Minetest logo

IRC log for #minetest-dev, 2014-10-09

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

All times shown according to UTC.

Time Nick Message
00:19 hoborific joined #minetest-dev
00:40 hoborific how do you police pull requests
01:15 kahrl joined #minetest-dev
01:16 kahrl https://github.com/minetest/minetest/issues/1728 fixed by https://gist.github.com/kahrl/778860462bfb6e43243c
01:16 kahrl since the fix is so trivial I'll push it in an hour if nobody complains
01:20 ShadowNinja Seems fine, provided that the engine prepends PROJECT_NAME + minetest_version_hash + " / " to it.
01:20 kahrl yeah, it doe
01:20 kahrl s
01:21 kahrl (and skips the " / " part if the argument to set_topleft_text is empty)
01:31 kahrl can anybody explain https://github.com/minetest/minetest/issues/1723?
01:31 kahrl why would the game be in draw_load_screen while playing?
01:49 NakedFury joined #minetest-dev
01:49 ShadowNinja kahrl: He probably means "while connecting to play".
02:12 Miner_48er joined #minetest-dev
02:23 hoborific left #minetest-dev
02:24 Zeno` joined #minetest-dev
02:27 GrimKriegor joined #minetest-dev
02:37 Zeno` joined #minetest-dev
02:50 Hunterz joined #minetest-dev
03:44 Hunterz joined #minetest-dev
03:45 Hunterz1 joined #minetest-dev
04:03 MikeFair_ joined #minetest-dev
04:30 kaeza joined #minetest-dev
05:01 SmugLeaf joined #minetest-dev
05:01 SmugLeaf joined #minetest-dev
05:15 sol_invictus joined #minetest-dev
05:40 Hunterz joined #minetest-dev
06:20 GrimKriegor joined #minetest-dev
06:22 ImQ009 joined #minetest-dev
06:25 alexxs joined #minetest-dev
06:41 hintss joined #minetest-dev
07:13 Zeno` how does this project like commits squashed? e.g. https://github.com/minetest/minetest/pull/1707 ?
07:14 Zeno` git rebase -i HEAD~3 ?
07:18 kahrl yeah, would be best in this case
07:18 kahrl but generally don't squash if it would produce a commit that does multiple unrelated things
07:21 kahrl (but do squash if the second commit just fixes a variable name introduced / heavily changed by the first, etc.)
07:21 kahrl s/variable name/name of a variable
07:24 Zeno` so in this case I *should* squash but only HEAD~2 ?
07:25 Zeno` the bug fix is already merged... so maybe I'd be better to squash all I think
07:25 Zeno` sorry, I just saw you already said that :3
07:26 kahrl wait, which one is already merged?
07:26 kahrl oh, the "Fix use ..." one
07:27 kahrl don't squash with that, just remove it
07:27 kahrl otherwise git would try to apply that patch again and everyone would be crying
07:27 kilbith joined #minetest-dev
07:27 kilbith since today, I can't see the server list in the client
07:28 Zeno` 27eff75, but as another request. The mistake I made was fixing 27eff75 as a separate PR
07:28 Krock joined #minetest-dev
07:29 Zeno` I'll squash all 3 and see what the diff looks like
07:30 kahrl don't
07:30 kahrl Zeno`: in these cases, since I don't know git well, I just make a new branch, cherry-pick the (non-merged to upstream) commits from the first branch, delete the first branch, rename the second back to the first, and git push -f
07:30 Zeno` ok, so leave as is?
07:30 Zeno` or can I squash e5d1da2 into 934e1c2?
07:30 Zeno` even if there is a commit in between them
07:31 kahrl yeah, just change the order in rebase -i
07:31 proller joined #minetest-dev
07:31 Zeno` well I may as well try... won't learn any other way. Making backup heh
07:32 Zeno` thanks
07:33 kahrl you can also delete 27eff75 by removing that line in git rebase -i
07:33 Zeno` just read that comment in vim
07:33 kahrl but I never trust myself to not mess up by doing something like that so I go the cherry-pick way
07:35 * Zeno` does git rebase --abort... lol, that worked out well
07:35 Zeno` I will cherry pick... easier
07:44 Zeno` seemed to work. Thanks again, kahrl
07:50 sfan5_ joined #minetest-dev
08:03 darkrose joined #minetest-dev
08:07 sfan5_ joined #minetest-dev
08:29 kahrl https://github.com/minetest/minetest​/blob/master/src/drawscene.cpp#L524
08:29 kahrl why are the clouds rendered before beginScene?
08:30 kahrl how does that even work?
08:45 Krock and I wonder about this use: https://github.com/minetest/minet​est/blob/master/src/sky.cpp#L497
08:48 kahrl Krock: that seems fine to me
08:49 kahrl perhaps a bit hacky to use recursion like that but should be ok
08:49 Krock kahrl, just interesting why a flood is needed to solve a problem
08:50 kahrl the alternative would be to duplicate update() into an initialize function
08:51 kahrl which does everything update() does except the smoothing
08:51 Krock :/
08:55 kahrl just timed it, took <1ms to do the 100 updates
08:58 ninnghazad joined #minetest-dev
09:00 FR^2 joined #minetest-dev
09:11 RealBadAngel joined #minetest-dev
09:12 blaise joined #minetest-dev
09:13 Zeno` it's probably tail-end optimisation and/or unrolled with optimisation anyway
09:13 Zeno` optimised*
09:14 Zeno` I was actually looking at that same line the other day :-o
09:15 Zeno` when I was looking for an unititized variable... weird
09:23 ImQ009 joined #minetest-dev
10:00 ninnghazad is there a limit to size of log in memory? or can i blow the server's mem by producing large amounts of logging?
10:02 kahrl the log is written to disk, not memory
10:02 kahrl unless debug.txt happens to be on a tmpfs, I suppose (but why would you do that?)
10:03 ninnghazad nono, i was wondering if the buffer used for logging may not be cleared (even if written to disk)
10:03 kahrl any signs of that happening? would be strange
10:05 ninnghazad not really, but i have a memleak and am looking for causes. there isn't much happening but logging, so i thought i'd ask
10:05 kahrl tried valgrind?
10:07 ninnghazad not yet, thought i might be able to pinpoint it by selectively disabling stuff in lua that is running when it happens. i was under the assumption that the lua-env gets destroyed when leaving the game, but the mem stays high when back in menu. does the lua persist until that point?
10:07 kahrl I'm pretty sure it is destroyed
10:08 kahrl maybe someone missed some userdata objects or something though
10:12 jin_xi joined #minetest-dev
10:12 ninnghazad hmk, has to be something in lua then that leaves stuff around outside of it. valgrind dumps me with: /tmp/minetest/src/script/cpp_api/s_base.cpp:74: ScriptApiBase::ScriptApiBase(): Assertion 'm_luastack' failed.
10:12 Zeno` you need to disable LuaJIT
10:13 ninnghazad aah.. ok
10:13 jin_xi ninnghazad: is this with dronetests? because minetest leaks the textures for the screens
10:13 Zeno` valgrind and LuaJIT do not work nicely together unless you have a special build of LuaJIT
10:14 ninnghazad yes, i found it while coding for dronetest. hm, its a lot of screen-updates, maybe thats just it.
10:14 ninnghazad using 640x640 screens, so that may explain the speed of leaking.
10:16 kahrl aren't the textures unloaded when you exit to menu though?
10:20 ninnghazad hm, i don't think thats it, printing lots of text alone is not enough to make it leak.
10:21 ninnghazad i think it happens around get_objects_inside_radius and using those.
10:23 PenguinDad joined #minetest-dev
10:30 jin_xi i found collisionMoveSimple generates lots of spurious hits on node borders
10:30 jin_xi i wonder if its a problem for some mods as its used for all movement in mt basically.
10:34 Krock joined #minetest-dev
10:43 Zeno` oh man, I don't wannna make a branch for this :9
10:55 Zeno` Krock!!! Best pal and buddy
10:55 Krock I feel nothing good
10:57 Zeno` No, no, this is fantastic. Can you check this is fine on MSVC? https://github.com/Zeno-/m​inetest/tree/minor_cleanup
10:57 Zeno` It's terribly exciting
10:58 Zeno` I only did it for memcpy(new_str, intermediate.c_str(), len * sizeof (wchar_t) + 1);  in gettext.h :(
10:58 Krock Zeno`, you really want to wait 25 minutes just to (maybe) get an error message?
11:00 Zeno` probably not. It looks fine so I will trust it
11:01 Zeno` actually forget it altogether
11:09 sfan5_ joined #minetest-dev
11:14 SudoAptGetPlay joined #minetest-dev
11:14 proller joined #minetest-dev
11:17 proller joined #minetest-dev
11:51 ninnghazad hmk, i was confused. get_objects did not leak, must really be the textures or something related to that. when not updating dronetest_display's texture, memory stays cool.
11:55 speak joined #minetest-dev
12:02 mos_basik joined #minetest-dev
12:06 Zeno` umm, kahrl...
12:07 Zeno` https://github.com/minetest/minetest/commi​t/d1ccc64e1ecfd00df36d2b8a8b3420b70434fefd <-- Why is that necessary?
12:07 Zeno` It makes no difference, I'm just interested
12:07 hintss joined #minetest-dev
12:13 SudoAptGetPlay joined #minetest-dev
12:16 hax404_ joined #minetest-dev
12:17 Megaf sfan5: I will try again with the following flags, cmake ../../ -DVERSION_EXTRA=MinetestPi -DRUN_IN_PLACE=1  -DENABLE_GLES=1 -DIRRLICHT_INCLUDE_DIR=../../../irrlicht/include -DIRRLICHT_LIBRARY=../../../ir​rlicht/lib/Linux/libIrrlicht.a  -DENABLE_SOUND=0 -DENABLE_LEVELDB=0 -DENABLE_CURL=0 -DENABLE_REDIS=0 -DCMAKE_EXE_LINKER_FLAGS="-lGLESv1_CM -lEGL"
12:17 Megaf and I recompiled Irrlicht too
12:21 Amaz joined #minetest-dev
12:22 PilzAdam joined #minetest-dev
12:24 SudoAptGetPlay joined #minetest-dev
12:31 ImQ009 joined #minetest-dev
12:51 CraigyDavi`` joined #minetest-dev
13:01 shadowzone joined #minetest-dev
13:03 SudoAptGetPlay Hi, is there an easy way to link server's players credentials to an external mysql db ?
13:18 rickmcfarley joined #minetest-dev
13:21 sfan5 SudoAptGetPlay: you can generate an auth.txt from a mysql db, but linking? no
13:26 SudoAptGetPlay the usecase is : users register on a site to create an account, and those credentials are used by minetest for user login
13:37 sfan5 thats even easier
13:45 hmmmm joined #minetest-dev
13:47 rickmcfarley joined #minetest-dev
13:49 Salbei joined #minetest-dev
13:52 Salbei hi all
13:53 Salbei sfan5: can u tell me more about SudoAptGetPlay question?
13:54 sfan5 sure
13:54 Salbei I want to link users from a mysql db to minetestserver
13:55 Salbei how do that?
13:55 sfan5 mysql is a bit more complicated
13:55 sfan5 <SudoAptGetPlay> the usecase is : users register on a site to create an account, and those credentials are used by minetest for user login
13:55 sfan5 <sfan5> thats even easier
13:57 Salbei i have to write with php in the auth.txt right?
13:57 sfan5 correct
13:58 Salbei ok i can do that
13:58 Salbei I have an another question
14:00 Salbei what is the solution for put 2 serverlist_url in the minetest.conf?
14:00 proller joined #minetest-dev
14:00 sfan5 you can only put one url there
14:01 Salbei ok
14:05 sfan5 Salbei: http://sprunge.us/bWIG?php
14:07 Salbei cool sfan5 thanks u :)
14:11 Krock Does the bus still exist in the newest Git, where the selection box for nodes is not shown up when pointing through water?
14:11 Krock *bug
14:20 casimir joined #minetest-dev
14:21 proller joined #minetest-dev
14:28 Amaz joined #minetest-dev
14:48 VargaD_ joined #minetest-dev
15:00 Salbei left #minetest-dev
15:05 proller joined #minetest-dev
15:23 ninnghazad is there a callback for when entities get unloaded cause they are too far away?
15:29 Krock why would you need that?
15:30 ninnghazad cause i have entites communicating, when a and b talk, and i move away to unload b, a will not known that b is gone.
15:30 ninnghazad like with digilines, what if an object is unloaded while processing a call between two objects?
15:31 ninnghazad with a callback, i could remove that entity from the list of talkers, and prevent calling methods of the removed object (which return garbage it seems)
15:32 mos_basik joined #minetest-dev
15:35 ninnghazad hm, or can temporarily keep entities from being removed, so that they can finish what they are doing without crashing?
15:39 ninnghazad hm there was some talk about unloading-prevention like 2 years back.
15:41 Krock you can forceload blocks
15:43 NakedFury joined #minetest-dev
15:44 mos_basik joined #minetest-dev
15:46 ninnghazad yeah, i guess that'll be the only way right now. at least that ensures that entities not only stay alive, but also the nodes around them they might want to access.
15:46 mos_basik joined #minetest-dev
15:47 ninnghazad from the irc-logs it seems the discussion about implementing isUnloadAllowed did not end fruitfull.
15:49 proller joined #minetest-dev
15:59 rubenwardy joined #minetest-dev
16:00 kilbith joined #minetest-dev
16:00 kilbith Megaf: what's the result of your compiling ?
16:02 Megaf joined #minetest-dev
16:02 * rubenwardy wishes compiling didn't take so long...
16:03 kilbith Megaf: what's the result of your compiling ?
16:05 * Krock slaps rubenwardy
16:10 Hunterz joined #minetest-dev
16:35 rubenwardy joined #minetest-dev
16:40 kahrl Zeno`: oh yeah, I wasn't quite sure if the C++ standard guarantees that the first enum value corresponds to 0 in absence of an initializer, but it does (7.2.1)
16:41 kahrl doesn't hurt to be explicit though
16:44 petrusd987 joined #minetest-dev
16:51 Calinou joined #minetest-dev
16:52 zat joined #minetest-dev
17:00 Calinou_ joined #minetest-dev
17:04 Anchakor_ joined #minetest-dev
17:16 asl joined #minetest-dev
17:23 proller joined #minetest-dev
17:32 petrusd987_ joined #minetest-dev
17:37 kaeza joined #minetest-dev
18:00 harrison joined #minetest-dev
18:01 petrusd987 Does anyone know how I can link Koneki to Minetest for debugging?
18:03 proller joined #minetest-dev
18:04 Megaf sfan5: well, it compiled
18:04 Megaf it have compiled
18:04 Megaf does anyone heve have a Raspberry Pi to test my build?
18:04 Megaf [100%] Built target minetest
18:04 Megaf real    109m10.320s
18:05 petrusd987 I can, I'd have to find the power cable though
18:05 sfan5 can't you test it on your raspberry pi?
18:05 sfan5 I have a raspi too, but it's busy with other stuff
18:06 Megaf No I can't, I don't have any monitor or keyboard here
18:08 Megaf petrusd987: Ok, I will find a way to upload to somewhere
18:08 Megaf upload it*
18:15 petrusd987 I'll be right back
18:21 VargaD_ joined #minetest-dev
18:27 kilbith joined #minetest-dev
18:42 petrusd987_ joined #minetest-dev
18:48 OldCoder joined #minetest-dev
18:54 Megaf sfan5: proller: any idea on the cause of this? http://paste.debian.net/125419/
18:54 Megaf ^ kilbith
18:55 kahrl Megaf: npot2 problem?
18:55 Megaf libEGL warning: DRI2: failed to authenticate
18:55 proller google it
18:59 kahrl Megaf: number 7 http://pi3d.github.io/html/FAQ.html
19:00 Megaf Ok, thanks
19:01 Megaf kilbith: try that
19:02 kilbith ok
19:04 Megaf [15:07:23] <kahrl> nonetheless at the rate Megaf gets them I would suggest doing a memtest on that machine
19:04 Megaf Lets see for how long minetestserver can run without crashing
19:08 kilbith ok I entered the commands and it still not works
19:08 kilbith I should remove some libs iirc
19:09 kilbith manually
19:11 kilbith /usr/lib/arm-linux-gnueabihf/libEGL.so.1
19:11 kilbith /usr/lib/arm-linux-gnueabihf/libEGL.so.1.0
19:11 kilbith should just keep libEGL.so ?
19:28 Krock joined #minetest-dev
19:40 Miner_48er joined #minetest-dev
19:41 w00tc0d3 joined #minetest-dev
19:43 VargaD_ joined #minetest-dev
19:55 jin_xi joined #minetest-dev
20:06 VargaD_ joined #minetest-dev
20:11 Amaz joined #minetest-dev
20:15 Krock Any chance to get #1715 merged in this year?
20:28 kilbith Megaf: also that when trying to get the public server list : 22:28:04: ERROR[AsyncWorkerThread_0]: httpfetch_sync: unable to fetch servers.minetest.net/list because USE_CURL=0
20:29 Krock ^ curl missing
20:29 paramat joined #minetest-dev
20:34 paramat sfan5, (and other MTgame devs) please can boats mod be fixed by merging my pull https://github.com/minetest/minetest_game/pull/328 it's been tested and fine tuned for hours, works like a dream. best to test in my 'canyon' river mod as that has extreme water level changes
20:34 sfan5 I'm fine with merging it
20:34 sfan5 ''tell BlockMen <paramat> sfan5, (and other MTgame devs) please can boats mod be fixed by merging my pull https://github.com/minetest/minetest_game/pull/328 it's been tested and fine tuned for hours, works like a dream. best to test in my 'canyon' river mod as that has extreme water level changes
20:34 sfan5 where's the bot
20:34 sfan5 argh
20:43 gravgun joined #minetest-dev
20:43 gravgun Hi everybody. @sfan5: some early work: http://pastebin.com/NFB031Rp
20:44 sfan5 cstdint is C++11?
20:44 gravgun Yes
20:44 gravgun cstdint.h is C++ < 11
20:44 gravgun eh i meant <stdint.h>
20:45 sfan5 TIL
20:45 gravgun and <cstdint>, although it existed before C++11 on some libc's is really defined in the C++11 standard
20:45 sfan5 >signed int types
20:46 sfan5 why do we even have different types?
20:46 sfan5 and "int addr;"
20:46 sfan5 int is not guaranteed to be the size of a pointer
20:46 gravgun not supposed to be
20:46 gravgun supposed to be a stack pointer
20:47 gravgun and stack will be an array, or a std::stack of some kind
20:47 gravgun as i said it's early work
20:47 gravgun and I've never made any VMs beforehand
20:47 sfan5 not having the stack be an array will make it slow
20:48 gravgun stl is slow in general you know...
20:48 sfan5 are you aware that you also need to serialize instructions?
20:48 gravgun at least compared to bare C arrays
20:48 gravgun yes I am
20:48 sfan5 and class Program should be class ProgramState (at least it's more like that)
20:49 gravgun it's both actually...
20:49 gravgun as I don't intend the code to be self-modifying, your suggestion is right
20:49 sfan5 Program is just std::queue<Instruction> instructions;
20:49 gravgun typedef?
20:49 sfan5 having a stack and pc if you want to pass around a program is a waste of resources
20:49 sfan5 typedef would be goo
20:49 sfan5 d
20:51 gravgun updated paste
20:52 sfan5 what does emit<something>() do?
20:52 paramat thanks
20:53 gravgun Those are therefore supposed to be in some kind of Program factory, they just emit instructions and append them to the std::queue
20:53 sfan5 which std::queue?
20:53 sfan5 there is none
20:53 gravgun typedef std::queue<Instruction> Program;
20:53 sfan5 that doesn't define a variable
20:54 gravgun not anymore according to my changes...
20:54 sfan5 it didn't define one before either
20:55 gravgun it did, it was in "class Program"
20:55 gravgun changed to ProgramState
20:55 sfan5 wat
20:55 sfan5 you can't just <a class>.<a variable>
20:55 sfan5 you need an instance of the class
20:55 sfan5 ProgramState.pc = 0; is not going to work
20:55 gravgun there was an instance
20:55 gravgun
20:55 gravgun class Program {
20:55 gravgun std::queue<Instruction> instructions;
20:55 gravgun std::stack<Arg> stack;
20:55 gravgun int pc;
20:56 gravgun };
20:56 gravgun
20:56 sfan5 thats not an instance
20:56 gravgun Program pgm;
20:56 sfan5 thats a class def
20:56 gravgun
20:56 sfan5 what
20:56 sfan5 where
20:56 sfan5 hm
20:56 sfan5 didn't see that
20:56 gravgun And what is "Program pgm" then?
20:56 gravgun :P
20:56 gravgun got removed too
20:57 gravgun Basically, those emitX() methods are here to create the program, step by step, to be called by a future asm syntax parser
20:58 gravgun But i'll need to move them somewhere else
20:58 shadowzone joined #minetest-dev
20:58 shadowzone joined #minetest-dev
20:59 gravgun and they'll possibly be exposed to the Lua API, in case someone writes a compiler of some kind
21:00 gravgun the same ways GCC directly feeds its assembler (as) when not using (temporary) .s files
21:05 ShadowNinja ninnghazad: You can use get_staticdata().
21:06 ShadowNinja ninnghazad: Really a #minetest question though.
21:07 zat joined #minetest-dev
21:15 iqualfragile joined #minetest-dev
21:16 gravgun BTW isn't this commit pretty useless? https://github.com/minetest/minetest/commi​t/d1ccc64e1ecfd00df36d2b8a8b3420b70434fefd   Because ISO C++ (98) 7.1§1 says "If the first enumerator has no initializer, the value of the corresponding constant is zero."
21:17 PenguinDad gravgun: we already figured that out :P
21:19 gravgun Well people, have a good night/day (depending on where you are :P ). Already 23:20 in France ;) Bye
21:19 gravgun left #minetest-dev
21:31 proller joined #minetest-dev
21:42 RealBadAngel hi guys
21:43 RealBadAngel im at the final testing stage for meshnodes
21:43 RealBadAngel http://i.imgur.com/ECELl3Y.png
21:43 RealBadAngel i had to learn blender a bit ;)
21:49 rickmcfarley joined #minetest-dev
21:57 DuDraig joined #minetest-dev
22:01 DuDraig joined #minetest-dev
22:06 petrusd987_ joined #minetest-dev
22:12 paramat Hi RealBadAngel, if meshnodes eventually replace nodeboxes (i'm keen to see the speed increase), how will you achieve faces-shading that matches normal nodes? and how will you map 6 textures onto top, base, N, S, E, W as currently? =)
22:14 RealBadAngel paramat, meshnodes are shaded basing on surface normals (so its working). bumpmapping and parallax is working too
22:15 RealBadAngel about UVing objects: tiles are used to describe materials of the object
22:15 RealBadAngel if you have more than one object in the mesh you can have different textures for them, described in tiles
22:16 RealBadAngel but one tile can hold whole texture for a object
22:16 RealBadAngel you need to define uv for it in blender
22:17 RealBadAngel http://i.imgur.com/nuYwyFY.png
22:17 RealBadAngel this is texture file for mushroom
22:18 ShadowNinja paramat: With models you can map a face of the model to an arbritrary section on an image, that's how the default character.x works.
22:18 paramat ah so the surface normals define a 'blend' of our current top, base, N, S, E, W shades ...
22:19 RealBadAngel theres no top, bottom etc anymore
22:19 RealBadAngel all has to be done in blender
22:20 RealBadAngel you just cut a piece of texture and assign it to vertices
22:21 RealBadAngel thx to that you dont care anymore bout it. and can scale it, rotate or whatever
22:21 RealBadAngel meshnodes doesnt need textures rotating code like nodeboxes do
22:24 paramat if you want meshes to replace nodeboxes, people will not want to use blender for texturing, they will want a simple way similar to how it is now
22:24 paramat also, people will want to define a nodebox in a simple way (as it is now) and have code convert that into a mesh
22:25 RealBadAngel nodeboxes will be turned into meshes on startup
22:25 paramat cool
22:25 RealBadAngel folks can wait for game to load a bit longer
22:26 paramat because the current nodebox definition is simple and easy, even i wouldn't want to use blender for that
22:26 RealBadAngel but if a stairs nodebox with just 2 boxes is slower than 7k vertices r2d2 model then its worth it
22:26 paramat yes i agree
22:27 RealBadAngel my target is to leave in custom drawtypes only those which are changing, are dynamical
22:27 RealBadAngel rest will be turned into native irrlicht ones
22:29 RealBadAngel but anyway, one with a bit of blender experience can replace almost any drawtype with meshnode
22:30 paramat so meshnode shading (by surface normals) roughly matches the current shading of nodes? (top brightest, base darkest, EW faces slightly darker than NS faces?
22:30 RealBadAngel with far more better results
22:30 RealBadAngel yes
22:30 paramat good
22:30 RealBadAngel mesh comes with precalculated normals
22:31 blaise joined #minetest-dev
22:31 RealBadAngel whats more interesting can be cloned into mesh with tangents
22:32 RealBadAngel but thats for laters
22:32 RealBadAngel atm i will add static meshes
22:33 RealBadAngel i know how to make also animated ones but this is far more complicated than static
22:33 paramat for example, if i make a simple mesh which has all faces axis-aligned (like stairs) it would be good to be able to simply assign 6 textures to top/base/n/s/e/w as currently, without using blender
22:33 RealBadAngel and will require static in the first place
22:34 RealBadAngel paramat, you should check out terasology
22:34 paramat aha i have
22:34 RealBadAngel they are using clean models without textures
22:34 RealBadAngel to define everything
22:34 RealBadAngel and adding just textures
22:34 paramat their shaders look horrible =)
22:34 RealBadAngel theyre good
22:34 RealBadAngel but damn slow
22:34 paramat lol i would say that
22:36 RealBadAngel but again, about stairs example
22:36 RealBadAngel you can define uv based on one texture
22:36 RealBadAngel like player model does
22:36 RealBadAngel and know which parts of it are sides
22:37 paramat ah okay
22:37 RealBadAngel then using the same model for another node and modified texture
22:37 paramat editing the character model texture is easy enough
22:38 RealBadAngel you know that theyre meshes?
22:38 RealBadAngel so why do you think meshnodes will be harder? ;)
22:38 paramat well now i understand, thanks
22:39 RealBadAngel also fun fact: size of texture will not matter
22:40 RealBadAngel mapping will stay the same
22:40 RealBadAngel you can have textures 1024x and 16x for the very same mesh
22:41 RealBadAngel and they will be properly mapped without any effort
22:43 paramat i want to see your work merged, so my advice is don't mix your speed-improving work with controversial changes to MT's visual style. i think this is why your pulls get blocked so often
22:43 RealBadAngel how folks will use meshes is not my concern
22:44 RealBadAngel accusing hammer producer that it was used to smash some1 in the head instead to a nail would be the same
22:45 blaise joined #minetest-dev
22:45 RealBadAngel drawtype itself doesnt change anything in the game
22:45 RealBadAngel it just offers unlimited possibilities
22:46 RealBadAngel rest is up to modders
22:49 Taoki joined #minetest-dev
22:49 jin_xi so, i think i just had a genius idea for dynamic textures: what if we create a [dynamic texture modifier, so it creates a texture with _two_ textures?
22:50 jin_xi so you can switch them instead of creating new ones endlessly
22:50 RealBadAngel ehm, we do have such mechanism already?
22:51 jin_xi so digilines and signs and similar mods could work better
22:51 jin_xi RealBadAngel: what?
22:51 RealBadAngel tiles and special tiles
22:51 jin_xi no, for dynamic textures which are changed in-game
22:51 shadowzone joined #minetest-dev
22:51 RealBadAngel ofc
22:51 RealBadAngel theres a PR waiting that does it
22:51 jin_xi oh
22:52 RealBadAngel you simply define one set in tiles, another in special tiles
22:52 RealBadAngel MSB of param2 switches them
22:52 RealBadAngel runtime
22:52 jin_xi but what if your texture is going to change?
22:52 RealBadAngel for any kind of drawtype
22:53 paramat RBA, for example, 'disable crack animation' pull is, i suspect, being delayed partly because you increased number of digging particles
22:53 RealBadAngel by 3
22:53 RealBadAngel ROTFL
22:53 RealBadAngel while dug one has way more
22:54 RealBadAngel some folks are just idk, sarcastic? dumb? ironic?
22:54 RealBadAngel no comments anymore ;)
22:55 RealBadAngel calling 3 particles 300% when dug one has a few k %?
22:55 blaise joined #minetest-dev
22:57 RealBadAngel https://github.com/minetest/minetest​/blob/master/src/particles.cpp#L242
22:58 RealBadAngel so following, dug one has 3200% more particles, and you can live with it :P
22:59 RealBadAngel not to mention SN proposed once to increase that number already
22:59 RealBadAngel by 25600% :P
23:01 RealBadAngel https://github.com/minetest/minetest/pull/943
23:04 proller rewrite particles first
23:04 RealBadAngel not my job, jin_xi is doing that
23:06 RealBadAngel proller, and btw, you should know im developing the code on rather low end hardware
23:06 RealBadAngel 1,8ghz dual core, gallium drivers amd
23:06 proller like rpi ? ;)
23:06 RealBadAngel and i wont allow anything that would slow the game for me
23:07 RealBadAngel unlike the other devs im rewriting the code all the time until i get satysfing results
23:07 RealBadAngel and i do measure it
23:08 RealBadAngel i am still a demoscene coder
23:08 proller then you must rewrite most ugly places
23:08 RealBadAngel lol, you never saw really ugly code then ;)
23:09 RealBadAngel but you will, soon
23:09 RealBadAngel with a guarantee it couldnt be done faster
23:11 RealBadAngel point is that coding demos, and games have one goal
23:11 RealBadAngel speed
23:11 RealBadAngel not the nice code
23:11 mos_basik_ joined #minetest-dev
23:11 paramat indeed particles on dig is much higher at 32, however, this is one of those things celeron may not like ;)
23:11 RealBadAngel if some1 thinks otherwise propably he should ask bill to hire him at mc
23:12 proller i'm about slowest places
23:12 jin_xi #of particles and their gravitation and such should not be hardcoded imo
23:12 mos_basik_ joined #minetest-dev
23:12 RealBadAngel proller, 3 particles more doesnt slow anything
23:13 RealBadAngel you know what is slowing your game? like really?
23:13 RealBadAngel cracks
23:13 proller no
23:13 RealBadAngel test the pull, disable the cracks
23:13 RealBadAngel and enjoy :P
23:13 proller cracks and particles slows ONLY on player interact
23:14 proller but here is other slow things
23:14 RealBadAngel you dont get
23:14 RealBadAngel cracks are causing mesh updates for a block
23:14 RealBadAngel so whole mesh is recalculated
23:14 RealBadAngel for 16x16x16 nodes
23:14 proller "it was solved in freeminer" ;)
23:15 RealBadAngel no it wasnt
23:15 RealBadAngel stop trolling :P
23:15 RealBadAngel i did that but you never saw the code :P
23:16 paramat personally, for speed reasons, i want to disable cracks, and not use highlighting, because i don't need to see dig progress. so i want to see your pull merged RBA
23:16 RealBadAngel cracks and highlighting will become independent scene nodes
23:16 proller mesh update - one of ugliest places
23:16 RealBadAngel without need to refresh the mapblock
23:17 RealBadAngel they will just use a copy of single node mesh
23:17 paramat as you probably know, i have similar AV tastes to c55, so if you can keep me happy, there's a chance c55 will merge your work =)
23:18 RealBadAngel paramat, you can define with mesh most simple and dumb voxel node
23:18 RealBadAngel and it will be faster than nodebox
23:18 paramat i would ask to be AV manager but i'm not qualified
23:18 paramat cool
23:18 RealBadAngel but
23:18 RealBadAngel you can use it for something like this:
23:19 RealBadAngel http://i.imgur.com/fUon82U.png
23:19 RealBadAngel its only up to you
23:29 paramat to me that looks ridiculous in a voxel game, however, i think very simple geometric shapes (like sloping roofs) will look okay in MT, and meshes will be good for this
23:30 RealBadAngel also micronodes, nice furniture, complex machines
23:30 RealBadAngel whatever you can imagine
23:31 RealBadAngel without any limitations of nodeboxes
23:32 paramat the digging particles argument: indeed 3 more particles is not much compared to breaking the node, however i suspect c55 and others will not want that changed (even though there is no performance hit), and that should really be a separate pull request
23:34 shadowzone That is one interesting texture.
23:40 RealBadAngel paramat, i wont be changing tested and proven to be fast code basing on some imaginations
23:42 RealBadAngel using the phrase "300%" in the pull was just a pure nonsense
23:42 kahrl the number of particles is really more artistic than performance based
23:42 RealBadAngel and smelled monthy python
23:43 RealBadAngel i raised the number of particles for the valid reason
23:44 Taoki joined #minetest-dev
23:44 RealBadAngel if cracks are disabled the process has to be more visible
23:44 RealBadAngel and it plays nice together with highlighting
23:44 RealBadAngel which also changes when digging
23:44 kahrl but the number is just ridiculously large with the default settings
23:44 kahrl i.e. crack animation, no node highlighting
23:45 RealBadAngel dug one is a bit too much
23:45 RealBadAngel i set it to 4 on punch (digging)
23:45 paramat RBA separating number of dig particles from 'disable crack animations' will not affect your peformance improving code. such an attitude may be why your pulls are often not merged ;)
23:46 RealBadAngel paramat, i dont care anymore if my pulls are merged or not
23:46 RealBadAngel i know im doing good code
23:46 paramat well i care, i want them merged
23:46 RealBadAngel if mt wont have it? pity
23:46 RealBadAngel i will
23:47 paramat separating artistic matters allows your performance code to be merged quickly, unobstructed by artistic arguments
23:47 RealBadAngel coding is an art
23:48 RealBadAngel whoever thinks its a job should use the hammer. to kill himself for sake of humanity
23:49 RealBadAngel any other way we will see M$ each time we will open the fridge
23:51 paramat as i said in the comments, each punch releases 4+ particles already, not 1, so doesn't need increasing much, perhaps double it? i actually think 32 on node break is too much =)
23:52 RealBadAngel before it was just 1
23:52 RealBadAngel now its 4
23:53 RealBadAngel that 32 could be lowered, by half propably
23:53 RealBadAngel hey, thats what Blockmen could understand! MINUS 1600% lol
23:55 kahrl so what about making the number dependent on the crack animation setting?

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