Time Nick Message 00:24 gregorycu est, good to see ya 00:25 gregorycu Let's drive this ring buffer discussion to conclusion, then I can go back to fallout 4 00:25 est31 okay 00:26 gregorycu I think you are talking about the scenario at() THEN push_back() 00:27 gregorycu Is that correct? So, T& element = at(0); push_back(something); // element now points to something different 00:28 est31 why should it point to something different? 00:28 est31 it doesn't 00:29 gregorycu hmmm.... 00:29 est31 or well 00:29 est31 it does 00:29 est31 but other containers have similar properties 00:30 gregorycu I agree 100% 00:30 gregorycu If you do an at(), then modify the thing, you may not have the same element 00:30 gregorycu We can agree on that 00:30 gregorycu So, let's look at the push() THEN at() case 00:31 gregorycu m_offset = 5; push will make m_offset = 6. then when you do at(6) you get the next element that will be push on-top of (aka, the oldest element) 00:32 gregorycu the push call has basically overwritten at(N-1) 00:33 gregorycu It's all a bit confusing, maybe that statement didn't helop 00:39 gregorycu This is eating into my fallout 4 time 00:40 est31 well whats the problem then? 00:41 gregorycu ShadowNinja said: It seems that pushing a value causes the indexes for at() to shift over by one, which seems like an odd way to do things. 00:41 gregorycu You said: Agreed, its different from vector behaviour, where the value from at(0) always points to the first element you have push()-ed. 00:41 est31 yes ... ? 00:41 gregorycu And you also said: No, the behaviour here is that at(0) points to the element last pushed. 00:42 est31 ah 00:43 gregorycu so, I was just saying that I think the implementation as it stands makes sense, and is consistent with vector and other std containers 00:44 est31 yes, I'm ok with current implementation 00:44 est31 and yes my statements contradict 00:44 est31 no time to find out which one is the right one :) 00:46 gregorycu Can you just say you revoke your objections? That way we can ignore them :) 00:46 gregorycu And I can argue with Shadow instead 00:47 gregorycu I don't want this dude to be discouraged, he seems pretty good 00:47 est31 yeah 00:47 est31 wanted a newbie task, got a bit less of a newbie task :) 00:47 gregorycu He's up for it 00:47 gregorycu That's the good thing 00:49 est31 and my statements do not contradict 00:49 est31 the behaviour here is that at(0) points to the element last pushed. 00:49 est31 the behaviour for vector is that at(0) points to the element first pushed. 00:49 est31 so its different 00:49 est31 which is not bad by itself 00:50 est31 I only agreed to the fact that it is different from vector 00:50 est31 and see, I just +1ed it 00:57 gregorycu Ahh ok, cool 00:57 gregorycu Awesome, Fallout 4 time 00:57 gregorycu Thank you 01:47 kaeza haven't looked at the commit log, but now the rendering is mmuch faster. almost 50 fps with 150 draw range on my crap intel integrated! 01:48 kaeza last time it was so 01:48 est31 in the master branch? 01:48 kaeza mething like 30 fps 01:48 kaeza yep 01:49 est31 idk 01:49 est31 I cant recall anything graphically changed 01:50 est31 what was the commit you were on before? 01:50 kaeza dont know maybe a month or more ago 01:51 est31 can you paste git reflog 01:51 est31 on pastebin 01:52 kaeza it is also less choppy, and map loading/gen looks faster too 01:52 kaeza cant form here on mobile, but will take note to report later 01:53 kaeza yes touch keyboards suck 02:15 Wayward1 I'm observing what appears to be a phantom player on one of VE's servers: http://imgur.com/nyyk1sD 02:15 Wayward1 There is no visible player named "unknown" on the server. Any thoughts? 02:18 est31 do /status again, is it still there? 02:18 Wayward1 yep 02:55 Wayward_One It's still there, if anyone's interested 02:56 est31 well, unless you can reproduce after a restart ... 02:57 Wayward_One Actual server restart, or relog? 03:03 Wayward1 Either way, I just relogged and it's still there, but the server won't restart until 10:00 am UTC 03:40 diemartin ~tell est31 FWIW, https://gist.github.com/kaeza/1f02b45b677065667db3 03:40 ShadowBot diemartin: O.K. 05:14 gregorycu kaeza must have been talking about my massive changes 07:10 paramat i'll merge https://github.com/minetest/minetest/pull/3377 'Mgfractal: Revert unnecessary duplication of parameters' soon 07:11 est31 one thing about that pr 07:11 est31 you seem to edit minetest.conf.example manually 07:11 paramat yes 07:11 est31 be noted that from time to time it will be updated from settingstypes.txt 07:11 paramat aha i didn't know 07:12 est31 so any update you do in minetest.conf.example that doesnt match settingstypes.txt will be removed sooner or later 07:12 est31 you can do these kinds of updates yourself as well 07:12 est31 just edit 07:12 paramat i see 07:12 est31 lemme see... 07:13 est31 Edit builtin/mainmenu/tab_settings.lua, change the "false" at line 727 to a "true" 07:13 est31 then start the client, mainmenu is enough 07:17 paramat nice, thanks 07:18 est31 I'll merge this in ten minutes https://github.com/est31/minetest/commit/bdfb761dd06261e71f96b731ad423338a9c01008 07:18 paramat ..then change the true back to false i guess 07:18 paramat ok go ahead 07:19 est31 is that a +1 ? 07:19 est31 otherwise I wait for you 07:20 paramat i'll look 07:20 paramat seems simple so +1 07:23 est31 thanks, pushed 07:28 Hunterz hi, est31 something broken wit the console... when serever running some days without restart, console looks like stop showing all info from the server 07:28 est31 Hunterz, what do you mean? 07:29 Hunterz when press some key and then enter, console show one new line of 2 days old event 07:29 est31 you might have scrolled up 07:29 est31 try pgup / pgdown 07:29 Hunterz ah works 07:30 Hunterz thanks, will be possible autoscroll? 07:30 est31 yes it is autoscroll 07:30 est31 just keep scrolled at bottom 07:31 Hunterz how? 07:31 est31 well press pgdown as often as possible 07:31 est31 then it will stay there 07:32 Hunterz ok 07:33 Hunterz will be possible add / instead of chat to console as defaut behavior? 07:34 Hunterz i think server console is primarily for server control, not chatting with players 07:34 est31 it can be added 07:34 est31 perhaps one mode for / one for no default 07:39 Hunterz Minetest 0.4.13-dev-0876623 | Game 4575726 Time of day 19:00 -- Number after Game is seconds server online? Will be possible change it to some human readable? (Server online 2 days 8 hours...) Is possible add to this line something about memory consumption, server lag etc? I think will be useful. 07:40 est31 its not uptime 07:40 est31 its the time since the first start of the server 07:40 est31 so its the sum of all uptimes 07:40 est31 if you restart, the value continues, doesnt start with 0 again 07:41 est31 And I dunno what it can be changed to 07:41 Hunterz I think actual uptime (latest restart) will be fine add too 07:41 est31 well too many time values won't look nice I think 07:42 Hunterz for example bukkit (minecraft) 07:42 Hunterz [08:18:55 INFO]: CONSOLE issued server command: /mem 07:42 Hunterz [08:18:55 INFO]: Uptime: 47 minutes 36 seconds 07:42 Hunterz [08:18:55 INFO]: Current TPS = 20 07:42 Hunterz [08:18:55 INFO]: Maximum memory: 989 MB. 07:42 Hunterz [08:18:55 INFO]: Allocated memory: 241 MB. 07:42 Hunterz [08:18:55 INFO]: Free memory: 177 MB. 07:42 Hunterz [08:18:55 INFO]: World "world": 272 chunks, 18 entities, 27 tiles. 07:42 est31 use a pastebin dude 07:43 Hunterz sorry 07:45 Calinou server could have /uptime command, yes, definitely 07:45 Calinou replies uptime in days/hours/minutes/seconds 08:06 paramat will merge 3377 after checks 08:12 paramat now merging 08:17 paramat done 08:36 est31 any core dev want to review #3354 #3355 #3369 ? 08:39 est31 kahrl ? nore ? sfan5 ? celeron55 ? ^ 08:43 paramat nore sfan5 i'll merge https://github.com/minetest/minetest_game/pull/728 'Default/mapgen: Use sidelen 16 for low density decorations' later, fairly trivial biome system maintenance 09:17 paramat now merging game 728 09:22 paramat done 12:02 nore paramat: looks good 12:03 est31 nore, I pinged you earlier about engine PRs. they are fairly simple, but only have my approval 12:03 est31 as hmmm is at mt vacation, a second person to review stuff is missing 12:04 est31 If you have time can you look at #3369 #3355 #3354 12:04 nore est31: ah, I had not seen those 12:04 nore looking at them now 12:05 nore #3369: not tested, but looks fine if it works 12:06 nore (the commit looks simple enough to me so if you tested it, +1) 12:07 est31 I didn't. Ill ask if the pr creator tested 12:07 nore #3355 looks fine too 12:12 nore about #3354, it looks fine, but I don't seen what it is used for 12:12 nore s/seen/see/ 12:13 est31 its to replace an std::queue in profiler code 12:13 nore ah, ok 12:14 * nore had not seen that while looking at the code 12:14 est31 yes, that pr doesnt change it 12:14 est31 it only adds the data structure 12:14 nore ok 12:15 nore I did not check this one as much as the others though 12:15 nore but the code looks fine; it however needs testing 12:16 nore so let's say I give +1 to #3369 ans #3355, and to #3354 if it is tested 12:17 est31 great thanks 12:35 nrzkt est31 i added a +1 for the item groups craft replacements 12:40 est31 fine thanks 12:49 Hunterz Hi, anyone have rendering delay when first render page of unified inventory? I made video: http://www.twitch.tv/hunterzcz/v/25917402 12:50 Hunterz its on the linux 12:52 Calinou Hunterz, don't use HD texture packs, it will be much faster 12:52 Calinou if it's still too slow, there is not much you can do, except enabling "Preload item visuals" which will make the delay occur when starting the world, rather than opening the inventory 12:52 Hunterz i have no texture pack 12:52 who_wants_some i have that delay when changing pages for the first time on unified invenory 12:53 Hunterz yes only first time 12:53 Hunterz i think some preload to memory will be solution 12:57 Hunterz hmmm when try enable to "Preload item visuals" - during login to server sreeen flickering like hell 13:52 Sokomine hi :-) i could use some feedback. mg_villages is mostly ready to be released. the documentation can be found under https://github.com/Sokomine/mg_villages/wiki feedback regarding the docu is welcome 13:56 Sokomine ups. sorry. wrong channel :-( 16:37 who_wants_some http://pastebin.com/Bd4r5y0J bad allocation error on Krock's build, 32bit, windows 7, 1.6gb of ram was taken by mt at the moment of crash 16:50 who_wants_some get this error after very long walks (hour or so) 18:50 celeron55 who_wants_some: how much ram and swap do you have? i mean, did you run out of memory? 18:50 celeron55 that kind of error can occur due to one of two reasons 18:51 celeron55 either you actually ran out of memory, or the program erroneously tried to allocate a ridiculously large chunk of memory 19:08 who_wants_some i made a topic about it, feel free to look at it https://forum.minetest.net/viewtopic.php?f=6&t=13585 / i have 8 gig or ram, and mt 32 bit dies at 1.6gb, that is reproducable 19:09 hmmmm well I think that's your answer 19:10 hmmmm how do you expect a 32 bit process to make use of your 8gb of ram? 19:11 who_wants_some i think if mt is a 32bit process, it should unload some blocks if it goes close to a "limit" 19:11 hmmmm right but it doesn't know that 19:11 hmmmm i think all lawn should cut itself too but it doesn't 19:11 hmmmm i think my lawn is bugged 19:12 who_wants_some not possible to detect if os is 32 bit and enable some dirty hacks? :( 19:14 hmmmm even if it were simply to do in a platform independent manner, there's a 99% chance you're running windows so you'd have to wait until the next release to get it 19:14 hmmmm your best shot is to just download a 64 bit build and play that instead 19:14 who_wants_some i know, i thinking about average user who does not know about this 19:14 who_wants_some i'm* 19:15 who_wants_some ultra large multiplayer servers like LinuxGaming2 crash 32 bit client in a metter of minutes :) 19:16 hmmmm you should just lower client_mapblock_limit 19:19 who_wants_some yes, that is good too, too bad it is not lowering automaticly when reaching mem limit in 32bit os :) I don't want to argue, I want to inform, thanks for response 20:23 celeron55 well i guess we should have a long-term goal of automatically unloading stuff if the memory limit comes close 20:23 celeron55 but it's not trivial because it's completely non-standard 20:24 celeron55 the OS could just stop allocating stuff even if there is free memory, or somehow create more virtual memory after the current RAM and swap are exhausted 20:25 celeron55 and i don't think we could manage to maintain a system where any allocation could fail and the game could continue running 20:25 celeron55 this is what hmmmm is referring to :P 20:27 celeron55 i mean, if you have 8GB of RAM and it's a 32 bit process, exactly what kind of a limit is 1.6GB? 20:27 celeron55 i don't see any way of detecting that there would be a limit at that point 20:28 hmmmm FWIW I am guessing who_wants_some is measuring memory usages from peak working set and not reserved 20:28 hmmmm if the reserved memory gets around 2gb, then you have problems 20:28 celeron55 that's very true 20:28 celeron55 (and still very OS-dependent) 20:29 hmmmm yeah i think this is totally wrong for Linux 20:33 celeron55 i guess a system that checks through an os-specific wrapper how much reservable memory is not yet reserved could be created and called before each larger allocation and be set up to trigger unloading of the oldest of things; but i bet nobody is going to attempt that 20:33 celeron55 it might get too fiddly and would be a very rarely used code path 20:34 celeron55 basically it would probably just crash anyway :P 20:41 who_wants_some some servers need [64 bit] tag, so user will know it is heavy load down there 21:06 celeron55 maybe some kind of a memory limit suggestion setting could be made that could default on each 32-bit platform to something that usually avoids address space exhaustion 21:07 celeron55 and which would speed up mapblock unloading when it triggers 21:07 celeron55 that could be simple enough, and fairly simple to test 21:08 celeron55 it cannot provide 100% protection against it though because of how memory works at OS level 21:08 hmmmm you know I wonder how many mapblocks contain meshes that are just solid cubes 21:08 hmmmm especially underground 21:08 celeron55 none 21:08 hmmmm is that case optimized already? 21:08 celeron55 mapblock edges are not generated into the mesh 21:08 hmmmm ahh ok 21:08 celeron55 if they are not needed 21:09 celeron55 the mesh generation isn't actually fed the data of a single mapblcok 21:09 celeron55 mapblock* 21:09 celeron55 it is fed the data of 5 mapblocks in order to give it the information it needs for doing that 21:09 celeron55 and it's also needed for avoiding z-fighting issues 21:09 celeron55 wait, i mean 7 mapblocks? 21:10 celeron55 actually no... 21:10 celeron55 it's four 21:10 celeron55 the middle one, and each one towards the three positive (or maybe negative, idk) coordinates 21:11 celeron55 anyway; faces are generated only towards a certain side so that when mapblocks are rendered next to each other, only one covers the edge 21:12 celeron55 and it will have nothing there if the other one contains something that doesn't make a face there 21:13 celeron55 (i did make FarMap work in such a way that it only generates based on a single FarBlock and it does make solid cubes underground; however it doesn't seem to be a problem) 21:13 hmmmm heh 21:13 VanessaE celeron55: did you see my z-fighting glitch regarding that? 21:14 celeron55 it isn't necessarily related to that 21:14 celeron55 a FarBlock couldn't carefully cover the individual faces of mapblock 21:14 celeron55 you'd have a huge hole in there instead 21:14 VanessaE oh ok. 21:14 VanessaE (for reference: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/random/Screenshot_2015-11-12_00-15-02.png ) 21:14 celeron55 maybe they should be offset so that z fighting wouldn't occur 21:16 who_wants_some i see z-fighting in OGL at least since 0.4.5 (i mean that texture tear I've posted on forum) 21:16 VanessaE who_wants_some: I don't think this is related to that. 21:17 who_wants_some maybe not