, better code won't fix luajit stupidities like needing to be at the start of address space. 21:59 sofar shure shure 22:00 VanessaE (the code that needs help is way above my proverbial pay grade, anyway) 22:12 Fixer gc64 mode is not stable it seems, at least on minetest 22:13 Fixer i've also seen some bugreports about crashes in this mode 23:21 VanessaE ..aaaaaaand I have to turn JIT back on. the server's just too slow otherwise :( 23:31 Fixer VanessaE: try with usual luajit without gc64 23:31 VanessaE doesn't matter, it was crashing before, too. 23:31 Fixer VanessaE: what error? 23:31 VanessaE OOM errors in random mods 23:32 VanessaE most recently mesecons. 23:33 VanessaE 2017-04-01 01:32:41: ERROR[Main]: ServerError: AsyncErr: environment_Step: OOM error from mod 'mesecons' in callback environment_Step(): not enough memory 23:33 Fixer totally random or there are some champions? 23:33 VanessaE for example. 23:33 VanessaE lately it's been blaming mesecons, I know you personally tend to see it happen with technic_worldgen 23:34 rubenwardy #5302 23:34 ShadowBot https://github.com/minetest/minetest/issues/5302 -- Store legacy metadata separate from new item meta data by rubenwardy 23:34 rubenwardy doesn't affect serialization 23:35 Fixer VanessaE: it is inevitable, how often it happens? 23:35 VanessaE repeatedly 23:35 VanessaE as in startup, run for several seconds, crash OOM 23:37 Fixer VanessaE: thats what I had with milan's subgame + darkage, very fast oom 23:37 Fixer when i tested it last time 23:39 Fixer VanessaE: probably cheapie with dat mesecons... 23:39 VanessaE I doubt it. 23:40 VanessaE can someone at least patch the engine to show the last requested amount of RAM before the crash? 23:40 Fixer VanessaE: how much mem was at the moment of oom in that error report? 23:40 Fixer VanessaE: it shows... 23:40 VanessaE Fixer: typically in the 970 MB range 23:40 VanessaE but that doesn't tell me how much was last requested or by which mod 23:40 VanessaE (saying mesecons crashed doesn't mean mesecons was what requested the RAM) 23:40 Fixer yeah, this needs improving 23:40 Fixer also want mod profiling as in memory usage 23:41 Fixer we have cpu usage but mem usage is also needed 23:41 VanessaE what good is mod profiling if I can't run it before the crash? 23:41 Fixer VanessaE: and it OOMs each time? 23:41 VanessaE yep. 23:42 Fixer VanessaE: right after start up? 23:42 VanessaE any second now it's gonna OOM 23:42 VanessaE there it goes. 23:42 VanessaE 2017-04-01 01:42:19: ERROR[Main]: ServerError: AsyncErr: environment_Step: OOM error from mod 'mesecons' in callback environment_Step(): not enough memory 23:42 VanessaE 2017-04-01 01:42:19: ERROR[Main]: Current Lua memory usage: 947 MB 23:42 Fixer VanessaE: linux? nothing changed? anything memory intenfive nearby? 23:43 VanessaE of course linux :P 23:43 VanessaE and nothing much changes from day to day 23:44 Fixer VanessaE: if it is mesecons i would probably look through the logs who interacted with mesecons last time and maybe shut down the mod or delete offensive mapblock 23:44 Fixer VanessaE: not many options here :( if you are not developer 23:44 Fixer developer can probably debug it, profile it, etc 23:44 Fixer mod itself 23:45 VanessaE I'll be quite honest here. I'm really getting tired of waiting for people to FIX SHIT (I don't mean Minetest engine) 23:45 Fixer VanessaE: in short, you are screwed, if you have some backup, try to revert back 23:45 Fixer VanessaE: revert back to preOOM time 23:46 Fixer revert entire map from backup 23:46 VanessaE oh hell no 23:46 Fixer VanessaE: maybe you updated some mods recently? 23:46 VanessaE fuck no 23:46 VanessaE sure, I update mods from time to time 23:46 Fixer VanessaE: updated mesecons recently? 23:46 VanessaE but this has been an ongoing problem forever 23:47 Fixer VanessaE: try older mesecons versions 23:47 VanessaE I've just updated everything, engine, game, re-compiled luajit too. 23:47 Fixer VanessaE: use simple lua 23:47 VanessaE it's running again, but who knows for how long 23:47 VanessaE simple Lua is too slow. 23:47 VanessaE 431 minetest 20 0 780936 374984 13420 S 81.4 2.3 1:24.61 /usr/bin/minetestserver --worldname Survival_World --confi+ 23:47 Fixer i know it is slower, but there is not much options here, i can't even run a dreambuilder on windows on luajit 23:47 VanessaE (virtual, resident, shared) 23:48 VanessaE it's not just "slower". it's "horrendously slower" 23:48 VanessaE my server relies heavily on technic, which is a CPU pig to beat all. 23:48 Fixer VanessaE: it is about luajit 1gb limit, if GC64 mode will work, that should fix OOM problems 23:49 VanessaE -DLUA_LIBRARY=/home/minetest/luajit-2.1-gc64/src/libluajit.so \ 23:49 VanessaE and that IS compiled with GC64 enabled. 23:49 Fixer strang 23:49 Fixer e 23:49 Fixer VanessaE: 64 bit library? 23:49 VanessaE of course. 23:50 VanessaE no one uses 32bit anymore 23:50 VanessaE especially on a dedicated server. 23:50 Fixer VanessaE: and you compiled 2.1.0 tree with make XCFLAGS=-DLUAJIT_ENABLE_GC64 ? 23:51 VanessaE I edited src/Makefile to turn it on, as per its instructions. 23:51 Fixer VanessaE: i have impression that it is not turned on, last time I checked I eated 4 gigabytes in GC64 before luajit segfault ._. 23:52 Fixer and you have 1gb limit 23:52 Fixer that means it is not turned on 23:52 Fixer how i did it on windows: https://github.com/minetest/minetest/issues/2988#issuecomment-285090239 23:54 Fixer gn 23:55 VanessaE ok, `git reset --hard HEAD` and recompiled like that ^ 23:56 * VanessaE rebuilds minetest against it. again.