Time Nick Message 01:25 ANAND Re #8817 - Windows builds pass, but non-Windows Travis builds fail with a SettingNotFoundException. 01:25 ShadowBot https://github.com/minetest/minetest/issues/8817 -- All Settings: Prevent unnecessary writes by checking for modification by ClobberXD 01:26 ANAND Gotta rename the PR first... 13:22 ANAND WIP PR: #8834 13:23 ShadowBot https://github.com/minetest/minetest/issues/8834 -- Fix LocalPlayer::isDead always returning false if player is immortal by ClobberXD 13:23 ANAND IcyDiamond: See ^ :) 13:24 ANAND Unfortunately it isn't complete, and requires some substantial changes in another area before it works. 13:24 ANAND Here goes another "NO SQUASH" PR... :'( 13:25 IcyDiamond :( 13:26 IcyDiamond Simple fix turned into a nightmare, huh 13:29 sfan5 pushing revert of fb6f1fdcbe3ee2321cc33eb00f2c6da86744d00a in 10 minutes, to fix the breakage 14:04 ANAND Thanks, sfan5 14:18 ANAND Does our code-style allow inlined function defs in a single line? 14:18 ANAND e.g. bool func() const { return boolean; } 14:19 ANAND ... in class declarations 14:20 sfan5 i don't know about the style guide but the existing code does it plenty 14:20 ANAND Yea, exactly. Wanted to know if I'm allowed to define new functions that way. :) 14:29 ANAND Rebased #8271, one more approval required. 14:29 ShadowBot https://github.com/minetest/minetest/issues/8271 -- Fix breath statbar scaling; defer breath_bar hiding by one second by ClobberXD 16:25 pmp-p what luajit function one would use in lua server/client context to do precise timing of function execution ? 16:27 pmp-p nvn i guess it's implemented by minetest.get_us_time 16:40 ANAND Just curious, is there an automated way of adding the "Rebase needed" label? 16:41 Krock ask a dev 16:41 Krock automated from your perspective 16:42 ANAND No, lol. I mean I never notice merge conflicts in my PR and yet, the label is added (if applicable) even to the oldest of my PRs. 16:42 Krock conflicts are listed on the very bottom of the page 16:42 ANAND But you'd still have to open each PR page to know that 16:43 ANAND Do you guys check all open PRs for merge conflicts, every single day? 16:43 Krock yet no way to sort or show PRs that need rebase 16:43 ANAND True 16:44 Krock DS-minetest: while (true) check(); 16:45 DS-minetest and minetest is not so multithreaded 17:07 Krock btw, the occlusion culling PR now ignores nearby mapblocks. It's now the same underground as master 17:35 pmp-p i wonder why execution times in client/server lua code are closer to lua5.1 than luajit2 17:40 pmp-p could be wrong but that means a sandboxed wasm vm client side would only be 2x slower, not 20x as expected 17:45 sfan5 Krock: well not exactly https://a.uguu.se/tMuvHSvjMZ4O_D4KxSsc.png 17:46 sfan5 oh wrong screenshot 17:46 Krock was about to say 17:46 Krock because they look way too similar to me 17:46 sfan5 here https://a.uguu.se/QqLrmCp5Gapi_uRK03Pq.png 17:47 Krock hmm 17:50 Krock it's entirely broken 17:52 * Krock is haunted by stupid errors 17:53 sfan5 what is broken this time? 17:56 Krock dunno what's the cause yet. map stops loading, player respawns always at the same position and /status shows that the server is still working 17:56 Krock s/working/responding/ 17:56 * twoelk has spent a week of making more complicated stupid errors while trying to avoid simple stupid errors when calculating materials quantities needed for a building job 17:57 Krock even happens without occlusion culling. what the heck is going on 17:57 Krock never mind me. outdated branch that includes a recent bug 17:58 twoelk still damaged dead players issue? 17:58 Krock twoelk: yeah, no outdated branches 17:58 Krock *on 17:59 Krock *happy entire-minetest-source-recompiling noises* 18:02 * BuckarooBanzai just pulled the latest updates and regrets it now after reading the last few messages... :/ 18:03 Krock BuckarooBanzai: no probs if it's baked fresh 18:04 Krock i.e. < 5 hours or > 1 day 18:04 BuckarooBanzai i'm at https://github.com/minetest/minetest/commit/bf22184d6e16f21b9001322aad1bf8f6dbaa372b (8 days old, just checked:) 18:06 Krock safe 18:17 Krock #8832 could also help to implement #8818 due to light source positions being always given 18:17 ShadowBot https://github.com/minetest/minetest/issues/8832 -- Mapgen: Run light spread algorithm iteratively by sfan5 18:17 ShadowBot https://github.com/minetest/minetest/issues/8818 -- Approximate spherical lighting 18:19 Krock misleading conclusion. light might be occluded by nodes 18:21 twoelk missleading lights with irrlicht? sounds legit 18:54 sfan5 i have debug output that shows isOccluded returns false, yet the mapblock is not being drawn 18:54 sfan5 ? 19:20 sfan5 https://github.com/minetest/minetest/blob/master/src/mapblock.h#L231 19:20 sfan5 http://irrlicht.sourceforge.net/docu/aabbox3d_8h_source.html#l00201 19:20 sfan5 Krock: am i correct that getBox() is incorrectly implemented since aabbox3d uses inclusive coordinates? 19:20 sfan5 or wait 19:20 sfan5 no nvm it's correct 19:21 sfan5 now only if I had realized earlier so I wouldn't have bothered you 19:21 Krock heh no problem 19:25 sfan5 gah why can't c++ do three way comparisons 19:25 sfan5 this just cost me 30 minutes 19:29 sfan5 Krock: ok here's my solution the problem with narrow tunnels https://0x0.st/zWpp.txt 19:30 sfan5 it even improves the situation over minetest without the pr https://a.uguu.se/BVPGSbzLK0aV_ephlgQw.png 19:32 sfan5 basically this ensures any block the camera can see with a straight line will be occlusion checked using exactly that straight line 19:34 Krock so basically optimizing it for x/z corridors 19:34 sfan5 yea 19:35 Krock I also thought about doing so, but it didn't seem consistent for mapgen tunnels that go diagonally down 19:35 Krock well, those tend to be rather short but still 19:37 Krock and diagonal corridors aren't that uncommon since advtrains 19:38 sfan5 how about finding the node in the mapblock that's closest to the camera and doing the occlusion check for that? 19:39 sfan5 the x/z corridor optimization is essentially that principle but simplified 19:42 Krock vector intersection with a pane given by the mapblock 19:45 Krock I'll need to sketch that to work out the formula to calculate that 19:45 sfan5 hmm cmake is still looking for freetype with -DBUILD_CLIENT=0 19:45 sfan5 ? 19:45 Krock set(ENABLE_FREETYPE ${BUILD_CLIENT}) 19:47 sfan5 updated the lightSpread PR btw 19:48 pmp-p why the hell is irrlicht still grabbing mouse+kbd events when X11 focus is lost when not in the game menu ? 19:48 Krock how about light == 1? 19:50 sfan5 it was like that before so ¯\_(ツ)_/¯ 19:51 Krock mhm yes 19:51 sfan5 it is correct though, light will be decremented first so if light == 1 then n.param1 would be set to MYMAX(0, ) 19:51 sfan5 which is a no-op 19:52 Krock right 19:52 Krock will test tomorrow 20:04 Krock using weighted average of light-propagating nodes to calculate a possible pointable corridor to the target mapblock 20:05 Krock 1) get nodes in a range of 4m at 5m away from player 20:05 Krock 2) sum all nodes that propagate light. divide by results 20:05 Krock 3) use that as new shootline 20:05 * Krock cries in node lookup counts 20:06 Krock BUT it would work 20:07 sfan5 dunno about using light levels, should occlusion culling work in entirely dark areas? 20:07 Krock only if there's no sunlight calculation 20:07 Krock because gray borders look awful 20:20 pmp-p meh USE_LUAJIT=0 i guess that is very bad 20:20 sfan5 is it? 20:21 pmp-p well it would explain why luajit2 is performing faster in terminal than in minetest 20:21 sfan5 it would, yes 20:21 pmp-p i knew i should not have installed all possible lua flavour from synaptic 20:22 sfan5 you really just need the dev package for luajit 20:22 pmp-p pfff liblua5.1-0-dev:amd64 *not the good one* 20:38 pmp-p yeah now i have full speed on rpc execution <3 luajit