Time Nick Message 00:11 TBC_x I made a bash script that will put your directories in a *.tar.7z http://sprunge.us/GLKW 00:11 TBC_x asynchronously 00:12 TBC_x just made it so I could free up some space and stash away stuff 00:48 Tesseract hmmmm: So, return in macro or inline functions? 01:16 aerth client hangs at item definitions , what am i doing wrong for setting up the debian server 01:21 hmmmm Tesseract: Return in macro 01:21 hmmmm Tesseract: It sucks but there's not much better you can realistically do 01:33 paramat okay hmmmm, thanks for the good input, here's the more subtle commit that doesn't break mgv7 worlds #2935. back in an hour to push if approved 01:33 ShadowBot https://github.com/minetest/minetest/issues/2935 -- Mgv7: Use density noise + density gradient for mountain terrain by paramat 01:34 hmmmm what do you mean by "breaking worlds" exactly? 01:34 hmmmm mapgen v7 never really promised a stable terrain output like mapgen v6 does (the "stable" mapgen) 01:35 hmmmm I'm just saying, that drastically new generation algorithms probably deserve their own mapgen 01:35 hmmmm i made this mistake with mapgen v7 before, and it ended up pissing a lot of people off. they liked the old one better. 01:36 hmmmm since you took over development of v7, I think it's ultimately your own artistic decision of what to do with the noise params 01:37 hmmmm just don't make them too crazy since it is a default 01:40 cornernote hi, im trying to extract item cube images from the game, i had some code that worked, but that was quite some versions ago... 01:40 cornernote right now im trying this - https://github.com/minetest/minetest/compare/master...cornernote:patch-1 01:41 cornernote it compiles with no error, but it doesnt extract anything, nor does the infostream appear on screen 01:41 cornernote any chance someone can give me a kick in the right direction ? 02:20 RealBadAngel cornernote, preload_item_visuals = true ? 02:20 RealBadAngel its false by default 02:27 cornernote made it true in minetest.conf, still the same 02:32 RealBadAngel hold on for a few minutes, compiling your code 02:38 RealBadAngel first of all change infostream do dstream 02:38 RealBadAngel at first sight your code does something, it makes inventory cubes upside down ;) 02:42 cornernote lol, yes i noticed that 02:42 cornernote it did the same on the old version, but it did extract them the right way up 02:43 RealBadAngel where it was supposed to save files? 02:43 cornernote bin/itemcubes 02:49 cornernote yes dstream helps 02:49 cornernote at least now i can get some debug info 02:49 cornernote but no files extracted 02:49 cornernote driver->writeImageToFile(baseimg, filename); 02:49 cornernote its relative to the bin/ filder, right ? 02:49 cornernote *folder 02:54 RealBadAngel when i skipped folder name it saves images but not in bin but in main folder 02:55 RealBadAngel it saves it to folder you were executing mt binaries 02:55 RealBadAngel so if you run it with ./bin/minetest it will save it here 02:55 RealBadAngel if you were in /bin files will be in /bin 02:56 RealBadAngel also creating a folder helps 02:56 cornernote yes, i did that 02:56 cornernote did it work for you ? 02:56 RealBadAngel yes 02:56 cornernote holey moley 02:57 RealBadAngel first of all, create a folder inside bin 02:57 RealBadAngel then run binaries from bin directly 02:57 cornernote i see 02:57 cornernote ok 02:58 cornernote awesome! thanks RBA ! 03:04 cornernote all works except upsidedown stairs 03:05 cornernote they look like cobble with an unknown overlay 03:09 cornernote any ideas why the upsidedown stairs dont work ? 03:09 cornernote they were actually one of the main ones i wanted changed 03:09 cornernote *generated 03:09 cornernote brain not fully functional :) 03:10 cornernote upsidown slab not working either... anything upsidedown i guess 03:17 RealBadAngel first of all, its strange that your code makes images upside down 03:23 aerth stairs can be upside down? 03:28 RealBadAngel aerth, just roatate them using screwdriver 03:29 TBC_x Oh and btw, we're using uninitialzed memory from the very beginning 03:30 RealBadAngel TBC_x, what are you talkin about? 03:30 TBC_x http://sprunge.us/Mhjb 03:30 TBC_x haven't lookead at it yet 03:30 paramat hmmmm by 'worlds are not broken' i mean this commit re-arranges the 3D terrain equation but without changing terrain shape for a particular set of noise parameters. so for example the xanadu mgv7 server will not end up with discontinuities 03:31 TBC_x gonna do it when i get back from work 03:31 paramat effectively this commit is just subtle tuning of noise parameters, stayng close to the original character og mgv7 03:31 paramat (of) 03:34 paramat this commit is mainly fixing 2 problems with mountains: unsatisfyingly short, and too fragmented into small floatlands. so height is boosted a little and persistence reduced to 0.63 03:38 paramat mountain height stays reasonable, occasionally up to 256, very rarely up to 400. still 'in character' and proportioned for mgv7 03:44 paramat added a screenshot to the thread 04:05 cornernote RBA - its strange that it makes them upsidedown, but it doesnt effect the image that is saved 04:06 cornernote its stranger that it cant save the upsidedown stairs, but it can save the normal stairs 04:35 hmmmm lol. 04:36 hmmmm "i can't possibly be this new to image file formats" 04:36 hmmmm in most image formats, the top left corner is the origin. in a 3d space like irrlicht uses to render, the origin is at... well, the origin 04:44 hmmmm oh shit. i think i found the source of a lot of nerzhul's network bugs 04:44 hmmmm incrOffset() is plain wrong. 04:45 hmmmm i think the fact that most things seem to work okay is a fluke 04:46 hmmmm this is why unit tests should be required for new code 04:47 celeron55 how is it plain wrong? 04:47 celeron55 it looks okay to me 04:47 celeron55 what doesn't look okay to me is that the reads aren't properly range checked with the size of the buffer 04:48 celeron55 checkReadOffset() apparently pretends to be doing that but actually it doesn't do it 04:49 hmmmm right 04:49 hmmmm incrOffset assumes that the serialized length of a field is always the same as its length in memory 04:49 hmmmm this is obviously not the case for v3s16 04:49 celeron55 that's a problem, yes 04:50 hmmmm or V3F1000 04:50 hmmmm there are a lot of laws to be honest 04:51 celeron55 this is what i was saying about using template types for serialization back when this was originally discussed 8) 04:51 hmmmm and the entire NetworkPacket interface is kind of lame 04:51 hmmmm it'd be so much more flexible if there were a std::vector variant of the serialization functiosn 04:51 hmmmm s/laws/flaws/ 04:55 cornernote ok, i see why it cant get the upsidedown images... they don't seem to have their own textures... its just a rotation on the right-way-up images 04:55 cornernote i guess i dont need them, because you cant craft them anyway 04:55 cornernote you just use the screwdriver to rotate them 04:55 cornernote if you dig them, you get the right-way-up one back to your inventory 05:21 nrzkt will push https://github.com/minetest/minetest/pull/2940#issuecomment-123155208 in 30 minutes 05:21 nrzkt #2940, sorry :) 05:21 ShadowBot https://github.com/minetest/minetest/issues/2940 -- Remove profiler.h include where it's not needed and some unreachable code. by nerzhul 05:22 nrzkt hmmmm, are you there ? 05:41 hmmmm yes? 05:45 nrzkt https://github.com/minetest/minetest/blob/master/src/environment.cpp#L1174 05:45 hmmmm ? 05:45 nrzkt i don't understand why we loop multiples times over ABM 05:45 hmmmm i don't see any loop 05:45 hmmmm aside from iterating over active blocks 05:46 nrzkt i think this do while is totally useless, i removed it on my yesterday 05:46 nrzkt the do while 05:46 nrzkt on my server* 05:46 hmmmm do { ... } while (0) is equivalent to a bare code block like { ... } 05:47 hmmmm just that with the former, you're able to break; to the end of it 05:47 hmmmm and it's ANSI C conformant 05:47 nrzkt oh yes, it's while(0), i misread it 05:47 nrzkt it's strange to have a do while(0) there and never in other places 05:48 hmmmm in general, if you see a do { ... } while (0) outside of a macro definition, it's a good candidate to be made into a separate function 05:49 hmmmm that could be made into ServerEnvironment::applyActiveBlockModifiers() or something 05:49 nrzkt right 06:11 nrzkt pushed 07:37 aerth how do we map users like minetestmapper does blocks 07:38 hmmmm huh 07:39 hmmmm sounds like a #minetest question. 11:50 Player_2 mornin' all 11:52 nrzkt hi 12:33 PilzAdam hi; Zeno`what would be needed to fix #2145 ? should I bisect it? 12:33 ShadowBot https://github.com/minetest/minetest/issues/2145 -- Multi-second lag in default singleplayer game 12:34 Zeno` nrzkt, I disagree that removing the profiler stuff would gain any performance (none that anyone would not notice anyway) 12:34 Zeno` PilzAdam, I guess. I tried for weeks on that bug but couldn't reproduce at all 12:35 nrzkt Zeno`: as a concept a profiler doesn't have to be integrated into a release it's only used for debugging purpose 12:35 Zeno` PilzAdam, do you get the same issue if you run a local server and connect to it (rather than simple_singleplayer where the client and server are in the same process?) 12:36 nrzkt for the performance problem in singleplayer can this be related to the mapblock not unloaded clientside , not regarding to player distance to mapblocks ? 12:36 Calinou check the client and server mapblock timeout before 12:36 Calinou try reducing them to 30 seconds or so 12:36 Calinou and check if there's a performance difference 12:36 PilzAdam Zeno`, testing... 12:37 Zeno` nrzkt, but... it if you do a performance profile the profiler in minetest isn't even in the top 100 of things that are consuming CPU 12:37 Zeno` (well, last time I checked that was true) 12:37 PilzAdam Zeno`, yes, it does happen when I run the server in a separate instance 12:38 Zeno` PilzAdam, ok cool 12:38 Zeno` I wish we could work out why :( 12:38 Zeno` I don't experience the issue 12:38 Calinou me neither 12:38 Calinou OS-specific perhaps? 12:38 Zeno` perhaps 12:38 Zeno` but PilzAdam uses Fedora? 12:38 PilzAdam do you have luajit? 12:39 PilzAdam Zeno`, ubuntu 12:39 Calinou all players should use LuaJIT :P 12:39 Zeno` I don't use luajit because valgrind doesn't like it 12:39 Zeno` (not without fancy tricks anyway) 12:39 Zeno` and, no, I don't experience the issue without luajit 12:40 Zeno` I'm glad that someone else has finally experienced it though 12:40 Zeno` Maybe some progress can be made now :) 12:40 PilzAdam Calinou, #server_unload_unused_data_timeout is set to 29 by default 12:40 PilzAdam and I guess the client isn't a problem if the server lags 12:42 nrzkt Zeno`: i agree with you, but each CPU percentage we win on each useless part can, combined, reduce the CPU usage and be viewed by users :p 12:46 Zeno` nrzkt, I have changed my mind. I use valgrind to profile because it's more accurate and the profiler doesn't really show much that's interesting at all because it's probably a bit "coarse" 12:46 Zeno` the graphs are pretty though 12:47 Zeno` even though I have NFI what they actually show LOL 12:58 PilzAdam Zeno`, it seems to happen 0.4.11, too, but not as bad 12:58 PilzAdam this suggests that it got worse since then 12:58 PilzAdam and it complicates bisecting 13:02 Zeno` :( 13:03 nrzkt Zeno`: i also agree with you, valgrind permit to profile and isn't selective :p 13:05 Zeno` nrzkt, I mentioned yesterday to hmmmm that'd I'd probably make a PR from my valgrind hooks in the code which would make the profiler pretty much redundant 13:09 nrzkt what is the interest of your hooks ? can we disable it with a make variable? :) 13:10 Zeno` yeah of courswe 13:11 Zeno` otherwise most people would not be able to compile anyway. The hooks are just to make controlling valgrind instrumentation easier to handle 13:11 Zeno` (e.g. start profiling here, stop here) 13:12 Zeno` Not really sure how I'm going to make it "clean" enough for a PR yet, but it's in the works 13:13 nrzkt cool ! 13:15 PilzAdam Zeno`, what else could be done instead of bisecting? 13:28 Zeno` PilzAdam, you could provide some additional callgrind data on top of what wayward1 provided. I guess :( 13:31 Zeno` PilzAdam, Wayward used something like: https://github.com/Zeno-/mtutils/blob/master/profile_client.sh and used "callgrind_control -i on" and "callgrind_control -i off" to profile the code after init was done and before shutting down 13:31 Zeno` I'm really out of suggestions :( 13:32 est31 perhaps it would be great to find out what component is lagging 13:32 est31 server vs client 13:34 Wayward_One I've noticed noticeably less lag when running a local server vs. just using singleplayer 13:36 Zeno` est31 you can instruct callgrind to break everything up into threads 13:36 Zeno` I can't remember the command line offhand though 13:37 Zeno` --separate-threads=yes 13:38 Zeno` might be useful *shrug* 13:38 Zeno` it's a bit harder to interpret though :) 14:02 * Sketch2 wonders what Wayward_One means by "running a local server vs. just using singleplayer" means, as they are essentially the same thing. 14:04 Wayward_One I start a local server from the terminal, then connect to it. Anyways, I'm not saying it does reduce lag, just that I have noticed lower lag while playing that way 14:41 Zeno` Sketch2, it *is* different 14:41 Zeno` the singleplayer scenario uses threads; the local server uses processes 14:42 Zeno` I assume that process switching by the OS is more optimised than the multithreading 14:55 Sketch2 I thought the singleplayer just sets up a local server in the background for the client to connect to. That's what the window of text is... 14:56 nrzkt yes 14:56 nrzkt maybe forking to create a server could be better than starting the thread ? :) 14:57 Sketch2 singleplayer is the only one connected, of course, but the game mechanics are no different 14:57 est31 Zeno`, can you have a look at #2922 14:57 ShadowBot https://github.com/minetest/minetest/issues/2922 -- Optional reconnect functionality by est31 15:18 est31 what do ppl think can I push this https://github.com/est31/minetest/commit/403e6e6c9cf6be5e16e3f1f9bd7805fe9b3006cd 15:19 nrzkt est31: ok 15:20 est31 pushed 16:07 est31 ok bugfix pr #2942 is out 16:07 ShadowBot https://github.com/minetest/minetest/issues/2942 -- Fix srp default password by est31 16:20 est31 man, default passwords didn't work with srp, and nobody noticed 16:20 est31 well, at least its fixed now 16:50 RealBadAngel merging #2934 in half an hour. last chance to comment it 16:50 ShadowBot https://github.com/minetest/minetest/issues/2934 -- Add wielded (and CAOs) shader by RealBadAngel 16:52 est31 RealBadAngel, who has +1ed it? 16:52 hmmmm I did 16:53 est31 then its ok 16:53 hmmmm did anybody else other than me look at it though? 16:54 est31 Zeno has commented "Looks good to me" 16:55 est31 question is: fsaa works without shaders enabled, right? 16:55 est31 so the bugfix is only if shaders are enabled?? 16:56 Calinou fsaa is not shader-dependent 16:56 Calinou FXAA would be though 16:57 RealBadAngel hmmmm, paramat actually compiled it and tried 16:57 TBC_x So, what was decided on how to handle nls? 16:59 hmmmm great 16:59 hmmmm NLS? 16:59 TBC_x native language support 16:59 hmmmm *shrug* 17:00 est31 wats that 17:00 TBC_x translations 17:01 est31 btw its "National" language support, not "Native" language support 17:01 est31 its not just for indians :) 17:05 RealBadAngel https://github.com/minetest/minetest/pull/2934#issuecomment-122815609 17:05 RealBadAngel is that +1? 17:05 RealBadAngel yes 17:05 RealBadAngel thx 17:05 RealBadAngel about FSAA i know how to fix the plants, i will push fix for this asap as previous one is in 17:07 RealBadAngel Calinou, problem with fsaa is the the same i was fighting lately for not tiling textutes 17:08 RealBadAngel ive found my own way, kahrl found another 17:09 RealBadAngel both are based on that modder (designer) should know what hes doing 17:09 RealBadAngel theres no magic way for shading pipelines to decide, whats seamless 17:10 TBC_x msgid "Toggle fast" 17:10 TBC_x msgstr "Turbo" <---- <3 17:14 RealBadAngel theres no "turbo" switch on pc boxes for quite a while... 17:15 TBC_x Who takes care of czech translation? 17:16 est31 TBC_x, do it yourself :) https://hosted.weblate.org/projects/minetest/ 17:16 Krock RealBadAngel, there's one on my 133 MHz super speed PC 17:22 TBC_x what do simple leaves toggle? I don't want to try on my hardware. 17:23 RealBadAngel u should 17:23 RealBadAngel they are toggling the drawtype 17:24 RealBadAngel glasslike is way faster than allfaces 17:25 RealBadAngel a bit worse lookin, but does not cause such dramatic fps drops as allfaces can 17:25 RealBadAngel its something in the middle 17:27 TBC_x is there a mod that has mobs running in circles? 17:28 RealBadAngel TBC_x, do you have more strange questions prepared for today? 17:28 TBC_x yes, what is this if (m_view_bobbing_anim <= 0 || m_view_bobbing_anim >= 1 || 17:28 RealBadAngel if i see bobbing i think taoki 17:29 RealBadAngel ask him 17:31 TBC_x because it looks like someone forgot to add .0 17:35 RealBadAngel yeah, its f32 17:40 RealBadAngel but thats not an error for c++, glsl by various vendors can make problems 17:40 TBC_x are you talking about the bobbing? 17:40 RealBadAngel using a dot in glsl for amd is like definin the type of the variable 17:41 RealBadAngel so 1 would mean int in glsl 17:41 RealBadAngel 1.0 would be a float 17:41 TBC_x well... Doesn't it upcast the comparsion to int? 17:41 TBC_x so the condition is always true? 17:42 RealBadAngel no, it throws an error, and makes shaders unusable 17:42 RealBadAngel for nvidia what was defined before is important 17:43 RealBadAngel missing a dot is not 17:43 RealBadAngel amd is smarter and thinkin that 1.0 != 1 17:44 TBC_x I'm very unfamiliar with shaders 17:44 RealBadAngel shaders language is mostly c++ clone 17:45 RealBadAngel but each vendor does it in its own way 17:45 RealBadAngel thats a real pain in the ass 17:45 RealBadAngel because something you wrote may be working perfectly for you 17:46 RealBadAngel and for some other user it will fail on such a stupid thing 17:46 TBC_x yeah, in yards for you, in meters for them 17:47 RealBadAngel i dont have many platforms to test my code on 17:47 TBC_x me neither 17:47 RealBadAngel so im usally blamed for the code that is failing 17:47 TBC_x but better make it PIC compatible :) 17:47 RealBadAngel i would rather go back to Z80 times ;) 17:48 TBC_x I found a Z80 17:48 TBC_x but I don't know where the mouse disappeared 17:48 RealBadAngel i would LOVE to have z80 here in minetest 17:48 RealBadAngel we could have real computers in game 17:49 RealBadAngel eloraam chosen 6502 for redpower2 17:49 TBC_x yeah 17:49 RealBadAngel we can easily have z80 17:50 RealBadAngel notch have chosen 16bit for his next project 17:50 TBC_x I played MC a lot before the betrayal 17:51 TBC_x I'd love to put together a 6502 based computer though 17:51 TBC_x just for lulz 18:00 RealBadAngel its not just for "lulz" 18:00 RealBadAngel for coders its lotsa fun 18:00 RealBadAngel thats the reason to build miracles in the game 18:00 Calinou why are people so excited at old computers they can emulate in their games? 18:00 RealBadAngel not just simply put one block on another 18:01 RealBadAngel Calinou, its memories and unbeliveable fun, believe me 18:02 RealBadAngel not to say that fullfills the "sandbox" definition 18:02 TBC_x I've got no such memories, our first computer was just 8086 with ugly DOS 18:05 TBC_x But I'm very likely to do more stuff with little piece of hardware than with a full blown CPU 18:05 TBC_x i mean PC 18:07 RealBadAngel TBC_x, with z80 lib here we can have CP/M 18:08 RealBadAngel with all the soft written for it 18:08 RealBadAngel mt run on "fairly old hardware" is able to run z80 too 18:11 TBC_x we were taught 8051 assembly in school 18:14 TBC_x I sticked very quickly to indirect threading 18:15 TBC_x given me enough time and I would've reinvent forth 18:29 RealBadAngel TBC_x, oh cmon, asm is all the fun in the world 18:30 RealBadAngel there are no whitespaces, conventions etc. only a machine doing exactly what your order it to do 18:41 RealBadAngel btw. /../builtin/mainmenu/textures.lua:157: bad argument #2 to 'random' (number expected, got nil) 18:41 RealBadAngel sounds like https://github.com/minetest/minetest/commit/899491325938d0b1e5bf701d33d24bb4283affaa is broken 18:52 TBC_x the check should be for nil, and math.random in the else branch 18:53 TBC_x or just init n to 0 18:54 TBC_x If I knew how is it supposed to work, I would fix that 19:12 TBC_x systemd is such a nusiance 19:24 TBC_x don't try to make coredumps of binaries with sanitizers 19:56 sfan5 pushing http://sprunge.us/ZCbT in 5 minutes, thanks to Sokomine for reporting this 20:31 nrzkt hmmmm: i think you life bug fix re-enable another bug: #2945 re-appear 20:31 ShadowBot nrzkt: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request." 20:31 nrzkt #2945 20:31 nrzkt bot is HS ? :o 20:31 ShadowBot nrzkt: Error: Delimiter not found in "Page is too big or the server took too much time to answer the request." 20:31 nrzkt Immortal/Zombie Glitch #2945 20:31 ShadowBot https://github.com/minetest/minetest/issues/2945 -- Immortal/Zombie Glitch 20:37 sfan5 oops, totally forgot 21:09 TBC_x virtual classes suck 21:47 paramat will push #2935 soon when checks are done 21:47 ShadowBot https://github.com/minetest/minetest/issues/2935 -- Mgv7: Use density noise + density gradient for mountain terrain by paramat 21:51 paramat sfan5 i'll add acacia wood to stairs mod soon, please let me know sometime if you approve of steel block being added too, see game#581 21:51 ShadowBot https://github.com/minetest/minetest_game/issues/581 -- New stairs and slabs discussion 21:51 sfan5 dont we already have steel blocks? 21:52 sfan5 or do you mean steel slabs 21:52 paramat i mean registered in stairs mod 21:52 sfan5 well 21:52 sfan5 hm 21:52 sfan5 we don't have bronze or gold stair 21:52 sfan5 s 21:53 sfan5 so i don't think steel stairs would fit 21:54 paramat well yes we should be restrained, steel block only, as it's an essential stair where stone or wood stairs are unsuitable 21:56 paramat modern/space/futuristic/vehicles/airships where wood or stone stairs would be ridiculous 21:57 sfan5 hm 21:57 sfan5 i think i can agree on that 21:57 sfan5 but we should ask the other _game devs too 21:58 paramat okay 21:58 RealBadAngel sfan5, cornernote is updating now skyblock, could you include then it in your builds? 21:58 sfan5 link? 21:58 RealBadAngel its not rdy yet i think, he was working today on skyblock page 21:59 RealBadAngel what i thought skyblock is one definitely worth including 21:59 TBC_x oh boy, my mt binary has 235M 21:59 sfan5 you should really disable debug symbols 21:59 sfan5 RealBadAngel: yeah, including skyblock is a good idea 22:00 RealBadAngel http://cornernote.github.io/minetest-skyblock/ 22:00 sfan5 ok 22:00 sfan5 noted 22:00 sfan5 now: good night 22:00 TBC_x i compiled irrlicht with -fsanitize and -g 22:01 TBC_x are routines in noise.cpp supposed to overflow? 22:11 paramat now pushing 2935 22:11 RealBadAngel paramat, https://encrypted-tbn2.gstatic.com/images?q=tbn:ANd9GcSn_ewtYUiDLDpN1p4qZP-x9xe7k_h781FpfgPh-En_d79Hj9Tq 22:11 RealBadAngel such metal surface is commonly used for machines 22:11 RealBadAngel for ppl to safely walk on 22:12 * paramat looks 22:12 RealBadAngel so metal stairs are not such bad idea imho 22:12 TBC_x where can i stretch the timeout delay? 22:13 paramat yeah perhaps we could have a dedicated steel stair node 22:18 paramat push complete 22:20 paramat mgv7 tuned and higher mountains =) 22:26 paramat wield looks really good now 22:27 paramat when stairs were changed to meshes they used to appear dark as a wield, thats fixed 22:29 TBC_x mapgen needs to get wilder IMHO 22:30 paramat heh i'm working on 2 new ones 22:31 paramat and you can use custom noise params in the current ones 22:31 TBC_x btw, are routines in noise.cpp supposed to overflow? 22:31 paramat that's one for hmmmm 22:33 hmmmm ?? 22:34 hmmmm depends how crazy the params are 22:34 hmmmm but no, generally they are not 22:36 TBC_x I got signed int overflow errors 22:36 TBC_x noise.cpp:159:23: runtime error: signed integer overflow: 2076139895 * 1013 22:37 TBC_x at client startup 22:38 paramat which mapgen? all params are sensible currently 22:42 TBC_x I don't know, this code runs at client startup 22:43 hmmmm oh integer overflow 22:43 hmmmm i thought you said memory overflow 22:44 TBC_x If we'll be nice, we may see some 22:45 hmmmm ahhh 22:45 hmmmm the code where it hashes up a position 22:45 TBC_x I haven't looked at it, just interpreting sanitizer results atm 22:46 hmmmm this is perfectly normal, i'm not entirely sure how to silence the warning 22:48 paramat phew 22:57 hmmmm yeah. for that matter i'm surprised the pseudorandom algorithms didn't get marked for integer overflow 22:59 hmmmm paramat: the primary issue with those "vast" mapgens is that their majestic effect is kind of ruined when you can't see more than 5 feet in front of you 22:59 hmmmm which is a client->server block transmission problem mostly 22:59 paramat yep 23:00 hmmmm this is why mapgen v6 held up so well 23:00 hmmmm its features with the default params are quite local 23:03 kahrl hmmmm: pseudorandom didn't get flagged because it is using unsigned types, I think 23:12 TBC_x boom #2946 23:12 ShadowBot https://github.com/minetest/minetest/issues/2946 -- Fixed minimap memory leak by t0suj4 23:23 TBC_x oh wow, minetest takes up 20TB of virtual memory 23:28 paramat cd .. 23:28 paramat sorry 23:32 Routh Is there a way to get the minetest server to say why it refuses to generate a biome? I reduced my entire biome defs down to one biome to see it render and it refuses too but I cannot identify why 23:34 paramat link me to your github? i could take a look 23:35 paramat game#585 23:35 ShadowBot https://github.com/minetest/minetest_game/issues/585 -- Flowers: Add mushrooms to mgv6 by paramat 23:36 Routh paramat: https://github.com/Routhinator/tekkest/blob/master/mods/base/biomes.lua 23:36 Routh It's the last biome, Tarpits, that's giving me issues. 23:37 Routh I've removed all the biomes and set just that one in there with 50/50 heat/hum and y_min/max -31000/32000 - still not generating 23:37 paramat you're using latest 0.4.12dev? 23:39 paramat make sure to have 'clear registered biomes' and 'clear registered decorations' at the start of your biomes file 23:39 Routh Yep 23:39 paramat otherwise default biomes interfere 23:39 Routh paramat: Have a code sample of that? 23:39 paramat my biomesdev mod 23:39 Routh kk 23:40 Routh Must have missed that. 23:43 Routh paramat: unfortunately after adding that, same result. 23:43 paramat =/ 23:45 hmmmm TBC_x: nice. two things, however. 23:45 TBC_x http://sprunge.us/iKFU Leak sanitizer output after walikng around on vanessa's server 23:45 TBC_x i listen 23:46 hmmmm if (foobar != NULL) delete foobar; is redundant because delete does no operation if foobar is NULL 23:46 TBC_x k 23:47 hmmmm second, looking up the value in a std::map using operator[] adds that item to the map with the default constructor if it doesn't already exist 23:47 hmmmm what would be better is to replace that with 23:47 hmmmm std::map::iterator it = m_blocks_cache.find(update.pos); 23:48 hmmmm if (it != m_blocks_cache.end()) { delete it->second; m_blocks_cache.erase(it); } 23:48 hmmmm hrmm.. 23:49 TBC_x I would approach it by swapping pointers and deletion then 23:49 hmmmm yeah 23:50 TBC_x do you want me to fix it? 23:50 hmmmm actually 23:50 hmmmm looking at that code again it might be a better idea to do 23:50 hmmmm std::pair::iterator, bool> result = m_blocks_cache.insert(std::make_pair(update.pos, update.data)); 23:51 hmmmm if (result.second == false) { delete result.first->second; result.first->second = update.data; } 23:52 TBC_x better check whether this code is also redundant 23:52 hmmmm this code does exactly one lookup (and insertion, if the item does not exist already)