Minetest logo

IRC log for #minetest-dev, 2013-03-20

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:05 VanessaE http://forum.minetest.net/viewtopic.php?pid=77032#p77032
00:05 VanessaE \o/
00:08 VanessaE I need to make a "dyeing" table at some point.
00:08 VanessaE something that can be used to color objects like unified bricks, colored woods, etc etc
00:08 VanessaE so I can get these four million nodes out of the inventory :-)
00:09 Exio and four millions seconds for loading the game
00:50 hmmmm erm what
00:51 hmmmm vanessa, the problem with the multi-threaded mapgen is that one thread generating a block doesn't know about that same block's present state when it's being used as a neighboring chunk in another thread, causing solid terrain to be generated on top of a cave in some cases, causing caves to get "walled in"
00:52 hmmmm ditto with trees and air
00:52 hmmmm causing trees to get sawed down the middle
00:52 hmmmm i have an idea of how to fix that completely, but it sucks
00:53 VanessaE ah, guess I was oversimplifying
00:53 VanessaE btw, the idea I just had: if overflow from one block to another is the issue, then give each thread every *other* block, never genrate two side by side in separate threads
00:54 hmmmm lol you can't do that, if you're thinking of a checkerboard pattern of generation you can't do that either, since chunks being generated would still share corners
00:54 VanessaE so if you have two threads generating a 3x3 area, the center one thread might go to each corner, then when those have finished, one thread to each of the top/bottom/left/right middle ones
00:54 VanessaE shit.
00:54 hmmmm yes i know what you mean
00:54 hmmmm i considered all of that already
00:55 VanessaE (why didn't I think to say "checkerboard" in the first place?)
00:55 VanessaE oh well
00:57 Exio hmmmm: i don't know how the values works for the mapgen_v6_noise and so, but is there any "way" to make it less predictable?
00:58 hmmmm not really
00:58 Exio hm k
00:58 Exio btw, this is what i got from my random-seed-experiment http://exio4.com.ar/_/mt/map.png
00:59 hmmmm you were expecting something different?
01:00 Exio yes, chunks more random, because chunksize=1
01:00 ecube joined #minetest-dev
01:00 Exio now that is fixed, no?
01:01 hmmmm huh?  what does chunksize have to do with random
01:01 Exio how is the size of the chunk defined?
01:01 hmmmm in the config
01:01 Exio i mean, how you know the 'chunks' will have that size?
01:01 hmmmm because that's what i generate
01:02 hmmmm it's a setting for a reason..
01:03 Exio so, if i replaced every thing where you can use 'seed' with 'rand()', why are the "edges of chunks" not 'very near'
01:03 Exio ?
01:04 hmmmm because of how perlin noise fundamentally works
01:04 Exio hm, how does it work in that way?
01:04 hmmmm the only reason why it's as coherent as it is with your experiment is because the seed doesn't change for the entire chunk
01:05 hmmmm you should probably read up on perlin noise
01:05 Exio i will
01:22 Exio i see what you meant with the mapgen
01:22 Exio the multithreaded stuff ate a whole desert stone wall.. in .. "nothing"
01:24 Exio and removed the ocean where i was swimming :(
03:17 loggingbot_ joined #minetest-dev
03:17 Topic for #minetest-dev is now Minetest core development and maintenance. Chit-chat goes to #minetest. Consider this instead of /msg celeron55. http://irc.minetest.ru/minetest-dev/ http://dev.minetest.net/
06:30 darkrose joined #minetest-dev
07:42 darkrose joined #minetest-dev
08:57 darkrose joined #minetest-dev
09:00 proller joined #minetest-dev
09:14 kaeza joined #minetest-dev
09:14 kaeza hi all
09:15 kaeza would it be possible to add support to control client's FOV from Lua?
09:16 kaeza I mean, for the server to control client's FOV
09:50 jin_xi joined #minetest-dev
09:55 Taoki joined #minetest-dev
10:23 jin_xi joined #minetest-dev
10:27 celeron55 joined #minetest-dev
10:31 serengeor joined #minetest-dev
10:36 jin_xi joined #minetest-dev
10:49 PilzAdam joined #minetest-dev
11:05 darkrose joined #minetest-dev
11:05 darkrose joined #minetest-dev
12:06 kaeza1 joined #minetest-dev
12:06 kaeza1 left #minetest-dev
12:20 Traxie21|Kindle joined #minetest-dev
12:20 Traxie21|Kindle Hello
12:44 Taoki joined #minetest-dev
13:17 hmmmm joined #minetest-dev
13:28 celeron55_ joined #minetest-dev
14:12 RealBadAngel hi all
14:25 PilzAdam joined #minetest-dev
14:30 RealBadAngel hmmmm, are you there?
15:06 iqualfragile joined #minetest-dev
15:09 serengeor joined #minetest-dev
16:08 Calinou joined #minetest-dev
16:45 Exio joined #minetest-dev
16:46 celeron55 joined #minetest-dev
17:02 hmmmmm joined #minetest-dev
17:02 hmmmmm y0
17:03 hmmmmm why should the server control the client's FOV?
17:03 hmmmmm 1).  i can't see any reason for that, and 2).  it'd add yet another network packet
17:10 celeron55 the answer to why is probably quite simple: to make binocular-like stuff
17:10 celeron55 which is useless-ish though
17:14 VanessaE I could see one use for it.
17:14 VanessaE well, skip that.
17:15 VanessaE (wouldn't be of much use in practice, not in MT)
17:22 serengeor #opengl
17:23 serengeor oops
17:28 thexyz binocular-like?
17:28 thexyz i think we can achieve the same with metology hud api
17:29 VanessaE I thought of quake/openarena's "zoom" (middle mouse) feature when he said that.
17:29 hmmmmm about that
17:30 hmmmmm the metology hud conflicts with my plans for the bar api
17:30 thexyz there's no point in doing that server-side then
17:30 thexyz hmmmmm: do we really need the bar api then?
17:30 hmmmmm yes
17:30 hmmmmm the bar api is much more flexible
17:31 thexyz how so?
17:31 hmmmmm read about it on the TODO dev wiki page
17:32 thexyz so why can't we implement a generic client hud api and then write a simple wrapper for such bar api?
17:34 hmmmmm look at how overcomplicated it is....
17:36 hmmmmm plus, didn't we want lua to not have that sort of power over clients, to draw arbitrary figures?
17:36 thexyz but it's already done
17:36 thexyz +statbar~<image>~<amount>~<X>~<Y>
17:36 thexyz +^ Draw a statbar.
17:36 thexyz +^ amount is in half-points.
17:37 thexyz hm..
17:37 thexyz well, I didn't want that
17:44 Calinou joined #minetest-dev
17:51 hmmmmm woah.  so i'm compiling minetest to play on my laptop, and it's taking forever compared to my desktop... my new computer has spoiled me
17:52 Exio what cpu in the laptop?
17:52 hmmmmm Sempron SI-42
17:53 VanessaE a 486 SX25 ;-)
17:53 Exio hmmmmm: time it!, it takes 30 minutes in my intel atom n270
17:56 Calinou about 20-25 mins on my netbook's N455 :P
17:56 Calinou 40 seconds to 60 seconds on my desktop
17:57 Calinou (make -j8)
17:57 Calinou no noticeable difference with make -j9
17:57 VanessaE Calinou: fix yer busted-ass circular saw ;)
18:00 Exio Calinou: with time?
18:00 hmmmmm omg
18:00 hmmmmm 717.065u 26.304s 13:38.68 90.7%7751+2947k 365+54io 33pf+0w
18:01 hmmmmm also it seems like it compiled the same stuff twice
18:01 VanessaE gah, posted that on the wrong channel, sorry.
18:27 proller joined #minetest-dev
18:32 PilzAdam I added some labels to the issue tracker of the engine
18:40 Exio :D thanks
19:02 proller 50+% external mobs crashes server at random time
19:03 proller maybe report lua error to log and continue working?
19:13 khonkhortisan joined #minetest-dev
20:10 sapier joined #minetest-dev
20:28 jin_xi joined #minetest-dev
20:33 jin_xi joined #minetest-dev
20:56 sapier1 joined #minetest-dev
21:02 celeron55 hmmmm: that is why everyone should seriously aim to keep header bloat as minimal as possible 8)
21:04 celeron55 and use interface classes for things, as those are very good at making sure a single change won't recompile half of things
21:05 celeron55 (well, that was already included in the first statement; *goes away*)
21:23 ecube joined #minetest-dev
21:28 ecube_ joined #minetest-dev
21:36 hmmmm but, at the cost of needing a vtable, adding another layer of indirection, which is less efficient
21:36 hmmmm at runtime
21:38 sapier1 vtable dereferencing is quite fast as far as I know only rtti is performance critical
21:38 iqualfragile1 joined #minetest-dev
21:39 sapier1 vtable usage might even be optimized by compiler
21:40 sapier1 btw I've just assembler debugged vtable usage for two days ;-) but of course compiler may not always be able to optimize that good
22:10 proller joined #minetest-dev
22:12 proller initial float islands - https://github.com/minetest/minetest/pull/555
22:37 iqualfragile joined #minetest-dev
22:43 thexyz Land Up, Canyon & Chasm ported in core would be great too
22:44 PilzAdam thexyz, maybe there is a bug caused by stl that breaks nick completion
22:44 PilzAdam need to research
22:44 sapier1 yes ... imho we need to make core more modular in order to improove maintenance for things like that
22:44 thexyz PilzAdam: seems so
22:45 VanessaE nick completion?  in minetest?
22:45 thexyz wow!
22:45 thexyz so, nothing is broken
22:45 thexyz as no one used it
22:46 Taoki joined #minetest-dev
22:46 PilzAdam it only works for my own nick
22:47 Guest88189 joined #minetest-dev
22:47 VanessaE what's the completion key?  tab?
22:47 thexyz PilzAdam: it doesn't work at e204bedf1d781e43b8caa334a99319efc5b7ce46 either
22:47 sapier1 wait you mean list of all logged in players is transfered to any client no matter if client needs to know or not?
22:47 VanessaE ah no wonder I never used it, it only works in the F10 screen, not in the 't' tialog.
22:47 VanessaE dialog*
22:48 thexyz sapier1: yes? /status ?
22:48 thexyz hm, compiling minetest while updating @world is not fun
22:48 thexyz should probably get hexacore
22:49 sapier1 puuhhh I don't like the idea someone can write a simple client creating a login profile of any player
22:49 Exio hahaha
22:49 Exio poor computer
22:49 Exio thexyz: what cpu atm?
22:50 thexyz Exio: i5-2400
22:50 thexyz sapier1: hmm?
22:50 thexyz i didn't get it
22:50 proller seems landup possible with mapgen params now - change scale
22:50 Exio ah
22:51 proller and already in indev mapgen far mountains higher
22:51 sapier1 I wasn't aware of minetest yelling out my presence to everyone ;-)
22:52 sapier1 in most mmos for example you can talk to everyone of course (at least if you know the name) but you only see online status of your friends
22:52 thexyz sapier1: well, it's quite obvious
22:52 thexyz I doubt minetest can be counted as MMO
22:52 thexyz PilzAdam: was this feature ever working?
22:52 PilzAdam dunno
22:53 sapier1 so minetest assumes all players in a game are friends ... hmm while this might be a possible interpretation I suggest thinking about changing this behaviour if minetest servers start to get more players ;)
23:27 proller PilzAdam, fixed initial size, coal+iron at minimal, minimum height
23:27 thexyz PilzAdam: so, the issue with tab completing is related to Environment
23:28 thexyz as in non-local game it only knows about local player
23:28 thexyz and tab completion code calls its getConnectedPlayers() method
23:28 thexyz which, obviously, returns only the local player
23:30 thexyz i wonder what's the right way to fix it
23:31 sapier1 :-) I suggest not fixing it at all but I assume this is only my opinion ;-)
23:31 thexyz why?
23:31 sapier1 for privacy reasons
23:32 thexyz what privacy reasons?
23:32 thexyz it's not like /status doesn't work anymore
23:32 thexyz so your argument is invalid
23:32 sapier1 ok then it's useless :-(
23:32 thexyz oh, also, today /status showed 2 "unknown" guys
23:32 thexyz what is useless?
23:33 sapier1 useless not to fix it if there's another way to get that information
23:33 thexyz yes, for example, you may as well look around to see other players' names
23:33 proller PilzAdam, commit please https://github.com/minetest/minetest/pull/555 - its in dev mapgen,  change in minimal game will not affect current mapgen, later will make better no-edge way
23:34 sapier1 yes but it's a difference if you see all players at once or get a well formated guaranteed to be complete list ;-)
23:34 PilzAdam proller, I would prefer that hmmmm looks over it first
23:34 sapier1 -or + and
23:34 proller PilzAdam, ok
23:35 proller hmmmm, https://github.com/minetest/minetest/pull/555/
23:45 thexyz hmmmm: http://forum.minetest.net/viewtopic.php?id=5188
23:45 thexyz (you may be interested i think)
23:48 PilzAdam celeron55, Id like to hear what you think: http://forum.minetest.net/viewtopic.php?pid=77359#p77359
23:57 hmmmm the only thing that could be computed with a GPU is the perlin noise, and the benefit would only be great enough if you're generating the entire world at once

| Channels | #minetest-dev index | Today | | Google Search | Plaintext