Minetest logo

IRC log for #minetest-dev, 2013-12-07

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

All times shown according to UTC.

Time Nick Message
01:02 smoke_fumus joined #minetest-dev
01:36 VanessaE hmmmm: https://github.com/minetest/minetest/blob/master/src/game.cpp#L824  <--- when you pointed this out, no one was sure what you meant... but now RBA and I think this may be why cheapie can't use shaders anymore.  Might possibly explain Sokomine's issue also.  Care to fix it?
01:37 VanessaE it was not intentional.
01:39 VanessaE also, can someone explain to me why the same structure isn't being used for lines 845-853?
01:39 VanessaE (namely wrapping the code in {} )
01:40 VanessaE and for that matter everywhere else in this general section where shader stuff is involved.
01:43 VanessaE actually forget that last part.  I'm just having a bit of trouble understanding the code.  but that semicolon has to go.
02:07 sapier left #minetest-dev
02:11 NakedFury joined #minetest-dev
02:55 bas080 joined #minetest-dev
02:58 hmmmm ?
02:58 hmmmm I went over that with RBA in PM
03:00 VanessaE no matter, it took the two of us to figure out what you meant anyway.
03:00 VanessaE it did change things slightly, btw:
03:01 hmmmm I thought it was pretty obvious myself
03:01 VanessaE [12-06 21:49] <cheapie> Now [with the semicolon removed], if I have none of the middle column options on, there are still no dots. If I turn on Mipmapping, they appear if I enter the game, return to the main menu, and then enter the game again. With any texture filtering on, they appear any time I go into the game.
03:01 VanessaE [12-06 21:51] <cheapie@VE-S> Before, any mipmapping or texture filtering would cause it to display the dots every time, and they wouldn't go away until I restarted the client.
03:02 VanessaE by "dots" he refers to an effect similar to old film or light tv static, always black.
03:02 VanessaE before, shaders worked fine for him.
03:11 ShadowNinja Whew, tcp_connection is rebased now. Now to see if it works...
03:15 ShadowNinja :-( UDP works, but when connecting with a client with TCP support blocks don't load and commands are ignored.
03:19 ShadowNinja Pushed to tcp_connection on my fork. This is a bit beyond my knowledge.
03:21 ShadowNinja I'll push that semicolon removal in a few minutes if there are no objections.
03:24 ShadowNinja Um, so apparently shaders lower the water leven below the top of the node now again.
03:24 ShadowNinja laval*
03:24 ShadowNinja RealBadAngel: ^ Why?
03:24 ShadowNinja level*
03:39 VanessaE https://github.com/minetest/minetest_game/pull/226  <--- this seems to fix the player texturing/model bug for me
03:41 ShadowNinja If there's anyone to approve it I'll merge it.
03:47 ShadowNinja hmmmm: ^ Tested and works.
03:51 hmmmm sure
03:51 hmmmm sure suure suure
03:52 hmmmm I'd like to clarify something about the "2 core dev approval" guideline
03:53 hmmmm that's just a rule of thumb and it's only actually helpful if a). the people doing the approving understand the problem and have read the code and b). are actually knowledgable about the subsystem
03:53 hmmmm like for example that change could trigger a nuclear detonation for all I know
03:54 hmmmm the code itself looks perfectly fine but I have no idea what it does in the context of the rest of the model loading subsystem
04:00 ShadowNinja Alright. /me pushes
04:05 Miner_48er joined #minetest-dev
04:09 ShadowNinja celeron55: I rebased your tcp_connection branch, however TCP connections stopped working at "Setup a TCP connection in Connection" or "Enable automatic usage of TCP". Do you think you could fix it? The rebase is in the tcp_connection branch on my fork.
04:10 hmmmm the whole reason why it never made it into minetest was because it was broken
04:10 iqualfragile ShadowNinja: its like 5:11 around here and celeron lives in europe, too
04:10 ShadowNinja (TCP client doesn't load things on a TCP server, but UDP clients work fine)
04:10 iqualfragile good night
04:10 ShadowNinja iqualfragile: I expect him to respond when he comes back.
04:21 OldCoder joined #minetest-dev
04:25 us_0gb joined #minetest-dev
04:42 bas080 joined #minetest-dev
05:31 bas080 joined #minetest-dev
05:34 RealBadAngel joined #minetest-dev
05:44 VanessaE joined #minetest-dev
06:04 kahrl ShadowNinja: I (rarely) seem to get a hang at "Connecting to server" when I start a singleplayer game under valgrind
06:04 kahrl I started lots of games without valgrind, no hang
06:05 kahrl maybe it was just exceedingly slow for some random reason
06:16 kahrl ShadowNinja: I tried starting a server from the menu and was surprised that it doesn't use bind_address. Is that intentional?
06:20 kahrl do we still need ipv6_server btw, could "bind_address = ::" be used in place of "ipv6_server = true"?
06:25 kahrl I mean, from the user's perspective you get a weird error "ERROR[Connection]: Bind failed: Socket and bind address families do not match" if you mismatch those settings
06:27 VanessaE another common failure a LOT of people have been reporting is "Access violation"
06:27 VanessaE unrelated, I'm sure
06:32 VanessaE joined #minetest-dev
07:03 Zeitgeist_ joined #minetest-dev
07:40 VanessaE RealBadAngel: does the automatic fps tuning code take the added load caused by shaders into account?
07:41 VanessaE it doesn't seem to (I get lower fps but more view range than what I should)
07:44 celeron55 ShadowNinja: i don't have anything to add to hmmmm's comment
07:44 celeron55 ShadowNinja: it's just utterly broken
07:50 emptty joined #minetest-dev
08:09 Zeitgeist_ joined #minetest-dev
08:19 darkrose joined #minetest-dev
08:40 bas080 joined #minetest-dev
08:47 emptty joined #minetest-dev
09:03 Akien joined #minetest-dev
09:16 john_minetest joined #minetest-dev
09:29 RealBadAngel joined #minetest-dev
09:41 troller joined #minetest-dev
09:47 emptty joined #minetest-dev
10:07 RealBadAngel joined #minetest-dev
10:15 ImQ009 joined #minetest-dev
10:38 thexyz anyone remembers what was the issue when Windows displayed all blocks images as they were build on game start?
10:38 thexyz and how was this one fixed
10:40 thexyz oh
10:40 thexyz I thought it was because previously I was getting this stuff in my Windows builds and now I don't
10:41 thexyz hm..
10:41 thexyz it happens now on Android so it'd be cool to know what was the solution
10:41 thexyz although it's not critical at all
10:44 VanessaE it's irrlicht's render-to-texture code.
10:45 thexyz and is there any fix?
10:45 VanessaE it doesn't bother to render it off-screen in some circumstances
10:45 thexyz ah
10:45 VanessaE no clue why though
10:48 thexyz the problem is that at the same time progress bar updates
10:48 thexyz so it jumps between both
10:48 thexyz which.. doesn't look very nice
10:52 sfan5 john_minetest: assuming the left-bottom corner always has the sky color as bg that could work
10:59 Jordach joined #minetest-dev
11:11 nore joined #minetest-dev
11:16 ImQ009 joined #minetest-dev
11:30 VanessaE a feature has been off-handedly requested...  public key authentication for client<->server communication similar to Quake, in lieu of passwords
11:39 sapier joined #minetest-dev
11:40 RealBadAngel joined #minetest-dev
11:52 RealBadAngel_ joined #minetest-dev
12:32 werwerwer joined #minetest-dev
12:44 iqualfragile joined #minetest-dev
13:12 hmmmm joined #minetest-dev
13:16 jin_xi joined #minetest-dev
13:57 kahrl that's a good idea, although it would mean it would rain through glass
13:58 kahrl a node is either sunlight_propagates or not sunlight_propagates, no 99% in that :P
13:59 iqualfragile is there some path_free method?
13:59 hmmmm isn't there an airlike node type?
13:59 iqualfragile no, forget this question
14:02 kahrl light is good because it can be checked locally
14:02 kahrl anything that requires iterating through a bunch of nodes is a performance killer
14:03 hmmmm so when are we getting hardware lighting
14:03 hmmmm oh my god minetest could become 500x better if only we hard hardware lighting
14:03 hmmmm i almost might do it myself if nobody else is willing
14:03 EvergreenTree joined #minetest-dev
14:04 iqualfragile hmmmm: wasnt there a pull request?
14:04 hmmmm i do not know
14:04 iqualfragile but: yes, hardware lighting would be awsome, especically as it would enable bumpmaps to actually look good
14:04 kahrl iqualfragile: I don't think so, maybe you're thinking of wield_light?
14:05 hmmmm lol, that's it?  you could do way more
14:05 iqualfragile kahrl: no, i am going to search for it
14:06 hmmmm like lighting would look way better, be relatively glitchless, remove a huge burden on the CPU, allow us to do really cool things like colored lighting, arbitrarily stretch the range of lighting
14:06 hmmmm make maps smaller or give more options
14:06 hmmmm for nodes
14:06 iqualfragile hmmmm: but it would need to be done the right way
14:06 iqualfragile if you do it wrong you might get a high complexity if you have multiple lightsources
14:23 NakedFury joined #minetest-dev
14:48 Akien joined #minetest-dev
15:06 zat joined #minetest-dev
15:07 thexyz sapier: l_get_texturepath uses porting::path_user but according to CMakeLists.txt textures/base/pack installs to ${SHAREDIR}; is this correct?
15:18 nore thexyz, IIRC l_get_texturepath is for texture packs, etc
15:18 nore so it should be in the user path
15:25 thexyz but then the menu uses it
15:25 thexyz to display images from textures/base/pack
15:26 thexyz maybe I just don't understand how this stuff works
15:28 nore ... dunno what it is used for then...
15:28 nore IIRC I coded it for the texture pack thing
15:29 nore and it has been used for other things later...
15:46 sfan5 who would merge #958 in it's current state?
15:46 sfan5 s/merge/agree to a merge of/
15:47 nore sfan5, does it work now?
15:48 sfan5 the one without lua-customizability works
15:48 sfan5 that is what #958 is supposed to add
15:48 nore but the lua one? have you found where the bug came from?
15:49 sfan5 http://pastie.org/8503458
15:49 sfan5 that is the backtrace
15:50 hmmmm eh, #958 could use a LOT more polish
15:50 thexyz i've been playing with the Lua one for some time
15:50 hmmmm I wouldn't approve of it right now
15:50 sfan5 hmmmm: what should I change?
15:51 hmmmm well for one, white text on a white translucent background is a BAD idea.
15:51 hmmmm I don't know why you thought it'd be fine
15:51 sfan5 that's grey
15:51 thexyz lol
15:51 hmmmm sure doesn't look it
15:51 hmmmm make it black translucent
15:51 thexyz just make it black
15:51 thexyz with the same alpha
15:51 sfan5 hm.. maybe I should've used dark grey
15:51 hmmmm black
15:52 sfan5 ok, ok
15:52 hmmmm second, it should appear more or less in the middle of the screen
15:52 hmmmm third, it should say "this is the list of currently connected players"
15:52 sfan5 that is the middle of the screen
15:52 hmmmm change #3 brings to mind a long running problem with minetest that ought to be addressed first I believe
15:52 hmmmm sfan5, vertically middle.
15:53 hmmmm we should have another font that's larger
15:53 hmmmm like... a lot larger
15:53 hmmmm as it is right now, the fact that we can only draw a single size of the font on the screen is absolutely nonsensical
15:54 hmmmm I understand that this is more or less an irrlicht problem but the least we can do is try to work around it by including a small, medium, and large variation
15:55 hmmmm also I can't help but feel that there should be more information on the player list than the player name, like for example amount of time logged on, or ping, or whatever
15:55 hmmmm john_minetest, we're waiting on that for the culling to get fixed
15:56 nore can someone explain me why all textures are black with the new shaders?
15:56 sfan5 video::SColor(128,255,255,255) is black right?
15:56 werwerwer joined #minetest-dev
15:56 hmmmm .....
15:56 hmmmm no.  no, it's not.
15:57 nore john_minetest, it worked well before...
15:57 sfan5 hmmmm: shall I use solid black?
15:57 thexyz sfan5: 0,0,0
15:57 thexyz is black
15:57 sfan5 oh
15:57 sfan5 ..
15:57 thexyz 255,255,255 is white
15:57 sfan5 I'm stupid
15:57 hmmmm why don't you make the alpha something more like 190
15:57 hmmmm try that
15:58 sfan5 note to self: why did I use white?
16:00 hmmmm sfan5, could you use console_color and console_alpha instead?
16:00 sfan5 yes
16:01 hmmmm thank you
16:03 emptty1 joined #minetest-dev
16:06 hmmmm oh no what the hell
16:06 hmmmm i just looked at the actual code
16:06 hmmmm you make an entirely new packet for this... seriously?
16:06 hmmmm can't you integrate this with the hud?
16:07 sfan5 sorry, didn't think of using the hud packet
16:07 hmmmm alright let's go over this
16:08 bas080 joined #minetest-dev
16:08 AllegedlyDead joined #minetest-dev
16:08 celeron55 joined #minetest-dev
16:08 hmmmm player holds down tab key, if the IGUIListBox that contains the playerlist is there, you remove it
16:09 hmmmm oh if it's not key down
16:09 hmmmm alright that makes sense then, nevermind, I didn't see the !
16:10 hmmmm so the point of the packet is to set a custom formspec for the player list
16:11 sfan5 yeah
16:11 hmmmm that's kinda heavyweight to set a single string, don't you think?
16:11 sfan5 yes
16:13 hmmmm alright so this is different from a normal dynamic HUD element because:
16:13 hmmmm 1). the client initiates this HUD to open by pressing the tab key
16:14 hmmmm and 2). the string contents are predefined, dynamic, but known always by the client
16:17 hmmmm so we turn the packet into a TOCLIENT_HUD_SET_PARAM
16:18 hmmmm wait, did I write that code?  I don't remember writing that code
16:18 hmmmm anyway make it HUD_PARAM_PLAYERLIST_FORMSPEC
16:19 EvergreenTree joined #minetest-dev
16:19 hmmmm actually would you rather have me write it exactly the way I envision it instead of having you guess
17:12 sfan5 why do we transfer the hashes in base64 instead of raw?
17:14 Calinou joined #minetest-dev
17:18 ShadowNinja #1036 looks OK. Any comments? sapier?
17:24 sfan5 hmmmm: what am I doing wrong? https://github.com/sfan5/minetest/commit/3347d7c9c66b23ba9e5e4f2243959ee69b8e0598 ; the hud event is not happening
17:24 hmmmm no no no, don't make a new hud field....
17:25 hmmmm I just said send a TOCLIENT_HUD_SET_PARAM
17:25 sfan5 oh..
17:25 hmmmm you should read what I write
17:25 sfan5 I did
17:25 sfan5 but I forgot
17:26 hmmmm and the other part of this is to put the code where you draw the playerlist into hud.cpp instead
17:26 hmmmm are you sure you don't want me to do this?
17:27 sfan5 if you want to do it, go ahead
17:27 hmmmm well it's going to take a while, I'm working on something else
17:32 sapier thexyz paths are a mess in minetest I'm not even convinced there is a way to fullfill all requirements at same time ;-)
17:33 sapier 1036 is just cleanup if it's tested on win32 and some linuxes it's ready for merge
17:34 hmmmm wow holy crap, the dungeon code needs a crazy amount of cleanup
17:34 hmmmm when I made it into what it is now, I kinda left it alone as much as I could
17:34 sapier :-) if that one is still based on same thing celeron once released for lua it's quite ugly broken :-)
17:35 hmmmm I think he did, i dunno
17:35 sapier that code did spawn multiple vaults on top of each other ... up to about 6 or seven at amost or even exactly same position
17:35 hmmmm oh no that code you defined a dungeon by text file or something
17:36 sapier yes that one :-)
17:36 hmmmm if you wanted a mod that made nothing but procedurally generated dungeons I'd change singlenode mapgen to respect the MG_DUNGEONS flag and just set the node to brick or something
17:37 sapier I was wondering why mobf dm's kept despawning telling there already is one at exactly same position .. till I realized that vault code doing things like that
17:37 hmmmm that's because celeron coded that for fun, not to make it a work of art
17:37 hmmmm 90% of work is polish, or something like that.
17:38 rubenwardy joined #minetest-dev
17:38 sapier Maybe I remove it completely ... I'm about to split mobf right now, some of the features mobf includes are valuable for others too and are just to complex to not have a clean interface
17:39 sapier hmmmm don't we all code for fun? ;-)
17:39 hmmmm sapier, I'd like to closely collaborate with you on the dungeon stuff because I'm soon about to add that decoration spawner like I was saying
17:39 hmmmm like a half year ago but it never happened
17:40 sapier I don't do that much at dungeon code, I'm on advanced spawning mod right now
17:40 sapier it's meant to be a prototype for later minetest core implementation
17:40 sapier https://github.com/sapier/adv_spawning/blob/master/README.txt is the current (WIP) api
17:40 hmmmm hold off on that, putting mobs back in core might turn out to be useless when the client side lua starts
17:41 sapier no not mobs
17:41 Exio4 it is advanced spawning of all entities
17:41 sapier only spawning code
17:41 hmmmm well alright
17:41 hmmmm but it's just that I don't want this to collide with decorations
17:42 sapier I don't think so it's a completely different usecase
17:42 hmmmm decoration spawning was originally meant to be the same generic interface for spawning things
17:43 sapier is decorations meant to spawn continously?
17:43 hmmmm no, only at mapgen time
17:43 hmmmm it's just that so much of this looks like it's the same
17:43 sapier ok then there ain't much collision adv_spawning is for low frequency continous spawning
17:43 hmmmm except the light hting
17:44 sapier entities around isn't quite usefull at mapgen ;-)
17:44 hmmmm I know, that'll just spawn them once :/
17:44 kaeza joined #minetest-dev
17:45 sapier the mapgen spawning feature may collide to decorations ... but it's not exactly mapgen too
17:45 sapier it's more "spawn to a new generated map once there is time to do so"
17:45 hmmmm I figured it'd be a thing like you register some decoraiton callback and then in on_generate or whatever you can grab a list of locations within that recently generated chunk that are where the decorations were placed that you were interested in
17:46 hmmmm so like say if you were interested in all tree decorations that were spawned, you'd do something like list_of_v3s16s = mapgen:get_generated_decorations("regular_trees")
17:46 sapier that doesn't fix the main issue with spawning ... it's hard to predict how long a single environment check may take, doing multiples at once increases risc to add lag drasticaly
17:47 sapier but that feature is usefull to have a starting point for checks
17:47 hmmmm yeah
17:47 hmmmm and that's it
17:48 hmmmm see that was intended for things like chests, and then if you need to run Lua code to set metadata on that like chest contents
17:48 sapier so I could avoid to check random positions and use those positions instead
17:48 hmmmm no, this still checks random positions
17:48 hmmmm and only at map generation time
17:49 hmmmm if you want to do this at runtime after generation continuously, then you're right, it has an entirely different usecase
17:49 sapier ok I guess I have to wait till it's finished and see if I can use things to improve adv_spawning
17:49 hmmmm right
17:50 sapier I guess I need to add some functions to core for this to become really performant too .. e.g. light checking is  a nightmare, you need to fetch light value for each node separate
17:51 sapier maybe vmanip could be improved to contain light values per nodes too
17:51 hmmmm you could use a vmanip
17:51 hmmmm what?  it does that
17:51 sapier if I understood api correct I can update light add light and everything but I didn't see a way to just read it
17:51 hmmmm you're talking about LuaVoxelManip right?
17:52 sapier yes
17:52 hmmmm yeah
17:52 hmmmm VoxelManip:get_light_data()
17:52 hmmmm it returns a flat array of all the param1 values just like get_data
17:52 hmmmm except get_data returns the param0 values
17:53 nore hmmmm, and is there a param2 one?
17:53 hmmmm nope
17:53 sapier so I could possibly write a VoxelManip:get_light_at(pos,time) ?
17:53 hmmmm what's the time thing about?
17:54 sapier I use this to spawn mobs in areas that have day/night light cycle
17:54 hmmmm why not just compute that yourself
17:55 hmmmm you know, blending the light banks
17:55 sapier that's what I meant with I could write a function
17:56 hmmmm don't add an unnecessary VoxelManip api
17:56 sapier I just didn't know all required information is there?
17:56 hmmmm it returns the literal param1 value
17:56 sapier I don't consider functions that make use possible unnecessary ;-)
17:57 sapier but I won't add it to core of course that's a thing for builtin
17:57 hmmmm oh yes
18:00 hmmmm found it in light.h
18:01 hmmmm blended_daynight = math.min((factor * light_day + (1000 - factor) * light_night) / 1000, LIGHT_SUN)
18:01 Jordach joined #minetest-dev
18:02 hmmmm where light_day = light_vals[i] % 16; light_night = light_vals[i] / 16; of course
18:02 sapier wow :-) there wasn't a way to do that more complex was it?
18:02 hmmmm or is LIGHT_SUN not defined in lua?
18:02 sapier light max is defined
18:02 hmmmm then LIGHT_MAX - 1
18:03 sapier light sun is LIGHT_MAX+1
18:03 hmmmm oh
18:03 sapier yes max isn't exactly maximum :-)
18:03 hmmmm i keep confusing the two
18:04 hmmmm oh, the factor value (between 0 and 1000) is more complicated than I thought
18:04 hmmmm that's in daynightratio.h
18:05 sapier I think I'll find out that's performance tuning first features need to work
18:06 hmmmm it's pretty poorly written to be honest, you could do a better job.  it's actually not complicated if you look at what it does
18:06 PilzAdam joined #minetest-dev
18:06 hmmmm it just looks bad
18:08 sapier yes mod handling itself needs cleanup too
18:08 sapier that's more important
18:09 sapier I need a clean interface to use it from modstore too I don't want to implement dependency handling in core and lua
18:14 celeron55 there are certain things to take note of regarding to light
18:14 celeron55 the thing is, light can't be handled simply linearly
18:14 celeron55 when using it as a multiplier for color values to create brightness
18:15 celeron55 that's why there is a lookup table for converting stored light values to a multipliable brightness value
18:16 celeron55 then... ehm... i don't even remember everything i probably should say
18:17 sapier Isn't there a proper documentation? *smile*
18:17 celeron55 i'm not sure if "daylight factor" and "day-night ratio" are a different thing; some things are simply assigned to another directly or using a linear multiplier as hacks to make things work somehow
18:18 celeron55 sapier: of course not, i wrote this stuff largely before i or anyone even considered that someone might want to contribute something
18:18 celeron55 8)
18:20 celeron55 john_minetest: extremely hard to choose, and long-time players will hate it; it would be welcoming to new players though
18:21 PilzAdam we should definetly support games to play sounds, currently you cant have large sound files
18:23 Jordach some music would be good, but it should only be used if a player has got something like a jukebox (for atmosphere, see the mc matmos mod)
18:26 thexyz since the channel seems to be more alive now, I'll ask my question one more time: http://irc.minetest.ru/minetest-dev/2013-12-07#i_3474795
18:27 PilzAdam it doesnt happen if you disable preload item visuals
18:27 PilzAdam that was the "fix" for it IIRC
18:28 PilzAdam john_minetest, the plan for minetest_game is to not touch it and let the community develop better games
18:31 thexyz PilzAdam: alright, thanks
18:33 PilzAdam john_minetest, IIRC thats in kahrl's formspec_table branch
18:45 emptty joined #minetest-dev
18:48 hmmmm please leave out music
18:48 hmmmm music could be done far better by an external media player and it'd only serve to annoy people
18:50 PilzAdam the problem with an external player is that it cant depend on the gameplay; e.g. playing different music in caves than when the sun rises
18:52 hmmmm i'm thinking of how to store DungeonParams, hrmm
18:52 hmmmm not saving it with map_meta.txt because it won't cause an inconsistency if it were modified
18:52 hmmmm on the other hand what about biomes
18:56 OldCoder joined #minetest-dev
19:06 ImQ009 joined #minetest-dev
19:17 Gethiox2 joined #minetest-dev
19:25 hmmmm huh?  we're not synchronizing with freeminer at all
19:26 hmmmm besides, this doesn't modify mapgen v7 itself at all, only biome.cpp
19:28 hmmmm I am the only person working on v7
19:30 hmmmm writing biome definitions and games that utilize it
19:31 Exio4 yes, it was him
19:34 hmmmm probably not
19:34 hmmmm minetest simply wasn't built for performance, and at this point there's poor design baked right into the whole thing
19:35 thexyz i think we had about ~150 (bots mostly)
19:35 thexyz with proller's patches
19:36 hmmmm I suppose we can do something like those patches except in a better manner
19:36 AllegedlyDead joined #minetest-dev
19:36 hmmmm ..
19:37 hmmmm gotcha
19:37 sapier does anyone have an idea about the chat segfault?
19:37 thexyz http://i.imgur.com/yus2klV.jpg
19:37 hmmmm why is it that nobody turns off that ugly biome blend
19:38 thexyz uh?
19:38 PilzAdam hmmmm, people also enable trilinear filtering on 16px textures...
19:42 thexyz sapier: was that clang address sanitizer issue fixed?
19:42 kaeza is that 4chan server? :P
19:42 thexyz where it reports an error every time you write anything to chat
19:44 sapier I don't know thexyz
19:45 sapier I guess that one might be the one I was talking about
19:45 thexyz well then if it wasn't then someone could investigate it
19:55 sapier randomly on chatting
19:58 pitriss Guys about music ingame.. I think client need some improvements prior making ingame music.. for example some way how can be music volume adjusted, ditto with ambient sounds.. I'm listening my own music while playing MT and some Ingame screams/music will be just pain..
20:00 us_0gb joined #minetest-dev
20:11 * Sokomine still stares at black textures when shaders are activated
20:12 Calinou joined #minetest-dev
20:16 Sokomine btw, what do you think of a client-side map that just shows everything on eye level of the player (only chunks the client knows about)?
20:16 Sokomine i'm sure you've thought about it already
20:22 Sokomine trouble is that server maps are pretty large. they might easily exceed even the cache size of vanessas servers
20:22 diemartin joined #minetest-dev
20:22 diemartin joined #minetest-dev
20:24 hmmmm hmmmm
20:24 Sokomine so my suggestion was to let the client do it. provided the client has the mapdata it received still at hand. if up/downward is limited to a few blocks, and x/z does not grow too large, it ought to be doable i think
20:24 hmmmm minetest is now freezing up on me when I go to start a game
20:24 hmmmm when I break in a debugger it seems that it's stuck in kevent() all the way up in Curl_getaddrinfo_ex()
20:24 Sokomine doesn't sound good, hmmm
20:28 Exio4 clang problems?
20:28 Exio4 how did you try?
20:29 hmmmm is there a way to request that lua threads get shut down, and then don't wait for them to join before starting the game?
20:29 hmmmm then you set a flag or something when it gets done saying "whoops, we don't need your stuff any more"
20:29 hmmmm that's really about the only decent way to fix the problem i just mentioned
20:30 hmmmm (it would also fix the minor freezup when you press play)
20:46 emptty joined #minetest-dev
20:46 thexyz no, there are no clang problems
20:46 thexyz what I was talking about is clang build with -fsanitize=address
20:46 thexyz which is like valgrind but the game is actually playable
20:46 thexyz (typical slowdown is 2 times IIRC)
20:47 thexyz er, minetest build with clang with -fsanitize=address
20:47 Sokomine regarding my shader problem: realbadangel said my gpu might be too bad for it. which may be true (it's pretty low end)
20:48 thexyz Sokomine: were shaders working before?
20:49 hmmmm i hate emoticons
20:49 thexyz eh?
20:52 Calinou Sokomine, you're on linux?
20:54 Jordach I csll serious bull on that
20:55 Jordach OpenGL 2.1.2 works fine for me...
20:57 nore thexyz, I had shaders working before
20:58 nore and now same problem, all black
20:58 kaeza ^ same
21:00 ecube joined #minetest-dev
21:02 thexyz does disabling individual shaders help? like waving thingies
21:02 thexyz although I'm not sure what has RBA done to them
21:02 nore thexyz, no, is doesn't: it dosn't work even with all boxes unchecked
21:03 kaeza dat feeling when a single addition+deletion (semicolon fix) causes 30 source files to be recompiled ._.
21:03 nore however, the waving leaves/plants/water works well, even if the textures are all black
21:03 thexyz hm.. I guess you should ping him: RealBadAngel_ and ask to fix this?
21:03 Sokomine thexyz: they where; but rba said previous shaders did much less than the new ones
21:03 Sokomine calinou: yes. debian wheezy
21:04 Calinou Sokomine, shaders pretty much need proprietary drivers, blame RBA
21:04 Calinou they would anyway be quite slow on your IGP
21:04 sapier1 joined #minetest-dev
21:04 thexyz Calinou: but it's a regression anyway
21:05 Exio4 so what? it worked before
21:05 thexyz if some of them worked fine, no?
21:05 Calinou due to shaders rework?
21:05 hmmmm dammit sapier
21:06 Sokomine the old shaders where extremly slow on my antique nvidia onboard card. they worked fine on this one. but really complex gpu stuff might be too much for this admittedly simple "gpu"
21:07 Exio4 i still don't get why they shouldn't work
21:07 Exio4 Sokomine: do you have logs w/ shaders + --verbose or so?
21:08 hmmmm you see this?  https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_async_events.cpp#L57  this causes minetest to freeze up if one of the threads are too busy waiting to time out while you want to just play the game
21:09 ShadowNinja I moved script_run_callbacks to Lua. Comments? https://gist.github.com/ShadowNinja/7848732
21:11 Sokomine exio: i put the logs on pastebin and showed them to rba
21:12 nore ok, I did some tests
21:13 blaaaaargh joined #minetest-dev
21:13 nore deleting everything in the solid shader that should be run only with bump/parallax mapping makes it work again
21:13 nore so the problem is there...
21:14 nore wouldn't it be best to compile different shaders then? perhaps using some #ifdef, etc (I don't know if that can be used in shaders, though)
21:17 nore it works for liquids and leaves/plant waving shaders too
21:20 Miner_48er joined #minetest-dev
21:21 Sokomine nore: if that's possible it would be fine. there are those few points i mentioned where shaders are desirable, while real complex shaders might be too much for the hw
21:22 nore for me, they work well now
21:28 PilzAdam nore, Sokomine, have you this commit compiled in? https://github.com/minetest/minetest/commit/96f753a108600619f94015bf889a73d97e44ce3a
21:28 nore PilzAdam, yes
21:29 nore and as I said, deleting the bump/parallax mapping related code is enough to make the shaders work again
21:41 ShadowNinja PilzAdam: Can you check ^?
21:41 PilzAdam check what?
21:41 ShadowNinja PilzAdam: https://gist.github.com/ShadowNinja/7848732
21:44 Sokomine nor: thank you! your suggestion works. now the world has got color again!
21:46 Miner_48er joined #minetest-dev
21:47 nore ok, very strange: this is the line that makes the shader black: if ((enable_bumpmapping == 1.0) && (use_normalmap > 0.0)) {}
21:47 PilzAdam ShadowNinja, "fatal: corrupt patch at line 176"
21:47 Sokomine the trees wave at me now, and the water level height changes slightly up and down - so those fancy new effects do work!
21:49 hmmmm only geometry shaders work with the default texture pack
21:50 ShadowNinja PilzAdam: It isn't designed to apply.
21:50 PilzAdam ShadowNinja, seems to work
21:51 nore hmmmm, I mean: replacing the test in the shader by the bit of code that would be executed if the test would fail works
21:51 PilzAdam ShadowNinja, should I check all callbacks?
21:51 nore but adding the test, even if the test is _empty_, makes it black again
21:51 ShadowNinja PilzAdam: If you have some time. :-) It should work identically.
21:53 Zeg9 joined #minetest-dev
21:58 thexyz nore: can you replace those == 1.0 with > 0.5 and try it?
21:58 thexyz wait, no
21:59 thexyz well, anyway, i don't think this is a good way to compare floats
21:59 ShadowNinja PilzAdam: I tweaked it a bit.
21:59 ShadowNinja http://ix.io/9i6
22:00 PilzAdam whats the point of moving it to Lua btw?
22:01 ShadowNinja PilzAdam: Fewer C++ -> Lua calls.
22:01 ShadowNinja (And simpler code)
22:03 sapier +1 for moving callback handling to lua
22:04 PandemoniuM joined #minetest-dev
22:49 ShadowNinja PilzAdam: So, is it good?
22:49 PilzAdam I guess so
22:59 OldCoder joined #minetest-dev
23:21 zat Could not find EKEY_CODE, using orig. X11 keycode instead: 25
23:21 zat ??????????
23:21 zat (I have set different keyboard layouts under XFCE)
23:32 john_minetest left #minetest-dev
23:34 damiel joined #minetest-dev
23:34 damiel joined #minetest-dev

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