Time Nick Message 12:27 proller https://github.com/proller/minetest/compare/wingsuit - good for float islands jumping 13:24 proller Zeg9, can you commit wieldlight to find error? 14:08 Zeg9 proller: see Zeg9/wieldlight2 14:08 Zeg9 just pushed 14:09 Zeg9 Note that this just crashes the minetest client 14:28 proller - } else if(version >= 2) { 14:28 proller + } else if(version == 2) { 14:28 proller bad idea 14:28 Zeg9 proller, I hadn't seen that 14:28 Zeg9 probably happened during the merge 14:31 proller http://paste.org.ru/?ver6v7 14:31 proller fix 14:39 Zeg9 https://github.com/minetest/minetest/pull/816 14:49 proller >= 3 better than == 14:50 Zeg9 proller, I dont think it belongs to that pull request to change this 14:50 Zeg9 brb 14:51 kahrl why isn't the setting called enable_wieldlight? 14:53 Zeg9 because its enabled by default 14:54 kahrl there a lots of setDefault("...", "true") in defaultsettings.cpp 14:54 kahrl are 14:55 kahrl http://blogs.msdn.com/b/oldnewthing/archive/2008/06/24/8644367.aspx 14:57 PilzAdam why is there even a client side setting to disable it? 14:59 thexyz kahrl: http://darkpatterns.org/ 15:16 proller Zeg9, if i get torch - anything around gets sunlight 15:16 proller with no range limitation 15:17 Zeg9 thats the problem with shaders 15:17 Zeg9 we never get the same thing on different GPUs 15:18 Zeg9 anyway, even when it works, the wieldlight goes through walls, ignores the top face (sides are as bright as the top)... 15:18 Zeg9 one should just implement real dynamic lighting (or use irrlicht's lighting system? Whats wrong with it?) 15:20 proller can you fix unlimited range? 15:49 proller https://github.com/minetest/minetest/pull/897/files 16:02 ironzorg кто антон? 16:03 thexyz ironzorg: это англоязычный канал, а еще данное сообщение не относится к разработке Minetest 16:05 ironzorg whatever 16:23 celeron55 ai venäjääkö täällä puhutaan; osallistunpa kieliesittelyyn 16:26 thexyz agreed 19:06 proller sfan5, can you make fresh -next build? 19:07 sfan5 yep 19:11 sfan5 proller: building.. 19:25 sfan5 proller: https://forum.minetest.net/viewtopic.php?pid=108572#p108572 19:26 proller thank you! 19:29 proller now base-jumping on sky server much simpler 19:29 proller but need to optimize blocks sequence 19:32 proller and now possible to reach ground from float islands at 20000+ 21:22 proller anyone trying to fix speed loss when loading bottom block when falling? 22:25 Jordach proller, air and content ignore when unloaded are considered solid 22:27 proller collision have flag standing_on_unloaded 22:27 proller maybe dont update vertical speed if it true 22:44 kahrl we tried that some time in the 0.2/0.3 era 22:44 kahrl iirc then you got stuck forever if you touched the edge of the world 22:45 kahrl so it was removed 23:10 kahrl aside from the insanity that is JSON, does this look sane? https://gist.github.com/kahrl/6407927 23:13 thexyz this is unrelated but I think menu_lua_api.txt should say "engine.parse_json → see minetest.parse_json" 23:13 kahrl that's reasonable 23:16 thexyz > int maxdepth = 42; 23:20 thexyz why do you lua_checkstack with depth*2? how does depth correlate with number of lua stack slots? 23:20 kahrl ok I should put a comment there 23:21 kahrl the maximum number of stack slots used at each level of push_json_value_helper is 2 (case Json::objectValue) 23:22 kahrl the idea behind maxdepth was to make it impossible to let push_json_value_checkdepth run out of (C++) stack space 23:23 kahrl but now that I think of it, there isn't a way to limit the stack usage of Json::Reader::parse(), so it's kind of futile 23:24 thexyz how many slots would something like [1, 2, 3, 4, 5, 6] use? 23:26 kahrl 2 23:26 kahrl (plus the two used by the parameters of parse_json) 23:27 kahrl the first additional is the lua_newtable in push_json_value_helper (case Json::arrayValue) 23:27 kahrl the second additional is the lua_pushinteger in push_json_value_helper (case Json::intValue) 23:27 kahrl uintValue* (?) 23:29 thexyz hm, I guess I don't understand something then 23:29 thexyz shouldn't it take at least 6 for all those numbers? 23:30 kahrl no, because it's like push 1, pop, push 2, pop, push 3, pop, ... 23:31 thexyz oh, okay 23:34 thexyz then it's good 23:42 kahrl changed menu_lua_api.txt, removed the useless maxdepth, renamed some thing, added debug: https://gist.github.com/kahrl/6408049 23:43 kahrl s/some thing/push_json_value_checkdepth to push_json_value_getdepth/ 23:43 kahrl so the debug shows it pops the things from the stack as expected 23:49 kahrl should I commit it upstream now (minus the debug stuff) or leave it as another commit in httpfetch? 23:50 kahrl (I need it in httpfetch as I'm going to move the serverlist and modstore handling to lua) 23:53 thexyz I guess for you it'll be more convenient to keep it in your branch? 23:54 thexyz I don't think that matters, just do what's easier for you 23:55 kahrl both are equally convenient for me 23:55 kahrl well actually pushing upstream now is a little easier because there's less to keep track of in the future, and there's less to review for everyone when httpfetch is done