Time  Nick       Message
00:06 * PilzAdam closes oldest open issue
00:46 kahrl      wow, very promising
00:46 kahrl      I implemented the software renderer for ^[inventorycube
00:47 kahrl      partially: it only draws the top face so far and there's no antialiasing / texture filtering
00:47 kahrl      but according to TimeTaker it takes 0ms
00:49 Exio4      after some work it will be an under zero time! :P
00:49 PilzAdam   it can pass a flag to the mapgen to pregenerate a world before you start the game then ;-)
00:59 kahrl      draws all three sides -> still 0ms
01:00 kahrl      dangit, not optimized enough yet :P
01:01 VanessaE   heh
01:01 kahrl      sneak preview: http://paste.dy.fi/BmC
01:01 PilzAdam   why dont you have a ns timetaker?
01:05 kahrl      didn't know it could do that. "inventorycube took 139014ms"
01:05 kahrl      ^ I assume that should be ns
01:07 Exio4      what cpu?
01:07 Exio4      force it to run at a lower speed ;P
01:09 kahrl      i5 760 @ 2.80GHz
01:10 Exio4      kk
01:11 Exio4      why is the singlethread performance of all the cpus around better than mine :(
01:11 Exio4      :P
01:17 VanessaE   https://github.com/minetest/minetest/issues/831
01:18 VanessaE   finally caught that bastard. :)
01:24 kahrl      looks complicated...
01:27 hmmmm      kahrl, TimeTaker timetaker("inventorycube", NULL, PRECISION_MICRO);
01:27 hmmmm      also great work :)
01:29 hmmmm      of course 0ms doesn't mean too much when that needs to be run ~5000 times
01:30 Exio4      lets say it is 1ms, that is "only" 4s~ for 4000 nodes
01:30 Exio4      nothing compared to now
01:31 hmmmm      that code looks like something that could easily be SSE'd if you wanted to waste some time
01:32 Exio4      some fallback "C" code should be available though
01:33 hmmmm      is that ever not the case?
01:55 kahrl      if anyone wants to try it and see what it does to load times: https://github.com/kahrl/minetest/tree/optimize_invtex
01:56 VanessaE   ok, I'll bite.
01:56 kahrl      ah yes, I'd be interested in what it does in your case
01:57 kahrl      since it only optimizes nodes with drawtype "normal"
01:59 VanessaE   on my server, that's still about, oh, two thirds of the nodes, give or take a bit
02:00 VanessaE   ok, first run to build the cache, then I'll time it.
02:04 Exio4      being able to load the "whole" game in less than a second with default is amazing ;P
02:05 VanessaE   26 seconds to get through the sign-on + media cache.  Add roughly 99 seconds to render the textures (512px).  total of 125 seconds from "Connect" to seeing the world.
02:06 Exio4      kahrl: wouldn't be possible to make it multithreaded "now"? ;P
02:06 Exio4      yes, i love multithreading
02:06 Exio4      VanessaE: from where was i able to download the game?
02:06 sokomine   if it helps: i tracked a stack of (this time) flowers down to the bottom and collected 58 of them that where embedded in stone. the same has happened with other kinds of blocks that where dropped
02:07 VanessaE   my game?
02:07 Exio4      yes
02:07 VanessaE   http://digitalaudioconcepts.com/vanessa/hobbies/minetest/VanessaE_Game.tar.bz2
02:07 Exio4      for reaching those 4100 nodes
02:07 Exio4      thanks
02:07 VanessaE   it's slightly out of date but recent enough to be relevant
02:07 VanessaE   but it only has ~4090 or so nodes.
02:09 kahrl      VanessaE: how does that compare to without the patch?
02:09 kahrl      in case you know or want to test
02:09 VanessaE   kahrl: let me try it and see.  hold
02:10 kahrl      Exio4: multithreading would certainly be easier :) but TextureSource isn't written with that in mind
02:10 Exio4      hehe
02:15 VanessaE   kahrl: with current HEAD, it's 24 seconds to get through the sign-on/defs/cache, 161 seconds total.
02:15 VanessaE   so a rather LARGE boost in fact.
02:15 VanessaE   125 vs 161 seconds
02:16 kahrl      nice
02:16 VanessaE   I wonder if there's anything that can be done to improve that ~24 seconds it takes to sign in before the rendering step begins
02:16 kahrl      I guess the rest is mostly due to extruding
02:16 VanessaE   yes
02:18 VanessaE   (most of the 24 seconds is spent waiting in the "Media..." step even though it's all cached)
02:19 kahrl      no idea but probably
02:19 VanessaE   btw, my server officially has >4096 nodes now (blame sokomine :P )
02:36 VanessaE   Exio4: the first time you run that game, btw, it will create a file in your world folder for pipeworks' settings.  Set the last line (mese sand tubes) to 'true' and it'll push you well over the 4096 mark, if you wanna mess with it.
02:36 Exio4      kk
02:36 VanessaE   (actually it may already be set as such)
02:41 Naked      is that the node limit of the game?
02:42 VanessaE   Naked: by default, 4096.  but a one-line tweak can raise it higher if necessary.
02:50 Naked      there is no negative side effects by raising it?
02:50 Exio4      not now
02:50 Exio4      ^^
02:50 Naked      then why isnt it even higher normally?
02:51 Naked      should be in the hundred thousands without negative effects to the engine and game
02:51 Naked      and why so high? because the game can without negative effects
02:51 Naked      just cuz it can
02:52 VanessaE   Naked: nope, no apparent ill effect as long as folks remember to update their clients for kahrl's commit
02:53 Exio4      VanessaE: your hate is insanely resource intensive ;P
02:53 VanessaE   hate?
02:53 Exio4      game
02:53 VanessaE   o.O
02:53 VanessaE   wow talk about a typo
02:53 Exio4      sorry, i'm not right
02:53 Exio4      game => hate
02:54 VanessaE   yes I know it's resource-intensive.  I don't cater to anemic laptops :P
02:54 Exio4      <- minetest
06:05 celeron55  kahrl: are you intending to extend that to nodeboxes?
06:06 celeron55  i figure it wouldn't be impossible
12:17 kahrl      celeron55: should be possible, though I'd have to find a way to pass the nodebox from ItemDefManager to TextureSource
13:23 Exio4      seriously though, would be a 100HP health system be possible to get merged?
13:24 PilzAdam   whats the benefit of it?
13:25 Exio4      more control for the damage, you'll be having 10 times more "accuracy" in that
13:26 proller    and make damage depend on high/low temparature!
13:26 Exio4      small damage would be easy
13:26 Exio4      +to do
13:27 proller    but it will be better with protecting clothes
14:38 PilzAdam   https://github.com/minetest/minetest/pull/833
14:53 proller    want!!! https://github.com/proller/minetest/commit/3a2f331bbc79293a7a230051706774cdede6f71c
14:53 proller    and than https://github.com/proller/minetest/commit/35d5be75c8810bb6d692be1b0084e95642d81afb
14:54 proller    PilzAdam, 833 - strange
14:55 proller    its better be based on viscosity
14:55 PilzAdam   hm?
14:55 PilzAdam   why restrict modders in their abilities?
14:56 proller    why restrict? viscosity already exists
14:56 PilzAdam   I also dont want to break maps
14:56 PilzAdam   people might want slow liquids that flow very far, or fast liquids that dont spread far
14:57 proller    and this flooding logic is very strange
14:57 proller    why liquid increase volume ?
14:57 PilzAdam   so what? dissalow strange things is a rule for the API?
14:57 proller    PilzAdam, its not liquids
14:58 proller    liquid can be fast and far OR slow and  dont far
14:58 proller    its one property
14:59 proller    and must change levels to ~64 to really far
14:59 proller    max levels 8=>64
15:00 PilzAdam   thats not possible with the current liquid mask
15:00 proller    need to change mask
15:01 proller    its will bit break maps
15:01 PilzAdam   that would break compatibilty of maps, wouldnt it?
15:01 proller    better no use new liquid 8)
15:01 proller    to use
15:02 proller    i want to make one water block instead of source & flowing with 65 levels
15:02 proller    64
15:02 Exio4      hmm, what?
15:02 proller    its possible with finite_liquids
15:03 Exio4      finite_liquids != normal liquids
15:03 proller    normal liquids != liquids
15:04 proller    they much more normal
15:04 Exio4      do you see a health bar irl?
15:05 proller    do you see stupid liquids in other games?
15:05 PilzAdam   proller, yes, MC
15:05 Exio4      i actually like the behavior of those "stupid liquids"
15:05 PilzAdam   I actually dont know a game that has realistic liquids
15:06 proller    terraria
15:06 proller    dwarf fortress
15:06 proller    Exio4, its not liquid
15:06 proller    PilzAdam, why MT must copy stupidests things from MC ?
15:07 Exio4      what are they? (if not liquids)
15:07 PilzAdam   hmm... maybe because MT copied everything else from MC?
15:07 proller    now liquids - most ugly
15:07 proller    PilzAdam, not everything
15:07 proller    why do nt copy 255 height limit?
15:08 Exio4      i don't like the idea behind finite_liquids
15:08 proller    Exio4, why?
15:08 PilzAdam   I like the finite liquid idea, but its poorly implemented
15:08 proller    its better to rename to new_liquds
15:08 Exio4      well, that too 8)
15:09 PilzAdam   proller, hows it better?
15:09 Exio4      proller: do you know work for redhat now?
15:09 proller    PilzAdam, its inherit all poor from old liquids
15:09 PilzAdam   finite liquids are unusable currently, people may think that we develop into a wrong direction
15:09 proller    unusable what?
15:10 PilzAdam   compared to the current liquids, that is
15:10 proller    what is unusable ?
15:10 PilzAdam   I made a list a while ago
15:10 PilzAdam   AFAIK nothing of this list is fixed yet
15:10 proller    its all problems in old liquids too
15:10 PilzAdam   no
15:11 proller    yes
15:11 PilzAdam   lol
15:11 Exio4      PilzAdam: i guess he works for redhat now
15:11 Exio4      oh, i see
15:12 Exio4      proller: reasking, do you work for redhat? :>
15:12 proller    NO
15:12 PilzAdam   https://gist.github.com/PilzAdam/5687050
15:12 PilzAdam   look at the finite liquid mod for minecraft
15:13 proller    i have no minecraft
15:13 PilzAdam   there are videos on youtube
15:13 proller    1 - somebody fix that? - but its must be in old too
15:14 proller    2 - some graphical must be fixed, but it all from old liquid
15:14 proller    3 - old too!
15:14 proller    4 - old too!
15:15 proller    and 4 partialy solved in v7?
15:15 PilzAdam   4 means that caves can actually remove complete oceans, that is definitely not in the current liquids
15:16 PilzAdam   and why do you even base the finite liquids on the current ones when you say they dont fit together?
15:16 PilzAdam   implement the finite liquids correctly and dont care about the current ones
15:16 proller    by default now ocean is renewable, and it will fill any cave
15:16 PilzAdam   thats against the finite liquid idea
15:17 proller    now finite compatible with old
15:17 PilzAdam   I think that the finite liquids were merged way too early in the core, they are hardly usable at all
15:18 proller    its used on my server, and peoples like it
15:18 proller    old liquids is not liquids and totally unusable it that way
15:19 PilzAdam   stop saying that the current liquids are "no liquids", we make the game so we define what liquid means in the world
15:19 proller    its make world ugly
15:19 PilzAdam   you can say that the current liquids are not realistic, but a world consisting of cubes isnt realistic either
15:20 proller    cubes - its main game idea
15:20 proller    its cool
15:21 proller    every who place old liquid on ground say WTF!!
15:23 proller    peoples makes dam on my server
15:28 proller    part of MT finite liquids already working, rain&snow under progress
15:29 proller    pressure is hard to calculate on huge volumes
15:29 proller    initial sponge in pull to _game
15:29 PilzAdam   proller, why dont you do all the stuff in a seperate fork and merge it back in master if its completly done?
15:30 proller    flowing push - in todo
15:30 proller    and current liquids works on huge oceans, i think MC liquids cant
15:30 proller    because nithing is "completly"
15:31 proller    nothing
15:31 proller    why celeron55 dont write complete game and than not publish it ?
15:32 proller    long forks is hard to support
15:32 PilzAdam   I mean for the weather stuff, I have no idea what you plan and what your goal is
15:32 Exio4      proller: c55 wrote the game 'just for the lulz'
15:33 Exio4      i'm pretty sure he never thought about people commiting or even playing it
15:35 Anchakor1  aren't finite liquids boring?
15:35 Calinou    they are boring and laggy
15:36 Calinou    :)
15:37 proller    PilzAdam, rain/snow
15:37 proller    like irl 8)
15:38 proller    Calinou, did you use rollback ?
15:38 Calinou    no
15:38 proller    now its laggy falling
15:38 Exio4      fix it then
15:39 proller    i tried to make fallung water, but falling code too slow and buggy and must be rewrited in.... core! 8)
15:39 proller    and always 49 objects error
15:41 proller    Calinou, old liquid not laggy?
15:43 Calinou    less laggy and glitchy, obviously
15:44 Anchakor1  why make finite liquids when they are boring and even use more resources?
15:46 Exio4      that second thing, too
15:46 Exio4      they are more resource-intensive
15:49 hmmmm      i don't like finite liquids
15:49 proller    Anchakor, its boring not for all
15:49 hmmmm      but at least they're there
15:50 proller    hmmmm, why you dislike?
15:50 hmmmm      not sure, regular water is more predictable and 'standard'
15:50 hmmmm      i know how to work with it
15:51 Exio4      hmmmm: you just don't like how it behaves
15:51 Exio4      it is a different way to "manage" liquids, i don't like how finite_liquid works in a voxel-game either
15:52 proller    how it must works?
15:53 Anchakor1  I think from gameplay perspective finite liquids don't add anything and cost more CPU
15:54 hmmmm      does it use more CPU?
15:54 hmmmm      i never checked and nobody ever talked about that
15:56 proller    Anchakor1, its add flooding
15:57 proller    you cam make pool, dam
15:57 proller    pipes with real liquid transfer
15:57 Anchakor1  you can make it with infinite water too
15:57 proller    no
15:57 Anchakor1  it is just more realistic
15:57 proller    old water cant flood cave
15:58 Anchakor1  you mean like completely, yes that it can't
15:58 proller    its make sloped wall from blue blocks
15:58 proller    finite is flooding as real
15:58 Anchakor1  but that would mean all underwater caves will be full of water
15:58 Anchakor1  that would be really annoying
15:59 proller    and its good and make more diverse gaming
16:00 proller    sometimes you must fight with water/lava
16:01 proller    and its makes possible like real snow-rain
16:01 Anchakor1  I don't mind if there was a game mode like this, but not the normal one
16:02 proller    its now disabled by default
16:08 proller    its can disturb in creative mode, but must be in survival
16:09 Anchakor1  lol I was imagining it exactly the opposite way
16:21 celeron55  i propose that you take on this fight by making youtube videos with the liquids
16:21 celeron55  then people will vote on which is more interesting!
16:23 celeron55  i do think that a very good implementation of finite liquids would be best
16:24 celeron55  and yes, i imply that currently stuff isn't particularly impressive
16:28 celeron55  (or not even good by pretty much any standard)
16:29 proller    i cant make video, maybe somebody can
16:31 proller    it can be bit more impressive with rain-snow, but it needs ~week to complete
16:31 proller    very good - liquids must fall as sand - need to make faster nodeupdate (core?)
16:32 proller    and must be fixed visual,
16:32 celeron55  currently when i start a finite liquid world, it just... doesn't work
16:32 proller    and better to make 64 levels and broke backward compatibility
16:33 proller    how?
16:33 celeron55  nothing flows anywhere
16:33 proller    strange
16:33 celeron55  also one thing is that the finiteness of water should be a per-world setting
16:34 proller    maybe, but finite_ can start work on old world
16:35 celeron55  there's no way MT would be moved completely to finite liquid until it works 100% perfectly
16:35 proller    ant with it possible to start making 64 levels
16:36 celeron55  and at the current rate, i can predict that it won't be that perfect until after years
16:36 celeron55  so well-made coexistance of both is a must
16:37 proller    maybe year
16:38 hmmmm      celeron, pilzadam pretty much broke regular liquid transforming by setting the default max number of loops to 1000
16:38 proller    maybe just slower ?
16:38 hmmmm      you need to set it to 10000 if you want things to actually move
16:38 hmmmm      definitely not.... it's at a complete halt
16:38 proller    better to set update interval to 0.1
16:38 hmmmm      I have my update interval at 0.2
16:38 proller    and halt ?
16:39 hmmmm      it's just completely stopped
16:39 proller    its 5000 per second
16:39 hmmmm      this means that there are liquids stuck in the queue
16:39 proller    maybe need enable flowing touching abm ?
16:39 hmmmm      or the same ones keep getting readded
16:39 hmmmm      that's pretty hackish, proller
16:39 hmmmm      i'd like to fix the actual problem
16:39 proller    btw finite have no this problem 8)
16:40 hmmmm      obviously not, but that's for a different reason
16:40 proller    its better algorithmically
16:41 Exio4      the implementation isn't that good though
16:41 hmmmm      i do believe that regular water could be done better too
16:41 proller    its have convergence
16:42 proller    Exio4, i making it better, but i cant do it in one time
16:42 celeron55  finite water has convergence? lol
16:42 celeron55  in a finite world with caves, for sure
16:42 celeron55  ehm
16:42 celeron55  infinite*
16:42 celeron55  infinite world, that is
16:43 proller    celeron55, by default 0- level is self-renewing, and it make always 0 queue at time
16:43 proller    all caves is finite
16:44 proller    very rare water can flow to -100 in caves
16:44 proller    most caves flooded per 2-5 minutes
16:44 Jordach    what if people made lakes at level 0
16:44 Jordach    or even infinite pools at that
16:45 proller    now is problem with huge caves in indev mapgen - its disaster, and can take hours to fill
16:45 proller    bit it fill!
16:45 celeron55  for sure, with 100% CPU consumption and 2 seconds of lag for 10 hours? 8)
16:45 celeron55  well, at least our players are used to such
16:46 proller    no lag with 1000 nodes per cycle
16:46 celeron55  a rare privilege for developers
16:46 Exio4      it is not a valid reason though
16:46 proller    on slow cpu can set queue to 300
16:47 celeron55  do you realize how little 300 nodes is?
16:47 celeron55  it's a 7x7x6 piece
16:47 proller    300 x 10 times per second
16:48 celeron55  anyway, currently nothing happens in a finite liquid world for me
16:48 proller    active area around player filled in 5-15 minutes, for fill huge cave you must moving around
16:49 proller    celeron55, its enabled ?
16:49 proller    maybe menu bug ?
16:50 proller    its must fill an caves on new area
16:50 proller    all caves
16:50 proller    immediate
16:50 celeron55  hmm, now something happens
16:52 proller    you can force it by digging around water
17:02 proller    still want to https://github.com/proller/minetest/commit/3a2f331bbc79293a7a230051706774cdede6f71c
17:03 celeron55  this appears to be somewhat better than i expected
17:04 celeron55  really the most annoying thing is that it looks like it's flowing often while it really practically isn't
17:06 celeron55  (except reeaaaallly slowly)
17:07 Jordach    celeron55, like real water
17:07 Jordach    the bottleneck is the space for the water to flow into
17:07 Jordach    </physics>
17:08 proller    its need in something like flowing flag, draw animation depend on it, and push player on flowing
17:08 hmmmm      proller, just to let you know, I might need to change getHeat and getHumidity in the future
17:08 hmmmm      well
17:08 hmmmm      you use heat in a different way than i do actually
17:08 proller    and delete this flag maybe when queue =0
17:09 proller    hmmmm, no problem, its not final, needs day-night changes, more adjusting, ....
17:10 hmmmm      see, for biome calculations, temperature is supposed to be static but instead you use 3d perlin noise to make it vary with time
17:10 proller    hmmmm, but close to your way
17:11 hmmmm      actaully same thing with humidity
17:11 hmmmm      it's completely different
17:11 Jordach    proller, use things as intended
17:11 Jordach    not differently as it will break at some point
17:11 proller    and later add biome-defined corrections to getHeat and getHumidity
17:12 hmmmm      if you want temperature and humidity to vary, but not completely change, that's a bit more complex
17:13 hmmmm      like you use the static heat/humidity perlin noise values and then vary that by 3d noise on a limited scale
17:13 proller    yes, i want to get something basicaly works, and then tune it
17:14 hmmmm      also it's important to note that 3d perlin noise with one dimension held constant isn't equal to 2d perlin noise, since the 3d variant doesn't use easeCurve() before interpolation
17:15 hmmmm      I really don't like how all the 3d perlin noise is angled and I am considering changing this by *gasp* adding a noiseparams field
17:15 hmmmm      it's going to be quite a bit of re-working but i intend to maintain full backward compatibility
17:18 proller    my features is very wip and you can change anything around
17:19 proller    i want to make working rain collector hills first and than long tune it
17:20 celeron55  have you considered how you are going to handle it at large scale+
17:20 celeron55  ?*
17:20 hmmmm      if you intend for your implementation of weather to become part of mainstream minetest, there are quite a few things that need to be changed beforehand that you'll probably never get to
17:20 celeron55  because it really doesn't make sense if a player needs to walk somewhere to get water in somewhere
17:20 celeron55  from the sky, that is
17:21 proller    now you can stay in one place and get rain or snow after some time
17:21 hmmmm      like, you poll each individual point for temp/humidity, which only makes sense if these are being done randomly
17:21 hmmmm      you have that on an ABM or something?
17:21 celeron55  proller: do you know the system that is used for slowly running abms?
17:22 celeron55  mapblocks store a timestamp of when they were last unloaded, and when they are loaded, stuff can be simulated based on that
17:50 celeron55  when you add stuff to the engine, first consider whether it's useful or at least not a deadweight in this: http://c55.me/random/2013-05/screenshot_1726683685.png
17:50 celeron55  8)
17:51 celeron55  it gives some perspective
17:51 celeron55  (yes, we have a lot of deadweight for that; some of that isn't really doable otherwise though so it's kind of reasonable)
17:53 Anchakor_  wtf is this, meteor shower?
19:05 sapier1    PLZ test https://github.com/minetest/minetest/pull/825 if I didn't miss any of your thousands of requests everything fixable should be fixe
19:05 sapier1    d
19:31 proller    sapier, good
19:32 sapier     don't tell me good unless you tested it ;-P
19:33 proller    its hard to test before merge
19:35 sapier     but it should be tested by someone except the one who wrote it ... I know how it works so I tend to do things as they are intended to be done
19:36 proller    its contain important fixes ;)
19:36 proller    and https://github.com/minetest/minetest/pull/825/files#L11R315
19:36 proller    it must be done via function
19:37 proller    json_vaue_to_number
19:37 sapier     no
19:37 proller    5+ copypastes in file
19:37 proller    why>
19:37 sapier     there only is a toInt in json which doesn't convert
19:38 proller    no, move this strings to function and use
19:38 sapier     didn't you volonteer to rewrite that file anyway?
19:38 proller    315..321
19:38 proller    yes, i want to rewrite
19:38 proller    but now have no time
19:39 sapier     ok so no need to do partial work now
19:39 proller    ok
19:39 sapier     it's working the way it is and rewriting twice isn't really usefull
19:44 sapier     ok last thing done ... modmgr and gamemgr are now enabled by default ... 6 weeks after starting finaly the main purpose for doing this shows up
22:39 PilzAdam   hmmmm, http://forum.minetest.net/viewtopic.php?pid=100302#p100302
23:19 hdastwb    I get the impression that the implementation of minetest.dir_to_facedir hasn't been updated for 6d facedir yet; should it be updated to support +/- y as well?
23:20 hdastwb    also, is minetest.facedir_to_dir a viable addition to the API?
23:20 PilzAdam   no, it would break placing of stairs
23:20 hdastwb    placing stairs calls it with +/- y parameters?
23:20 PilzAdam   facedir_to_dir would be useful