Time  Nick         Message
02:02 ShadowNinja  kahrl: Hehe.  :-P  That's pretty ridiculous.
02:03 ShadowNinja  If you can't read past the first sentence you shouldn't be let near ANY code.
03:01 ShadowNinja  celeron55: Can you rebase set_sky/override_day_night_ratio?
03:02 VanessaE     I can't really read that code, was my concern ever addressed about there being one "default, no mod needed" skybox?
04:20 RealBadAngel if merging that code, there shall be added textureable sun/moon to make it complete
04:28 VanessaE     YES
04:29 VanessaE     bbl
06:27 sfan5        ShadowNinja: good use? I couldn't think of any good use for getting the version of the server.
14:41 ShadowNinja  sfan5: For display.  You might be able to move the status string generation to Lua for example.  Or the CTCP VERSION thing, or a menu...
19:05 xyz          sapier: apparently serialization of functions doesn't work in luajit 2.0.0_beta7
19:05 xyz          sapier: is it supposed to work with luajit at all?
19:11 elliot_      Evening all :)
19:16 ShadowNinja  Hello elliot_!  This channel is strictly on-topic, #minetest is our social channel.
19:16 * specing    changes topic to random chatter
19:18 elliot_      ShadowNinja: Okay :)
19:18 elliot_      I was just going to ask what made you guys choose Irrlicht over writing your own opengl rendering code?
19:19 sapier       good question xyz I thought everyone except me uses luajit on regular base. I use 2.0.1
19:19 ShadowNinja  You'd have to ask c55, although he regrets that decision now.  Irrlight does include some batteries to make things easier though.
19:21 elliot_      To me, voxel rendering ought to be it's own thing
19:22 elliot_      I haven't yet looked at the code, but I'm imagining a horrible for loop creating many cube objects...
19:23 sapier       I suggest looking at code and make suggestions how to improve we know this code is suboptimal but by now noone really tried to improve it
19:23 ShadowNinja  The rendering code isn't known for it's speed...  But you're welcome to improve it.
19:24 xyz          sapier: so what will we do?
19:24 xyz          I don't think serializing luajit function is a defined behavior
19:24 ShadowNinja  However Minetest currently works with DirectX, you'd have to break that if you use OpenGL.
19:24 ShadowNinja  (Or add a compat layer)
19:25 elliot_      I thought irrlicht is graphics API independent anyway?
19:25 sapier       I don't se a reason why it shoudln't work, serialization doesn't use anything non standard compliant
19:25 ShadowNinja  And OGL ES is needed for Android support.
19:25 sapier       what exactly happens?
19:25 ShadowNinja  elliot_: Yep.
19:25 xyz          elliot_: irrlicht is good enough and there aren't many devs anyway
19:26 xyz          sapier: the deserialized function returns nil
19:27 xyz          http://lua-users.org/lists/lua-l/2011-01/msg00255.html could be related
19:27 sapier       what did you try to serialize?
19:27 sapier       serializing will fail for anything containing minetest.*
19:28 xyz          it happens in server list
19:28 xyz          because you serialize functions here
19:28 xyz          I don't know why you do that though
19:28 sapier       ok so it's something quite simple
19:28 sapier       because the server list needs to be fetched async in order to not stall the menu
19:29 sapier       as everything doing network operation
19:30 xyz          yes, so what are we going to do about it?
19:30 sapier       try to find out whats really wrong and fix it ... what else?
19:31 sapier       shouldn't be that hard to find out why marshal returns nil on luajit
19:31 xyz          what really wrong is — function serialization doesn't work
19:31 xyz          for example, test.lua from marshal distribution fails
19:32 xyz          the message I linked to says "The attempts to marshal and unmarshal functions of course fail on LuaJIT 2" so I thought this behavior is not defined
19:32 sapier       your bug report is almost as bad as "it doesn't work" ... I can't say what's wrong right now too but I we need a way better error report than this
19:33 sapier       that's be a real problem as it would stop any attempts to do client side lua prior even starting
19:33 xyz          ugh
19:33 xyz          what else can I say?
19:33 xyz          function serialization doesn't work
19:33 sapier       I hope you're wrong
19:34 xyz          and what you expect me to say?
19:34 ShadowNinja  Why not just use a string and then loadstring() it?  You can use [[ ]] multi line quotes.
19:34 xyz          bug report: there's a bug on line XXX in file ololo.cpp
19:34 xyz          that's not how bug reports usually work
19:35 ShadowNinja  That will mess up syntax highlighters, but it seems fine otherwise.
19:35 sapier       might be an option ShadowNinja but very very ugly ... I'd prefere to find a solution to make serialization work
19:35 xyz          well we can just say we don't support luajit versions lower than 2.0.2 and hope that this behavior is supposed to work
19:35 ShadowNinja  And the Async API isn't usable by mods yet, so there's no need to work about compatability.
19:35 xyz          since it works now
19:36 xyz          with some versions of luajit
19:36 sapier       xyz you're not a plain user ;-)
19:36 ShadowNinja  Seems a lot simpler than marshal too.
19:36 sapier       it works with 2.0.2?
19:36 xyz          guess what!
19:37 xyz          it does!
19:37 xyz          also it's very strange that you don't use LuaJIT since you're the author of the most cpu-hungry mod
19:37 sapier       oh, ok I didn't realize this before. Is there a way to check the luajit verison on startup?
19:37 ShadowNinja  sapier: Not all that ugly.  Creating the function and serializing and deserializing it is a lot slower, depends on a external lib, and doesn't always work.
19:38 sapier       well ShadowNinja if it's as simple as requireing 2.0.x (x yet to find out) I don't want to use any ugly hack
19:39 xyz          we don't know, maybe we should ask people who do
19:39 xyz          is that even supposed to work
19:39 sapier       no shadow as we do only serialize/deserialize once on client connect ... network transmission is way slower
19:39 ShadowNinja  sapier: But it's not a ugly hack.
19:39 sapier       it is as you always have to escape your code which is quite a pain in the ass
19:40 ShadowNinja  Only if you use [[ and ]] in your code.
19:40 sapier       you can't use any ide support to find syntax errors ... just the first issue
19:40 ShadowNinja  And you can easily add them after writing the function.
19:40 sapier       WHY?
19:41 sapier       if irrlicht is broken and can't display polygons do we switch to asciart for everyone?
19:42 ShadowNinja  What?
19:42 xyz          haha, that's so typical for this channel
19:42 sapier       by now we only know luajit beta4 is broken ;-)
19:42 sapier       is it really an issue if a beta version of an optional lib is broken?
19:42 ShadowNinja  That's a rather obvious fallacy.
19:43 xyz          sapier: well it's not really optional
19:43 ShadowNinja  ^
19:43 sapier       I'm sorry but imho we can't fix any bad thing in world
19:43 xyz          it's automatically enabled when found and you can't disable it if it's found
19:43 xyz          and yet
19:43 xyz          you're missing the point, read my messages once again please
19:43 sapier       ok then we could check if a working version is found too
19:44 xyz          understanding is not very strong, I see
19:44 xyz          well whatever
19:44 sapier       I don't use luajit because it's useless to debug anything with luajit enabled
19:45 sapier       maybe I don't understand because of two different converstions at same time ;-) ok back what's the point I missed?
19:46 sapier       you can't disable it? who the hell changed this?
19:47 sapier       how is debugging supposed to work? ... wait I forgot minetest programmers don't use debuggers :-)
19:47 xyz          no, you didn't get it
19:47 xyz          how about this
19:47 xyz          you spend 5 minutes reading my messages and if you still can't understand it then I'll explain
19:48 ShadowNinja  sapier: I fixed the C++ exception issue.
19:48 sapier       I already said it doesn't work, you haven't been interested
19:49 sapier       or did you do additional fixes recently?
19:49 ShadowNinja  sapier: My patch?
19:49 sapier       the one prior I told you assert() doesn't work and causes follow up exceptions hiding real issues?
19:50 ShadowNinja  Because C++ exceptions in Lua -> C++ calls show up for me now.
19:50 sapier       I'll try again later, but I don't have much hope It's going to work
19:51 ShadowNinja  I don't know what you mean by that assert() sentence.
19:51 sapier       just do assert(flase) in lua
19:51 sapier       false
19:51 ShadowNinja  My patch didn't add or remove assert()s, Lua or C++ side.
19:52 sapier       ok xyz you thin serializing functions isn't defined ? it's as less defined as serializing in general.
19:52 sapier       I didn't say you added it, assert(false) is a lua function call
19:52 ShadowNinja  Yes, and?
19:52 xyz          sapier: yes, I think it's possible that it's not defined for luajit for whatever reason, it's better to ask someone familiar with stuff
19:52 sapier       if you use it in lua you actually expect to stop lua code immediately and see that state
19:53 sapier       but that doesn't work
19:54 ShadowNinja  Sounds like a Lua bug that I haven't experienced.  :-)
19:54 sapier       if it doesn work for luajit 2.0.1 2.0.2 and all other "stable" versions why do you assume it's not "supported"  at all instead of the way more likely explanation "bug in 2.0.0 >>beta4<<". Do you have any other information that cause you thinking this?
19:56 sapier       if it wasn't supported at all I'd expect it to fail in luajit beyond any fixability
19:56 ShadowNinja  > assert(false) print("test") -> stdin:1: assertion failed! stack traceback: ...
19:57 ShadowNinja  Same with non-JIT Lua.
19:57 sapier       it failed for me in on step callback I'll check later. I'd be glad to be able to use luajit
19:57 ShadowNinja  No "test" printed.
19:58 xyz          sapier: read my messages once again and you'll find the truth
19:58 sapier       and obviously I don't check every day if it works now
19:58 xyz          sapier: why is this always happening when I chat to you? it seems to me as you're ignoring at least half of what I'm saying
19:58 sapier       I'm sorry xyz, I don't know why I always missread your messages
19:59 sapier       do you know anyone we could ask about it? maybe the one who wrote the marshalling code?
19:59 xyz          nope
19:59 xyz          maybe we should open an issue as a question or something
20:00 sapier       yes we could document which versions work in there too
20:01 sapier       if it's really beta versions only I'd suggest fixing cmake to not support those versions. It's way more easy then any other solution. At least on debian that's not an issue as there ain't even a luajit package in repos. Does anyone know about other distributions?
20:02 Calinou      no luajit package in debian? I have one on Ubuntu, so I would be surprised
20:02 Calinou      do you mean in debian stable?
20:03 sapier       yes
20:03 sapier       another reason why I don't have luajit installed
20:03 sapier       hmm ok so ubuntu has?
20:04 sapier       what version is ubuntu luajit calinou?
20:11 Calinou      Version: 2.0.2+dfsg-1
20:11 Calinou      in saucy
20:11 Calinou      (libluajit-5.1-dev)
20:12 Calinou      raring is dead, quantal will be dead soon
20:12 Calinou      also look at http://packages.debian.org/search?keywords=libluajit
20:12 sapier       ok so I guess ubuntu based distributions wont be an issue
20:12 Calinou      and http://packages.ubuntu.com/search?keywords=libluajit
20:12 Calinou      Debian testing looks fine too
20:12 Calinou      and unstable
20:12 sapier       jessie ... next version
20:13 sapier       ok debian based is not an issue to as first you don't have to compile has a valid version and if you compile yourself you can choose a valid version
20:13 sapier       what about redhat suse fedora and bsd?
20:17 sapier       proller you have bsd what's bsd's luajit version?
20:17 xyz          you're fun
20:18 sapier       for what reason? what did I miss this time?
20:20 sapier       xyz did you have time to look for other issues then the concurrent connect limit?
20:21 xyz          nope, not interested
20:21 sapier       anyone else?
20:22 Exio4        why is a concurrent connect limit an issue?
20:22 Exio4        some high value for avoiding random DoS seems a nice thing
20:23 ShadowNinja  8 should be plenty high.
20:23 sapier       xyz and proller don't like it and vanessae doesn't care about it ... without anyone who want's it I will not spend time on persuading people to add it
20:24 ShadowNinja  And maybe the default max_users should be bumped to 32 now.
20:24 ShadowNinja  (Or something around there)
20:24 Exio4        you like base2 numbers, don't you? ;
20:24 Exio4        ;P*
20:24 sapier       like any programmer ;-)
20:25 Exio4        true true
20:25 sapier       maybe not lua and java programmers :)
20:25 ShadowNinja  Yep. :-)
20:25 Exio4        probably
20:25 sapier       or perl ... guess the vm programmers don't care about 2
20:32 sapier       xyz I may have missed it but for what reason did you choose 2.0.0beta4?
20:32 sapier       pure chance or is there something wrong with later versions?
20:32 xyz          sapier: I didn't, also it's beta7, not 4
20:32 sapier       ok why beta7?
20:33 xyz          one FM user reported this issue
20:33 sapier       I see
20:43 ShadowNinja  Wouldn't max_lag be a better value to announce as lag than the step time?
20:44 sapier       Can someone explain to me what max lag is supposed to be?
20:45 PilzAdam     sapier, AFAIK its the time spend in Lua code
20:47 ShadowNinja  sapier: http://pastebin.ubuntu.com/6845876/
20:48 ShadowNinja  Basically dtime that decreases only slowly.
20:49 sapier1      hmm ok guess we need to think about this way of determine lag later as it doesn't honor network lag
20:51 Calinou      sapier, a lot of desktop users use Jessie already, not necessarily a bad choice.
20:51 Calinou      some server users do too.
20:51 Calinou      but not much
20:52 Calinou      https://admin.fedoraproject.org/pkgdb/acls/name/luajit?_csrf_token=be21ce0f2450c29ae4d69bf0fd47d352f701f382 ?
20:52 Calinou      for fedora
22:11 Sokomine     hmmmm: are you there?
22:12 Sokomine     or, more general map-related question: blocks consisting solely of air. do they take up storage space?
22:17 xyz          yes
22:24 ShadowNinja  ^ But they're zlib compressed, so they should be pretty small.
22:26 xyz          that too
22:39 ShadowNinja  This happens an awfull lot, usually in pairs: collisionMoveSimple: WARNING: maximum step interval exceeded, lost movement details!
22:43 hmmmm        perhaps that has something to do with lagging
22:56 Sokomine     ShadowNinja: i just got afraid because a player sounded as if he intended to map out the entire map :-)
22:57 Sokomine     fortionately, he later explained that his mapping-bot sticks to ground level; that seems far less threatening
22:57 Sokomine     ShadowNinja: with mobs running around?
22:59 NakedFury    https://www.humblebundle.com/weekly?utm_source=Humble+Weekly+Sale+Newsletter&utm_campaign=82107c4317-newsletter_codemasters&utm_medium=email&utm_term=0_a406d06e11-82107c4317-108957521
23:08 sapier       ShadowNinja for some reason your server step is laging way to bad
23:08 sapier       that warning is in collision handling to avoid collision handling killing your server by adding even more lag based uppon huge steps
23:22 sapier       https://github.com/minetest/minetest/pull/1116 final call for review and comments, I squashed and cleaned this one in order to merge it this weekend
23:52 proller      NOOOOOOOOOO