Time  Nick          Message
00:17 VanessaE      ShadowNinja: you never gave me your opinion on the revisions to #959 ...
00:25 proller       i think.. i found...
00:25 proller       IT
00:25 VanessaE      IT?
00:25 proller       IT!!!
00:26 proller       source of server LAG
00:26 VanessaE      wht what?
00:26 VanessaE      spill it dammit
00:26 proller       5 min and i will push
00:28 VanessaE      wait, what IS the problem?
00:28 proller       server process only one packet per step
00:29 proller       every client generate 10 only for move
00:31 VanessaE      so 10 packets come in from a client, but only one of them gets processed?
00:31 VanessaE      are you absolutely sure that you're about to change this for the better?
00:32 VanessaE      I could see this causing a rise in CPU load on the server if the packet flow isn't reduced a bit to compensate.
00:34 proller       usually step faster than 0.1s
00:34 VanessaE      well
00:34 VanessaE      10/sec is the current default
00:34 proller       but if you use lot of mods with liquid_finite - step can be longer
00:34 VanessaE      riiiight
00:34 proller       10 clients generate 100 packets per second
00:35 VanessaE      eek
00:35 proller       if all moving
00:35 VanessaE      I can see that affecting a lot more than just finite liquid servers
00:35 proller       every server with lot of clients
00:55 proller       runned on sky
00:55 VanessaE      ?
00:57 proller       https://github.com/proller/minetest/commit/a1da6a871c35ed71be94367a8797c3faf7d60076
00:58 VanessaE      so you'll process up to 100 packets per server step?
00:58 proller       yes
00:59 kaeza         that could be made configurable
00:59 proller       but usually less
00:59 proller       no need to configure it
00:59 VanessaE      it might be a good idea to make it configurable, yeah
00:59 VanessaE      in case someone has a high-end server that can handle say several thousand packets/step
01:00 proller       who want to test - welcome to "Sky"
01:00 pitriss       i second that ide a to make it configurable
01:02 proller       ok
01:03 VanessaE      it might be prudent to test out the 100 limit on a high-traffic server, in case that much is still too high of a default
01:03 VanessaE      (slow machines, etc)
01:05 proller       now i will try on my odroid
01:08 proller       seems much better
01:09 VanessaE      connect to it from several clients
01:09 VanessaE      try to flood it with connections - simulate a very busy server
01:10 VanessaE      even to the point of *almost* DOS'ing it.
01:16 proller       my opengl broken
01:16 proller       i use software render
01:19 proller       four of me, need more
01:20 VanessaE      four ain't enough :)\
01:20 VanessaE      you need about 20 or 30 of you ;)
01:23 proller       its problem for my notebook
01:23 VanessaE      address?
01:23 VanessaE      I'll flood it ;)
01:24 proller       80.240.216.69:30010
01:24 proller       or "Sky" in servers
01:26 proller       cooool
01:26 VanessaE      there, is that enough? :D
01:27 VanessaE      I have a lot of RAM to burn ;)
01:28 VanessaE      <proller>  STOP!  MY MODEM IS BURNING!!!!!!!
01:28 VanessaE      :D
01:28 kaeza         Mine's still downloading media :<
01:29 proller       100% cpu
01:29 VanessaE      all of mine are still awaiting media also
01:30 proller       blocks not loading
01:30 kaeza         100% CPU is not very... informative
01:30 kaeza         which CPU? OS?
01:30 VanessaE      kaeza: I nailed him with 13 live minetest clients :D
01:31 kaeza         O_O
01:32 proller       i core of corei7 freebsd
01:32 proller       problem with media
01:32 proller       need to join slower
01:32 kaeza         VanessaE, random_input = true?
01:32 VanessaE      nope
01:32 VanessaE      I didn't even think about that
01:33 proller       but game playable
01:33 VanessaE      shall I cancel them all out?
01:33 proller       instant blocks placing
01:33 proller       wait please
01:33 kaeza         that setting wil come in handy here, since it simulates a real live client
01:33 VanessaE      thus far none of them have completed the login process
01:33 VanessaE      (one of them appears to be crashed)
01:34 proller       sending files via udp is.. is....
01:34 VanessaE      is...slow until kahrl finishes his http_fetch work
01:34 VanessaE      which will be soon, I[m sure
01:34 proller       now better
01:35 * VanessaE    kills off the crashed client
01:35 proller       it still playable
01:36 VanessaE      shall I add a few more clients?
01:36 proller       can 10 more
01:36 proller       now 18
01:38 VanessaE      I now have 22 clients open :)
01:38 VanessaE      my CPU is screaming bloody murder at me :)
01:38 proller       but there are already max_users=30 players.
01:39 proller       block loading problem
01:39 proller       chat instant
01:39 VanessaE      shall I kill these off yet? :)
01:40 proller       dig-place instant
01:40 proller       packets  errs idrops      bytes    packets  errs      bytes colls
01:40 proller       677     0     0      58151        645     0     349476     0
01:40 proller       689     0     0      51810        681     0     326439     0
01:40 proller       691     0     0      48230        652     0     304881     0
01:40 VanessaE      and leave-flood :D
01:40 VanessaE      all but one (which seems to be the only one still alive)
01:41 VanessaE      ha!
01:41 VanessaE      I killed all those clients already
01:41 proller       blocks loading ok
01:41 VanessaE      not for me
01:41 VanessaE      blocks are not reliably loading
01:42 proller       need to try with http media
01:43 proller       or slowly one-by-one join
01:49 kaeza         oh also small typo: s/recieved/received/
01:50 proller       and playable on arm!
01:50 proller       kaeza, ok
01:53 proller       finite liquds flow fast on cpu from galaxy s2
01:55 VanessaE      proller: have you done any work with the infinite/regular liquids to make them behave better?
01:56 proller       only "range" removed
01:56 proller       and levels adjustable too
01:56 proller       and small visual fixes
01:56 proller       and its not merged
01:57 VanessaE      well I meant in terms of the water flowing readily without consuming tons of CPU for things like those huge fountains pilzadam and I tested with
01:57 proller       and no sense touch old liquids because they not liquid 8)
01:58 proller       smaller queue and longer liquid starting flow
02:00 VanessaE      note to self:  buy proller a book on learning to write complete sentences ;)
02:00 proller       8(
02:09 VanessaE      aw I was only kidding :)
03:50 proller       VanessaE, switched to remote media 8)
03:50 VanessaE      ok
03:51 proller       and max 100 users
03:51 VanessaE      I'll attack in a few minutes :D
04:03 proller       screen minetest --address 80.240.216.69 --port 30010 --go --name t1
04:04 VanessaE      is THAT how it works??
04:04 VanessaE      ready? :)
04:08 VanessaE      MUAHAHAHAHA!
04:09 kaeza         dis gon b gud
04:10 VanessaE      here come a few more
04:12 VanessaE      proller: do I stop or keep going?
04:12 proller       wait
04:12 * VanessaE    waits.
04:13 VanessaE      lemme bring up the last few that timed out.
04:13 proller       they not fall
04:13 VanessaE      I noticed that.
04:14 proller       strange
04:14 VanessaE      they're all default, plain unmodified client
04:14 VanessaE      (null video, random input turned on)
04:14 proller       maybe still not recieved spawn block
04:14 VanessaE      maybe
04:14 VanessaE      a lot of them are dieing with floating point exceptions (!)
04:14 VanessaE      [66]   Floating point exception(core dumped) minetest --address 80.240.216.69 --port 30010 --go --name test-user-$i --config ~/.minetest/minetest.conf-flooding
04:15 VanessaE      why is your spawn point so high up, anyway?
04:16 proller       its fun
04:16 proller       and anti - grief
04:16 proller       cpu ~50%
04:17 proller       net 100-200 pps
04:17 VanessaE      looks like I can top out at 29 clients
04:17 VanessaE      after that I seem to get lots of float point exceptions
04:18 proller       mem 158356k
04:18 VanessaE      wow
04:19 VanessaE      not bad.
04:19 proller       but need to optimize block-sending
04:19 proller       i thing 100 players is not limit
04:19 proller       think
04:20 proller       after optimize
04:20 VanessaE      yep, I'm maxing out at a clean 29 clients.
04:20 ShadowNinja   Hmmm, better talk to someone else familiar with the net code, but that sounds great!
04:21 VanessaE      oh wow, my CPU's maxed out completely.
04:21 VanessaE      all 6 cores at 100%
04:22 VanessaE      BAM!  leave flood ;)
04:22 proller       player action instant
04:22 VanessaE      here comes the next wave.
04:23 proller       most lag on joins
04:23 VanessaE      yeah
04:23 VanessaE      and now I seem to have botched my password.  mind resetting it?
04:24 proller       on my server?
04:24 VanessaE      yeah
04:24 proller       need to restart
04:24 VanessaE      ok
04:25 proller       i think enough
04:25 VanessaE      ok
04:46 VanessaE      80.240.216.69:30010   1/100, 45
04:46 VanessaE      haha
04:54 VanessaE      I'm off to bed.  good luck, proller
05:40 celeron55     why don't you just put it on VanessaE's server
05:40 celeron55     real-life testing tends to give useful results
05:41 celeron55     and you didn't even test it without the patch
05:44 celeron55     now that i look at it, that fix does exactly nothing
05:45 celeron55     if you look at what AsyncRunStep() does, it doesn't do anything unless enough time has passed since it was last called
05:51 celeron55     the thing wouldn't work even in singleplayer if it wasn't so
05:56 ShadowNinja   VanessaE: #959 looks better, although it needs some tweaks(shorter lines, foolproofing for unknown nodes, and a few others). I still would like infinitestacks to be detected in the function, but I don't know how that should be done.
05:59 ShadowNinja   I notice there are three nick-quiets and a nick-ban in here, which are completely ineffective. <-- celeron55?
06:01 ShadowNinja   Night for me too, bye.
07:22 nore          any thoughts on #976?
07:47 proller       celeron55, my server is real and runned without patch more than half year
08:04 proller       and it works even in singleplayer
08:41 celeron55     obviously it works because it changes nothing
08:42 proller       grep lag debug.txt | wc -l
08:42 proller        208
08:43 proller       it solves problem
08:43 celeron55     no it doesn't
08:43 celeron55     look at the god damn code
08:43 celeron55     i'm not going to even argue with this
08:43 proller       okay
08:44 nore          celeron55, about the ABMs in Lua
08:44 nore          I made speed tests
08:44 nore          dirt and dirt_with_grass ABMs take each less that 0.25ms to execute for all nodes
08:45 nore          and saplings take 150ms for each sapling that grows whereas it was 130ms before
08:45 celeron55     proller: it merely prioritizes incoming packets over server processing and outgoing packets resulting from the processing
08:45 celeron55     which lessens lag in certain things and increases lag in other things
08:46 proller       celeron55, try to play in your game, try to join servers, not in your imaginary spacecraft simulator, in minetest
08:47 celeron55     nore: i saw that; i have no comments
08:49 celeron55     proller: make a clone of your server and run the other one without the patch and the other one with it
08:49 nore          so you reckon it is fast enough?
08:50 celeron55     nore: i think there is no real alternative anyway
08:51 celeron55     (but it still remains a problem that there is duplicated tree-growing code)
08:52 nore          about duplicated tree-growing code, trees could be changed to be made in Lua too
08:52 nore          added by an on_generated on something like that
08:52 nore          but I reckon it wouldn't be very efficient, though
08:52 celeron55     they should be changed to use the decoration functionality of the mapgen
08:52 celeron55     but is it ready yet?
08:52 proller       celeron55, you can reproduce it in single player, liquid_finite=1 + 2 mobs mods, or just slow cpu (arm)  - fast move to new area, debug packets - and you will see server process client moves after 1-2...10 ...  seconds
08:53 nore          yes, but decorations don't produce good trees
08:53 nore          if changed, there should be a callback for decoration placement
08:53 celeron55     have you tried pilztest?
08:54 nore          celeron55, I have, and I find the trees look all the same
08:54 nore          they all have the same height, IIRC, and the same leaves pattern
08:54 celeron55     but it's not worse than the current trees
08:55 nore          the current trees are all different...
08:55 proller       maybe use 5-10 predefined schemes?
08:55 nore          perhaps the best to do would be to wait for a mapgen gecoration placement callback, and then add the trees
08:55 celeron55     uhm... well, there definitely should be some way to make good trees with the mapgen
08:56 celeron55     without speed decrease
08:56 nore          proller, yes, but that does not fix the issue of game-specific trees...
08:57 celeron55     what if some optionally enableable "tree-mangling" functionality was added to schematics placed as decorations
08:57 proller       make tree api with lot of growing variants? random, scheme, ...
08:57 nore          if a game wants different growing trees, how can we change the mapgen *for that game only* ?
08:57 celeron55     it would use some heuristics to make variation to the schematic
08:58 proller       like minetest.define_tree({...params...})
08:58 proller       register_tree
08:58 celeron55     proller: the L-system thing exists for that purpose, really
08:58 celeron55     maybe it should be just fixed
08:58 celeron55     (fixed for speed)
08:59 celeron55     there are rumours that it's slow because of how it does lighting but i haven't looked into it
09:00 nore          celeron55, the mapgen would have to be able to support trees registered in Lua
09:00 nore          and the tree API will never be enough to support all kind of trees
09:01 proller       l-tree now place only one tree, i'm about like-ore generator, just register your trees and place conditions, and it will spawn them automaticaly
09:01 nore          so a tree growing function should be called each time a tree grows...
09:02 nore          does anyone know how costly is a Lua call?
09:03 nore          I mean, the time to make the call
09:04 proller       depend on params
09:06 nore          vm data, a voxelArea, a pos, a bool, an int
09:07 nore          what would that cost?
09:08 celeron55     the mapgen uses a vmanip in itself, i guess hmmmm has some optimization in there to not require copying it?
09:09 nore          dunno, perhaps tree code should be made one when this is done...
09:10 proller       one for tree or 100 per node?
09:23 nore          what I said while I had network problems:
09:23 nore          <nore> ok, perhaps it would be better to have a function that is called for all trees at once
09:23 nore          <nore> but anyway, the tree code is far too specific
09:23 nore          <nore> they may only grow on grass and dirt on mapgen
09:23 nore          <nore> so a game that wants to put trees in other places has to make its own tree generation
09:24 nore          celeron55, ^
09:25 celeron55     the decoration placement has been made to provide that kind of placement
09:25 nore          perhaps I could try to put treegen in default...
09:25 nore          yes, but decorations do not provide callbacks when placing
09:25 nore          so one could place saplings with that, but no trees...
09:25 celeron55     let's add a callback-type decoration? 8)
09:26 celeron55     hmmmm might hate it but if it gets stuff done, it might be good
09:26 nore          I guess hmmmm should do that...
09:26 nore          why should he hate it?
09:26 sapier        is it only my feeling or are current discussions more suitable to a merge window then a prerelease?
09:26 celeron55     speed problems are the only probable issue i guess
09:27 nore          celeron55, yes
09:27 nore          the idea would be to make all callbacks at once, from Lua and not C
09:27 celeron55     sapier: that's true
09:27 nore          and only one callback from C
09:27 nore          but that would still be slow
09:28 celeron55     how about leaving this tree and grass stuff to after 0.4.8?
09:28 sapier        while I'm absolutely positive about improving tree generation I don't think this is something to be called a blocker for 0.4.8
09:28 nore          well, I would like the ABMs to be removed from core before 0.4.8, or at least be disableable...
09:28 celeron55     and focusing on just adding all necessary fixes and releasing it, and then continuing from this after it
09:29 nore          but the mapgen things could wait a bit
09:29 celeron55     at least proller and sapier have patches that are necessary to make things work
09:30 nore          celeron55, perhaps add a setting in minetest.conf to disable core tree growing and core dirt/dirt_with_grass conversion
09:30 nore          so games could change it if they want
09:30 nore          and leave the other things to after 0.4.8
09:30 celeron55     nore: if you add a comment in the duplicate tree growing code that it's duplicate and will be replaced with something in the future, i guess the patch is fine
09:30 sapier        and we have multiple ways to fix the modstore stalling right now we need to discuss and decide what's best to do ... both aren't "small" fixes
09:30 nore          in the core one?
09:31 celeron55     nore: i was thinking of the minetest_game one, but maybe explain it in bot
09:31 celeron55     +h
09:31 proller       after my night patch speed of abm not very important
09:31 nore          the one that should be removed is the core one, isn't it?
09:31 nore          anyway, I will add that in both
09:31 celeron55     i don't know
09:31 celeron55     really both will be removed eventually
09:33 nore          WARNING: this code is duplicate and will get removed in the future. <-- I added that
09:33 nore          is it ok?
09:35 celeron55     more like "NOTE: Tree-placing code is currently duplicated in the engine and in games that have saplings; both are deprecated but not replaced yet."
09:36 nore          both deprecated?
09:37 celeron55     s/Tree-placing code/This tree-placing code/g
09:37 celeron55     yes
09:37 nore          how will tree be placed then?
09:38 celeron55     with schematics or the L-system or something; not with that old code
09:38 nore          if it is Lua decoration callback, it will be the same code...
09:38 celeron55     what invokes the tree-placing code isn't relevant
09:53 BlockMen      #928 is ready to merge now
09:53 BlockMen      and here the testing code: http://pastebin.com/YRGHVHk1
09:54 * BlockMen    switch to hexadecimal colors and added kahrl's hexcolor-parser
09:59 nore          celeron55, so, can it be merged?
10:06 celeron55     i think it's good now (in the context of the future plans)
10:09 sapier        https://github.com/minetest/minetest/pull/977 ok here it is I did additional cleanup and documentation so it's ready for review
10:12 nore          sapier, can it be called in_game too?
10:12 sapier        not for 0.4.8
10:13 sapier        it's planed to be added later but I don't want to open up a risky thing like that for upcoming release
10:13 nore          ok...
10:14 sapier        and of course prior adding all of those scriptapi functions need to be checked if they can be called from async ... quite some work
10:15 nore          yes, but I mean can Lua-only things be called?
10:15 nore          i.e., a voxelmanip data, and some parameters to do a lot of work
10:15 sapier        of course those would work at once but I guess there isn't much use for it (except voxelmanip)
10:16 sapier        and you should consider a async call is a costly operation so if you just switch some nodes it may be even more performant to be done synchronous
10:24 celeron55     should we add something to the topic mentioning this semi-feature-freeze until 0.4.8?
10:25 nore          perhaps...
10:25 celeron55     i guess not
10:25 nore          but if we do, everyone will ask when will 0.4.8 be out
10:25 celeron55     it even isn't a semi-feature-freeze because many features are really about to go in
10:25 jin_xi        hey, my two cents to trees and deco discussion: i talked to hmmm about deco callbacks and he said he planned to add them. i wanted to do it myself, but idk how... anyways i think the lsystem treegen should really be done in lua for more flexibility. in my experience speed should not be an issue, as its about the same
10:26 nore          jin_xi, about deco callbacks, I guess this will be done to remove that duplicated  code for default trees
10:26 celeron55     sapier: you should publicize your membership on https://github.com/minetest?tab=members
10:26 nore          and you say you have coded the spawn-tree thing in Lua?
10:27 celeron55     sapier: you not being publicly in there already caused confusion 8)
10:28 jin_xi        hmm, i did a turtle thing that is not tree specific but does all sort of structures, just look at my posts in the forum...
10:29 nore          so perhaps the spawn-tree code should be done in Lua?
10:29 jin_xi        i think 'trees' with both schematics and custom lua on deco callback should be able to do everything anybody could ever want to do
10:30 celeron55     BlockMen requested himself to be a core dev; objections?
10:31 nore          I don't have any
10:33 PilzAdam      celeron55, not that Im against it, but hes not very active in this channel
10:34 celeron55     he has said 906 lines here this year though
10:35 celeron55     it's about the same as ShadowNinja, and 10% of you or me
10:35 nore          celeron55, how do you get the number of lines?
10:35 celeron55     i grep my logs 8)
10:35 nore          so, how many lines I have written this year?
10:36 * celeron55   has always logs available for grepping facts out
10:36 celeron55     1550
10:36 PilzAdam      I liked the form you had a while ago to find the first core devs
10:36 * nore        likes the verb "to grep"
10:36 PilzAdam      it makes sure that all core devs share the same goals
10:37 nore          what was that form?
10:37 BlockMen      PilzAdam, core dev can also be thrown out agan and i would like to know what you make think im having that different goals
10:37 BlockMen      s/dev/devs
10:38 PilzAdam      BlockMen, I dont think you have different goals, I just dont know yours
10:38 celeron55     PilzAdam: related to this, i said:
10:38 celeron55     13:00:20 < celeron55> i'm basing this addition on the hope that if people follow  what i wrote here, it's possible to avoid huge conflicts in  direction and unmergeable work: http://c55.me/blog/?p=1491
10:38 celeron55     anyone is free to disagree of course
10:40 nore          celeron55, there are some suggestions of adding more "official" games...
10:40 nore          should that be done?
10:41 celeron55     i left that open in that post... because i don't know
10:41 nore          I reckon we need several very different games...
10:41 PilzAdam      well, we obviously dont "throw out" core devs who dont agree with our goals
10:41 nore          but I don't know if I am the only one to have that opinion
10:43 celeron55     my opinion is that competing with minecraft in it's own game is boring, which is why we should stretch ourselves a bit because our technology allows doing it with reasonable effort
10:45 celeron55     but there are many ways to organize that and it is the actual problem
10:45 BlockMen      well, since my goals seem to be unclear: my main goal is make minetest more stable(if necessary) and allow as much individuality as possible (with focus on game modes)
10:45 BlockMen      and if you say no, that does not fit, ok. but i hope its more clear now what my focus is on
10:46 celeron55     one question is, where are the people who would make content of publishable quality
10:46 celeron55     we clearly have people who can work on the engine; that's not an issue
10:47 nore          celeron55, do you mean modders?
10:47 celeron55     and... with content, even quality isn't a problem; there just should be... something
10:47 celeron55     some people suggest shipping minetest with no games at all
10:47 nore          we perhaps need more people to work on minetest_game...
10:47 nore          and other games too...
10:47 celeron55     and make it possible to easily get a subgame in it
10:48 BlockMen      a game manager would be nice
10:48 nore          yes, perhaps if a game is good enough, allow to ship it with minetest could be nice
10:48 nore          (and put it in default builds)
10:52 PilzAdam      BlockMen, you seem to think in the same direction as I do: games are the future
10:52 PilzAdam      the problem with mods is that they all depend on a single game (which is minetest_game currently), and there are obviously problems when we try to make one game for everyone
10:52 celeron55     so a thing that would work out technically just fine currently is choosing multiple games to put into minetest packages; the issue is, who chooses them and by what criteria
10:53 PilzAdam      so the Minetest engine should become a generic engine, and allow multiple different games to run
10:55 celeron55     and the other issue is that there currently aren't games that are built to take advantage of this way of doing things
10:55 celeron55     eh, let's start talking about subgames
10:55 BlockMen      celeron55, i would prefer a good way to offer games instead bundle them with packages. because i expect many games to come in future and that makes your question more difficult
10:55 celeron55     (this "game" term is confusing even myself)
10:56 celeron55     BlockMen: that's the future for sure; but what about the close future when we don't have the systems for that set up?
10:57 celeron55     ideally 0.4.8 would have something to show off this new way of doing things
10:57 BlockMen      for the close future:packages and which could be decided on#-dev or by devs
11:03 celeron55     okay so anyway; BlockMen is approved
11:04 nore          I reckon we need something to replace hacky_swap_node code
11:04 nore          because now, each time it is done, the clients thinks meta has disappeared
11:04 nore          and formspec closes, etc
11:04 BlockMen      yay :)
11:05 nore          and hacky_swap_node, is, as its name says, hacky
11:06 celeron55     so does someone want to make a proposal of what to put in the minetest 0.4.8 distribution, or do i make one?
11:07 BlockMen      i think pilztest would be nice
11:07 celeron55     the proposal can really be left as-is from our part and we can just throw it as a recommendation at the face of packagers
11:07 celeron55     (which some of us are, but anyway)
11:09 celeron55     my suggestion-suggestion is: 1) something that has a large amount of stuff to build with, 2) something that has some kind of challenging survival, 3) something that is as much different from those other two as possible
11:10 celeron55     "just fill in the names" 8D
11:12 nore          3) a space game... :)
11:12 nore          2) perhaps realtest, but it would need to be worked on a lot
11:12 nore          1) technic_game...
11:12 celeron55     this is one option for 3 https://github.com/celeron55/dungeon
11:12 BlockMen      1) technic_game + "moreblock", 2) pilztest, 3)like nore said
11:13 celeron55     but would need work
11:13 thexyz        BlockMen: now please register @ freenode
11:15 thexyz        if by 3) you meant something original then I think https://forum.minetest.net/viewtopic.php?id=6636 would fit
11:15 celeron55     could work
11:16 celeron55     if they are small games (like <0.5M), it's not like there couldn't be many of them
11:16 PilzAdam      I still need a better name for pilztest
11:17 thexyz        yeah, OCD is unfinished sadly
11:18 BlockMen      thexyz, done
11:19 BlockMen      PilzAdam, Voxelion?
11:20 celeron55     should this plan be published on the forum? maybe someone gets motivated to quickly finish some fun oddball project
11:20 celeron55     PilzAdam: my suggestion still stands
11:21 PilzAdam      celeron55, I would prefer a "real" name
11:21 celeron55     games these days sometimes have names like that too
11:21 PilzAdam      BlockMen, thats actually pretty nice
11:22 celeron55     that name is infinitely undescriptive
11:22 celeron55     the names of these subgames must make it somehow clear what kind of things they are
11:24 BlockMen      we need something like "about" for games
11:24 BlockMen      were are creator, version and descripts in shown
11:24 BlockMen      s/were/where
11:24 celeron55     probably
11:25 celeron55     but still the name can't be that undescriptive
11:25 BlockMen      but then you cut the creativity
11:26 celeron55     there shouldn't be any reference to voxels or blocks in subgame names unless they are especially focused on voxels and blocks
11:26 celeron55     minetest is about those things in any case so it's no news to anyone
11:26 celeron55     it's like calling a car "Car"
11:27 PilzAdam      pilztest actually is focused on voxels
11:28 celeron55     what if a subgame comes out that's more focused on voxels than pilztest?
11:28 celeron55     then your name falls flat
11:29 sapier        ahh didn't realize I have to do something to become visible as member -->doe
11:30 PilzAdam      yea, the survival part needs to be clear in the name too
11:33 sapier        there already is a game manager focused on modifying different games ... maybe it's not exactly what you meant with "game manager"
11:34 PilzAdam      sapier, its not about modifying games, users shouldnt do that
11:34 PilzAdam      its about installing games, like the modstore
11:35 sapier        if users shouldn't modify games modstore is a little bit useless ;-) maybe we should decide about a consistent way of handling mods/games first (again)
11:38 celeron55     games are constant and they can be amended with mods
11:38 celeron55     it's like it's always been
11:39 sapier        constant as of games you download yes but maybe a user wants to configure different "private games"
11:39 PilzAdam      but do we really need a GUI for that?
11:40 BlockMen      sapier, that does not need a manager, etc
11:40 sapier        in best case user could do this within minetest and doesn't have to copy mods around
11:40 sapier        of course it doesn't "require" those things but modstore isn't required too but yet a big benefit
11:40 PilzAdam      I mean, there could be a 3rd party program for that, but it doesnt need to be in the Minetest main menu
11:41 sapier        hmm maybe I should add a way to modify main menu less invasive than replace it completely e.g. add custom main menu tabs (for 0.4.9 of course)
11:41 BlockMen      but with modstore you dont edit the mods, so with gamestore you dont change the game ;)
11:42 sapier        there isn't a gamestore yet ... but that could be the naming gamestore for downloading and game(manager/editor) for modifying
11:43 sapier        and I don't think we'll get gamestore ready for 0.4.8 unless we want to wait until 2014 ;-)
11:45 celeron55     (that's why we're preselecting a couple of games for 0.4.8)
11:47 BlockMen      another suggestion for 2) wasteland :P
11:51 sapier        https://github.com/minetest/minetest/pull/968 any comment on this one?
11:54 celeron55     pilztest's name could refer also to the limited inventory
11:56 celeron55     blockbag? 8)
12:03 PilzAdam      nore, your broke grass removing ABM
12:03 PilzAdam      sunlight_propagate in Lua has a different meaning than in the engine
12:03 PilzAdam      in the engine its sunlight_propagates || paramtype = "light"
12:03 PilzAdam      currently you remove the dirt_with_grass under default:grass
12:06 sapier        could some (2) coredevs plz agree to 968 ? ;-)
12:08 kahrl         why do we have to define ZLIB_WINAPI at all? shouldn't this be in zlib.h depending on how it's been compiled?
12:08 kahrl         or detected by the build system or something like that?
12:09 sapier        yes but there's no autodetect within the headers
12:09 kahrl         well that sucks a bit
12:09 sapier        you have to define in your project what api to use so if you compile zlib the winapi style use that define if not the other one
12:10 kahrl         is it the case that one is compiling with the stdcall version if and only if one uses msvc?
12:10 kahrl         s/compiling/linking
12:10 sapier        if stdcall means winapi style then yes
12:11 celeron55     nonstdcall 8)
12:11 kahrl         std as in the first definition on this page? :P https://en.wikipedia.org/wiki/STD
12:13 sapier        yea there are a lot more ppl having sex just this moment than are programming right now so this priority may be even correct ;-P
12:14 sapier        actually reading zlib.h it's exactly oposite you only use WINAPI calling convention if msvc compiler is used
12:15 sapier        after my patches. current state is using winapi calling convention once built on windows
12:15 VanessaE      ShadowNinja: line length is easy enough to fix (well, to a degree, too much obsession with it will make the formatting look like crap).  Checking for unknown nodes is already in there (lines 21, 22 as they now exist in the pull).  What "few others"?
12:16 kahrl         I thought nothing would prevent you to use e.g. zlibwapi.dll with mingw
12:16 kahrl         (which requires ZLIB_WINAPI)
12:17 sapier        you can do this of course still it may result in having to add two different zlib dlls as you require to compile all libs to use winapi zlib
12:17 sapier        while zlib dll is already basic component of mingw
12:17 sapier        yet this one isn't built winapi style
12:19 kahrl         we could use the wndproc hack :D
12:19 kahrl         described here http://blogs.msdn.com/b/oldnewthing/archive/2004/01/15/58973.aspx
12:19 sapier        we could ... but why use a complicated solution if there's a small fix? ;-)
12:20 sapier        it's not like you'd have to do anything more if you do a mingw build. you actually need to do less ... not download zlib dll separate ;-)
12:20 kahrl         yeah well I guess if the fix works for everyone currently offering windows builds then commit it
12:21 sapier        I tried building mingw32 and vs2012 ... I expect it to work with vs2010 too
12:23 celeron55     kahrl: i guess they need to change the dll they're using
12:23 celeron55     otherwise they get a million obscure link errors
12:23 celeron55     or something
12:24 celeron55     or maybe it looks to them link it works and the end result works only on computers without mingw installed? 8-)
12:24 celeron55     like*
12:24 sapier        ppl using msvc don't get any difference
12:24 celeron55     with*
12:24 sapier        ppl using mingw would have had a problem atm
12:24 kahrl         unless they fixed it by downloading a separate dll... now they have to fix it again
12:25 sapier        either they don't use gettext/freetype or have multiple dlls in there
12:26 sapier        still it's way more consistent to use non winapi calling convention on mingw
12:28 kahrl         how is this stuff done in irrlicht?
12:29 sapier        good question I'll have a try with non self built irrlicht dlls
12:29 kahrl         oh I think they bundle it
12:29 sapier        I think so too but I'm not sure
12:30 celeron55     it's statically linked into there for sure
12:30 celeron55     like many other libs
12:30 sapier        ok so  I don't need to test
12:30 kahrl         so at least using a different zlib dll won't break irrlicht in some obscure manner
12:32 sapier        as I said actually current way of doing it is broken if someone managed to work around this issue she/he didn't give back that fix by now
12:33 celeron55     the official build instructions put the msvc zlib dll into both builds
12:33 celeron55     so that's what most people have currently
12:33 sapier        did anyone actually do this?
12:33 celeron55     probably almost everyone
12:34 celeron55     (who has built a mingw build)
12:34 sapier        in forum msvc build is way more common than mingw
12:36 sapier        still you can't do a clean mingw build without doing that change
12:36 celeron55     if you're prepared to help people who get problems out of it, it's ok 8)
12:36 sapier        of course I always sort out issues I cause ;-)
12:37 sapier        I guess I should update the "official build instructions" ... which didn't work for me either ;-P
12:38 kahrl         for example PilzAdam's build has zlibwapi.dll and is made with mingw
12:39 sapier        PilzAdam ? how did you do that? ;-)
12:39 celeron55     the mingw instructions worked as-is for me but i've never used gettext or freetype with them
12:42 sapier        official build instructions don't cover freetype and curl at all
12:42 sapier        maybe that's why nobody noticed by now
12:42 VanessaE      can someone look at the current state of #959 and tell me perhaps what those other "things" might have been that shadow was referring to, so I can at least work on them?
12:45 nore          PilzAdam, about that bug with dirt_with_grass
12:45 kahrl         VanessaE: https://github.com/VanessaE/minetest/commit/f8bb9ccfb3c3d2a5e5d4afc0c95b4de1be63ef8c#diff-19b5801567ac89eb6f0459c3c0cf2a47R29
12:46 kahrl         I think it's possible that this indexes a nil value
12:46 VanessaE      look at line 21.
12:46 VanessaE      that catches the unknown node
12:46 sapier        PilzAdam do you have any issues with compiling your mingw build against mingw zlib dll instead of separate dll?
12:46 VanessaE      oh wait.
12:47 nore          should I push the fix?
12:47 kahrl         VanessaE: if the node is unknown that if is executed leading to line 29
12:47 VanessaE      right, I missed that.
12:48 kahrl         34 probably to be made safer too
12:48 kahrl         +has
12:48 VanessaE      yah, on it.
12:48 VanessaE      yeah*
12:51 VanessaE      how's that look?
12:54 kahrl         good
12:54 VanessaE      excellent
12:55 kahrl         now about the "other things" shadow mentioned... you'll have to ask :)
12:55 VanessaE      heh, I just realized I call it a "library" yet it's only one function :P
12:55 kaeza         one small bit: you could cache minetest.registered_nodes[name] locally for a tiny performance improvement
12:56 kaeza         not sure if worth the effort though
13:02 VanessaE      like that? :)
13:02 VanessaE      (yes, I redefine it half-way through the call, but it's just a temp var so I didn't care :P )
13:03 VanessaE      this could be written a tad better actually.
13:04 VanessaE      there, that's better.
13:06 VanessaE      hrm.  could line 34 possibly fail out with the right combination of nil fields?  I can never remember how that works out in practice.
13:06 proller       full set of my patches https://github.com/minetest/minetest/pull/895/files - they do absolutely nothing and useless ;)
13:06 proller       ^ about speed
13:07 nore          VanessaE, 34?
13:08 nore          34 can not fail
13:08 VanessaE      ok good.
13:10 nore          about that space game perhaps shipped with Minetest... who would help doing it?
13:10 sapier        did we discuss usage of "signef off" git feature by some time?
13:11 sapier        I'll provide space mobs
13:11 celeron55     nore: it's not really possible to make a reasonable one at the moment due to a number of reasons
13:11 celeron55     or, it really depends on what can be considered reasonable...
13:11 nore          yes, you may be right...
13:11 proller       we have no darth vader mob 8(
13:12 sapier        we don't?
13:12 VanessaE      somehow I don't think LucasArts will accept tht :)
13:12 sapier        we call it darktest
13:12 VanessaE      (or whatever LucasFilms game division was)
13:14 sapier        yet I'd like to merge the zlib mingw fix can I interpret celeron55's and kahrl's comments as agreement?
13:16 PilzAdam      nore, how about posting the fix here first?
13:17 PilzAdam      sapier, umm..... dunno about that
13:17 sapier        PA does your build contain curl as well as freetype?
13:17 PilzAdam      I just build wit buildbot.sh and copy the .dll files into the bin/ folder when it says that its missing them
13:17 PilzAdam      yea, it does
13:18 sapier        hmm interesting so at least building minetest without any dependencys might work the way it is ... with outdated libs of course
13:18 nore          PilzAdam, too late...
13:18 sapier        so at least I have to check if buildbot needs a fix too
13:18 nore          I pushed it (anyway, the only change was to write nodedef.sunlight_propagetes or nodedef.paramtype == "light"
13:24 proller       VanessaE, today i will make liquid porcessing limited by time, it helps to all liquids
13:24 VanessaE      cool
13:32 celeron55     i just realized the receive priorization patch probably does improve response time in cases when the server is at 100% CPU use
13:33 celeron55     and in those cases the processing would lag in any case
13:34 proller       slow abm and env not big issue
13:34 sapier        100% cpu load? I'm not sure if this is a normal state of operation
13:34 proller       waiting 10 seconds to open inventory - big
13:34 celeron55     sapier: it shouldn't be, but sadly it happens with crappy liquids and crappy ABMs and whatever
13:35 proller       cureently all servers is laggy when 5+ players
13:35 sapier        do we have any option to fix this e.g. by limiting those things to lets say 30% of cpu capacity?
13:35 proller       you can wait minute+ for loading block mor move forward
13:35 sapier        if there was proper map locking those things could run in their own (low prio) thread
13:35 proller       not i will try liquids queue processing linited by time
13:36 proller       now
13:36 proller       with abm it harder
13:36 celeron55     i'd say let's put the priorization in, but it's purpose needs to be explained well in a comment; it's purpose is VERY easy to misunderstand
13:37 proller       ok
13:38 celeron55     what the code does is that it prioritizes processing of received packets over timed processing when there is 100% CPU use
13:38 proller       also lot of falling items can kill server (make sand fload island and touch)
13:39 VanessaE      I can vouch for that one.  and it is wont to kill the client to :)
13:39 celeron55     when there is no 100% CPU use, it does nothing
13:39 VanessaE      too*
13:39 proller       also mapgen can eat 100% cpu
13:39 sapier        sounds like a reasonable workaround until proper rework is done
13:55 PilzAdam      we need to decide if we merge #911 or #882
13:55 PilzAdam      one this needs to be merged before 0.4.8
13:56 PilzAdam      #911 just fixes the current code, while #882 dumps the current code and (tries) to apply that "leveld" stuff to the normal liquids
13:57 PilzAdam      while the additions of #882 to finite_liquids might be nice, it doesnt really work well with the normal liquids
13:58 sapier        in current state "try to get 0.4.8 out" I'd prefere fixing instead of rewrite
13:58 sapier        does 911 work reasonably well?
13:58 PilzAdam      yea
13:59 proller       my pull 3 months old
13:59 sapier        would there be a problem if 911 is in 0.4.8 and 882 is added later?
13:59 proller       maybe some compatibility glitches
13:59 PilzAdam      we would need to alias "range" to "level" then, but thats not a big problem
13:59 proller       maybe not
14:00 proller       nobody will use range 8)
14:00 sapier        was that name in 0.4.7 before?
14:00 PilzAdam      range wasnt used before, no
14:01 celeron55     everything can be changed that wasn't in 0.4.7
14:01 sapier        ok what about using level as name with old code if it's clear we add 882 later?
14:01 celeron55     (as long as loading worlds work)
14:01 proller       making strange feature for one version is... strange
14:01 sapier        -clear + sure
14:01 proller       882 extends "leveled" stuff added after 0.4.7
14:01 PilzAdam      sapier, the problem with "level" is that it doesnt make sense with the normal liquids
14:02 PilzAdam      from the views of modders
14:02 sapier        we could explain it in docs for this version
14:02 PilzAdam      also I dont like dumping range anyway
14:02 VanessaE      crap, I think shadow's change to the 4k+ nodes did not work right
14:02 sapier        ok if we want both there's no compatibility issue
14:03 nore          VanessaE, why?
14:03 PilzAdam      #882 should be rewritten to a) not dump the range feature and b) not break maps like it currently does
14:03 VanessaE      nore, lemme double-check firt.
14:03 VanessaE      first*
14:04 proller       882 is incompatible with range
14:04 proller       it not big break
14:05 VanessaE      nore: maybe it isn't that.  something is wrong.
14:05 sapier        big or not big is no difference ;-) broken is broken
14:05 proller       but if not move flowing flag - make more than 8 level for old liquid impossible
14:05 proller       sapier, small liquid flow is not "broken"
14:06 PilzAdam      also we already decided to add "range" when we merged that pull request, #911 just fixes some stuff
14:06 sapier        I don't have a good fealing adding this to 0.4.8
14:06 VanessaE      nore: lemme check.  helps if I actually build *and install* it...
14:06 proller       PilzAdam, who "we"?
14:06 sapier        it still seems to be a good improvement for 0.4.9
14:06 PilzAdam      proller, the core team
14:06 proller       and then for 0.5, and then....
14:07 proller       PilzAdam, me - not.
14:07 sapier        no proller I'd suggest merging changes stage for 0.4.9 once 0.4.8 is released to get them tested as soon as possible
14:07 celeron55     0.5 doesn't come after 0.4.9
14:07 celeron55     after 0.4.9 comes 0.4.10
14:08 VanessaE      ok it wasn't shadow's change, but something is now preventing me from starting my servers -- C++ Exception
14:08 sapier        we're sorting out issues with things merged months ago and haven't even fixed all of them yet
14:08 proller       sapier, and test new incompatibility..
14:08 PilzAdam      VanessaE, gdb
14:09 sapier        yes but I'm almost sure if we add this now we're add additional delay
14:11 VanessaE      http://pastebin.ubuntu.com/6347162/
14:12 sapier        No symbol table info available. :-(
14:12 VanessaE      yeah, I stopped making debug builds, I got tired of the slowdown.
14:12 PilzAdam      that looks like and assert() failed
14:12 PilzAdam      anything in the normal logs?
14:13 VanessaE      http://pastebin.ubuntu.com/6347174/
14:13 VanessaE      (and no, it isn't actually technic doing it)
14:14 PilzAdam      throw technic out and try again
14:14 VanessaE      no.
14:14 sapier        why don't we se a debug stack?
14:15 VanessaE      PilzAdam: if I remove technic, it'll just be some other mod that triggers it.  When I get "C++ Exception", it is universally a > 4096 nodes issue.
14:15 VanessaE      I'm double-checking against unmodified master just to be sure.
14:17 VanessaE      nope, no good even with vanilla.
14:18 PilzAdam      "C++ exception" also happens alot if you deal with e.g. meta
14:18 sapier        you could try a debug build and run within gdb
14:18 VanessaE      that's stupid.
14:18 PilzAdam      you cant say htat its because of a node limit
14:19 VanessaE      I'll play with it a bit, but "C++ Exception" when it should be throwing a traceable error is just stupid.
14:19 sapier        true
14:21 PilzAdam      sapier, so you would agree to merge #911 and deal with #882 after 0.4.8?
14:21 sapier        yes that'd be my prefered way to handle it
14:21 VanessaE      PilzAdam: good call on the metadata issue.  it's a technic bug all right.
14:22 sapier        do we have a tag "moved to next version" or similar?
14:22 sapier        wait maybe milestones are better?
14:22 PilzAdam      we have a  0.48 milestone
14:22 PilzAdam      +.
14:22 celeron55     VanessaE: it's better to handle exceptions like that in release builds because they're practically never run in a debugger; it at least sometimes gives some information
14:22 celeron55     a debug build will pass the exception through
14:23 celeron55     build a debug build with -O2 if you want
14:23 sapier        any objections to creating 0.4.9 milestone?
14:23 VanessaE      celeron55: but how is a modder supposed to trace such bugs?
14:23 PilzAdam      sapier, what will go in there?
14:24 sapier        pull requests we decided to merge next version
14:24 celeron55     VanessaE: such bug is not supposed to happen and if we knew what it is, we could work towards making them more understandable
14:24 nore          sapier, I don't have objections for that
14:25 celeron55     hmm
14:25 PilzAdam      sapier, well, we havent exactly decided to merge #882 as it is
14:25 celeron55     it seems like it's a C++ exception that has occurred when lua has called something though
14:25 VanessaE      I'm checking on it, celeron55
14:26 celeron55     that's a tough path to pass any information about them i guess, lua being made in C
14:26 sapier        doesn't imply they are finished yet only that their basic features are to be added
14:26 PilzAdam      ok, that sounds good
14:26 sapier        ok I' creating the milestone now
14:28 VanessaE      celeron55: this technic commit throws the exception:  https://github.com/minetest-technic/technic/commit/5ed47abee6354509ca46fefe5740872b028f083e
14:28 PilzAdam      I dont like that proller has added his pull requests to the 0.4.8 milestone
14:28 VanessaE      so the craft prediction function is bugged.
14:28 PilzAdam      they are marked as "low-priority-enhancement"§
14:28 celeron55     it seems like lua would handle such an exception somewhat more reasonably
14:29 celeron55     luajit catches C++ exceptions and hides their what() like that
14:29 proller       PilzAdam, you set this mark?
14:29 PilzAdam      no
14:29 sapier        ok as 0.4.9 exists now we can flagg 882 0.4.9 agreed?
14:30 nore          VanessaE, it works fine with my computer, though
14:30 PilzAdam      I guess 891 goes into 0.4.9, too
14:31 sapier        891?
14:31 PilzAdam      892*
14:31 celeron55     VanessaE: can you build a non-luajit version? it seems like the only way to know what the exceptions is about
14:32 sapier        ok I'm gonna tag 882 0.4.9 if noone complains now
14:33 celeron55     it looks like we need to add exception handler blocks to all lua->c++ calls
14:34 sapier        great :-/
14:34 celeron55     or deal with undecipherable errors
14:34 sapier        I'm gonna file a issue
14:34 proller       and dont crash server on lua error
14:36 celeron55     i'm trying now what regular lua does for C++ exceptions
14:36 sapier        a lua error is a showstopper you can't tell if it's safe to continue
14:37 celeron55     i assume luajit always shows just "C++ exception" and quits
14:37 nore          celeron55, no it doesn't
14:38 nore          when the error is inside Lua, it shows a normal error message, with where it comes from
14:38 sapier        PilzAdam 892 seems to be way to heavy for a prerelease commit too yes
14:38 celeron55     nore: that's not a C++ exception
14:39 celeron55     i guess someone would do this test faster than my laptop compiles minetest
14:39 celeron55     building since many minutes, and building until many minutes more
14:40 celeron55     we should maybe try out precompiled headers
14:40 sapier        any quick way to cause a c++ exception?
14:40 PilzAdam      sapier, minetest.get_meta(something_that_is_not_a_pos) IIRC
14:40 celeron55     http://pastebin.com/gLwy51E2
14:40 celeron55     well, apparently i wouldn't have had to even rebuild this then
14:41 celeron55     i guess i'll just do that thing instead, this is still building...
14:42 sapier        ok I need to rebuilt too but I guess it's faster for me
14:42 celeron55     hey what the actual fuck
14:42 celeron55     why does this rebuild like all of minetest if i just modify l_craft.cpp?
14:42 celeron55     >.cpp
14:42 celeron55     it shouldn't definitely work like that
14:43 sapier        of course not that's one of the reasons why I split scriptapi.cpp
14:43 nore          celeron55, a lot of files depend on it
14:43 sapier        why do files depend on a cpp?
14:43 celeron55     it's a .cpp file, obviously nothing depends on it except the link stage
14:43 celeron55     but it seems like everything depends on it
14:44 sapier        did you modify one of the build scripts?
14:44 celeron55     no
14:44 celeron55     this is vanilla minetest
14:44 VanessaE      C++ exception cause found:  too many braces in the shapeless register craft recipe
14:44 sapier        I'll ty once my build has finished
14:44 VanessaE      shouldn't that have thrown a more readable error?
14:45 PilzAdam      hmm.. that get_meta() thing doesnt seem to "work" anymore...
14:45 sapier        so someone fixed the bug
14:45 sapier        VanessaE it should
14:45 celeron55     VanessaE: give the errorneus code
14:45 sapier        can you pinpoint the c++ file where the exception is thrown?
14:45 celeron55     VanessaE: we want to test it
14:46 nore          minetest.register_craft({type="shapeless", recipe={{"a","b"}}}) crashes
14:46 VanessaE      celeron55:  https://github.com/minetest-technic/technic/commit/5ed47abee6354509ca46fefe5740872b028f083e
14:46 nore          minetest.register_craft({type="shapeless", recipe={"a","b"}}) that does not
14:46 VanessaE      nore, I'll push a fix to technic.
14:46 celeron55     16:46:22: ERROR[main]: ServerError: LuaError: Crafting definition (shapeless) is missing an output
14:46 nore          VanessaE, already done
14:46 celeron55     non-luajit Lua gives this
14:46 celeron55     so it's luajit's problem really
14:46 nore          celeron55, forgot output
14:47 nore          you should add output to that too
14:47 PilzAdam      sapier, I merge #911 then, or do you have any objections?
14:47 sapier        no merge it
14:47 celeron55     nore: what?
14:47 nore          so it is type = "shapeless", output = "c", recipe = {{"a", "b"}} that crashed
14:47 nore          I forgot output when I wrote this
14:48 sapier        still this should cause a more speaking error
14:48 nore          the problem is with the fact that recipe is a list inside a list, instead of a list
14:48 celeron55     anyway, it gives a reasonable error message either way
14:48 nore          even with the output added?
14:48 celeron55     what do i put in output?
14:48 nore          anything
14:48 nore          call it "c"
14:49 sapier        http://pastebin.com/BrJ66uib result without luajit
14:49 celeron55     http://pastebin.com/C2ehkh2S
14:49 sapier        are you sure this is same thing causing vanesaE's error?
14:50 celeron55     with lua, with output
14:50 sapier        ok celeron gets same error
14:50 nore          yes, I am sure it is that
14:50 celeron55     i guess luajit guys say they can't pass a C++ exception through because it would screw up the stack
14:50 nore          if you remove the additional braces, it should work
14:51 VanessaE      ERROR: An unhandled exception occurred: ServerError: LuaError: error: /usr/share/minetest/builtin/vector.lua:5: Invalid vector   <--- unrelated problem, and quite frequent on my servers.  No traceback is printed.
14:52 sapier        didn't I add improved traceback some time ago? are there still situations where this doesn't work?
14:52 VanessaE      apparently so.
14:52 VanessaE      or your code was not pushed.
14:53 sapier        assert(type(v) == "table" and v.x and v.y and v.z, "Invalid vector") thanks whoever wrote that assertion
14:54 sapier        if you look for that bug you could replace it by a if (assert) do backtrace assert
14:55 sapier        if (assert reason) of course
14:55 VanessaE      how am I supposed to find it without a valid traceback?
14:55 sapier        obviously there are some situations where no traceback is created but you can create it manually
14:56 sapier        it's not a solution of course only a workaround until fix is done
14:57 PilzAdam      nore, another thing: do you catch that minetest.get_node_light() can return nil?
14:59 VanessaE      sapier: right.
15:00 VanessaE      sapier: but as I simply don't know how to fix this, I'm stuck.
15:00 nore          PilzAdam, can it?
15:00 BlockMen      can someone agree on #928?
15:00 PilzAdam      nore, similiar code just crashed in pilztest for me, fix: https://github.com/PilzAdam/pilztest/commit/4c188a8ec36981b26f0a8c5723b6196102000a35
15:01 nore          ok, I will push a fix
15:01 sapier        http://pastebin.com/69wYhFtr VanessaE replace the assert in vector by this variant (I use it in mobf)
15:02 nore          PilzAdam, pushed
15:03 PilzAdam      nore, stop pushing things if nobody has looked at them
15:03 nore          that was a 4-chars fix...
15:03 PilzAdam      that line is too long now
15:03 nore          ok, so I should split it?
15:03 PilzAdam      yes
15:04 PilzAdam      you can overwrite the history if the commit you are overwrting isnt older than 10 minutes
15:04 VanessaE      sapier: done,  servers going up.
15:05 nore          PilzAdam, like that?
15:06 PilzAdam      no
15:06 nore          why?
15:07 PilzAdam      https://gist.github.com/PilzAdam/7279851
15:07 PilzAdam      because of our unwritten Lua code styleguideline
15:08 nore          ok
15:08 nore          like now?
15:09 nore          (for the second one, mainly)
15:09 PilzAdam      oh god, its still wrong
15:09 nore          why?
15:10 nore          the second being more indented?
15:10 PilzAdam      can you just stop pushing everything to upstream directly?
15:10 sapier        :-) may I have a look too? I'm curious ;-)
15:10 nore          yes, I should do that...
15:10 PilzAdam      the 2 line in the if statement should have the same intendation
15:10 PilzAdam      s/t/d/ s/d/t/
15:11 nore          yes, but it is inside a parenthesis started in the previous line
15:11 PilzAdam      that doesnt matter
15:12 nore          ok
15:12 nore          so I remove that indent, and I push?
15:12 PilzAdam      yea
15:12 nore          (and I am more careful next time...)
15:12 PilzAdam      and fix minetest_game too
15:13 nore          yes, done
15:13 PilzAdam      BlockMen, havent you removed the "Leftclick: take all ..." stuff in that pull request for a while?
15:14 BlockMen      yeah, but there was no agreement that it gets removed so i left it in when i rebased it
15:14 VanessaE      sapier: if your code works, but is not a "proper" fix, what would be?
15:14 PilzAdam      nore, what a wonderful second day as core dev: commiting something that needs 2 "fix-commits" and pushing wrong things 3 or 4 times so you need a history rewrite
15:15 thexyz        what a usual day, PA being picky
15:15 PilzAdam      :-p
15:15 sapier        a proper fix would be find out why the traceback isn't shown on any lua error I'm not exactly sure if this really can be done as there are multiple ways a error comes to notice
15:16 thexyz        I'm actually interested in what "minetest.get_node_light(above) >= 13 or 0" means
15:17 PilzAdam      its (get_node_light() or 0) means "take get_node_light(), or 0 if get_node_light() is nil)
15:17 PilzAdam      s/)/"
15:17 sapier        so if something isn't loaded yet we asume it's light is 0?
15:17 PilzAdam      the "or 0" part makes sure that you dont compare nil with 13
15:18 thexyz        okay, I guess it's just in the wrong place
15:18 PilzAdam      sapier, it returns nil for solid nodes
15:18 sapier        doesn't it return nil for unloaded too?
15:18 PilzAdam      maybe
15:19 PilzAdam      nore, umm... thexyz is right, you pushed the wrong code to minetest
15:19 sapier        I'm not sure but at least in mobf I assume it :-)
15:19 PilzAdam      its correct in minetest_game, though
15:19 sapier        come on guys don't make him even more nervous he already is ;-)
15:20 PilzAdam      :-p
15:20 PilzAdam      sapier, we all need to go through this
15:20 thexyz        yeah
15:21 PilzAdam      BlockMen, what about my suggestion to remove it, and add it to the default formspec?
15:21 sapier        of course I still have to proove I can handle it better I haven't merged anything yet
15:21 PilzAdam      Ill fix it minetest now (since nore disappeard)
15:21 thexyz        I don't think there's much point in raging because he mistyped "or 0"
15:21 nore          PilzAdam, what was the problem?
15:22 PilzAdam      fixed, people can pull now
15:23 PilzAdam      nore, you wrote "(minetest.get_node_light(above) >= 13 or 0)" instead of "(minetest.get_node_light(above) or 0) >= 13"
15:23 nore          ah, yes, I just saw that
15:23 BlockMen      PilzAdam, so adding a label to default formspec and kick it from engine, right?
15:23 nore          sorry, my bad
15:23 PilzAdam      BlockMen, yea
15:24 BlockMen      ok, im fine with that
15:24 nore          you should add "Middle click: move 10 items" too
15:24 nore          since that one is the one I would never have discovered alone
15:24 nore          (and for crafting too...)
15:25 PilzAdam      nore, that wouldnt fit into the default sized inventory with the default window size
15:25 proller       sapier, 892 fixes broken current weather
15:25 VanessaE      movew/craft 10 items
15:25 VanessaE      -w
15:26 nore          PilzAdam, perhaps something like: Left/middle/right click: move all/10/1 items
15:26 nore          that would fit
15:26 PilzAdam      could work
15:26 sapier        proller looks more like a rewrite than a fix ;-)
15:27 nore          and perhaps even move/craft instead of move
15:27 proller       its continue developing
15:27 proller       weather was broken after this pull
15:27 VanessaE      nore, that would be too many slashes.
15:27 VanessaE      (i.e. it would seem too repetitive)
15:27 proller       i cant support 90000 pulls with one thing
15:27 nore          ok, write move|craft then...
15:28 VanessaE      "move or craft"
15:28 nore          ok for move or craft
15:28 sapier        I'm not sure about it maybe delay decision until critical things are merged
15:28 VanessaE      "Left/middle/right-click:  move or craft all/10/1 item(s)"
15:28 proller       and i want to join 882 + 892 into one
15:29 sapier        then it's 0.4.9
15:29 nore          the only problem is that left crafts only 1...
15:29 PilzAdam      VanessaE, so leftclick is "craft all items"?
15:29 VanessaE      PilzAdam: ask nore :)
15:29 * BlockMen    still thinks its useless at all
15:29 sapier        I don't se any reason why weather and liquids need to be a single commit
15:30 VanessaE      I was just tweaking his line a bit, didn't pay attention to whether it was actually correct :D
15:30 nore          so perhaps move all/10/1 or craft 1/10/1 items
15:30 PilzAdam      it would be ok to just leave middleclick out (as it is currently), since peopel can always find the complete set of controls in the wiki
15:31 proller       sapier, because they have intersection and i cant make something new 3 months
15:31 VanessaE      "Left-click:  move all/craft 1.  Middle:  move/craft 10.  Right:  Move/craft 1."
15:31 VanessaE      PilzAdam: people don't read wikis
15:31 PilzAdam      I did
15:31 BlockMen      who came up with craft? it says move, that enough
15:31 sapier        proller still having multiple things that all need discussion in a single pull will reduce it's chance to be merged drasticaly
15:31 VanessaE      you're a geek.  geeks aren't people ;)
15:31 nore          the middle click should definitely be there...
15:31 nore          else, you don't have the idea
15:32 PilzAdam      VanessaE, ok, I decide that Minetest is for geeks then
15:32 proller       sapier, its not my problem, it already works on my server
15:32 VanessaE      BlockMen: actually I guess that was my suggestion - a lot of people don't know that you can craft 10x with a middle-click
15:32 VanessaE      PilzAdam: :P
15:33 sapier        proller so you really wanna fork?
15:33 proller       i make all what i can for merging, but.. nothing..
15:33 BlockMen      VanessaE, you should ppl try out a bit, not write everything down ;)
15:33 sapier        proller that's quite normal my security fixes have never been merged
15:33 proller       just one huge pull
15:33 sapier        async api is 3 version now
15:33 PilzAdam      proller, the problem that I see with your features is that it doesnt really fit the goals of the other devs
15:33 sapier        and entity linking I provided 2 years ago was rewritten by someone a year ago ;-)
15:34 VanessaE      BlockMen: well if you think it's fine, I won't stand in your way :)
15:34 PilzAdam      we mainly want to create a generic game engine, while you continue hardcoding stuff in the engine and you dont care much about the API
15:34 nore          BlockMen, at least write something about middle-click
15:34 proller       PilzAdam, making space shooter?
15:34 sapier        and a "ALL OR NOTHING" pull request isn't very polite too
15:35 proller       PilzAdam, i hardcode only slowest parts
15:35 proller       and change already existed things in core
15:35 PilzAdam      yea, thats not good
15:35 proller       and all my things disableable
15:36 proller       and disabled by default
15:36 sapier        that's why we're even talking about merging it proller ;-)
15:36 PilzAdam      that doesnt make them good, and as we see with finite_liquids you make it conflict with the normal liquids
15:36 proller       wat?
15:36 proller       conflict make you with "range"
15:37 proller       other was two way compatible
15:38 sapier        couldn't we fix that issue by using an alias? so this'd be a minor issue?
15:39 PilzAdam      I suggested a while ago that "finite" would be a flag in nodedef
15:39 proller       its one place in code
15:39 BlockMen      nore, but why? there are mouses without middle button, e.g touchpads of notebooks
15:39 BlockMen      so writing it always can be confusing too
15:39 proller       client place, he dont know about finite or not
15:40 proller       PilzAdam, no flag!
15:40 nore          BlockMen, left+right click = middle click
15:40 BlockMen      PilzAdam, removed from engine. fine to merge now?
15:41 nore          and since almost no games use middle button, if not written, no-one thinks to it
15:41 proller       flag = 2 queues and lot of shit in code, and nobody will use it
15:42 BlockMen      nore, oh right
15:42 PilzAdam      BlockMen, have you added it to the default formspec?
15:42 BlockMen      not yet because of the discussion and because its _game, so not the same commit
15:43 PilzAdam      the default formspec is in the engine
15:43 BlockMen      wut? where?
15:43 PilzAdam      player.cpp:67
15:44 PilzAdam      in minetest_game we need to add it to creative, chest, furnace and bones
15:45 BlockMen      ah, ok. would ' "LMB: Move all items, MMB: Move 10 items, RMB: Move single item" ' be fine?
15:47 PilzAdam      I guess so
15:48 sapier        proller how does this match to the recent abm changes?
15:48 proller       neet to rewrite about dirt
15:48 proller       need
15:48 proller       but change small
15:49 proller       and add dirt_with_snow куьщмук
15:49 proller       remover
15:49 sapier        what about this the weather things seem to be usefull for 0.4.8 as weather was broken after 0.4.7 but the liquid things have to wait till 0.4.9 PilzAdam what do you think about this?
15:49 VanessaE      куьщмук  ?
15:49 VanessaE      "English, dude.  English."
15:50 thexyz        VanessaE: he've already corrected himself
15:50 thexyz        "remover"
15:50 VanessaE      sapier: weather depends on finite liquids wrking *just* right.
15:50 thexyz        s/'ve/'s/
15:50 VanessaE      thexyz: oops, yep.  ninja'd
15:50 sapier        depends? why?
15:51 proller       weather was added after 0.4.7
15:51 VanessaE      sapier: because with weather comes the ability for it to rain, and the rain must be able to collect on the ground.  with infinite liquids, this can lead to flooding when it shouldn't.
15:51 VanessaE      that's part of why proller seems so obsessed with finite liquids :)
15:51 proller       VanessaE, but finite liquids now works in master
15:52 proller       my pull about make them better
15:52 VanessaE      proller: right.
15:52 sapier        ok so actually weather fixes could be merged without causing flooding ... true?
15:53 VanessaE      as long as the mod that actually makes it rain is not preloaded, probably.
15:53 proller       sapier, yes
15:54 sapier        is vanessae right too?
15:54 PilzAdam      sapier, dunno if thats good, but we definitely need hmmmm to review proller's pull request, since he is the one who "broke" it according to proller
15:55 proller       no problem with rain+current finite liquids
15:55 sapier        of course that's some code to check
15:56 sapier        so proller expect (minor) changes
15:56 PilzAdam      it would also be possible mark weather as "indev" and recommend to no use it
15:56 sapier        thats a minor change in menu
15:57 sapier        first change to do proller ;-)
15:58 PilzAdam      BlockMen, have you tested it well enough?
15:59 BlockMen      i tested any case i could think of so i would say yes
16:02 BlockMen      my suggestion for the formspec help text is too long
16:02 BlockMen      it does not fit
16:04 sapier        hmm that most likely means it's way to long in other translations
16:04 PilzAdam      sapier, we cant translate it anymore if its done in Lua
16:05 sapier        didn't we add translation support to lua?
16:05 PilzAdam      for the mainmenu only
16:05 * BlockMen    still want to kick at all
16:05 BlockMen      *that string
16:05 sapier        I assume adding it to scriptapi wouldn't be a big deal ;-)
16:06 PilzAdam      sapier, the Lua code runs on the server
16:06 sapier        oh ... ok so we'd have to make client tell it's localization to server ... guess thats a 0.4.9 thing
16:09 BlockMen      and another problem..the label breaks the text much earlier, so even the old string does not fit anymore
16:15 BlockMen      how about that: we remove the string and consider to re-add if ppl complain about?
16:16 sapier        I'd be fine with it
16:16 PilzAdam      how are newbies supposed to complain about it when they dont know that it existed?
16:16 nore          ok with it too
16:16 sapier        come one PA we don't have a hud telling punch right place left
16:17 PilzAdam      sapier, we have, press Esc
16:17 sapier        so what about a "?" button topright showing detailed help?
16:20 BlockMen      can buttons use tooltips?
16:20 BlockMen      or labels?
16:24 BlockMen      and PilzAdam, i think that ppl can find out how to use inventory. or do you think they are not as skilled as mc players?
16:26 VanessaE      BlockMen: buttons CAN use tooltips, yes
16:26 VanessaE      Unified Inventory does this
16:28 BlockMen      VanessaE, thx. just looked up, item_image_buttons can. maybe that would be an easy way to show info
16:29 VanessaE      a tooltip over a "[?]" button?  sure
16:29 VanessaE      I could see that.
16:29 BlockMen      damn, no. it needs an registered item (image)
16:30 BlockMen      that means changing button element then
16:30 BlockMen      yay
16:30 VanessaE      so include an image of a question mark :P
16:34 PilzAdam      of course 3 core devs who want to remove the string can just overvote me
16:34 BlockMen      well, i have to leave now so we talk about that next time
16:34 BlockMen      bye everyone
16:37 thexyz        celeron55: any updates from china?
16:37 celeron55     no; i would tell if there were
16:38 thexyz        right; did they specify any time frame?
16:38 celeron55     no
16:54 thexyz        what's the status of httpfetch?
16:54 sapier        difficult as it may not be required for mainmenu if the async things are added
16:55 thexyz        that's not the only thing I need it for; it's got nice media download support
16:55 sapier        yes that's why I said difficult
16:56 sapier        I assume it doesn't break non curl media download
16:56 thexyz        https://github.com/kahrl/minetest/commit/b7125e56b28751e30d2d2bc5fa7c6b4dc2d674a7
16:57 sapier        it's a good improvement yet I'm not sure if it's tested enough for 0.4.8
16:58 thexyz        oh, so we're releasing Very Soonâ„¢ now?
16:59 thexyz        then it should be merged after that
16:59 VanessaE      http_fetch will probably also fix a bug in the media fetching where some files are sent too late
16:59 sapier        that's base for my suggestions ... we haven't released for quite a while imho it's time to do
17:00 sapier        yes it may fix this bug but it may add additional concurrency bugs too
17:00 celeron55     the blocking windows localication thing was fixed (but not yet merged?) so it's okay to try to get 0.4.8 out now
17:01 celeron55     (as long as the pending problems are taken care of)
17:01 sapier        no it wasn't merged by now as the fix for msvc is quite ugly ... without a clear "yes we add it" by 2 core devs I wont add things like that ;-)
17:01 celeron55     well what options do we have?
17:01 celeron55     keep stuff broken or add it? not a hard choice...
17:02 sapier        If I knew a better solution I wouln't have come up wit that one ... but maybe others have better ideas
17:02 thexyz        I think celeron55 suggested some improvement to your solution
17:02 sapier        I don't remember any additional suggestions can you tell me where to find them?
17:04 thexyz        okay, let me check logs
17:04 sapier        hmm maybe you're talking about that "full localization" list?
17:06 thexyz        http://irc.minetest.ru/minetest-dev/2013-10-28#i_3398880
17:08 proller       https://github.com/minetest/minetest/pull/895/files - now liquid is fast
17:09 sapier        hmm the only way to use this would be creating a lookup table on startup I don't want to do api calls each time a string is translated
17:10 sapier        I have to go now but I'm gonna have a try if this works later
17:10 thexyz        that'd be great
17:11 sapier        maybe it's as easy as doing a name lookup ... not quite sure if I understood doc right ... but could be
17:11 sapier        bye
17:12 VanessaE      I don't advocate 0.4.8 going out just yet.  Not 'til the http_fetch and lib_6d both go in.
17:14 proller       VanessaE, all liquids fast ^^
17:14 PilzAdam      VanessaE, lib_6d uses sneak for placement, that is not good since its used to place nodes at other nodes that would usually have a different effect on rightclick
17:15 VanessaE      PilzAdam: well ShadowNinja advocated I do that check in the code.
17:15 VanessaE      for consistency, since we have no custom modifier keys
17:15 VanessaE      the original code took an 'invertall' parameter
17:16 VanessaE      (which, if omitted, would be treated as false)
17:16 VanessaE      er invertwall*
17:17 PilzAdam      wait, so its not possible to place nodes that use minetest.rotate_and_place() when pointing at a chest?
17:17 VanessaE      of course it is.
17:17 VanessaE      it's just that the invert flag would be triggered, causing the node to appear in 'floor' orientation.
17:17 PilzAdam      if you simply rightclick then it opens the chest instead of placing the node
17:18 VanessaE      there are only three ways to do this:  either I accept an incoming flag from the calling routine, or I check for a key like sneak, or we get proper modifier keys that aren't already bound to some feature.
17:19 VanessaE      the third option of course is the best, but requires more changes to the engine, and probably to the client for all I know
17:21 PilzAdam      ah, so you dont want nodes to override on_place() with that function?
17:22 VanessaE      the way the function is used now (well the version of it that is in current use), it checks explicitly for 'sneak' before calling the function, and passes that to the function.
17:22 VanessaE      so basically it does the first option -
17:23 VanessaE      it's normally used in an on_place callback actually.
17:23 PilzAdam      what Im talking about is that you dont do the check for sneak before calling on_rightclick()
17:23 PilzAdam      so if you just do on_place = minetest.rotate_and_place; then its not possible to place nodes while looking at a chest
17:24 VanessaE      oh, nonono
17:24 VanessaE      that's not how I normally use it
17:24 ShadowNinja   VanessaE: I meant things like t["valid_lua_identifier"] instead of t.valid_lua_identifier which I intended to fix be amending if the pull was acepted.
17:24 ShadowNinja   by*
17:24 PilzAdam      VanessaE, thats not how _you_ use it
17:24 VanessaE      ShadowNinja: ah, well that's easily fixed.
17:25 PilzAdam      also, why doesnt it apply the normal facedir if you place it at the floor or ceiling?
17:25 VanessaE      hadn't thought about that.
17:26 VanessaE      this was designed for nodes that explicitly use 6d though.
17:26 PilzAdam      this seems to be too ungeneric to be in the API
17:26 VanessaE      e.g. slabs, lighting panels, etc.
17:27 PilzAdam      slabs cant use this because they need the full block transformation
17:27 VanessaE      ShadowNinja: fixed.  I guess.
17:27 VanessaE      else screw it.  the code is there and I did my best on ioy.
17:27 VanessaE      it*
17:28 VanessaE      it works fine in moreblocks, and I use a variant of it (where I don't check for 'sneak') in homedecor's lights.
17:28 VanessaE      works perfectly in both.
17:28 VanessaE      or at least, it works exactly as intended.
17:28 ShadowNinja   VanessaE: BTW sapier's assert_backtrace function is wrong, it should be "if not x" not "if x == false" so that nil fails properly.
17:29 VanessaE      (in moreblocks, sneak is 'force wall' rather than 'invert wall')
17:29 PilzAdam      see? even moreblocks cant use the function that you want to add
17:30 proller       maybe old idea: little move object when punching, (for push inactive players to holes)
17:30 VanessaE      proller: offtopic.
17:30 proller       singletopic again?
17:31 VanessaE      PilzAdam: it can, if I pass invertwall as a separate flag.
17:31 VanessaE      I coded it differently in moreblocks because that's what Calinou wanted.
17:31 ShadowNinja   proller: Yes, but that isn't high priority now, we have to get 0.4.8 out.
17:31 VanessaE      ShadowNinja: check #959.  is that more like you were thinking?
17:32 VanessaE      PilzAdam: I originally wanted the moreblocks version to do invert-wall also.  calinou didn't like that, so I didn't do it.
17:32 VanessaE      so blame him :P
17:32 kahrl         does anyone mind if I set #962 to milestone 0.4.8?
17:33 PilzAdam      kahrl, I wouldnt mind merging if it works
17:33 proller       892 weather updated
17:33 thexyz        kahrl: why can't we just throw it in
17:33 kahrl         I can't test it since apparently it's a windows issue
17:34 ShadowNinja   kahrl: Or just merge it. :-)
17:34 kahrl         if somebody tested it it can be merged
17:34 PilzAdam      I guess sapier tested it?
17:34 ShadowNinja   sapier has, or at least so I assume.
17:34 kahrl         I think so but I don't exactly remember
17:34 PilzAdam      we wait for him then
17:34 thexyz        yeah it doesn't make things worse
17:34 PilzAdam      add it to the milestone if you fear that we will forget it otherwise
17:34 ShadowNinja   Well he should know better than to commit un-tested code...
17:35 VanessaE      PilzAdam: so what about it?  what's gonna be here?  do I check inside the routine or outside (for sneak)?  So I invert-wall, force-wall, what?
17:35 ShadowNinja   How about #965? I can't test it though...
17:36 VanessaE      Do*
17:36 PilzAdam      VanessaE, I just wanted to point out that there are different expectations how the function should work; that means its too ungeneric to be in the API
17:36 proller       965 looks ok, i can test on freebsd
17:37 VanessaE      PilzAdam: well if there are different expectations, it's because there is no consistency in these sorts of things.
17:37 ShadowNinja   Ok, push it if it works.
17:37 VanessaE      and it is NOT too 'ungeneric'
17:38 proller       ShadowNinja, push!
17:38 nore          and how about #856?
17:38 PilzAdam      dunno if "ungeneric" is a good word
17:38 VanessaE      you probably mean 'specialized'
17:38 ShadowNinja   proller: Hmmm? Is that a confirmation that it works on FreeBSD?
17:38 VanessaE      and if you want it to be consistent, then suggest a way to make it so!
17:38 VanessaE      (and yes I'm getting upset)
17:39 proller       ShadowNinja, to correct test opendsd needed
17:39 proller       ShadowNinja, but it now only one-user-feature 8)
17:39 proller       so, push!
17:40 PilzAdam      VanessaE, yea, the problem is that we cant force modders to our "consistency"
17:40 proller       or i can
17:41 VanessaE      PilzAdam:  well all I know is this routine is needed by most any mod that places a facedir object.
17:41 VanessaE      (except maybe default e.g. chests and furnces)
17:42 PilzAdam      not exactly this routine, but a similiar one
17:45 VanessaE      there, how about that>
17:46 VanessaE      optionally pass a table as the last param, "invert_wall = true" or "force_wall = true"
17:48 nore          what about merging #856?
17:48 VanessaE      need to add a flag for forcing on-floor facedir to 0, versus normal facedir, maybe
17:49 nore          is there anyone else agreeing with #856 (protection)?
17:50 VanessaE      PilzAdam: there, how's that look?
17:51 VanessaE      I suppose I need to do the same facedir thing with the ceiling mode.
17:54 nore          so, about #856? any thoughts?
17:55 nore          VanessaE, you need to update lua_api.txt
17:55 VanessaE      oh, yeah I'll get that
17:55 VanessaE      need to do something else first.
17:56 ShadowNinja   I can agree with #856, but I would like more opinions.
17:56 nore          I agree
17:56 nore          but we need another dev probably...
17:56 celeron55     #965 seems good to me
17:57 nore          and #856?
17:58 celeron55     if people really want it as-is, i guess i can't block it
17:58 nore          why, what is missing?
17:59 celeron55     hmm, actually, the documentation has been changed and now it roughly is how it should be
17:59 celeron55     i'm not objecting
18:03 rubenwardy    What do you guys think about #977?
18:03 proller       celeron55, #895 - with lot useless stuff!
18:04 ShadowNinja   Alright, I will push #856 then.
18:05 ShadowNinja   rubenwardy: Already discussed, it will be going in after 0.4.8.
18:05 nore          ShadowNinja, you need to write something about it in News... (for modders, they need to know that since there are a lot of protection mods)
18:05 ShadowNinja   nore: Alright.
18:06 rubenwardy    ok
18:08 VanessaE      PilzAdam: ok, check the code now.  It seems to work right anyway.
18:08 ShadowNinja   Hmmm, it seems that GitHub properly auto-closes and marks as merged pulls that are merged with rebase and "git merge --ff-only".
18:09 nore          yep, I saw that
18:10 PilzAdam      only if they are alraedy rebased, AFAIK
18:10 VanessaE      tested by substituting it into homedecor with some textures that clearly show the facedir
18:10 VanessaE      oh I guess I better fix the docs too
18:11 PilzAdam      cant you put this in misc_helpers.lua or so?
18:11 Evergreen     Can someone check this pull request to see if it's good for merging?   Can a core dev approve/disapprove this pull request?    https://github.com/minetest/minetest_game/pull/197
18:11 Evergreen     *whoops, accidently copied and pasted something
18:14 VanessaE      there, doc updated.
18:15 VanessaE      PilzAdam: who, me?
18:15 PilzAdam      yes
18:15 VanessaE      sure if you want.
18:15 VanessaE      it's a habit to keep things like this in its own file, is all
18:18 VanessaE      done.
18:18 VanessaE      oops
18:18 VanessaE      missed a line.
18:19 VanessaE      there.
18:20 PilzAdam      and why have you added it in lua-api.txt in that line and somewhere else?
18:20 ShadowNinja   Hows this? http://pastebin.ubuntu.com/6348417/
18:20 ShadowNinja   +'
18:20 PilzAdam      *not somewhere else
18:21 PilzAdam      ShadowNinja, why is the . part of the link?
18:21 ShadowNinja   PilzAdam: Because that was easier. ;-) fixed.
18:22 VanessaE      PilzAdam: I couldn't think of a better place to put it than near the original place_node() call
18:22 VanessaE      it seemed related.
18:24 ShadowNinja   Anthing else PilzAdam?
18:25 ShadowNinja   nore: What do you think ^.
18:25 VanessaE      ShadowNinja: seems reasonable to me.
18:26 VanessaE      PilzAdam: got an idea for a better location?  I put it where I did because I figured it would be easy to find.
18:27 PilzAdam      I would put it near item_place_node(), but it cant be in that category since its called "Defaults for the on_* item definition functions:"
18:28 VanessaE      yeah, that's not really a good place I don't thing.
18:28 VanessaE      think*
18:30 PilzAdam      put it in "Random:"
18:33 VanessaE      done.
18:33 VanessaE      (and somehow I managed to remove a couple of bogus spaces at the same time :P )
18:48 Evergreen     Sorry about that previous request, accidentally copy pasted something from my clipboard.   Can Someone check to see if this pull request is clear to go?    https://github.com/minetest/minetest_game/pull/197
18:49 proller       http://paste.org.ru/?0080bf WTF!
18:50 proller       do not touch history!!!!
18:50 PilzAdam      proller, do you read this channel?
18:51 proller       sometimes
18:51 PilzAdam      the history rewrite was announced and within 10 minutes after the commit that was rewritten
18:51 proller       for what?
18:54 Exio4         if you commit something, you are free to change it within a 10(~) minutes range, it is in the guidelines
18:56 proller       and nobody can merge master in this 10 minutes
18:56 PilzAdam      Exio4, only if you announce it here
18:56 ShadowNinja   History rewrites should be avoided of course.
18:57 Exio4         PilzAdam: that is in the "common sense" part i guess?
18:58 proller       895 and 882 conflicting 8( and i cant merge them in one
18:59 thexyz        ShadowNinja: shouldn't you tag your topic in the news section with [0.4.8]?
19:00 ShadowNinja   thexyz: Ah, yes, done.
19:01 PilzAdam      ShadowNinja, and fix everything tha rubenwardy mentioned
19:05 ShadowNinja   PilzAdam: Done.
19:06 ShadowNinja   All except "* [0.4.8]?" that is
19:06 ShadowNinja   And for mods "C++ API" == "builtin API".
19:07 ShadowNinja   And "That's not really a hack." == false.
19:11 kaeza         It's not a hack; it's called compatibility layer
19:11 VanessaE      PilzAdam, ShadowNinja:  any other changes to #959 ?
19:55 VanessaE      (oops)
20:01 ShadowNinja   Perhaps the serverlist flags should be stored in a list, eg, servers.list[0].flags.creative = true, rather than mixed with the other data.
20:01 ShadowNinja   I am trying to write a doT.js template for the serverlist now...
20:04 VanessaE      so any further changes to #959 ?
20:06 ShadowNinja   Not now.
20:06 VanessaE      ok.
20:06 * VanessaE    pokes PilzAdam
20:06 * general3214 punches PilzAdam
20:07 ShadowNinja   Hmmm, creative is set to a empty string, rather that true or 1.
20:15 proller       ShadowNinja, good idea
20:23 PilzAdam      ShadowNinja, when you create minetest.<function> article in the dev wiki then create <function> that redirects to the article too
20:25 ShadowNinja   PilzAdam: Perhaps there should be a auto-redirect.
20:26 ShadowNinja   Untested doT.js template: https://gist.github.com/ShadowNinja/7283108
20:26 ShadowNinja   It doesn't have the modlist dropdown yet (Because that should be done differently).
20:29 proller       http://minetest.setun.net:8000/ - now with penis-meter
20:30 proller       * pop = average number of players on server
20:30 sfan5         *cough*
20:30 VanessaE      10 users on realtest? O.o
20:31 proller       and average=pi
20:31 VanessaE      ha!
20:32 PilzAdam      is there a limit for description?
20:32 proller       no
21:31 ShadowNinja   Here's a version of the serverlist using a doT.js template, it may or ma not work. http://ix.io/8QT
21:32 ShadowNinja   (I have to load it on a HTTP server...)
21:34 VanessaE      PilzAdam: so, #959 is good?
21:35 ShadowNinja   \o/ It works! Although table borders are missing...
21:36 PilzAdam      VanessaE, what happens if force_wall and invert_wall are both true?
21:36 VanessaE      PilzAdam: force_wall overrides.
21:37 PilzAdam      I wonder if it would be better if minetest.rotate_and_place(flags) would return a function that can be used in on_place in nodedef, similiar to item_eat()
21:37 Evergreen     Is #197 on minetest_game good for merge?
21:37 VanessaE      PilzAdam: I don't know how to do that.
21:38 VanessaE      imho the pull as it stands now is good enough for general use :P
21:40 ShadowNinja   http://shadowninja.minetest.net/serverlist/ <-- Serverlist w/ template. Needed: human-readable times and modlists. (+style fixes)
21:41 thexyz        also needs some design
22:40 ShadowNinja   I fixed the serverlist. To anyone that knows how: How should I implement the modlist box and similar boxes? The old way reserved a lot of extra space at the bottom of the screen.
22:42 iqualfragile  @randomdevwithalotofsparetime sometimes things get out of sync between client and server
22:43 iqualfragile  more precisely: entities
22:43 iqualfragile  even if it is a local singleplayer server
22:49 PilzAdam      sapier, have you tested #962 ?
22:51 sapier        yes on windows as well as linux .. any issues with it?
22:51 PilzAdam      and it works?
22:51 sapier        at least for me ;)
22:52 PilzAdam      it can be merged then
22:52 sapier        I only tested irrlich 1.8(windows) as well as 1.7.3(linux)
22:55 ShadowNinja   https://github.com/ShadowNinja/minetest/commit/d3cf8a822ebd97d3ef790e210a25bde3d2c888ed
22:55 ShadowNinja   proller: ^ Thoughts?
22:57 ShadowNinja   Hmmm, I notice that my implementation of humanTime() works a lot better.
22:58 proller       16d 22h 25m 9s
22:58 proller       its better?
22:58 proller       9s very important
22:59 proller       and now  uptime, age - its 2 values
23:00 ShadowNinja   proller: I can remove seconds, but it's better than "10.8h 7.6M".
23:00 proller       0.4.7, minetest, undefined
23:00 proller       ShadowNinja, its not bettes
23:00 proller       r
23:00 * ShadowNinja disagrees.
23:00 proller       if value more than day - hours have no sense
23:00 ShadowNinja   I will fix the undefined mapgen.
23:01 proller       and where is xss transform?
23:02 ShadowNinja   proller: Hmmm? The server is expected to send valid data.
23:02 proller       no
23:03 proller       for server "alert()" is valid
23:03 ShadowNinja   proller: That will display as "alert()".
23:04 proller       <script>alert()</script>
23:04 ShadowNinja   But "<script>alert()</script>" probably won't.
23:04 proller       and you lost age value
23:04 ShadowNinja   There is probably some standard escapr function...
23:05 proller       and why -function get(refresh) {
23:05 ShadowNinja   No, age is just always undefined. (Why?)
23:05 proller       its defined here http://servers.minetest.net/
23:06 ShadowNinja   No need for it, it is called imediately afterward with refresh unfefined.
23:06 ShadowNinja   Odd.
23:09 proller       and where is more...
23:12 proller       and where is disableable columns
23:12 ShadowNinja   proller: Where are they in your version?
23:12 proller       http://minetest.setun.net:8000/ - press clients/players
23:13 proller       they can be adjusted from html
23:13 proller       http://minetest.net/servers - same script
23:13 proller       but without ping
23:15 ShadowNinja   Hmmm, that can probably be done with doT, if you really want that. The way it works there is a little odd.
23:17 ShadowNinja   Ah, I should add display.ping and the like.
23:18 proller       .. and template will be not better than now
23:22 ShadowNinja   Me and thexyz disagree.
23:27 ShadowNinja   With IPv6 support the IP colomn has to be shortened, perhaps the first few charachters and toe rest on hover...
23:34 proller       ipv6 with port must be [ip]:port