Time  Nick         Message
00:00 paramat      GetBiomeInfo what biome info are you requesting? biome at point?
00:00 GetBiomeInfo paramat: at point, why not? the one hmmmmmmmmmmmmm wanted
00:00 paramat      i'll add that if i can
00:01 paramat      it's much requested
00:01 paramat      i miss hmmmmm
00:02 paramat      ok it's on my todo list
00:03 GetBiomeInfo weather mods will use it for sure
00:04 GetBiomeInfo they can request biome info at point and decide snows it or rains it etc
00:04 twoelk       hm - so mobs can wander around and still keep to their biome not by looking for nodes but by requesting biome info?
00:04 GetBiomeInfo or when growing stuff
00:05 GetBiomeInfo say you trying to grow vegitables in arctic... > biomecheck > will not grow
00:05 est31        I just fail to see the point in organizing defaultsettings.cpp
00:06 paramat      i think it's organised enough
00:06 paramat      as long as those mapgen things are moved as requested
00:06 est31        but shrug, if he really wants to organize it...
00:06 twoelk       so some greenhouse mod will have to override biome info to grow stuff out of biome ;-P
00:06 paramat      yeah no prob
00:07 paramat      lem
00:08 ShadowNinja  paramat: I updated the defaultsettings pull.  Does it look good to you now?
00:09 GetBiomeInfo twoelk: you can add more challenge to farming for sure
00:10 paramat      looking
00:11 twoelk       so would each node carry biomeinfo or will it be recalculated at request?
00:12 paramat      there's already a function that calculates biome at point, it's used in mapgen, the API would just access that
00:12 paramat      that returns a biome ID, then in lua you can get the biome name and maybe other info
00:13 twoelk       so regardless of player or mod action the biomes always stay the same?
00:14 twoelk       what about mods that change mapgen largescale?
00:14 paramat      yes defined by the heat/humidity noise params and the 'heat/humidity points' of registered biomes
00:15 paramat      if mods change the nodes in the terrain, calculated biome locations don't change
00:17 twoelk       so all them built over lakes will stay biome corralsea and wont grow trees if they are told to stick to biomeinfo :-D
00:17 paramat      correct
00:19 twoelk       can't remember; did minetest have underground biomes?
00:19 paramat      it does, just 1 from y = -112 down
00:20 paramat      just default:stone
00:20 paramat      mainly because other stone types have no ores
00:21 twoelk       I wonder if  a cave biome might be usefull
00:22 paramat      perhaps other stone types should have coal, to make it easier to find
00:22 paramat      if i'm allowed to make coal a full-cube texture then we won't need multiple coal nodes
00:23 twoelk       actually biomes could be used to make largescale regions with different properties/ores underground
00:23 paramat      coal is suitable to be full-cube
00:23 paramat      biomes don't affect cavegen
00:23 paramat      ores can be limited to certain biomes
00:24 paramat      i do this for dirt blobs
00:24 twoelk       yeah, but you could have coal rich biomes or other stuff
00:24 paramat      yes
00:25 twoelk       you could have different types of caves according to biome
00:28 paramat      caves are cut after biomes are calculated and placed so theoretically yes
00:28 paramat      but i'm not enthusiastic
00:29 paramat      caves are part of 'base terrain', which is fairly independent of biomes
00:29 paramat      'base terrain generation'
00:29 zorman2000   paramat, I think it would be nice to have different types of caves by default
00:29 zorman2000   But I do understand what you are saying
00:30 twoelk       I was just musing this could give the underground some structured variety
00:30 paramat      well i'm going to add underground decorations
00:31 zorman2000   That would be very nice
00:31 paramat      (or try to)
00:31 zorman2000   A glowing mushroom
00:32 twoelk       like the red ones on Xanadu that grow only on exposed coal?
00:33 ShadowNinja  paramat: Good to merge defaultsettings then?
00:35 paramat      hm i will have to trust you on the changes, have you rechecked since latest update?
00:35 ShadowNinja  Yep.
00:35 paramat      +1
00:38 ShadowNinja  Just #5031 left now.
00:38 ShadowBot    https://github.com/minetest/minetest/issues/5031 -- Server list cleanup by ShadowNinja
02:51 paramat      sofar game#1513
02:51 ShadowBot    https://github.com/minetest/minetest_game/issues/1513 -- Tree log decorations: Replace lua tables with .mts files by paramat
04:24 twoelk       the more I think of it the more I like the idea of underground biomes
04:26 twoelk       adding some things that makes living underground possible for extended mining might also reduce the need of underground farming of normal crops and trees
04:28 twoelk       some underground source of sticks and something edible
04:33 twoelk       there could be biomes of more sediments or others more karst like with sinkholes near the surface, ore rich lava contact zones or large caves with some sort of flora feeding on energy rich ores, crystal rich crevices and deep shafts
04:49 paramat      there could be edible mushrooms
11:06 nrzkt        est31, i pushed a first for #5033
11:06 ShadowBot    https://github.com/minetest/minetest/issues/5033 -- BADLY broken: Recent checkins break Active Objects
11:06 nrzkt        a fix*
11:07 nrzkt        a missing const in parent ServerActiveObject
11:17 sfan5        nrzkt: clang still complains about this http://sprunge.us/DDMQ
11:18 nrzkt        sfan5, okay i'm looking for this too and pushing the fix when it's done
11:22 nrzkt        it's pushed, thanks
11:26 Thomas-S     I don't know, if this is already reported, but building the latest sources from gitub fails: http://pastebin.com/raw/7wexPs2S
11:27 sfan5        you can blame ShadowNinja for that
11:28 sfan5        Thomas-S: fix pushed
11:29 Thomas-S     Thank you very much!
11:32 nrzkt        sfan5, travis build are releases, right ?
11:34 nrzkt        sfan5, can you repush i did an error on master
11:36 nrzkt        i discaded your commit by pushing to wrong origin
11:38 nrzkt        sfan5, i take it from your origin, discard my request for push
11:40 nrzkt        sfan5, est31 sofar Zeno` please review #5035
11:40 ShadowBot    https://github.com/minetest/minetest/issues/5035 -- Add WARN_AS_ERROR flag to prevent non fixed warnings by nerzhul
11:52 Zeno`        why are warnings ignored in the first place?!
11:56 Zeno`        I don't think anything that adds warnings should even be merged
11:56 Zeno`        if it's a false alarm (and you're 110% sure) then suppress the warning, otherwise it should be taken seriously
11:57 Zeno`        I guess if we're not disciplined enough to take warnings seriously then -Werror should be add to ALL builds
11:57 Zeno`        not just release
11:59 Zeno`        sorry nrzkt but I can't +1 that PR
11:59 Zeno`        it's either for all builds or none at all
12:01 Zeno`        also doesn't LuaJIT have warnings?
12:01 Zeno`        i.e. stuff we can't fix?
12:02 Zeno`        (or Lua... one of the two)
12:08 Zeno`        this one: https://travis-ci.org/minetest/minetest/jobs/191892538#L596
12:08 Zeno`        does travis not build release versions?
12:18 nrzkt        Zeno`, i can make it enabled for all UNIX builds if you want
12:19 nrzkt        i tested on my machine keeping default build options, i removed a const in inheritance to make overloading and it make errors, now just tell on the  PR if we enable it everytime or just on release build ?
12:20 Zeno`        it makes error for debug builds?
12:20 nrzkt        no, as flag is only pushed to release build by default, but i can change it
12:21 Zeno`        why should debug builds be treated differently?
12:21 nrzkt        i don't think we have debug builds
12:21 Zeno`        I really don't like it because I'm not sure why we would ignore warnings anyway
12:21 nrzkt        i just kept WARN_ALL thing
12:21 Zeno`        oh, come on
12:22 Zeno`        nearly every build I do is a debug build :P
12:22 nrzkt        i don't want to ignore warning personnaly
12:22 Zeno`        i can't remember the last release build I did
12:23 Zeno`        nrzkt, how come the Lua warning is not treated as an error?
12:23 Zeno`        (with your PR I mean)
12:24 nrzkt        because it's applied on CMAKE_CXX_FLAGS in src/ directory
12:24 nrzkt        like every other MT flag
12:24 nrzkt        i changed the behaviour, it's now enabled by default on every UNIX build, not depending on debug/release
12:24 nrzkt        and rebased from master
12:25 Zeno`        where is the bundled Lua?
12:25 nrzkt        src/lua
12:25 nrzkt        and CMAKE_CXX_FLAGS not CMAKE_C_FLAGS are modified
12:25 nrzkt        i should go, see you, add comments on the pr
12:25 Zeno`        ahh ok
13:45 IhrFussel    Can someone tell me where in the code the engine renames users as "unknown" when they are no longer connected? I think there is a bug in there
13:47 IhrFussel    IMO the engine shouldn't even display such "unknown" clients, it just should remove them from the list...or was that implemented on purpose?
13:51 Krock        IhrFussel, grep helps. server.cpp L2900
13:52 Krock        this means the player are already removed from the environment - just not from the client list
13:53 Krock        the line number might be sligtly different. didn't update my repo
13:53 IhrFussel    Krock, and that doesn't get fixed because?
13:53 Krock        either because the source of this bug isn't known or nobody wants to fix it
13:54 Krock        just ignoring these unknown players isn't how you solve such a problem
13:56 IhrFussel    Exactly...well I'm experiencing "disconnecting peer" issues in the log and think those are the clients that are still shown as unknown in the server status...and btw that bug is also very annoying cause the diconnected client does NOT ever tell "connection timeout" when that happens, so the clients continue playing and playing until they realize that nothing loads anymore
13:57 Krock        maybe it's because the connection didn't timeout but the server doesn't handle its requests anymore
13:58 IhrFussel    Clients should see an error in that case too...should I file an issue on github regarding that?
13:58 Krock        there are at least two issues about that
13:59 Krock        players being deleted by accident, which causes these "unknown" entries in the list
13:59 IhrFussel    What should I look for as keyword?
14:01 Krock        I used the keywords "player delete" and got #4067 and #1425
14:01 ShadowBot    https://github.com/minetest/minetest/issues/4067 -- Canceling: No player for peer_id when too much objects on the map
14:01 ShadowBot    https://github.com/minetest/minetest/issues/1425 -- Deleting extraneous entities from a mapblock deletes PLAYERS also
14:01 Krock        I believe unknown players are also caused by one of these issues
14:02 Krock        basically, issue 4067 is just a duplicate
14:05 IhrFussel    But #4802 didn't seem to fix it then...my server version is new from Jan 3
14:05 ShadowBot    https://github.com/minetest/minetest/issues/4802 -- Don't clear excess entities on mapblock save (possible fix for #4067, #1425, etc.) by cheapie
14:07 IhrFussel    And the pull 4847 that referenced it didn't fix it either :/
14:10 IhrFussel    I get that error every few hours...sometimes it only deletes one client, other times up to 5
14:10 IhrFussel    I'll increase the object limit even more maybe that fixes it
14:15 Fixer        0.4.15?
14:16 IhrFussel    Fixer, server version is 0.4.15-dev (Jan 3)
14:16 IhrFussel    The client version shouldn't matter...right?
14:17 Fixer        no idea
14:22 IhrFussel    Well logic tells me no it should absolutely not matter since it's the server code that deletes the player
14:23 * Zeno`      hugs everyone
14:25 IhrFussel    Maybe the function that removes objects needs a certain check? Like "entity or player"
14:27 IhrFussel    Or player objects need some property that makes them unique compared to other objects
14:32 IhrFussel    But sometimes It even happens when the mapblock has 0 objects or only a few...so I kinda doubt it's the only problem
14:58 nrzkt        sfan5, i changed the PR to a simple thing: mark only overloaded-virtual as an error
14:58 nrzkt        #5035
14:58 ShadowBot    https://github.com/minetest/minetest/issues/5035 -- Mark overloaded-virtual warning as an error by nerzhul
15:10 sfan5        nrzkt: good idea if it's only for overloaded-virtual +1
15:46 IhrFussel    Is there a way to download the exe without zip?
15:47 sfan5        no
15:48 IhrFussel    sfan5, but if I only copy the .exe file would it work?
15:48 sfan5        depends
16:12 Calinou      .exe of what, IhrFussel?
16:12 Calinou      the Minetest executable needs some data files to function
16:12 Calinou      it won't work alone
16:15 IhrFussel    Calinou, I meant to post that in minetest channel oops^^ Umm well there is a player who apparently doesn't know how to extract .zip so I was just asking if it's possible to simply send it without the other files...but we already got the "not found" error haha
16:15 Calinou      nope, it's not possible
16:15 Calinou      extracing a .zip is really easy, come on
16:15 Calinou      by the way, Windows 10 will recommend extracing a .zip when you try to run an .exe file inside
16:18 IhrFussel    You know some players are VERY VERY young on my server so no it's not easy for them...but I still try to find solutions, this time around seems like there's none
16:22 Fixer        IhrFussel: i guess exe installer is good for them, it should create a Shortcut in Start menu somewhere
16:23 IhrFussel    Fixer, I honestly thought that the .exe in /bin is an installer
16:23 Fixer        it is not
16:24 IhrFussel    I figured that out now xD
16:25 IhrFussel    So Minetest on Windows never installs?
16:27 Fixer        lets move to #minetest
16:30 Fixer        0.4.15 has no installer right now
16:35 sfan5        varying packet serialization depending on prot ver in the Server::Send<something>() methods is a total clusterfuck
18:57 sapier       https://github.com/minetest/minetest/pull/5041 please comment
19:30 paramat      good to see you here
19:31 paramat      will merge game#1511 game#1513 later
19:31 ShadowBot    https://github.com/minetest/minetest_game/issues/1511 -- Explosions should extinguish fire by lhofhansl
19:31 ShadowBot    https://github.com/minetest/minetest_game/issues/1513 -- Tree log decorations: Replace lua tables with .mts files by paramat
19:40 red-001      rubenwardy do you approve #5000? It is the second commit of #4472 and you approved that pr
19:40 ShadowBot    https://github.com/minetest/minetest/issues/5000 -- Simpler serverlist search by red-001
19:40 ShadowBot    https://github.com/minetest/minetest/issues/4472 -- Make serverlist searchable. by red-001
19:42 red-001      also sofar you said that you approve the first two commits of #4472 does that mean that you approve #5000?
19:42 ShadowBot    https://github.com/minetest/minetest/issues/4472 -- Make serverlist searchable. by red-001
19:42 ShadowBot    https://github.com/minetest/minetest/issues/5000 -- Simpler serverlist search by red-001
20:48 paramat      i forgot to change the spreads of heat and humidity noises in settingtypes.txt and minetest.conf, will fix very soon
21:27 sapier       https://github.com/minetest/minetest/pull/5043 modders might be thankfull about this
21:28 VanessaE     holy shit
21:28 VanessaE     it's a ghost :)
21:29 nrzkt        merging #5023
21:30 ShadowBot    https://github.com/minetest/minetest/issues/5023 -- Only set material flag on rendered meshes by lhofhansl
21:32 sapier       buuuuuhhhhh ;-)
21:33 nrzkt        hi sapier
21:33 sapier       hello nrzkt
21:38 est31        sapier: server side security is something different than client side
21:38 est31        also, we broke actual features
21:39 est31        with clientside security, there is no problem if we "break" something because we won't promise to keep anything stable until its enabled by default
21:39 est31        in fact, I'd say that even if you disabled mod_security, the client side lua security should still be enabled
21:40 sfan5        sapier: what made you use a Makefile for android building, it fucking sucks
21:40 sfan5        also hello
21:40 sapier       yes that's why we can demand security to be added first, once it's merged without security we will "break features" again
21:40 est31        well idk, i doubt that duckbrain can implement the security I have in mind
21:41 sapier       sfan5 well I was completely new to android and android by that time was about to switch build systems anyway so I used what I already knew
21:41 est31        it should involve at minimum a separate sandboxed process for defense in depth
21:41 sapier       btw didn't google complete that switch and discontinue the old build system anyway?
21:41 est31        sapier: I think it was about maven vs gradle
21:42 est31        sapier: we have since switched to gradle
21:42 est31        for NDK nothing was deprecated yet, afaik
21:42 est31        and tbh, I've seen worse android NDK build scripts
21:42 sapier       I don't know I was hoping someone else would start continuing the android thingy once the initial pains have been done ... obviously that's been an error
21:43 est31        you know, libreoffice actually uses automake
21:43 est31        automake, can you imagine that
21:43 est31        https://github.com/minetest/minetest/commit/e41673820ffe200df78b1ec185ccb9d9ca962ae1
21:44 sapier       for what I remember rest of minetest uses cmake ;-)
21:44 sapier       I don't know but maybe cmake provides some kind of android support in between?
21:44 est31        yes
21:44 est31        it has android support now
21:44 sfan5        experimental cmake android support was added recently
21:44 sfan5        est31: btw please review #5039 if you have time
21:44 ShadowBot    https://github.com/minetest/minetest/issues/5039 -- Add particle animation, glow by sfan5
21:45 sapier       so what about just waiting a little bit and switch to this one once it's at least somehow useable?
21:45 sfan5        there's also the issue with building dependencies
21:45 est31        I'm okay with a switch
21:45 sfan5        that should be done seperately anyway
21:45 sfan5        switching to cmake for android too sounds good though
21:45 est31        as long as it still works
21:46 est31        cmake has functionality to download dependencies
21:46 est31        also, to specifically pin hashes
21:46 est31        that's very nice as it allows for a more verified build
21:46 sapier       sounds great ... just one open issue ... who's gonna do it? ;-)
21:46 est31        :)
21:47 paramat      lordfingle says he may be able to help out with android soon
21:48 paramat      he saw my request in the forum for new android maintainers
21:48 est31        nice
21:49 est31        its a quite big change though, not sure whether hes' ready for it
21:49 est31        he's
21:51 sapier       great there's no (satisfying) way to colorize a sheep using colors instead of separate textures atm
22:15 sofar        death to cmake, autotools all the projects!
22:16 sfan5        pls no
22:16 sofar        having worked on 4 linux distributions, of which 3 professionally, for over 10+ years, I can state, without hesitation, that cmake is a fucking nightmare every damn time
22:17 nrzkt        sofar, autotools are a nightmare themselves, what is your complain about cmake ?
22:17 sofar        in our latest distribution, we auto-package 95% of our packages. The remainder is surprisingly often? cmake
22:17 sfan5        auto-packaging with cmake isn't much harder than with autotools
22:18 sofar        yes it is
22:18 sofar        but hey, ignore me who am I doing a professional linux distro that autopackages 95% of their packages which is just anecdotal right?
22:19 sfan5        why is it harder?
22:19 sapier       https://github.com/minetest/minetest/pull/5045 what do you think about this one? Colorize behaves exactly as it's defined. But it's actually not what at least I would expect.
22:19 nrzkt        sofar, strangely FreeBSD packages cmake package easily without complaining :)
22:19 nrzkt        you add a port, add USES=cmake to port, and then... it just works
22:20 nrzkt        sorry linux packaging is shit
22:20 sapier       usually there are good reasons for using either of them ... most time it's "because we already have 90% this way" ;-)
22:20 sofar        sfan5: because every package can ship their own cmake templates making a mess of things (and it's not like they are attempting to avoid it... cmake rules are a disaster of undiscoverability and inconsistency)
22:20 sofar        mind you, minetest doesn't use cmake so badly
22:21 sofar        but the amount of cmake abuse it too damn high
22:21 nrzkt        in fact you can do shit with either cmake or autotools
22:21 sofar        we've spent almost a month with 2 people on hhvm
22:21 sofar        facebook decided to make a better php
22:21 sofar        they dropped autotools and went to cmake
22:21 * sofar      sounds of death
22:22 sfan5        huge projects pretty much always take long to configure correctly
22:22 nrzkt        hhvm should die, PHP 7.0 fixed problems pointed by HHVM
22:22 sofar        nrzkt: aye, that too
22:28 sofar        uh, I'm looking at #5000, but I don't see a search box appear in the menus?
22:28 ShadowBot    https://github.com/minetest/minetest/issues/5000 -- Simpler serverlist search by red-001
22:30 sofar        nvm, menu's don't update unless I install
22:36 sapier       nerzuhl is same as nrzkt if I remember correct?
22:36 sofar        yes
22:37 nrzkt        sapier, ofc
22:37 sapier       nrzkt I'd like to keep the reference for the getTextureMod thing but for some reason this breaks the behaviour
22:37 sapier       setTextureMod of course
22:38 sapier       once mod is a reference the stored values for the modificator somehow get messed up
22:38 sapier       to be honest I don't understand why and would be glad if someone could explain it to me ;-)
22:40 est31        sofar: we ship our own discover templates as well
22:42 est31        they are needed because distros ship broken templates themselves and minetest is supposed to work cross distro
23:18 sapier1      does anyone have knowledge about coloring? :-) more in detail about color subtractions and additions ;-)
23:40 sofar        sapier1: sure, yes
23:54 sapier1      MYMIN(256, dst_c.getRed() -   (255 - color.getRed())),                                                      +				MYMIN(256, dst_c.getGreen() - (255 - color.getGreen())),                                                      +				MYMIN(256, dst_c.getBlue() -  (255 - color.getBlue()))
23:54 sapier1      ups
23:55 sapier1      https://gist.github.com/sapier/a8ce7b49ef3529545fe96624b31cbd6f
23:55 sapier1      can someone explain to me why this results in coloring as you'd expect on painting to a white wall?
23:59 sapier1      255 - color is difference from max so the result is the actuel color if it's subtracted from a maxed out (white) image