Time Nick Message 00:04 Hijiri lock thing could be cool 00:04 Hijiri you put locks in an octree or whatever 00:05 Hijiri I just think putting locks in an octree feels like a weird mix 00:05 Hijiri (in a good way) 00:06 Hijiri wait, that's not really how it would work 00:06 rubenwardy you could make make C++ only mapgens run on a separate thread, and then just use the current way to run lua mapgens 00:06 Hijiri you would probably have a normal lock around an "octree lock" structure 00:07 Hijiri and the octree lock can notify threads when it clears a thing that was blocking some earlier ones 00:07 rubenwardy alternatively, iirc Lua map gens work by using a on_generated callback, so the separate C++ thread could just add a queue event to run the callbacks on the normal lua thread 00:07 rubenwardy Bet it's more complicated than this 00:07 rubenwardy :P 00:08 rubenwardy Like, passing LVMs 00:09 Hijiri what if LVMs automatically locked/unlocked part of the map when they wre created/destroyed 00:12 Hijiri maybe the octree lock thing is overcomplex 00:12 Hijiri since it's voxels could just have a lock for each block 00:12 Hijiri but then editing a mapchunk requires acquiring/releasing 125 locks 00:14 Hijiri or if it's targetting specifically mapgen then could have a lock for each chunk 00:24 Fixer this fog start config option is very useful 00:25 Fixer feels so different with less fog, like in winter, summer haze is so annoying tbh 00:26 lhofhansl Fixer: cool :) 00:29 lhofhansl The comparison of fps is also interesting. 00:30 Fixer don't take it seriously 00:30 Fixer and for some reason MT version of map is inverted 00:31 Fixer in one axis 05:20 lhofhansl We should document somewhere that generating normal map causes a significant drop in fps (with shaders) 05:51 OldCoder lhofhansl, generating normal map as opposed to any other kind? 06:14 Zeno` OldCoder, I think he's referring to bump mapping 06:14 OldCoder OKty 06:42 lhofhansl I just shaders for time mapping and the "wavy things" 06:42 lhofhansl use 06:45 sofar I really like the tonemapping and even the DoF stuff 06:45 sofar with a bit of tweaking it probably can be made very usable 06:49 lhofhansl1 The normal map generation costs 50% of the fps in my setup. 06:49 sofar https://github.com/minetest/minetest/pull/3623 06:49 sofar we should really salvage DoF from that 06:49 sofar and bloom 06:50 sofar lhofhansl1: seems like right up your interests 06:52 paramat updated saveschems 06:54 paramat can i also add in the tree logs and normal cactus? after release i would like to have all schematics as mts files, to reduce space used in mapgen.lua by the lua tables 06:54 sofar yes! please do! 06:54 paramat ok 06:54 nrzkt i added approval on some PR, please merge when you want 06:55 paramat when floatland biomes are added in, mapgen.lua is going to get big, need to save space 06:56 paramat yeah i'll merge #4857 within 2 hours 06:56 ShadowBot https://github.com/minetest/minetest/issues/4857 -- Fix range check for simple deco->deco_param2 by sofar 06:56 paramat any support for #4829 ? sofar? 06:56 ShadowBot https://github.com/minetest/minetest/issues/4829 -- Mapgen: Make mgv7 the default mapgen by paramat 07:00 paramat added a line comment to 4857 sofar 07:04 sofar honestly, I massively prefer v7 to any other mapgen at this point 07:05 sofar but the only reason I would prefer it as default is because it's what I use 07:05 sofar not because I care about half-generated maps :) 07:05 sofar (because frankly, I don't, really) 07:08 sofar wow someone needs to reassemble the skeleton that is #1686 07:08 ShadowBot https://github.com/minetest/minetest/issues/1686 -- Added "[sheet" to the texture special commands. by lupuchard 07:16 paramat for me it's mainly because i check all minetest videos on youtube, and almost all reviews and 'lets try' / 'lets play' videos use the default mapgen 07:29 sofar #4861 07:29 ShadowBot https://github.com/minetest/minetest/issues/4861 -- Added "[sheet" to the texture special commands. by sofar 07:30 sofar that's a rebased and cleaned up #1686 07:30 ShadowBot https://github.com/minetest/minetest/issues/1686 -- Added "[sheet" to the texture special commands. by lupuchard 07:30 paramat thanks 07:30 sofar if I'm good at something it's rebasing lol 07:33 sofar I might just dig through more PRs that are "probably approved" but just seem to need a little fix 07:33 sofar but right now, it's time to relax and watch the wonderful inside of my eyelids 07:34 sofar (and enjoy a snow day tomorrow - schools closed!) 08:34 paramat will merge #4857 #4858 in a moment 08:34 ShadowBot https://github.com/minetest/minetest/issues/4857 -- Fix range check for simple deco->deco_param2 by sofar 08:34 ShadowBot https://github.com/minetest/minetest/issues/4858 -- Buildbot: Update LuaJIT by sfan5 08:44 paramat now merging 08:47 paramat complete 17:50 rubenwardy Right, so I'm debugging "Failed to save block: attempt to write a readonly database" on Android, and it appears that /storage/emulated/0/Minetest is never even created, even though the textures load correctly??? 17:50 rubenwardy https://gist.github.com/rubenwardy/7302857872cb922e33c9a2d47382dac4 17:52 est31 rubenwardy: I think its done by java code 17:52 est31 in build/android 17:52 est31 somewhere 17:52 rubenwardy V/Minetest(13104): 2016-12-08 17:51:49: VERBOSE[Main]: Client: Loaded cached media: aefa4b0d034b32b1bdfe30b23c18d22687552e07 "tnt_side.png" 17:52 est31 build/android//src/main/java/net.minetest.minetest/MinetestAssetCopy.java 17:52 rubenwardy it loads from the cache 17:53 rubenwardy yeah, I'm reading that currently 17:53 est31 I'm not sure, I think the cache dir is separate 17:53 est31 uses android provided cache directories 17:53 est31 but not sure whether I implemented it, or I just wanted to 17:54 est31 nope it was sn : ea2964f5a168cb52d1b9f74a08f00c7c068c6649 17:54 rubenwardy yeah, the cache dir is separte 17:54 rubenwardy Failed to open asset file "/storage/emulated/0/Minetest/media/Shaders/COGLES2SphereMap.vsh" for size check 17:54 rubenwardy Maybe it takes that error as succesS? 18:04 rubenwardy E/MinetestAssetCopy(15183): /storage/emulated/0/Minetest/games/minetest_game/mods/carts/README.txt is: readble 18:04 rubenwardy Failed to open asset file "/storage/emulated/0/Minetest/games/minetest_game/mods/carts/README.txt" for size check 18:21 rubenwardy shell@E2303:/storage/emulated/legacy/Minetest/worlds/singleplayerworld $ ls -l 18:21 rubenwardy -rw-rw---- root sdcard_r 12288 2016-12-08 16:46 map.sqlite 18:25 rubenwardy that looks fine to me 18:26 Fixer readble ? 18:28 rubenwardy obvious typo :P 18:28 rubenwardy Ignore that anyway - it's saying the the phone has a file that the asset pack doesn't 18:28 rubenwardy not a bug 18:28 rubenwardy actually it may be, as some lua files are complained about 20:05 rubenwardy If I make the app write to its private storage then it doesn't crash 20:06 rubenwardy (by replacing getExternalStorageDirectory() with "/data/data/net.minetest.minetest/files", the app's private URI) 20:07 rubenwardy when the app is started, getExternalStorageState == MOUNTED 20:43 rubenwardy #4121 20:43 ShadowBot https://github.com/minetest/minetest/issues/4121 -- Failed to save block: attempt to write a readonly database 21:28 rubenwardy Found a fix to #4121, preparing to submit 21:28 ShadowBot https://github.com/minetest/minetest/issues/4121 -- Failed to save block: attempt to write a readonly database 21:29 rubenwardy it's an issue with mismatching data type length on 64bit CPU 22:21 est31 rubenwardy: any news? 22:21 est31 (on the fix) 22:21 rubenwardy had a break to make dinner 22:21 rubenwardy preparing now though 22:25 garywhite I flew around for a bit on the latest Win64 build of MT from sfan5, and made some TNT blow up, no issues 22:36 rubenwardy #4871 22:36 ShadowBot https://github.com/minetest/minetest/issues/4871 -- Fix sqlite databases being readonly on 64bit Android by using a patch… by rubenwardy 22:36 rubenwardy est31, 22:37 * est31 looks 22:38 est31 wow thats cool 22:38 rubenwardy I suggest transferring my repo to minetest/ before merge 22:38 est31 yeah 22:38 est31 good idea 22:39 est31 its quite sad though that we have to fork sqlite over this 22:39 est31 rubenwardy: there is a patch mechanism available, it looks like its a better canidate for it 22:39 est31 other projects use it 22:39 rubenwardy hopefully they'll update eventually 22:40 est31 no, I mean instead of putting the source into a git repo, that its still downloaded from upstream & then we apply the patches 22:40 rubenwardy build/android/patches? 22:41 est31 yes 22:41 est31 don't try manually generating a patch, you need to either find a git command that can make an appropriately formated patch, or you need to make two dirs, one with the patch applied and one without it 22:42 est31 (manually generating patches doesnt work usually) 22:43 est31 ooor 22:43 rubenwardy ik :D 22:44 rubenwardy https://github.com/rubenwardy/sqlite3/commit/536c10ab94013a8f6b0f247d0fb71806f7f861b5.patch (with modifications to remove the commit info) 22:44 est31 we could use https://android.googlesource.com/platform/external/sqlite/ 22:44 est31 that would work just as well 22:45 est31 then we dont have to maintain another fork of a project 22:45 rubenwardy they have a weird tree structure inside 22:47 est31 PhoneNumberUtils.cpp 22:47 est31 veeery convincing... 22:47 est31 well, let's go with a patch then 23:04 rubenwardy updated #4871 23:04 rubenwardy est31 ^ 23:04 ShadowBot rubenwardy: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request." 23:04 est31 okay 23:04 est31 trying it now 23:12 est31 rubenwardy: what should I test? 23:13 est31 I have a 32 bit phone 23:14 rubenwardy start the game and keep it running for 30 seconds. Build some things and make sure they persist 23:14 rubenwardy singleplayer 23:14 rubenwardy apart from that - not sure 23:14 est31 mhh 23:17 est31 yeah that works 23:17 est31 +1 to the PR then 23:18 est31 mhh do I actually have 32 bit? 23:18 * est31 double checks 23:18 rubenwardy also, makes sure the patch actually applied in deps/sqlite-ama*/sqlite3.c 23:19 rubenwardy afaict it only applies when downloading sqlite3 when it doesn't exist :D 23:19 est31 yes 23:19 rubenwardy gdgd 23:19 est31 the makefile is bit bad about this 23:19 rubenwardy was just making sure :D 23:19 est31 needs fixing :) 23:19 rubenwardy I'd quite like to make it use cmake 23:19 est31 I removed the entire directory :) 23:20 est31 idk, its quite tricky to rewrite the makefile 23:20 est31 lots of issues may appear 23:20 rubenwardy then hopefully I can then use Android Studio https://developer.android.com/studio/projects/add-native-code.html 23:20 rubenwardy yeah, it's a massive file 23:20 rubenwardy don't fix want ain't broken 23:21 rubenwardy *what