Time Nick Message 03:46 paramat will merge #5382 over the next few mins 03:46 ShadowBot https://github.com/minetest/minetest/issues/5382 -- Cavegen/Mgv5/Mgv7: Add optional giant caverns by paramat 03:56 paramat merged 06:55 burli hi 06:57 burli I have a question. How is the map stored in memory? 06:57 burli as array for each mapblock? 07:37 celeron55 burli: pretty much that 07:37 burli celeron55, ok, thanks. Because my pathfinder currently makes use of voxelmanip data 07:38 burli So if it is ported to C++ there will be some changes 07:39 celeron55 C++ has VMs too, and also the same world-coordinate node getters that lua has 07:40 burli I don't work with coordinates mostly. I use the index of the array 07:40 celeron55 it might make sense to test the performance on both of them as in C++ function calls and coordinates are much cheaper than in lua... dunno 07:41 celeron55 if you use indices of a VM array in luajit, i imagine it's almost as fast as the same thing in C++ 07:42 burli I don't have the experience to write that in C++. I need some help to port this to C++ later 07:44 burli I don't know how it is in C++ vm, but I need to load the vm data each time and depending on the size only loading of the data takes up to 5ms 07:44 burli I think this is to much 07:45 burli The algorithm itself needs between 0.5 and 120ms, depending on the terrain and distance 07:45 celeron55 yeah, the purpose of a VM is to make a copy of the data for very fast access later 07:46 burli I think it will be better to use the mapblocks directly 07:46 burli just put a layer above the mapblock that maps the data 07:47 celeron55 well it's what the Map class in the engine does :P 07:47 celeron55 it just re-routes node accesses to the mapblock they're found in 07:47 burli Does it work with coordinates? 07:47 celeron55 yes 07:48 burli ok 07:48 nrzkt 5ms per pathfind is inacceptable FYI 07:48 nrzkt the server loop should run every 50ms 07:49 burli I know 07:49 nrzkt celeron55: do you know why suggested interval for sending movements is calibrate on current server lag ? 07:49 burli But the runtime depends on the distance 07:49 burli brb 07:50 nrzkt because if a clietn connects when there is high lag, if lag is finished you send your movements with lag interval instead of current interval 07:50 celeron55 nrzkt: umm... i'm not aware of that 07:51 celeron55 maybe ask whoever added it 8) 07:52 celeron55 doesn't sound right for sure though 07:55 nrzkt celeron55: sorry: 742614180 (Perttu Ahola 2013-08-03 23:16:37 +0300 357) m_recommended_send_interval(0.1), 07:55 nrzkt it seems it's you :) 07:56 celeron55 where is that calibrated to be the same as lag? 07:57 nrzkt celeron55: in 0 07:58 nrzkt in 0.4.10 it seems it's the dedicated server step setting only 07:58 nrzkt i relook at current code but if i remember it recalibrate 07:59 nrzkt in fact no, i think i was tired this day :p 08:01 celeron55 it just sends g_settings->getFloat("dedicated_server_step") 08:01 celeron55 and the client puts that into m_recommended_send_interval 08:01 celeron55 in AUTH_ACCEPT 08:59 kilbith RE: https://github.com/minetest/minetest/pull/5502#issuecomment-291082521 08:59 kilbith I vehemently oppose an upgrade to Lua 5.3 until LuaJIT keeps up to the same version 09:00 kilbith if not, you can be sure I'll leave MT forever 09:04 celeron55 lol 09:14 nrzkt i don't said we should update, but it seems your forgot to read the parenthesis :p 09:14 kilbith I remember you were talking about upgrading in the past 09:15 nrzkt yeah but i'm not talking about it here, i don't care :) 09:15 nrzkt i just said, at a moment Lua 5.1 will be dead, and like every project we will face to a lib to update 09:15 kilbith it's already dead, so what's the point 09:16 nrzkt no it's not dead, it's quite old but not dead 09:17 kilbith people using 5.1 use JIT, otherwise 5.3 09:18 kilbith there's no 5.1 branch on the Lua repositories 09:18 nrzkt i don't understand why JIT is unmaintained as it seems to interess many people 09:18 nrzkt maybe a silly dev do it at a moment 09:19 celeron55 jit is pretty much a one-man project 09:19 celeron55 which is weird, but it just happens to be 09:20 nrzkt it's a problem , nobody maintain it,n it seems to be dead but it's useful and nobody understand or wants to maintain it 09:20 celeron55 anyway, lua is so simple it might not even have any security weaknesses or anything 09:20 kilbith and Mike Pall dislikes 5.3, so don't get your hopes high 09:20 celeron55 the official interpreter, that is; any version of it 09:21 nrzkt maybe, but Lua evolve, disliking it and not maintaining JIT for upper versions means he doesn't care about users and Lua future 09:21 celeron55 yeah, language features is the matter here 09:22 burli Hm, so what is the fastest possible access to map data over mapblock borders? 09:22 nrzkt burli: from CPP ? 09:22 burli yes 09:22 nrzkt getNodeNoEx 09:24 burli hm, ok. 09:25 celeron55 Map::getNodeNoex, to be precise 09:25 celeron55 E* 09:29 glut32 Hi all, I found a bug in the function "intlGUIEditBox::breakText()". The function font->getDimension(word.c_str()).Width counts the width of all characters of the enriched/colored text. And I would like that it counts only the displayed characters / brut text 09:32 glut32 ok I think that I must use unescape_enriched function 11:35 glut32 I have a question : the enriched text can be used without freetype ? 13:48 burli just to clarify: Map::getNodeNoex is similar to Minetest.get_node() in lua. Is that correct? 13:49 sfan5 yes 13:50 burli Ok, thanks. Than I will rewrite the pathfinder and us get_node. Than it is easyer to port it to C++ later. I use too much Lua magic 13:50 burli *use 13:55 sfan5 vmanip is also available from c++ as far as i'm aware 14:36 burli sfan5, yes, it is. But it takes a few milliseconds to load the data. And I assume that it will take the same time in C++ to load it. 14:56 celeron55 it will be interesting to see performance results from all versions of the pathfinder 15:54 rubenwardy #5506 15:54 ShadowBot https://github.com/minetest/minetest/issues/5506 -- Add Joystick type detection and Xbox controller support by rubenwardy 16:18 rubenwardy any opinion on #5506, sofar? I believe that you're the only one that's reviewed this code before apart from est 16:19 ShadowBot https://github.com/minetest/minetest/issues/5506 -- Add Joystick type detection and Xbox controller support by rubenwardy 16:20 sofar yeah, me and est31 debugged the original code (since I have a logitech) 16:22 rubywarden I think the best approach to this is to detect the most popular controller types and automatically bind for them, and then allow custom binding for other types 16:22 sofar rubywarden: I can test with my logitech later today 16:22 sofar happy to 16:22 rubywarden I'd like to wait for an input rework before adding the custom binding 16:22 rubywarden sweet 16:23 rubywarden not much should have changed 16:23 sofar the logitech isn't boxed up yet, so :) 17:18 red-001 #5508 17:18 ShadowBot https://github.com/minetest/minetest/issues/5508 -- [CSM] Add local node meta reference. by red-001 19:30 kilbith RE: https://github.com/minetest/minetest/pull/5457#issuecomment-291021948 19:30 kilbith open a new PR with your patch 19:47 sofar hehe, yeah, when I saw "needs OSX code" in the comments, I would have thrown in the towel myself 19:47 burli Hmm, is it possible to get a notification if a mapblock has changed? 19:48 rubenwardy yeah, that's why I'm hesitant to do it 19:48 rubenwardy easiest way would probably be to use the draw texture fallback 19:48 kilbith OSX is a PITA to deal with, we already annoyed at work with it 19:48 sfan5 do we support OS X at all? 19:49 kilbith at work we fixed the OSX build recently 19:49 kilbith we'll backport our patches soon 19:51 kilbith we fixed the unability to press 'M' on OSX also 19:54 nerzhul sfan5, as we don't have CI it doesn't seems supported 20:50 benrob0329-phone You know what STK needs? 20:50 benrob0329-phone A Tardis :P 21:46 nerzhul minetest uses too many streams.... the mapblock deserializaiton method is just a pain 22:14 nerzhul when someone has time, pelase review the nearly trivial #5510 22:14 ShadowBot https://github.com/minetest/minetest/issues/5510 -- Client handlers: Remove useless stringstream usage in two handlers by nerzhul 22:27 est31 rubenwardy: Just note that I didn't close #4153 back when I added support for my gamepad because of the missing formspec support, which makes it impossible to play the game entirely on gamepads. 22:27 ShadowBot https://github.com/minetest/minetest/issues/4153 -- xbox gamepad support 22:28 rubenwardy That's a different issue however 22:28 est31 well, your choice 22:28 rubenwardy Could just rename that to "Add formspec support to Gamepads" or just "Add complete gamepad support" 22:29 est31 rubenwardy: the issue was opened before I made my pr for gamepads support 22:29 est31 (and tbh I wasnt aware that xbox controllers have a different mapping) 22:30 est31 I just went to the nearest shop and bought some gamepad 22:31 est31 thought mapping was all the same, like for mice 22:31 est31 left button, right button, wheel 22:32 est31 *shrug* 22:32 rubenwardy yeah, it's all weird 22:32 rubenwardy there's no nice way to query for which button id a button is as well 22:33 rubenwardy you just have to have a look up table for all controller types 22:33 est31 indeed 22:33 rubenwardy renamed and added a todo to #4153 22:33 ShadowBot https://github.com/minetest/minetest/issues/4153 -- Add complete gamepad support 22:33 est31 thx 22:34 est31 now I'll go again 22:42 * sofar waves to est31 22:43 rubenwardy too late :P 22:44 rubenwardy removed "fixes #4153" from #5506 22:44 ShadowBot https://github.com/minetest/minetest/issues/4153 -- Add complete gamepad support 22:44 ShadowBot https://github.com/minetest/minetest/issues/5506 -- Add Joystick type detection and Xbox controller support by rubenwardy 23:33 paramat closed https://github.com/minetest/minetest/pull/2712 23:33 paramat #2712 23:33 ShadowBot https://github.com/minetest/minetest/issues/2712 -- Add settings for colors by Gael-de-Sailly 23:34 paramat sometimes it's difficult to decide whether to close because of no response, or keep it open for adoption 23:34 rubenwardy yeah 23:35 paramat problem is keeping open for possible adoption often results in lots of PRs hanging around for years 23:35 paramat need to judge it per-PR, but i think strict is needed 23:37 paramat anyway 2712 was not much needed 23:37 rubenwardy yeah 23:38 rubenwardy I mean, Minetest isn't particularly themable in terms of GUI atm anyway 23:38 rubenwardy especially in terms of the client