Time Nick Message 08:44 MTDiscord has anyone seen hecks recently? 09:01 MTDiscord @a floofy ralsei (ROllerozxa) nope. He was tagged on a few items on github, but did not respond 09:02 MTDiscord https://github.com/minetest/minetest/pull/11747 is ready and needs some testing. Extra pair of eyes would also be helpful 09:31 erlehmann > (Almost) Fix self-shadowing of banner entities in MCL2 (offsets?) 09:31 erlehmann please elaborate on that? 11:55 MTDiscord At some point banners had a huge dark spot in the middle because of self-shadowing of the banner surface. In the latest commit I observed that the spot only appears temporarily when approaching the banner at high speed and shadow map update is spread over multiple frames (I had 16). 11:56 MTDiscord In short, the situation is much better, but not fully resolved yet. 11:56 MTDiscord erlehmann: ^ 11:58 erlehmann x2048 self-shadowing seems like a dificult beast for complex structures, will it be possible to turn it off? 11:59 erlehmann x2048 also, does the mcl2 crystal self-shadow? do the slimes make green colored shadows? 11:59 erlehmann the latter would be hilarious (and cool) 12:01 MTDiscord The slimes should make green shadows, yes. And complex structures should self-shadow properly (checked with trains, drawers, leaves etc.) 12:02 MTDiscord The problem with banners it the combination of non-uniform SM texels, large flat surface and cloth physics. 12:03 MTDiscord The banner surface suffers from wrong offsets, and I cannot set them uniformly 12:03 MTDiscord Disregard the last statement :) 12:06 MTDiscord My hope with LiSPSM in #11782 is at least to have simpler logic for offsets and filtering and better performance 12:06 ShadowBot https://github.com/minetest/minetest/issues/11782 -- Light space perspective shadow mapping by x2048 17:03 sfan5 why are mods allowed to set a different name in mod.conf than their folders are 17:17 MTDiscord Wasn't this intended as a fix for the usererror of not renaming mod folders? 17:35 erlehmann pretty sure the reason is that you may have different mods that satisfy the same dependency 17:35 erlehmann doors redux vs doors redo 17:35 erlehmann regardless, why would anyone ever care about the folder name? 17:53 MTDiscord using folder name causes so many issues to 17:54 erlehmann yeah any code that uses folder names should use mod.conf imo 17:54 erlehmann except for, of course, code that reads that file xD 19:00 Pexin what is the current attitude/policy about extending the csm api? i gather there may be reluctance to consider it because potential for even more security or cheating risks and i guess the apparent desire to move toward sscsm? 19:07 MTDiscord Last time I proposed an extension, it was paramatted with the reasoning being "CSM features should not be accepted until SSCSM is implemented." 19:12 Pexin mhmm. how long ago? 19:12 MTDiscord When paramat was still a core dev 19:12 MTDiscord #10455 19:12 ShadowBot https://github.com/minetest/minetest/issues/10455 -- Implement get_item_defs() client-side call by appgurueu 19:13 Pexin heh 19:13 Krock will merge #12061 , #12023 and #12018 in 10 minutes 19:13 ShadowBot https://github.com/minetest/minetest/issues/12061 -- Correct world_format.txt specification by dennisjenkins75 19:13 ShadowBot https://github.com/minetest/minetest/issues/12023 -- Add support for 'seed' in disallow_mapgen_settings by Wuzzy2 19:13 ShadowBot https://github.com/minetest/minetest/issues/12018 -- Apply texture pack main menu textures immediately by rollerozxa 19:23 Krock merging 19:24 Krock done 22:47 MTDiscord How am I supposed to do https://github.com/MT-CTF/capturetheflag/blob/df09c107b7c9b33bb2927c037d41136208564b68/mods/ctf/ctf_rankings/init.lua#L13 after this? https://github.com/minetest/minetest/commit/9a12e4499ecf5c1a3467af9c831d0d350a21923d#diff-f65d96a13c649554a5df1643c51d66874c21ff698642949cf3b618b4ba3c68f5R252-R256 22:51 MTDiscord Why are you using env.core.get_modpath and not simply minetest.get_modpath? No need to involve the insecure env here. 22:51 MTDiscord Can't other mods override that? 22:51 MTDiscord Ahh, so that's the reason behind it 22:52 MTDiscord Then just do env.minetest I suppose? 22:52 MTDiscord env.minetest has never existed AFAIK 22:52 MTDiscord Although knowing MT, I wouldn't be surprised if env.minetest == env.core == minetest == core 22:52 MTDiscord So in the end it wouldn't matter 22:53 MTDiscord If you look at the comment in sfan5's commit they also say env.core isn't secure either, hence my first question here 22:53 MTDiscord I guess my 'after this' was misleading 22:53 MTDiscord Then the answer is simple: It isn't possible. The minetest table has always been insecure. Modders have just been mislead by it being available in insecure environments. 22:55 MTDiscord To secure it, it would either have to be copied - which would end up a hot, memory eating mess, especially considering the engine relies on indexing it in multiple places - or be protected using a proxy + metatable, potentially harming performance (and messing with raw indexing). 22:56 MTDiscord TL;DR: Due to this exact issue - having to rely on overridable functions - mod security + loading files currently can't be secured. 22:56 MTDiscord To secure stuff, you'd have to move it to init.lua. 23:24 MTDiscord looks like it's possible and even secure env.dofile(env.debug.getinfo(1, "S").source:sub(2, -9) .. "redis.lua")