Time Nick Message 03:10 repetitivestrain I came upon a trivial reproducer for the z fighting issue 03:10 repetitivestrain just place a mineclonia slime block underwater 16:35 ireallyhateirc would anyone be willing to get bribed into fixing this thing? https://github.com/minetest/minetest/issues/9357 16:37 MTDiscord lol 16:37 rubenwardy the trick is to loudly say that it can't be done 16:38 ireallyhateirc lol 16:38 ireallyhateirc yeah, only if we had professionals capable of fixing that bug! 16:50 ireallyhateirc the bug is the last thing keeping me from enabling more than one emerge threads on our servers which otherwise provide a great performance boost thanks to sfan5's mapgen env 17:22 sfan5 you're welcome to attempt a fix 17:22 ireallyhateirc I would if I knew C++ 17:23 sfan5 could also start smaller and try to debug this 17:23 ireallyhateirc There's something scary about 3 years long "learn C++ in one week" youtube playlists 17:23 sfan5 it requires some understanding but you can theoretically get by with adding a bunch of print statements and drawing conclusion 17:23 sfan5 +s 17:24 ireallyhateirc sfan5, I already did. See my "mapgen_test" game under the issue 17:24 ireallyhateirc I'm kromka-chleba on github 17:25 sfan5 seen it 17:25 sfan5 this is already good but I meant something close to the source 17:25 ireallyhateirc like for example? 17:27 sfan5 confirming the hypothesis that the affected parts are written twice and maybe in what order 17:28 ireallyhateirc hmmm well I could do that 17:34 sfan5 maybe even possible from lua actually 17:42 ireallyhateirc I should be able to elucidate the order of chunk generation provided that on_generated runs really in order of chunk generation 17:43 ireallyhateirc then I'll find chunks with c_stone and do some stats for their neighbors 17:55 ireallyhateirc Does 1 mapblock overgeneration happen before or after on_generated gets triggered? 17:56 ireallyhateirc I believe before 18:32 ireallyhateirc Interestingly when I spawn a VM in on_generated I don't see any of the c_stone nodes 18:36 ireallyhateirc hmmmm never mind, it started detecting them 18:38 ireallyhateirc but it only detects a fraction of them which probably means that first it generates stone correctly, that's what I see in on_generated and later that gets overwritten by another chunk 22:13 MTDiscord "well technically you could do the whole thing in just a week if you don't eat or sleep" 22:15 MTDiscord You will definitely learn C++ in a week that way, they just aren't telling you which week will be the one where it finally clicks. 22:40 MTDiscord I want to move away from depending on default. Currently, my mod checks for water nodes using the line if node.name == "default:water_source" then. Would a good alternative be if node.groups ~= nil and node.groups.liquid then? I've been looking at nodes API: https://api.luanti.org/nodes/ 22:41 MTDiscord The approach I use in NodeCore for that is generally (1) use a group to identify things that are definitely a match, (2) also offer a group to mark things as definitely not a match, (3) if both groups are absent, use heuristics to make a guess that matches correctly for all examples I know. 22:46 MTDiscord Sounds good to me. 22:50 MTDiscord Sure, or check the draw type, etc. Also see xcompat