Time Nick Message 07:02 sofar what the 07:02 sofar print(type(minetest.wrap_text("", 5, false))) 07:02 sofar guess what that returns? 07:02 sofar `table` 07:02 sofar o_O 07:22 sofar making decent looking text in a formspec is virtually impossible 07:23 sofar somehow any method I use makes it more ugly 07:23 nerzhul formspec is impossible :D 07:28 sofar nerzhul: can you try this for me: 07:28 sofar print(type(minetest.wrap_text("", 5, false))) 07:32 nerzhul no :p 07:32 nerzhul i don't have any lua runner atm 07:38 sofar sigh, it's 0.4.16 11:58 Megaf Taoki: ping 11:59 Megaf Taoki: any special reason you are not at #Minetest-Hub ? 11:59 Megaf sofar: did you get your wrap.text sorted? 12:13 shivajiva can confirm bad argument #1 string expected, got table 12:13 ShadowBot https://github.com/minetest/minetest/issues/1 -- GlowStone code by anonymousAwesome 12:23 sfan5 wrap_text is totally broken in 0.4.16 indeed 12:34 shivajiva ^ created my own word wrap on the last form 15:53 rubenwardy merging #6881 in 10 15:53 ShadowBot https://github.com/minetest/minetest/issues/6881 -- Fix buffer parameter not working in LuaPerlinNoiseMap::l_getMapSlice() by pgimeno 15:53 rubenwardy also, I'm disappointed with the person who named that method "getMapSlice" 15:54 rubenwardy lua_uses_underscores_for_functions 15:54 rubenwardy unfortunately 15:54 rubenwardy hmmmm, as I expected 16:22 nerzhul can somes take a time to review https://github.com/minetest/minetest/pull/6872 ? 16:36 red-001 #6872 16:36 ShadowBot https://github.com/minetest/minetest/issues/6872 -- Refactor Game class to extract GameUI by nerzhul 16:38 Krock nerzhul, haven't tested it yet but as far I can see there's the problem where a new status text only shows up for a short moment 16:38 Krock this happens when another is shown and another status text shows up within these 1.5s 16:38 Krock s/when another/when one/ 16:42 nerzhul on text replacement you mean ? 16:42 nerzhul what is the problem ? can you show me ? 16:43 Krock Enable free_move and disable it after 1s again. The "Disabled free_move" status text will now only show up for 0.5s and fade away 16:44 nerzhul oh 16:44 nerzhul there is missing reset timer then 16:45 Krock never mind. you're actually doing that in "showStatusText(". Didn't see that 16:45 nerzhul yeah the timer is reset 16:45 nerzhul it's strange to see 0.5s fade 16:46 nerzhul i don't see how can this be problematic the only way to modify statustext is using showStatusText 16:46 nerzhul do you see an error somewhere ? 16:47 Krock nerzhul, sorry, I only looked at the code for now. Will test it later. As already said, I seem to have missed that new function, which makes my solves my concern 17:12 rubenwardy #6882 trivial PR 17:12 ShadowBot https://github.com/minetest/minetest/issues/6882 -- Fix naming conventions of noise userdata by rubenwardy 17:55 red-001 hi book` 18:16 rubenwardy #6883 18:16 ShadowBot https://github.com/minetest/minetest/issues/6883 -- Convert Lua vectors into a proper Lua class by rubenwardy 18:17 rubenwardy for Lua classes, are we using lower case or upper case names? 18:17 rubenwardy LuaStack is uppercase 18:17 rubenwardy but vector is lower case 18:17 rubenwardy I'd prefer uppercase 18:21 nore rubenwardy: did you test the performance impact of this? 18:21 rubenwardy the overhead will be from calling the functions 18:21 rubenwardy so it's the extra new call 18:22 nore Well 18:22 rubenwardy I'm not that worried, as if you care about performance you shouldn't use the vectors anyway 18:22 rubenwardy and there's JIT 18:22 nore We used metatables before 18:22 nore And it was changed because it was like twice as fast without them 18:22 rubenwardy it would be better to have a non-copying API if you cared about performance 18:23 rubenwardy is that from settable or from indexing? 18:23 rubenwardy *setmetatable 18:23 nore I don't know 18:24 nore You should ask ShadowNinja or make some tests 18:27 rubenwardy isn't it used in VoxelArea and such? 18:27 rubenwardy is voxel area in Lua or C++? 18:27 * rubenwardy looks 18:27 red-001 c++ iirc 18:29 nore Lua 18:30 sofar if we make it Vector then we can keep `vector` for the fast stuff? 18:30 nerzhul rubenwardy, if i remember we don't break Lua API no ? 18:30 rubenwardy it doesn't break the Lua API 18:31 rubenwardy I do need to make sure I have tests for all the previous constructor methos 18:32 Krock speed defined by capitals? I won't imagine how slow VECTOR will be 18:33 rubenwardy if I were implementing meta tables, I'd have it do the same thing if there is an entry in the initial table. ie: if looking up x, it wouldn't touch or care about the meta table at all 18:33 rubenwardy so the slow downs would be from more function calls, ie: vector:new() in places 18:34 rubenwardy and from the more complicated new function 18:34 rubenwardy but still worth benchmarking than guessing 18:36 rubenwardy ok, can confirm vector.new hasn't broken compatibility 18:40 nerzhul Krock, i fixed #6872 points (except the timer bug i should reproduce) 18:40 ShadowBot https://github.com/minetest/minetest/issues/6872 -- Refactor Game class to extract GameUI by nerzhul 18:41 nerzhul i cannot reproduce it 18:45 nerzhul maybe you tested an intermediate commit ? 18:49 Krock yes, I have tested it. The only issue I can find now is that something seems to prevent the status text from fading out 18:50 Krock but that seems to be an older problem. it behaves the same way is a previous build 18:50 nerzhul yes it's older, i tested on regular MT build it's always the case 18:51 nerzhul i don't know why i was suprised to find this dead code 18:51 nerzhul or maybe an irrlicht bug ? 18:55 Krock *shrug* 19:04 nerzhul Krock, i added https://github.com/minetest/minetest/pull/6872/commits/e8362ade8227a5d99e1e525724bb81842632a2a3 commit, is this okay to you for a merge ? if okay i will do it later this evening 19:10 Krock testing 19:14 Krock nerzhul, that didn't fix the problem and the speed benefit is tiny. However, I'm fine with it 19:40 Krock nerzhul, ah right, I approve it. (for official reasons) 19:56 nerzhul nice 19:56 nerzhul merging in ~2 mins then, thanks for your time Krock 19:56 Krock no problem :) 19:57 nerzhul hmmm wait unittests segfaults i hsould verify before :p 19:57 nerzhul erf 19:57 nerzhul the guienv 19:58 nerzhul just adding guienv && guienv->getSkin() :) 19:59 nerzhul merging as 8 commits 19:59 nerzhul 6% less liens of code in game.cpp nice 19:59 nerzhul 4.3k remaining 20:07 rubenwardy woah 20:07 rubenwardy doing :add() is 10% faster than doing local v = vector; v.add() 20:08 Krock LuaJIT magic, obviously 20:09 rubenwardy Lua 20:09 rubenwardy and it was a bad test 20:09 rubenwardy I was constructing a table in one, and not in the other 20:09 rubenwardy it's now 10% slower 20:10 rubenwardy but :add() is still 10% faster than vector.add 20:10 rubenwardy but that's due to localisation 20:15 rubenwardy who approved the GameUI PR? 20:15 rubenwardy fuck sake 20:16 rubenwardy the author of a PR that big can't approve it 20:16 rubenwardy oh well 20:26 rubenwardy lol, LuaJIT is on average 2% faster with the metatable vs the namespace 20:27 Krock only 2%? that's quite much a redundant difference 20:30 red-001 we should write down the unwritten rules about PRs somewhere 20:30 Krock "somewhere" = http://dev.minetest.net/Git_Guidelines 20:31 red-001 unwritten == not written down there 20:31 red-001 like don't add the approval lable for self-approvals by author 20:37 rubenwardy https://github.com/minetest/minetest/pull/6883#issuecomment-355658738 20:38 nerzhul guyz, it's Krock and it's linked 20:53 Krock nerzhul, it's not about me. It's about PR authors and their self-approval 20:56 rubenwardy merging #6881 in 10 as trivial bug fix 20:56 ShadowBot https://github.com/minetest/minetest/issues/6881 -- Fix buffer parameter not working in LuaPerlinNoiseMap::l_getMapSlice() by pgimeno 20:57 nerzhul Krock, it seems we don't have quality problem with that for coredevs these days 20:59 nerzhul i want to have more unittests in core but current PR needs more unittest basis to use them 20:59 nerzhul it's why i started to add more thngs 20:59 nerzhul i will look if we can have the fake irrlicht device in unittests 21:00 rubenwardy that would be cool 21:01 rubenwardy What are the issues with LuaJIT's GC64 mode? 21:01 nerzhul i cannot do it alone, i would like all coredev to use that a little bit more 21:01 rubenwardy ie: why don't we have it enabled by default 21:01 nerzhul it's better for non regression 21:05 sfan5 rubenwardy: it's officially considered "experimental" IIRC 21:08 red-001 no way to toggle it without compiling again? 21:09 sfan5 it's a compile-time option