Time  Nick        Message
08:08 celeron55   http://paulmillr.com/posts/github-pull-request-stats/#lang-stats
08:09 celeron55   >C++
08:09 celeron55   8D
08:09 Calinou     lol
08:09 Calinou     you're making it go down :D
16:27 sweetbomber hello, anyone on?
16:27 sfan5       no
16:27 sfan5       as you can see the channel is completly empty [/sarcasm]
16:28 PilzAdam    sweetbomber, that question is quite stupid in IRC; we have logs
16:28 sweetbomber sorry, just asked because just like me, sometimes i have irc open but not paying attention to it
16:29 sweetbomber anyway
16:30 sweetbomber ive been analysing the code, particularly the spawn code, and i noticed two things in code that i think deserve attention and that im willing to improve
16:30 celeron55   all these replies meant is: just shout your question :P
16:31 sweetbomber the first thing, is about the maximum spawn height: I noticed that it assumes water_level+6 as the maximum spawning height
16:31 hmmmm       !calc ask
16:31 hmmmm       well, Just *ask* a precise question (while clearly describing your problem) in the channel. Then wait for an answer. (Do not ask to ask, just ask.)
16:31 celeron55   it's quite old code; it wants to place the player someplace near the origin in some low place
16:32 sweetbomber i have version 0.4.6 from github
16:32 hmmmm       how can you improve upon the spawning code?
16:32 hmmmm       it does what it needs to perfectly fine
16:32 sweetbomber the question is that certain biomes could just be high, just like highlands
16:33 sweetbomber and then the spawning point needs to be high as well
16:33 PilzAdam    one thing could be improved: sometimes you spawn in a tree
16:33 hmmmm       so you'd like to be able to spawn in every biome
16:33 hmmmm       pilzadam, that cannot be fixed
16:34 celeron55   hmmmm: of course it can, but it will be more complicated
16:34 sweetbomber i think there is not problem on spawning over a tree, i guess..
16:34 celeron55   sweetbomber: what is the second thing?
16:34 hmmmm       celeron, any ideas on how to fix that? :/
16:35 PilzAdam    sweetbomber, not over a tree, inside a tree trunk
16:35 celeron55   hmmmm: well, it'd need to wait until the map generator has generated the desired location and then fine-tune the spawn position to a final one
16:35 hmmmm       right, who wants to do that
16:35 PilzAdam    hmmmm, maybe in Lua register_on_spawnplayer()?
16:35 sweetbomber i have an idea to fix that, and if you agree, i can implement it. Basically inspect ground level, without any height limit
16:35 sweetbomber and set it as spawn level
16:35 sweetbomber the place to implement that i think is server.cpp
16:36 Exio        the code changed for that in upstream
16:36 Exio        when hmmmm fixed the "spawning too up
16:36 sfan5       if you would implement that and send a pull request, that'd be appreciated
16:36 Exio        " or so
16:36 hmmmm       sweetbomber, we're not really interested in having the player spawn in high places though.
16:36 sweetbomber I just wanted to discuss it with you, to know if it was a bug or a feature
16:36 hmmmm       if we specifically put a limit there, it's not a bug...
16:37 ssieb       hmmmm: why is that?  What does it do if there is no low location?
16:37 hmmmm       then it'll choose the 1000th randomly generated location anyway
16:37 hmmmm       regardless of ground height
16:37 celeron55   sweetbomber: i guess what you need to put special attention towards is that the ground is not yet generated in certain cases of player spawning but rather will be after some seconds
16:37 sweetbomber My question arouse because i am experimenting with mapgen, and in high places, the player simply spawns underground
16:38 hmmmm       sweetbomber, that might not be caused by what you pointed out
16:38 celeron55   so rather than a measured ground level, it currently then (or always?) relies on the guessed to-be-generated level reported by the mapgen
16:38 hmmmm       instead, it could be because Mapgen::getGroundLevelAtPoint() is no longer accurate
16:38 sweetbomber in my case, the terrain was all at height 20...
16:38 sweetbomber hmmm, that was my second point
16:39 sweetbomber that im about to explain
16:39 sweetbomber getGroundLevelAtPoint uses the terraingeneration functions
16:39 sweetbomber to get the ground level
16:39 sweetbomber but what happens in a mp server where some area of the map was severely dug?
16:40 sweetbomber or completely covered by nodes?
16:40 hmmmm       if it was completely covered by nodes, something else will be chosen
16:40 hmmmm       if there was a big hole, well, then the player will fall in
16:41 hmmmm       i don't get why people believe spawning position is such a big deal... is it because it's the first (minor) thing people have to complain about upon starting the game? :/
16:41 celeron55   well, first impressions do matter
16:41 PilzAdam    how does MC handle this? AFAIK you always spawn at x=0, z=o
16:41 PilzAdam    *0
16:41 sweetbomber sorry, i am just volunteering to help out
16:42 celeron55   PilzAdam: it first generates the full area (full height) around 0,0 and then figures out the top position... at least when i last played it
16:43 PilzAdam    there could be some kind of setting max_spawn so games can change it in their game specific minetest.conf if needed
16:43 hmmmm       i personally see it as too small of a detail to matter at all, but if you really want to change the height at which players spawn at, it might be a good idea to make it configurable
16:43 hmmmm       but there are already too many configurable things and this is bordering on ridiculous.  i believe there's a static spawn point potion already for mods and such
16:44 Exio        why is adding options "ridiculous"?
16:44 hmmmm       adding an option to set the height range people can spawn at would really just be satisfying a *single* person's request
16:44 PilzAdam    yea, but static spawn point is too static for a game
16:44 celeron55   derp
16:44 celeron55   if someone wants to fix something, he can fix it
16:44 sweetbomber besides, i think that some polishing is never unwelcome...
16:44 sweetbomber and since i am new to this code, i wanted to start with something easy, instead of going to hardcore stuff first
16:44 sweetbomber i am planning to contribute to mapgen afterwards...
16:45 sweetbomber anyway, do you agree i could implement code to inspect ground height accurately?
16:45 hmmmm       I don't agree you could, it's much more difficult than you probably think it is
16:45 Exio        he can always try
16:45 celeron55   yes; please just first be aware of all the nuances of what happens when you boot a world and want to put a player there (mainly the state of world loading)
16:46 hmmmm       celeron, this goes back to what i was talking about a long time ago where there was a world pregeneration period before players are allowed to emerge
16:46 hmmmm       rememeber that?  and you said "let's not discuss something complicated like that"
16:47 celeron55   my opinion is that there should be a solid system for just asking for a position and it would then generate or load it that you could just synchronously request a position to be loaded, no matter what
16:47 celeron55   it would be useful for a variety of situations
16:47 sweetbomber i agree that players should only emerge after the world, i think it makes sense..
16:47 celeron55   (to simplify code)
16:48 hmmmm       i removed that from emergeBlock for a reason ..
16:48 hmmmm       basically, now you're saying you want the serverthread to wait for something slow like the emergethread
16:49 hmmmm       i'd really rather have an asychronous event for these sorts of things
16:49 celeron55   well whatever; but it should be doable in some way; currently it's doable only by a hack
16:49 hmmmm       i would be okay with waiting on emergethread if it wasn't so slow
16:50 sweetbomber one thing fellows, regarding pull requests: I believe that your pull-request policy is one feature for pull, right?
16:50 sweetbomber *per
16:50 hmmmm       yes
16:50 celeron55   yes (i don't know anyone who's pull rquest policy would be any different)
16:50 sweetbomber nice
16:51 sweetbomber well...lets just say that ive already seen some strange things in the industry
16:51 hmmmm       anyway, if you'd like to have the player spawn higher, do make it a config entry
16:51 sweetbomber ok
16:51 hmmmm       with the default what it currently is
16:52 hmmmm       (0, 6)
16:52 hmmmm       (well, water_level + that
16:53 hmmmm       also you plan on doing things with the mapgen, but i just want to point out that you can't change anything visible with mapgen v6 anymore, mapgen indev is a separate mapgen for these sorts of things
16:53 hmmmm       you can't change anything that would cause an inconsistency from already generated v6 maps*
16:53 sweetbomber im studying mapgen v7...will it be adopted?
16:53 hmmmm       yes, but it's far from finished
16:53 celeron55   v7 is dictated by this hmmmm you are talking with
16:53 sweetbomber yes, i noticed that
16:53 hmmmm       it also does some terrain things that are stupid that i do not like
16:54 sweetbomber i noticed that the terrain as some glitches
16:54 sweetbomber like one-block towers
16:54 sweetbomber and other stuff
16:54 hmmmm       not glitches
16:55 sweetbomber i read some notch's statements, and he stated that he used some interpolation on perlin noise
16:55 hmmmm       oh, to remove 'towers'?
16:55 sweetbomber this can at least cancel singularities on mp
16:55 sweetbomber on map
16:55 sweetbomber towers, and floating blobs
16:55 hmmmm       i don't know, is it _really_ worth it..?
16:55 hmmmm       i like floating blobs and towers
16:55 sweetbomber i dont know
16:56 sweetbomber i will test some algorightms latter
16:56 sweetbomber as i said, im still getting used to the code
16:56 sweetbomber lol
16:56 celeron55   it was for speed, and it was for removing small features from the base terrain (to allow flat areas to be generated automatically); and as far as i know, it's 3d noise interpolation that notch talked about
16:57 celeron55   interpolating 2d noise is quite boring and doesn't really result in anything
16:57 sweetbomber he stated that 3d noise was quite slow...
16:57 hmmmm       he clearly hasn't seen our 3d noise implementation
16:57 hmmmm       heheh
16:57 sweetbomber the interpolation was about height, i think
16:57 sweetbomber wow, you implement 3d noise? nice!
16:58 hmmmm       huh?
16:58 celeron55   well, MT's noise implementation was very slow, but then hmmmm came and optimized it and now it's essentially free for non-crazy things
16:58 hmmmm       of course we did, how do you think did what we do?
16:58 celeron55   (MT has always had 2d and 3d noise)
16:58 hmmmm       yeah, previously minetest only had point-value noise, but i optimized it for batches
16:58 sweetbomber i will study that code
16:59 hmmmm       you still use point-value noise, obviously, use your own judgement to tell what's better where
16:59 hmmmm       your own judgement + profile it
17:00 sweetbomber in the long-term, do you think biomes should be hard-coded, or configurable, e.g. in a xml?
17:00 Exio        the biomedef thingy
17:00 hmmmm       they're already there in lua
17:00 hmmmm       did you miss biomedef? :/
17:00 Exio        is already here :P
17:01 sweetbomber oh, i only saw the default biome defined in code, and the other ones commented out
17:01 hmmmm       i don't think i have anything commented out... hmm?
17:02 hmmmm       what are you looking at?? surely not the current head
17:03 sweetbomber the createBiome function
17:03 hmmmm       in scriptapi?
17:04 hmmmm       oh
17:04 sweetbomber biome.cpp
17:04 hmmmm       that's a factory function, nothing real goes on in there
17:05 hmmmm       see, mapgen v7 was something i wanted to do right away, but it got pushed aside for other more pressing issues, and i just re-started working on v7 about 2 months ago and i changed basically everything
17:05 hmmmm       the past month or so i've been busy with finals, graduation, job seeking, etc.
17:05 hmmmm       i really am still active though, i promise
17:06 * Exio      lynches hmmmm
17:06 Exio        ;P
17:06 hmmmm       hey
17:07 * celeron55 mentions that's largely what kahrl said before he left for a full year some 14 months ago
17:07 hmmmm       sweetbomber, i'm sorry, don't let createBiome() throw you off, the original intent of that was to have each biome "class" to have its own generateColumn() method (at that time i actually, foolishly thought that i'd be able to do terrain generation decently a column at a time with a single pass)
17:07 celeron55   8)
17:07 hmmmm       s/foolishly/arrogantly/
17:08 celeron55   terrain generation is tough shit and notch was a god for a reason!
17:09 hmmmm       about that
17:09 celeron55   then they broke MC and now even we have a chance
17:09 celeron55   8)
17:09 hmmmm       i am going to toy around more with positive 3d noise
17:09 hmmmm       (but i honestly feel like it isn't completely necessary at this point)
17:10 sweetbomber ok
17:10 hmmmm       mapgen v7 won't be ready until it's been completed and extensively evaluated by others, and tuned, etc.  and even then it won't be the default for even longer
17:11 hmmmm       you can't help but look at projects that just change things for the hell of it and see how they become hated (like gnome, firefox)
17:11 hmmmm       but everything's okay if there's an option
17:12 celeron55   MT has had some things changed for the hell of it already
17:13 celeron55   not in the engine though, unless Lua is counted (it *was* done for the hell of it, largely)
17:14 hmmmm       in the end, lua was worth it, it transformed minetest into a minecraft clone into an increasingly-generic voxel engine, which makes it something special
17:45 hmmmm       from a minecraft clone*