Time  Nick      Message
00:09 VanessaE  screw it. wrote it out as an issue.  https://github.com/minetest/minetest/issues/1532
00:36 Zeno`     Hi. When I make a Debug release, TestSocket does an ipv6 socket test
00:36 Zeno`     however I have ipv6 disabled and therefore the test fails
00:37 Zeno`     So I have to #ifdef that code out. I wonder if there could be a way to make that test optional (on by default)
00:43 fu-fu     Zeno`: Did I do the test-code thing alright?
00:44 Zeno`     fu-fu, yeah. Unfortunately I'm not terribly familiar with that part of the engine. It looked like it should work
00:44 Zeno`     Maybe paste the link again so someone else might see it?
00:46 fu-fu     I'll throw it in a bug report I guess.
00:47 fu-fu     Aaand nvm
00:50 fu-fu     to thr forums.
07:14 sfan5     nore__: why would someone need that setting?
12:15 nore__    sfan5:; to provide other oregun parameters
12:16 nore__    s/gun/gen
15:33 Sokomine  hmmmm: is there a (good) way to find out which surface node will be the dominant one in one area? i mean those 80x80 mapchunk areas (or roughly that size). i want to create a more abstract map of the world
15:34 hmmmm     sokomine, perhaps you should just downscale a bitmap of each individual node on the surface map
15:38 Calinou   use mapper, resize the image down using no interpolation?
15:38 Sokomine  does not work in this context as i want that to be a on-the-fly map. perhaps there's really no general way other than iterating over the surface of each generated chunk (provided it's a "surface" one)
15:40 Sokomine  just thought that those perlin noise things in combination with the known algorithm of a particular mapgen might allow to create a generalized map without generating each chunk. but i don't really know much about that
15:47 Exio4     ?
15:54 Sokomine  exio4: i want a more abstract map. using one from a mapper and downscaling would be one way. what i'm after here right now is another approach. the question is: can the general structure of a map be predicted without calculating each mapchunk?
15:54 Exio4     abstract map? :P
15:55 Sokomine  yes. if you want to navigate, you usually don't use sattellite photos. maps are abstractions
15:57 Sokomine  coastlines would be particulary intresting i think. coastlines (perhaps also desert) combined with the current position of the player and a handful of points of intrest may be sufficient already. there's no need to have an accurate picture of each tree or cactus on the map
15:58 Sokomine  i'm not sure how the underlying mechanism of mapgens works, but it *seems* they use mathematical functions to determine height and biome type. those functions *might* be used more general to create a picture of how the map would look like where it ever fully explored (which would not fit on any storage anyway)
15:59 Exio4     oh
15:59 Exio4     like, biomes map somehow
15:59 Exio4     'biomes'
15:59 Sokomine  yes, like that
16:01 Sokomine  based on my very limited current knowledge of mapgens as such (extremly limited), i can't tell if something like that could be calculated easily. it just seems that mapgens use those mathematical functions anyway and that it might be possible to use them for what i have in mind (kind of creating a map with a way lower resolution)
16:03 Sokomine  a map of the entire world may not even be particulary helpful as most mapgens are rather small-scale and would not help much with navigation. still, small parts of such a map might help
16:08 hmmmm     that's not possible
16:10 hmmmm     map is generated by the result of several interacting systems that, while deterministic, cannot be predicted
16:14 Sokomine  what would happen if the "resolution" would be lowered? i.e. only every 4th or 16th node on a map beeing generated, but with x/z values from the current map, and the other nodes around them not computed? would those calculated be the same as if all where calculated?
16:15 Sokomine  perhaps i'll really have to look into this one day in order to ask more intelligent questions :-/
16:16 hmmmm     I just explained why
16:33 Exio4     ops/
16:33 Exio4     ?
18:12 paramat   Sokomine "perlin noise things in combination with the known algorithm of a particular mapgen" i think a simple coastline + biome map of the entire world can be done this way
18:16 Sokomine  ah! that would be great already. coastlines + player position(s) + poi + (probably) roads already help a lot. it does look strange if all the later things are put on an "empty" ground
18:18 Sokomine  such a map would be very helpful i think. i'm currently considering storing the dominant node (i.e. grass, water or whatever the biome has) for each 80x80 chunk and will try that out as a map. it has the additional benefit of showing which regions have been visited. i'm not sure how useful that will eventually be
18:18 paramat   i can reproduce mgv6 in simple form in a lua mapgen, then as you say reduce the resolution, perhaps by 16. because i'm no good at mapping i could generate a floating 1/16th scale map of the world with 1 node per pixel
18:19 paramat   then it would be up to you to convert that into a realtime HUD map
18:19 Sokomine  uff :-)
18:20 Sokomine  right now it's a formspec. villages are one big square with no internal texture other than that of the texture which happened to be my choice for that village's representation
18:21 paramat   yeah or formspec whatever. crazy idea just might get me inspired to work on it, you want a map of mg mod or mgv6?
18:21 Sokomine  it'll help a bit that it's (mostly) 2d
18:21 Sokomine  mg i can do myshelf i think. i've just counted the nodes at mapgen time by now. will store that and take a look at how it looks
18:22 Sokomine  but..such a map for mg would be very welcome as well :-) perhaps in a way that it can be adapted to most mapgens out there
18:22 sfan5     a map for mg?
18:23 paramat   also, such a method can generate a map for areas not visited yet
18:23 paramat   well i might stick to mgv6 only. 1 pixel per 80 nodes will lose a lot of detail, i might try 1 per 16, or variable
18:24 Sokomine  variable sounds very good
18:24 Sokomine  the question is also how fast that may be
18:25 paramat   i might be able to do this with perlinmaps, so fairly fast
18:25 sfan5     >i'm currently considering storing the dominant node (i.e. grass, water or whatever the biome has) for each 80x80 chunk and will try that out as a map.
18:25 sfan5     that sounds interesting
18:25 Sokomine  i tried gsmapper, but that basicly has the same problem as all mappers: its resolution is too fine and the area shown too small
18:26 paramat   i can't see the attraction of small local-area maps that just show you what you can already see :)
18:26 Sokomine  well, figuring out that mapnode type was easy - inside the chunk generation code of course. there i can count it. storing...is another matter. either i'll get too many write operations for my taste, or information about generated chunks will be lost
18:28 Sokomine  they have their use - to a degree. inside a village/complex of buildings, it helps to see the relation of the buildings. it just doesn't help on large-scale problems
18:29 Sokomine  don't want to know how many people lost their first house in mt or mc due to never finding back there :-) the home command in minetest_next definitely helps (once people know these commands)
18:55 gentoobro it would be interesting if there was a way to specify the depth of the map
18:56 gentoobro then you could have a map of the surface, as well as maps of your various mines
18:56 gentoobro perhaps the depth could optionally have a sliding window to account for variations in the mine depth
18:58 paramat   'depth' as in surface elevation? yes that could be found and converted to pixel brightness, i would have to calculate elevation anyway to get coastline
19:03 Sokomine  i don't know; for maps of mines, those small-scale mappers are perhaps best suited? to my knowledge, all those mapping attempts stick to the surface
19:07 paramat   ah i see mapping the depths, default cave generation is not defined by perlin noise so thats's not possible
19:09 gentoobro you'd have to call the map generation function from lua when desired since user-created mines are not in perlin noise either
19:35 Sokomine  what i suggested is not about player generated structure. it's for a background map (on which player generated structures might be shown independently)
22:06 hmmmm     sokomine:  no, there are still things such as caves and trees.
22:07 hmmmm     caves, especially in mapgen v6, play a very large role in the shape of the terrain
22:07 hmmmm     that's just one example of a procedurally generated layer that is deterministic but not predictable
22:11 cg72      what are the new mapgen settings in map_meta.txt for jungles as the old ones dont work now and keep getting kicked back out
22:11 hmmmm     sounds like a #minetest question.  please leave if you have nothing development related to ask.
22:12 cg72      leave well thats very rude as i login her alot
22:12 cg72      here*
22:14 cg72      just remember hmmmm, rudeness gets you nowhere in life!!!!!!!
22:20 proller   caves is deterministic ???
22:52 Sokomine  if there's a cave, does that mean that there is a cave (and, naturally, the hole it consists of), or is the landscape as such quite diffrent?
22:53 Sokomine  i thought caves where generated afterwards? would a landscape be the same (minus the caves) if generated with the same seed and caves disabled?
22:55 Sokomine  for now, i just count the surface nodes in nores mapgen and set the most dominant one for the entire 80x80 area. the loss in resolution is there, but as i'm displaying a 2000x2000 nodes area, it is not too dramatic
22:55 Sokomine  not perfect yet either
23:07 paramat   you could compare 2 worlds, one with cavegen disabled, same seed, there is a fair amount of difference
23:18 VanessaE  hmmmm: to be perfectly fair, cg72's question was relevant to here in that mapgen settings are poorly documented even from a dev standpoint, and not documented at all from a modder/end-user standpoint.  If a user has to read the engine code to understand a setting, the documentation has failed and therefore it falls to the devs to correct it.
23:19 VanessaE  and in fact I *think* I answered her question via the dev wiki some time back (before she knew she'd ask it, that is).
23:19 paramat   the new mapgen flags are documented in lua_api.txt
23:19 VanessaE  when was that added?
23:20 paramat   oops sorry in minetest.conf example
23:20 VanessaE  bad place to document them.
23:20 VanessaE  the .example file should not be considered documentation, as such
23:21 VanessaE  also, they're not documented in the .example so much as they're *listed* there with some very basic info
23:22 VanessaE  no explanation here about what a given setting does, or why
23:22 VanessaE  (with a couple of exceptions)
23:27 paramat   mg flags docs look fine to me :) clearly and simply explained
23:28 VanessaE  too simply
23:28 VanessaE  that's the issue
23:28 VanessaE  and where they're mentioned.
23:28 VanessaE  that stuff belongs somewhere other than the .example
23:28 Zeno`     As an example I didn't really know what 'mud' did until yesterday
23:29 cg72      rotf Zeno`
23:29 Zeno`     I didn't. I thought it must have been an unimplemented feature until you mentioned it and then I looked at the src code