Minetest logo

IRC log for #minetest-dev, 2014-08-13

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

All times shown according to UTC.

Time Nick Message
00:37 Zeno` joined #minetest-dev
01:00 AnotherBrick joined #minetest-dev
01:05 werwerwer joined #minetest-dev
03:16 Eater4 joined #minetest-dev
03:23 kahrl joined #minetest-dev
03:56 Miner_48er joined #minetest-dev
04:03 OldCoder joined #minetest-dev
04:45 OldCoder joined #minetest-dev
04:47 VanessaE_ !tell sapier please update dreambuilder and give it a try with the profiler again.  plants_lib got a major speed boost tonight and it/its mods should look much better now.
04:47 VanessaE_ fek.
04:47 VanessaE_ oh well.
05:45 Hunterz joined #minetest-dev
06:26 RealBadAngel https://github.com/minetest/minetest/pull/1553
06:26 RealBadAngel can we have this merged?
07:21 Hunterz is there any eta to implement server console?
07:21 RealBadAngel kahrl, what was the reason to copy finalColorBlend in shaders?
07:22 Hunterz or is possible remotely control server only using irc mod for now?
07:22 RealBadAngel Hunterz, idk
07:30 Zeno` You can remotely control the server using IRC, yes
07:31 Zeno` Everything except for restarting the server (you can shutdown, but your keep-alive script will have to restart unless you do it manually)
07:31 Zeno` To put simply, if you have the privs, you can do anything from IRC that you can do from the in-game console
07:36 Garmine joined #minetest-dev
07:42 kahrl RealBadAngel: what would be the alternative?
07:46 Anchakor_ joined #minetest-dev
07:48 grrk-bzzt joined #minetest-dev
07:52 Calinou joined #minetest-dev
07:58 mrtux joined #minetest-dev
08:02 RealBadAngel kahrl, keep it on c++ side, thats way faster
08:03 RealBadAngel and apply finalColorBlend no matter if shaders are used or not
08:05 RealBadAngel i just figured it out thx to the bug, in mapblockmesh::animate its done that way
08:05 kahrl RealBadAngel: but then you have to change almost every vertex in every frame
08:06 kahrl because the day night ratio changes all the time
08:06 kahrl well, not every frame, but you get the idea
08:06 RealBadAngel yes, but its done already anyway
08:07 kahrl only if shaders are disabled, right?
08:08 RealBadAngel refresh is triggered even on camera move
08:08 kahrl what's refresh?
08:08 RealBadAngel rebuild the mesh
08:08 kahrl huh
08:09 RealBadAngel so it happens more often than changing the time
08:09 kahrl I don't believe, can you point to code that does that
08:09 kahrl believe that*
08:09 RealBadAngel gimme a minute
08:14 RealBadAngel cant find now wheres the code responsible for that, i just put dstream<<"Updating\n"; at the start of MapBlockMesh::MapBlockMesh
08:14 RealBadAngel and it works all the time
08:16 RealBadAngel hmm, not so often that i thought
08:17 RealBadAngel digging triggers that and something else for sure
08:18 RealBadAngel because with time and no actions taken it hapenns all the time
08:19 kahrl it happens more frequently in minetest_game than minimal
08:19 kahrl I guess it's ABMs?
08:20 kahrl although I do see it triggering on rotating the camera as well, I don't know why that would happen
08:21 kahrl are the meshes rebuilt if they had been hidden by occlusion culling?
08:21 kahrl and/or frustum culling
08:21 sfan5 isn't there a profiler for num_mesh_rebuilds?
08:22 kahrl sure, but it doesn't tell you why
08:24 kahrl actually it seems there's just a profiler for the time spent, not the number
08:25 Zeno` how do code merges (pull requests) work? I mean, is there a reason why many pull requests are weeks (or months) old?
08:26 Zeno` I understand that they have to be reviewed of course
08:27 RealBadAngel Zeno`, at least 2 devs have to agree on the change for it to be merged
08:28 RealBadAngel btw, https://github.com/minetest/minetest/pull/1553
08:28 Zeno` So... they need to be online at the same time to discuss it?
08:28 RealBadAngel not really
08:28 RealBadAngel sfan5, are you ok with above?
08:29 Zeno` 2 devs need to agree even for bug fixes?
08:30 RealBadAngel for obvious ones rather not
08:30 RealBadAngel sometimes i push bugfixes directly, buth rarely
08:30 RealBadAngel *but
08:31 kahrl Zeno`: also http://dev.minetest.net/Organisation
08:32 RealBadAngel kahrl, are liquid drawtypes yours or c55's ?
08:33 Zeno` thanks kahrl
08:33 kahrl well, they are based on the water that existed in 0.3 and before, so c55's
08:33 RealBadAngel im askin because you were modifying flowing lately
08:33 kahrl no?
08:34 RealBadAngel i thought so, rotating the textures on the flowing ones wasnt yours?
08:34 kahrl oh, I didn't count that as "lately" :P
08:34 kahrl but that was me
08:34 RealBadAngel time is relative ;)
08:36 RealBadAngel https://github.com/minetest/minetest/blob/master/src/content_mapblock.cpp#L381
08:36 RealBadAngel heres some code that affects liquids that are lightsources at the same time
08:37 RealBadAngel i wonder why the heck only flowing ones not sources
08:37 sfan5 RealBadAngel: if that doesn't break anything it's ok
08:37 RealBadAngel sfan5, it fixes lotsa things
08:37 kahrl RealBadAngel: no idea
08:37 RealBadAngel including very very old issues
08:38 RealBadAngel like flowing water and lava having different brightness than sources
08:38 RealBadAngel that one i remember trying to fix more than a yr ago
08:40 kahrl water too? not just lava?
08:40 RealBadAngel current textures were adjusted in an attempt to match each other
08:41 RealBadAngel with #1553 theres no need to
08:41 RealBadAngel it was because of shading/boosting the faces
08:42 RealBadAngel in 4 different places and using different methods
08:43 RealBadAngel no smooth used to decrease light level for faces
08:43 RealBadAngel smooth did that based on normal.Y and boosting top face but horizontal only
08:44 RealBadAngel and shaders did the opposite by shading out everything that is not the top face
08:45 RealBadAngel and there was some more code for that i think
08:45 RealBadAngel to make story short, total mess ;)
08:45 kahrl that sounds more sensible than expected
08:47 RealBadAngel i just made it all at one place
08:47 RealBadAngel so the results are almost identical for all the modes
08:49 RealBadAngel http://i.imgur.com/V3Bphid.png
08:49 PenguinDad joined #minetest-dev
08:50 RealBadAngel hi, PenguinDad, i found the reason for acid liquids being fucked up, will fix that asap
08:50 RealBadAngel but thats not what you thought. that bug is here since always
08:54 RealBadAngel kahrl, so im gonna apply the very same code to sources too ( https://github.com/minetest/minetest/blob/master/src/content_mapblock.cpp#L381 ), ok?
08:55 kahrl RealBadAngel: I guess it's okay
08:55 kahrl I didn't write that part of the code but if the change produces good results, sure
08:56 RealBadAngel lava is not affected because it was LIGHTMAX
08:56 RealBadAngel but acid from carbone was, because it was 5
08:56 RealBadAngel erm, 8
08:56 RealBadAngel so sources were full bright, flowing darkened
08:56 kahrl I see
09:08 RealBadAngel kahrl, also what do you think about the idea of removing code for makin wielded models and inventory images and making tools out of them?
09:09 RealBadAngel included in main repo ofc
09:09 kahrl what do you mean by making tools?
09:10 RealBadAngel separate irrlicht app that would generate image/models for further use
09:10 RealBadAngel i mean modders tool
09:11 kahrl sounds fine, but why do you want to remove it from minetest?
09:11 RealBadAngel from the game code, not from the repo
09:11 kahrl that's what I mean
09:12 RealBadAngel because it fails miserably in many cases, is slow
09:12 kahrl it works for me, but then I don't use HD texture packs
09:12 RealBadAngel in external app one could prepare correct models with settings or whatever
09:12 kahrl why not fix it instead?
09:13 RealBadAngel i tried but failed ;)
09:13 RealBadAngel it worked to some point, i changed alpha trigger and got nice hd tools
09:14 RealBadAngel but then VanessaE_ popped with her textures and caused that function to work for MINUTES
09:14 kahrl giving HD texture pack authors a way to make their own wield models is good
09:14 kahrl but don't force everyone else to do that
09:15 RealBadAngel we can leave the code as failsafe
09:15 RealBadAngel that sounds reasonably
09:16 sfan5 uh
09:16 RealBadAngel but providing inventory images and wielded meshes should become good manner
09:16 sfan5 you're never going to get removing minetest.inventorycube past the rest of the rest dev team
09:17 RealBadAngel so ok, failsafe then
09:18 RealBadAngel but mods that have hundreds of nodes, dozens of models should take care of it
09:18 RealBadAngel to not overuse the engine
09:18 sfan5 why would modders include inventory images for simple nodes?
09:18 sfan5 wat
09:18 sfan5 thats like
09:18 sfan5 not an argument at all
09:18 sfan5 we should remove all texture modifiers and texture combining too
09:19 sfan5 thats overusing the engine too
09:19 RealBadAngel do you like screen blinking and lagging inventories?
09:19 RealBadAngel i dont
09:19 sfan5 preload_item_visuals = true
09:19 sfan5 if you don't like that set that setting
09:20 RealBadAngel neither i dont like enormous loading time
09:20 sfan5 then find a different solution
09:20 RealBadAngel i did
09:20 RealBadAngel premade textures
09:20 kahrl providing extra inventory images and wield models for everything = 3x as much data to download
09:20 sfan5 removing/deprecating dynamic features is not the solution
09:20 sfan5 just cache them to disk
09:21 sfan5 that would mostly solve it
09:21 RealBadAngel thats a solution
09:21 kahrl is loading a wield mesh from disk & parsing the file really faster than extruding the image
09:22 sfan5 it would solve the blinking RealBadAngel mentioned
09:22 sfan5 but the blinking is a bug in Irrlicht anyway
09:22 sfan5 rendering to RTT shouldn't affect whats displayed right now
09:28 celeron55 MC draws them in 3D in the UI
09:28 sfan5 isn't that .. like .. slow?
09:28 celeron55 without prerendering (afaik)
09:28 celeron55 no?
09:28 celeron55 isn't that, like, what GPUs are for
09:28 sfan5 hm, yes it is
09:29 sfan5 why does MT do it differntly then?
09:29 sfan5 different*
09:29 celeron55 because it just happens to be that way
09:30 kahrl has somebody tried finishing optimize_invtex?
09:30 kahrl or benchmarking it
09:31 kahrl I guess I should explain what that is
09:31 kahrl it's a branch I made ages ago that renders the inventorycubes on the CPU without using RTT
09:31 kahrl on my machine it's considerably faster than the RTT version
09:31 kahrl (and it would fix the blinking issue, I suppose)
09:31 kahrl but I never got around to finishing it, and at some point I didn't want to rebase it anymore
09:31 sfan5 renders them on the CPU?
09:31 sfan5 how?
09:31 celeron55 i don't even remember that existing
09:32 kahrl sfan5: software rasterizer
09:32 kahrl coded in about a hundred lines of code iirc
09:32 celeron55 if it's reliable and doesn't require constant maintenance (nobody will want to fix it), it could be fine
09:32 sfan5 I see, wouldn't it be easier to use Irrlicht's burningvideo/software driver?
09:32 kahrl maybe a few hundred
09:33 kahrl sfan5: I wasn't sure how to embed that
09:37 Garmine joined #minetest-dev
09:37 kahrl and my rasterizer avoids the overhead of creating a scene for every inventory cube
09:38 kahrl I think I looked into rendering the inventory cubes in real time like you said but I couldn't figure out how to configure the Z buffer properly
09:38 sfan5 couldn't the scene be reused in the current code w/o changing much?
09:38 kahrl in any case, the optimize_invtex branch is still around:  https://github.com/kahrl/minetest/tree/optimize_invtex
09:38 kahrl wow, over a year ago now
09:38 kahrl sfan5: it does that, at least it used to
09:41 kahrl sfan5: but the only call that costs any significant amount of time is endScene()
09:41 kahrl which can take tens of milliseconds even when you are just rendering a cube to a texture
09:41 kahrl now multiply that by the number of nodes
09:41 kahrl (the rasterizer takes on the order of microseconds per node, iirc)
09:44 proller joined #minetest-dev
09:48 Zeno` so 1000 calls takes on the order of... 1 second?
09:49 RealBadAngel it can take about a minute to prepare the images for big games
09:49 Zeno` When I looked at things a few weeks ago it seemed most of the time seemed to be spent in png_getline() (or something like that, I can't remember) for the client
09:49 RealBadAngel not a second
09:50 RealBadAngel for a second no1 would bother
09:50 Zeno` client side I wouldn't bother... server side I would
09:50 Zeno` well, not in start up
09:50 * Zeno` scrolls back
09:51 RealBadAngel a minute for a client to start a game is annoying as hell
09:51 RealBadAngel and we are talking bout item images only
09:51 Zeno` well of course
09:51 Zeno` a minute?
09:51 Zeno` it really takes that long?
09:52 RealBadAngel add other things that have to be done it appears one can go out to have a beer befoure connecting to the server
09:52 Zeno` yes, well that's a problem
09:52 Zeno` but I'll have the beer anyway
09:52 RealBadAngel lol
09:57 RealBadAngel 1m23s with item preload vs 12s without. (technic game)
09:57 RealBadAngel thats what im talkin about :)
09:58 RealBadAngel thats singleplayer ofc
09:58 kahrl_ joined #minetest-dev
09:59 RealBadAngel 1m23s with item preload vs 12s without. (technic game)
09:59 RealBadAngel kahrl, pasted that again for you ^^
10:00 RealBadAngel oops
10:07 Exio joined #minetest-dev
10:07 deltib joined #minetest-dev
10:21 diemartin joined #minetest-dev
10:30 Exio joined #minetest-dev
10:30 deltib joined #minetest-dev
10:32 PilzAdam joined #minetest-dev
10:33 alexxs joined #minetest-dev
10:36 Megaf joined #minetest-dev
10:36 Megaf can you please fix all this fucking seg faults and things like this?
10:36 Megaf https://www.youtube.com/watch?v=A8A5IMkW-Kg
10:36 Megaf pff
10:36 Megaf I mean, this
10:36 Megaf 10:18:44: ERROR[main]: ERROR: An unhandled exception occurred: C++ exception
10:37 Megaf Just stop implementing new stuff and begin to fix old stuff
10:38 Megaf /home/minetest/Server/src/main.cpp:1945: int main(int, char**): Assertion '0' failed.
10:41 Megaf https://github.com/minetest/minetest/issues/1556
10:41 Zeno` yes, assert(0)... lazy code
10:42 Zeno` and not a valid use of assert() anyway :/
10:45 Zeno` A fairly large amount of that code has not really been changed since MT was first put on GitHub (or shortly thereafter)
10:46 Megaf and by the way, main.cpp is huge
10:47 Zeno` over 1000 lines, yes
10:47 Megaf this should be totally rewritten from scratch
10:47 Megaf all minetest
10:47 Zeno` actually main() (the function) is over 1000 lines long!
10:47 Megaf and with moding and multithreading in mind from the begining
10:48 Zeno` take a look at the_game() in game.cpp (?)... it's fantastic
10:49 PenguinDad Megaf: you are aware how long it would take to rewrite minetest from scratch?
10:49 Megaf yep, one week
10:50 Zeno` It'd take a few hours to refactor main()
10:50 Zeno` if that
10:50 Megaf one person here did a thing like minetest without using the bloody irrlicht
10:50 Megaf alone
10:50 Zeno` at least then things might be able to be isolated :)
10:50 Megaf minetest is such a simple game, no need for that huge code
10:51 Megaf Zeno`: maybe half an hour
10:51 Zeno` yes, but the patch would never be accepted
10:52 Megaf I think they like meaningless crash messages and complex things
10:52 Zeno` there are nested conditions and compound statements over 9 levels deep in main()!
10:52 Zeno` for no reason at all
10:53 Megaf yep, they only add and add more lines of useless code
10:53 Megaf they need to stop adding and begin to take
10:53 Megaf or rewrite
10:54 proller your crash was fixed in freeminer long time ago ;)
10:54 * Megaf kills proller
10:54 Megaf proller: where do you live again?
10:54 Megaf if you live in Europe, I’m just a few hours from you
10:55 proller ok, im waiting you
10:55 Megaf proller: should I bring beer?
10:56 Zeno` The answer to that is that if you need more than 3 levels of indentation, you're screwed anyway, and should fix your program.   <-- Linus
10:56 Zeno` and this is 9 levels deep
10:56 Zeno` I think the_game() even has 14 levels deep in places
10:56 proller Megaf, good idea! ;)
10:56 Megaf Zeno`: well, the Linux hernel is not a good example of how to make a good software either…
10:57 Zeno` Megaf, true, but if I have to scroll up  400 lines to find an opening bracket then something's very wrong
10:57 Megaf indeed
10:58 Megaf and I’m not even a C++ developer, actually I’m not a developer at all, and I know the code is pretty wrong
11:03 proller here much better code  - https://github.com/Nocte-/hexahedra
11:09 PenguinDad The fences there have the same problem as fences in minetest have
11:10 init joined #minetest-dev
11:10 Megaf what problem is that?
11:10 Megaf proller: that looks nice
11:28 ImQ009 joined #minetest-dev
11:36 Amaz joined #minetest-dev
11:39 Jordach joined #minetest-dev
12:01 init joined #minetest-dev
12:04 Megaf joined #minetest-dev
13:18 hmmmm joined #minetest-dev
14:02 rubenwardy joined #minetest-dev
14:18 Amaz joined #minetest-dev
14:19 proller joined #minetest-dev
14:21 celeron55 proller: how is this supposed to work
14:21 proller what this?
14:21 celeron55 proller: hexahedra; i built it, and get into the menu; clicking any of the options does nothing
14:22 celeron55 ah now i found a log
14:22 proller i dont know
14:27 proller http://www.youtube.com/channel/UC5ZGtWbqB4Y762Z44Rt0h5A
14:32 celeron55 got it to work now
14:32 celeron55 seems like it barely works; everything except rendering is crappy though
14:33 celeron55 and rendering is crappy too on this modern intel chip with the --ogl2 option, have to use nvidia to even get pixels not blurred together
14:33 celeron55 this might develop into something useful though
14:35 celeron55 also i had to fix a path in its data files because it expected i would install it in /usr/local even while i set up the correct paths in cmake variables 8)
14:36 proller but it semi-abandoned
14:38 celeron55 seems so
14:39 celeron55 do you know anything about this? does it run the lua stuff on the server or the client?
14:41 celeron55 this is so much a minetest clone that it's adorable 8)
14:42 Zeno` wow
14:42 Zeno` it looks so much better though
14:42 celeron55 development started and stopped 1 year ago it seems
14:42 Zeno` nothing wrong with minetest clones
14:43 Zeno` that's the point of GPL, after all
14:51 Calinou joined #minetest-dev
14:52 NakedFury joined #minetest-dev
14:54 rubenwardy joined #minetest-dev
15:20 AnotherBrick joined #minetest-dev
15:53 rubenwardy left #minetest-dev
15:53 ^v joined #minetest-dev
16:17 Hunterz joined #minetest-dev
16:34 harrison joined #minetest-dev
16:50 VanessaE_ celeron55: what's your take on #1553 ?
16:51 VanessaE_ (from the earlier conversations, I don't recall seeing your opinion thereof, and I know you'll have one.)
16:51 proller joined #minetest-dev
16:53 celeron55 VanessaE_: eh
16:54 celeron55 the first thing i see when looking at this is some arbitrary constant modification with no explanation whatsoever in a comment or the commit message
16:55 VanessaE_ you mean the 13 -> 23?
16:55 Eater4 joined #minetest-dev
16:56 celeron55 i don't even care anymore; i consider this permanently fucked, hopefully people who actually play the game will test it properly and comment on it
16:57 VanessaE_ well if you meant that, this is why:  https://github.com/minetest/minetest/blob/master/src/mapblock_mesh.cpp#L328
16:58 VanessaE_ as for not caring... well nothing in code is ever permanently fucked you know :P
16:58 celeron55 okay that change is probably reasonable then
17:00 celeron55 i mean, i cannot check whether it is good or not in an objective way
17:00 celeron55 it's just wrong in a way or another
17:01 celeron55 it was pulled out of my ass originally, and changing something that was pulled from my ass based on someone else's ass is just that; the only way to judge it properly is to ask users
17:01 celeron55 and i am not a user
17:02 VanessaE_ haha well at least we know the origin of that value now :D
17:02 celeron55 my opinion is however that rba has no right to dictate it and he has to ask about it from many people
17:02 VanessaE_ RBA and I were trying to figure out if there was a reason, or if it was a typo
17:02 celeron55 i would guess that line was a typo; not sure though
17:03 celeron55 as i didn't attempt to make shader and shaderless rendering look the same
17:03 Amaz joined #minetest-dev
17:03 VanessaE_ personally, the way the patch behaves looks fine to me, though I'm used to the tops being a bit brighter -- but that's a problem with the normalmaps code, not the lighting, per se.
17:03 celeron55 i just wanted to make them look good each on their own
17:03 VanessaE_ of course.
17:07 zat joined #minetest-dev
17:20 rubenwardy joined #minetest-dev
17:33 grrk-bzzt joined #minetest-dev
17:33 grrk-bzzt joined #minetest-dev
17:49 grrk-bzzt joined #minetest-dev
17:53 sapier joined #minetest-dev
17:56 PenguinDad It looks like #1553 fixes #1501
17:57 sapier shadowbot doesn't work so links would be helpfull ;-)
17:57 VanessaE_ sapier, please update dreambuilder and give it a try with the profiler again.  plants_lib got a major speed boost tonight and it/its mods should look much better now.
17:57 VanessaE_ s/tonight/last night/
17:58 VanessaE_ also hi :)
17:58 VanessaE_ sapier:  https://github.com/minetest/minetest/pull/1553
17:58 PenguinDad sapier: https://github.com/minetest/minetest/pull/1553 & https://github.com/minetest/minetest/issues/1501
17:58 sapier hello, once I finished my current cleanup
17:58 sapier nice
18:04 khonkhortisan joined #minetest-dev
18:14 Amaz joined #minetest-dev
18:22 Garmine joined #minetest-dev
18:25 Megaf joined #minetest-dev
18:26 luizrpgluiz joined #minetest-dev
18:28 luizrpgluiz developers, I've been thinking of some ideas for mods minecraft that could be transformed into modifications or implementation for minetest :)
18:28 Megaf Oh, he said the word that shall not be said
18:29 * Megaf runs
18:30 luizrpgluiz sorry :(
18:31 luizrpgluiz but it would be so good ideas
18:49 loggingbot_ joined #minetest-dev
18:49 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/
18:52 VanessaE_ sapier: so what I did, in case you're curious.. I routed the main reg call through a routine to just log all the surfaces that need to be modified, had that also record all the biomes associated with them, then filtered out the dupes, and passed the list to a single register_on_generated call.
18:52 VanessaE_ except for calls where surfaces need to have air checking disabled.  those still go through the old code.
18:53 VanessaE_ then that one single on_gen call evaluates the surface list and then parses the list of biomes and populates the block.
19:02 VanessaE joined #minetest-dev
19:05 proller joined #minetest-dev
19:49 khonkhortisan joined #minetest-dev
20:21 sapier VanessaE: thanks I already use a similar mechanism but quite a different way ... and I found a solution too now I'm below 170ms may
20:21 sapier max
20:21 VanessaE nice
20:22 sapier unless you disable fishes then it's way less :-)
20:22 VanessaE of course, you don't have to evaluate surfaces quite to the same degree that I do ;)
20:23 sapier yes but for fishes I need to know what's around ... spawning a shark in a small lake is quite strange for example
20:23 VanessaE haha
20:23 VanessaE true that.
20:23 VanessaE so do you figure your profiler will be suitable for mainline?
20:23 VanessaE it seems incredibly useful
20:24 Megaf sapier: are you making a fish mod?
20:24 sapier true but I don't have a way to show the stats
20:24 VanessaE true.
20:24 sapier Megaf:  no but one of my mobs is a shark
20:24 Megaf interesting
20:24 sapier printing stats in 60s raster is usefull for debugging but not for mineline
20:24 sapier +a
20:24 VanessaE sapier: how about extra pages added to the F6 displays, with a command or key to dump them to the console?
20:25 sapier F6 display would require a lot of c++ code ... chat command to dump once would be quite small
20:25 VanessaE hm
20:25 VanessaE hm
20:25 VanessaE oops.
20:26 VanessaE well a chatcommand would work at least.
20:26 VanessaE wasn't there a key already defined that ..... wait, Home isn't it?
20:26 VanessaE it prints "something" but nothing useful that I recall
20:27 VanessaE hm, no wasn't home
20:27 VanessaE that's quicktune.,
20:27 VanessaE which does nothing anyway :D
20:27 VanessaE P prints debug stacks..
20:28 sapier :-)
20:28 VanessaE F8 does nothing.  Tie the print-profiler to that?
20:31 VanessaE or better, F8 to toggle enable/disable of the profiler, and F9 to print the output.
20:32 VanessaE when turned from off to on, it would reset its counters to 0
20:32 Megaf mplayer http://31.12.64.204/SlowRadioCom or vlc http://31.12.64.204/SlowRadioCom
20:32 VanessaE just brainstorming.
20:32 Megaf such a good radio
20:32 Megaf wrong channel, sorry
20:40 sapier I think I'm gonna do it the chatcommand way
20:40 asl joined #minetest-dev
20:41 sapier it's usually way to much data to print it in game
20:41 VanessaE sapier: nonono, print to the *console*
20:41 VanessaE not in-game
20:41 VanessaE er
20:41 VanessaE print to the *terminal* I mean
20:41 sapier action stream?
20:41 VanessaE but using a couple of hotkeys to trigger it.
20:41 sapier logfile + console
20:42 VanessaE well I dunno which stream is best; this seems like something that should go to errorstream so that you can be sure you get it regardless of log level
20:42 Jordach_ joined #minetest-dev
20:42 sapier action is the only one printed to console but not in game
20:42 VanessaE hm
20:42 VanessaE well ok
20:42 VanessaE what log level does that belong to?
20:44 sapier "action"
20:46 VanessaE heh
20:46 VanessaE I meant numerically.
20:47 sapier I don't know
20:47 VanessaE oh well no worries
20:47 VanessaE you know what you're doing.
20:47 sapier sometimes
20:48 VanessaE incredibly useful tool no matt.......  hahahah
20:48 VanessaE .... no matter how you decide to print the output.  As long as it can be copied from the terminal to somewhere, and the output is accurate, that's all that matters, right?
20:49 sapier define "accurate" ;-)
20:50 VanessaE accurate enough to use as a rough guide :)
20:51 VanessaE enough to say "this mod took 10ms over the past 5 minutes" and it really did take between 9 and 10 ms :)
20:52 sapier tracing doesn't show things like that
20:53 sapier it's average per server step thus it should be in submilisecond area at best
20:53 VanessaE I was kidding :)
20:53 sapier and even more important are usually the worst case values
20:54 VanessaE I just meant that if it says something like 10ms, it really was no more than 10ms for whatever the measurement interval was.
20:54 VanessaE rather than reading 10ms when the measure was really 200000 ms ;)
20:54 sapier that's not very likely right now ... at least not if I didn't add a bug
20:56 sapier I really hate java
20:56 VanessaE ?
20:56 sapier did you ever use java applications on linux?
20:56 VanessaE what, trying to port your profiler over to That Other Voxel Game? ;)
20:56 VanessaE oh sure, a few times
20:57 sapier java is unstable cpu eating memory consuming and just slow as hell
20:57 VanessaE this is true.
20:57 VanessaE what are you doing that demands java?
20:59 sapier eclipse ...
21:00 sapier great ide but very very bad choice on language
21:01 VanessaE ahhhhh
21:01 VanessaE am I ever glad I still code the old fashioned way
21:01 VanessaE text editor (+compiler as needed) + the program to test it with.
21:02 sapier well once you got used to all those nice and shiny things making life so much easier you don't wanna miss them
21:02 Zefram_Fysh oh wow, I thought you liked to do everything through web interfaces
21:02 sapier do you have a eclipse plugin for your storage ;-)
21:03 sapier still Zefram_Fysh I never did review or discuss code using eclipse ;-)
21:09 sapier joined #minetest-dev
21:13 Anchakor_ joined #minetest-dev
21:16 VanessaE sapier: wrong button? :)
21:18 Calinou make sure you have a recent version of Java (OpenJDK)
21:18 Calinou and make sure software slowness isn't your fault (happens too often)
21:19 sapier nope restarted
21:19 sapier didn't help ... and found out eclipse crashes due to some feature running havoc I didn't even know it exists
21:20 sapier well I usually don't expect a built in feature to cause an application to deadlock ... and this doesn't happen on windows too
21:20 sapier and by now I didn't experience any difference between different java versions on linux they're all crap
21:22 Calinou report bug
21:22 sapier which one? it's slow and crashes without usefull error on different seemingly unrelated locations?
21:23 sapier I know what I would do with a error report like that
21:23 sapier "nice to know but I can't do anything about it"
21:23 sapier does anyone know why chatcommands all return invalid command even if they work?
21:24 sapier hmm maybe a tracing issue
21:31 alexxss joined #minetest-dev
21:35 sapier https://github.com/minetest/minetest/pull/1557 here it is ... please test and give feedback everyone
21:41 realbadangel_ hi sapier, Calinou
21:42 VanessaE greetz.
21:42 realbadangel_ you had "hello" in pm ;)
21:42 realbadangel_ https://github.com/minetest/minetest/pull/1553
21:43 Calinou I had no PM on IRC recently
21:43 realbadangel_ sapier, Calinou going to merge that now, have enough votes
21:43 sapier rba that's a big promise ;-)
21:43 realbadangel_ but if you have some comments... youre welcome
21:43 Calinou does it change visuals much?
21:43 sapier are you absolutely sure you didn't miss a single case? ;-)
21:43 realbadangel_ a bit
21:44 sapier define "a bit" ;-)
21:44 realbadangel_ colors are slightly different but very same for each mode
21:44 realbadangel_ try it, its kinda cosmetic
21:45 sapier usually those changes are only visible on direct comparison ;-)
21:45 realbadangel_ VanessaE said theyre darker, paramat was saying too bright
21:45 realbadangel_ thats something in the middle
21:45 VanessaE only darker if normalmaps are turned on
21:45 VanessaE otherwise they're fine.
21:46 realbadangel_ normalmaps and bumpmapping brings the shadows
21:46 VanessaE you just need to tune the lighting in the normalmaps code to handle the changed brightness
21:46 realbadangel_ dark guys
21:46 realbadangel_ so no wonder ;)
21:46 sapier rba maybe you should stop changing it by some time as you'll never find something really matching everyones taste ;-)
21:46 realbadangel_ theres no taste that should fit everyone
21:47 realbadangel_ but taste is not the point of the change
21:47 realbadangel_ this code solves many issues
21:47 VanessaE sapier: in this case it's fixing a couple of things that actually did kinda go awry before.
21:48 realbadangel_ this is not cosmetic change
21:48 sapier well we're fixing this for about half a year now (at least felt) ... it's time to make it right for at least some time now
21:48 realbadangel_ it really deserves the name "unified"
21:48 realbadangel_ 4 modes, 3 different approaches for shading the tiles
21:48 realbadangel_ now unified
21:49 realbadangel_ and with speedup
21:49 realbadangel_ also solving a few bugs
21:50 realbadangel_ do you remember water/lava sources and flowing having different brightness?
21:50 realbadangel_ we were trtying hard to tune the textures to look the same
21:50 Calinou normal maps will always darken, by nature
21:50 realbadangel_ that code fixes that
21:51 Calinou realbadangel_, bump mapping and normal mapping are separate things
21:51 Calinou I prefer saying “parallax mapping” in order to not make things confusing
21:51 realbadangel_ no
21:51 Calinou so how are bump mapping and normal mapping the same things?
21:51 realbadangel_ parallax mapping doesnt use normal maps
21:51 Calinou yes, it uses parallax maps.
21:51 Calinou which represent altitude;
21:51 realbadangel_ it just need height
21:51 Calinou height maps, same thing.
21:51 sapier if it's fine now merge it but please check everything at least twice prior merge as there've been to many glitches last times
21:52 realbadangel_ which is usually stored in alpha channel of a normal map
21:52 Calinou <realbadangel_> do you remember water/lava sources and flowing having different brightness?
21:52 Calinou it's really cool, if you fixed that
21:52 Calinou does it bring any performance improvement?
21:52 realbadangel_ yes it does
21:52 Calinou realbadangel_, yeah, but that requires normal map to be .png, so sometimes the height map is separate
21:53 realbadangel_ there are different approches, i have chosen one compact
21:53 realbadangel_ normal = rgb + height in alpha
21:53 Calinou yes, it's generally a good thing
21:53 Calinou but for 512² or higher, you don't want .png which can be too large
21:53 Calinou thankfully, most Minetest textures are 256² at most
21:54 realbadangel_ i made lava with 2k
21:54 realbadangel_ this rocks the house
21:55 PenguinDad Would be funny if it's faster than 16² :D
21:59 sapier as of minetests pov it doesn't matter if you use png jpg or even bmp
21:59 sapier ok maybe time reading from disk differs :-)
22:00 VanessaE sapier: in this case it matters because you want the normalmap to be lossless
22:00 VanessaE otherwise it will introduce errors into the normalmap's gradients.
22:01 sapier ok but that's not (direct) related to it's size
22:01 PenguinDad But png would be a better choice than bmp anyways
22:03 sapier yes especally as someone is supposed to download them ;-)
22:04 VanessaE heh
22:05 RealBadAngel in time of SSD's havin them all premade at greater resolutions can be a good way
22:05 RealBadAngel see Tiled texture packs
22:05 RealBadAngel with a good normalmap you can make miracles even out of 16px
22:09 RealBadAngel Calinou, i will fix your acid liquids too, but with another pull
22:11 RealBadAngel before i have to push 2 different bug fixes to mapblock_mesh
22:12 RealBadAngel then i can fix your isssue and rebase the water/lava shaders again
22:24 ^v joined #minetest-dev
22:29 sapier left #minetest-dev
22:30 grrk-bzzt left #minetest-dev
23:19 fu-fu joined #minetest-dev
23:33 domtron joined #minetest-dev

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