Time Nick Message 00:16 kilbith paramat, can you merge https://github.com/minetest/minetest_game/pull/732 now, please ? i have something else on my plate for _game (and forgot to make a branch) 00:17 paramat yes sure 00:17 kilbith thanks, good night 00:18 paramat now merging to game 00:25 paramat merged 00:51 paramat ah the optional shadow bool is causing compilation to fail due to the 2 'calcLighting' functions which have differing numbers of arguments, guess i'll rename one 01:18 paramat weird, use of optional arguments isn't working, details in the thread, bbl 02:21 paramat think i'll remove the old 2-argument calcLighting function, then we can have just one form and each mapgen will specify the volumes for propagateSunlight and spreadLight 02:53 paramat PR updated 03:09 hmmmm what on earth 03:09 hmmmm okay first of all who changed calcLightingOld to calcLighting? 03:11 hmmmm wow I totally don't remember doing this 03:13 hmmmm oh how helpful. it seems like calcLightingOld was removed completely by onkrot 03:14 hmmmm might want this back at some point if we ever start unit testing light spread 03:16 paramat ah no i changed 2-argument calcLighting to calcLightingOld earlier in this PR 03:17 paramat then removed it 03:17 hmmmm there was another calcLightingOld 03:17 paramat aha 03:18 hmmmm the old one used voxalgo namespaced functions which were very slow but pixel perfect 03:18 paramat ah thanks for comments 03:18 hmmmm the new version was one of my few first commits to minetest and sped up lighting calculation an insane amount, at the cost of having some minor edge case bugs 03:18 hmmmm i am not really sure if they've been fixed 03:19 hmmmm if nobody's complaining i guess it's all good 03:20 paramat yeah seems good 03:22 hmmmm okay 03:22 hmmmm so I have a proposition 03:22 paramat if you want calcLightingOld put back in commented-out let me know i'll do that in this commit 03:23 hmmmm keep the 2 argument version of calcLighting, and make the 4 argument version of calcLighting take all 5 parameters non-optional 03:23 hmmmm we'll keep the 2 argument version of calcLighting as calcLighting, and make the "extended" version calcLightingEx() 03:24 hmmmm i wonder if using Ex is too microsoft-y 03:24 hmmmm the point of this is that the new parameter is used in exactly one place whereas all the other usages are the two parameter version 03:24 hmmmm so this would remove all ambiguity, while keeping actual changes to a minimum 03:25 paramat the 4 argument version is used in all mapgens other than mgv6 03:25 paramat to cope with overgeneration 03:26 paramat other than mgv6 and singlenode 03:26 hmmmm oh 03:26 hmmmm okay then nevermind about what I said 03:26 paramat ok 03:26 hmmmm adding an optional bool param to the 4 parameter version shouldn't cause any compilation errors, if it's done correctly 03:27 hmmmm the optional parameter goes into the .h file, not the .cpp file 03:27 paramat ok i'll update now 03:27 paramat the big crrent problem is the y = 63 lighting bug 03:28 paramat https://github.com/minetest/minetest/issues/2759 03:29 hmmmm hmm, well y=63 is the very top of the voxelmanip 03:36 paramat good it compiles now 03:47 paramat i'll check lighting in mgv6 and singlenode then it'll be ready 03:50 paramat bbl 04:59 paramat well i don't see a point in keeping 2-argument calcLighting, singlenode specifies different volumes to mgv6 so only mgv6 would be using it 05:01 paramat next job, i'd like to add a mapgen_air alias so that games can use vacuum nodes 05:03 hmmmm wait a minute, why does singlenode use calcLighting at all? 05:04 hmmmm that doesn't really make much sense 05:04 hmmmm I think this is a misunderstanding of the available lighting functions when singlenode was written; it should really be using setLight() 05:05 paramat hm it can be configures to be all-stone or whatever 05:06 paramat (configured) 05:06 hmmmm setLight(node_min, node_max, ndef->get(n_singlenode).propogates_light ? 0xFF : 0x00); 05:08 paramat yeah seems ok 05:08 paramat i'll try it 05:38 paramat updated, singlenode will be much faster now 05:50 paramat that 0 should be 0x00 though.. 14:46 yunfan hi, i was wondering why the client got so lag even in local server 14:47 yunfan for eg, when placing a item to the chest, it would made me waiting about 1 second to see the item displaying in the target position 14:56 Krock does it make any change when you start a local server and join it with an additional client? 17:15 Calinou yunfan, -ELONGSTANDINGBUG 17:15 Calinou :) 19:15 paramat hmmmm i'll merge https://github.com/minetest/minetest/pull/3418 later unless there are objections 21:32 paramat i'll merge this soon https://github.com/minetest/minetest_game/pull/734 'Default: Remove root from mapgen trees and large cactus' as it has been requested by devs and players 21:33 paramat ^ nore sfan5 ShadowNinja 21:37 sfan5 paramat: seems ok 21:37 paramat thanks 21:38 paramat sapling-grown trees don't replace the dirt below, they are seperate schematics 21:49 nore paramat: looks good 21:52 paramat ok 22:10 paramat now merging game 734 22:22 paramat merged 23:18 celeron55 https://github.com/KarlHegbloom/mmap_lowmem 23:18 celeron55 somebody e-mailed me about this 23:18 celeron55 actually i guess it's the author of the thing 23:20 celeron55 apparently it's an LD_PRELOAD thing that enables luajit to use "almost 4GB" of RAM on 32-bit platforms instead of 1GB 23:21 BlockMen "The main use of this wrapper is to allow LuaJIT to use more then 1Gbyte of ram on Linux." does this mean its only for linux available or that LuaJIT already useses more than 1 GB on windows? 23:21 celeron55 he suggests it to be used in the launchpad packaging script 23:21 celeron55 dunno 23:22 celeron55 i guess this is profile is public if someone is interested: https://launchpad.net/~karl.hegbloom 23:24 paramat hmmmmm, updated https://github.com/minetest/minetest/pull/3418