Minetest logo

IRC log for #minetest-dev, 2015-06-13

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

All times shown according to UTC.

Time Nick Message
00:55 paramat joined #minetest-dev
00:56 paramat hmmmm and all https://github.com/minetest/minetest/pull/2777 any comments?
01:01 Darcidride joined #minetest-dev
01:19 Wayward_Tab joined #minetest-dev
01:25 Wayward_One joined #minetest-dev
01:43 paramat i'll change one thing: disable decoration code in mg_decoration instead of in each mapgen. then i might push this later since i am apparently in charge of the biome API
01:47 hmmmmm joined #minetest-dev
01:48 paramat O/
01:52 hmmmmmm joined #minetest-dev
02:25 hmmmmmm oh no...
02:25 hmmmmmm paramat, why disable decorations below water level?
02:26 paramat they were already had a lower limit of water level, something i did a while back
02:26 paramat but sea floor decos..
02:26 hmmmmmm i'd really rather not have any restriction at all
02:27 paramat okay
02:27 hmmmmmm sea floor decorations, in-cave generations, etc.
02:27 hmmmmmm it's not really that helpful of an optimization either
02:27 hmmmmmm yikes i got highlighted a lot the past few days
02:27 hmmmmmm did something implode in minetest?
02:27 paramat decos can only be placed under open sky, so only if a cave intersects the mapchunk top
02:28 paramat i was busy and asked for reviews =)
02:28 hmmmmmm est31:  yes, that does look fine to me
02:28 hmmmmmm hmm
02:28 hmmmmmm about when to do a release
02:29 hmmmmmm well more people have to be involved to release things
02:29 hmmmmmm i learned the hard way that you can't just say "we're gonna have a release in X weeks"
02:30 hmmmmmm you have to track down all the people who make a build happen and then coordinate with them first, otherwise you have a fragmented half-assed release that people get upset with
02:30 paramat i enabled water-surface decos a while ago by creating a lower limit of water level for deco placement. but now i agree it's better to not restrict it, i will change that back in this PR
02:31 hmmmmmm huh?  what do you mean
02:31 * hmmmmmm looks...
02:31 paramat that might have been when you were away
02:32 hmmmmmm >y = MYMAX(y, mg->water_level);
02:32 hmmmmmm oh dear
02:32 paramat yeah
02:32 hmmmmmm that was never supposed to be the intent of biomes
02:32 hmmmmmm if you want to do that, please do it with a flag instead...
02:32 hmmmmmm errm, the intent of decorations* i meant to say
02:32 paramat it wasn't a great decision, will revert
02:32 hmmmmmm well don't revert the whole thing, i think there's value in placing decorations on water level
02:32 hmmmmmm like lilypads or so on
02:33 hmmmmmm just make it optional
02:33 paramat sure, can i work out how to do that later, after removing restrictions in this commit?
02:34 hmmmmmm paramat, you can do anything you want to do.
02:34 hmmmmmm the sky is the limit.
02:34 paramat lol
02:34 est31 hmmmmmm, paramat has said they want one more month. I wonder whether clientside scripting can be finished in 2 months?
02:34 paramat lol
02:37 hmmmmmm est31, the next version is definitely not going to be 1.0.0 or 0.5.0 or whatever
02:37 paramat noise defined biome blend has a new implementation and is ready for review https://github.com/minetest/minetest/pull/2764
02:38 est31 0.4.13
02:39 paramat (i'll be around on and off for the next few hours)
02:41 hmmmmmm lol
02:42 hmmmmmm paramat, you sure you want to have two separate noises for the biome blending?
02:42 hmmmmmm also I told you not to use a Noise object
02:42 hmmmmmm try comparing the two in speed and memory usage, you'll understand why :)
02:43 paramat a perlinmap seemed a faster method, guess i was wrong
02:43 paramat (lua habits)
02:44 paramat yeah i tried 1 noise but that resulted in straight borders in places
02:44 paramat seems to need 2
02:46 paramat explanation is in the thread
02:47 paramat what is different about blend noise that makes point-polling faster?
02:47 paramat anyway i'll change the PR
03:00 paramat oh yes, small spread is the difference, nevermind =)
03:02 hmmmmmm paramat: the spread factor
03:03 hmmmmmm below a spread factor of, say, 75, i'd say that's the threshhold at which buffered bulk noise loses its advantages
03:04 paramat okay
03:13 paramat back in an hour or so
03:13 paramat left #minetest-dev
04:20 Miner_48er joined #minetest-dev
04:32 est31 joined #minetest-dev
04:33 est31 man 100 PRs again
04:33 est31 bad bad
05:17 electrodude512 joined #minetest-dev
05:38 RealBadAngel joined #minetest-dev
06:01 Hunterz joined #minetest-dev
06:06 nore joined #minetest-dev
06:19 paramat joined #minetest-dev
06:22 paramat hi nore, is the implementation of bucket groups correct and backwards-compatible https://github.com/minetest/minetest_game/pull/528 ? any comments?
06:23 nore looks fine
06:23 nore ah, yes, there is a problem
06:24 nore wait, no actually
06:24 nore :)
06:25 paramat okay thanks
06:29 paramat hmmmmmm, see https://github.com/minetest/minetes​t/pull/2764#issuecomment-111679349
06:36 paramat perhaps i need to add '#include mapgen.h' to mg_biome.cpp/.h ..
06:57 paramat left #minetest-dev
07:05 OldCoder joined #minetest-dev
07:07 kilbith joined #minetest-dev
07:32 Calinou joined #minetest-dev
07:42 troller joined #minetest-dev
08:00 Yepoleb_ joined #minetest-dev
10:05 Krock joined #minetest-dev
10:30 Amaz joined #minetest-dev
11:10 Dartmouth joined #minetest-dev
11:27 RealBadAngel joined #minetest-dev
11:58 Taoki RealBadAngel: Woah... the new shaders can be tested now? Where can I get the branch please? I've been wanting to try them for over an year now, since you announced these shaders :)
11:59 Taoki Also, is there an estimation as to when extruded plants and torches will be in? I so, so want to see these happenig too! Last time you mentioned they had issues with waving however :(
11:59 RealBadAngel Taoki, atm new parallax shaders are rdy
12:00 Taoki Nice. Any screensgots of that? What's different?
12:00 RealBadAngel hold on a sec
12:00 Taoki Sure
12:00 * Taoki really wishes there was an easier way to keep up with your shader and lighting progress. That's like the #1 thing I'm interested in with Minetest right now.
12:00 Taoki **changes
12:01 RealBadAngel https://imgrush.com/h_F-0fk8FC8Z
12:01 Krock pls4 minimap
12:02 Taoki RealBadAngel: Just amazing! I can clearly see a great difference
12:02 Taoki Commit soon, please <3
12:03 RealBadAngel this is on the fly effect works with any tp
12:03 Taoki Anyway, where can I test the water shaders as well? And other new shaders you're making?
12:03 Taoki I know, that's even more awesome about them :)
12:03 RealBadAngel i will redo them soon, now i do have tangents
12:04 Taoki ok
12:04 RealBadAngel parallax PR is almost rdy, im just polishing it
12:04 Taoki Will we see them in master this summer vacation? Sorry if being asked feels pressuring... certainly it's something to be eager for :)
12:05 RealBadAngel i will do my best
12:05 * Taoki nods. Thanks!
12:05 RealBadAngel now the parallax, then minimap (it needs parallax)
12:06 RealBadAngel later on connected textures and back to shaders
12:06 Taoki Also, I again can't help my curiosity: Have you taken a look at implementing the basic framework for post-processing? Bloom is one thing I really want :)
12:07 RealBadAngel i do have  somwhere irrlicht water renderer
12:07 RealBadAngel it uses all the things  needed by that
12:08 Taoki Ah. So the water shaders add a post-processing pipeline too?
12:08 Calinou when do we switch to Vulkan? :-)
12:08 * Calinou runs
12:08 RealBadAngel minimap is also using shaders
12:09 Taoki Very nice
12:09 Taoki Calinou: Irrlicht might. If they do it, so might we
12:10 Taoki BTW: I have this idea to implement voxel smoothing, to allow for smooth random edges on terrain and such. Thinking about opensubdiv... not sure if that's a GLSL thing however.
12:13 RealBadAngel rather not
12:13 Taoki Yeah
12:13 RealBadAngel if you would like to make terrain smoother why not use meshnodes?
12:14 RealBadAngel theres no limit in number of vertices
12:17 Taoki I think they're more of a hack and less of a solution
12:17 Taoki Should be done in the code IMO
12:18 Taoki Been mostly thinking of marching cubes and the like
12:19 RealBadAngel you can do with meshes anything you want
12:19 kilbith smooth terrain is like travesty the nature of the game
12:19 kilbith play Rust instead
12:20 Taoki For one thing, you have to define each corner / terrain type as an individual node, which I think is a bit wrong. Another problem is I'm not sure if performance will be lower this way.
12:20 RealBadAngel http://i.imgur.com/eKgfuns.png
12:20 Taoki There is however this... closest to my idea: https://forum.minetest.net/vi​ewtopic.php?f=47&amp;t=11982
12:21 Taoki Would help a lot if there was a function to change the mesh of individual nodes on the run.
12:30 Taoki Actually, I'm not even sure if that was implemented. Can you currently change the nodebox of an individual node?
12:33 * Taoki really thinks that's something that should be implemented. If anyone can easily do it, of course.
12:34 RealBadAngel there were attemps to do so
12:35 RealBadAngel c55 started the code, nore was also working on it
12:37 Taoki I see
13:00 H-H-H joined #minetest-dev
13:37 LittleJoe joined #minetest-dev
13:45 Hunterz joined #minetest-dev
13:48 Hunterz joined #minetest-dev
13:52 Taoki Hmm... why is it that a lot of leafs decrease performance so badly?
14:06 LittleJoe1 joined #minetest-dev
14:09 proller joined #minetest-dev
14:10 LittleJoe2 joined #minetest-dev
14:23 LittleJoe1 joined #minetest-dev
14:27 RealBadAngel Taoki, because of transparency and lotsa more things to render
14:32 LittleJoe2 joined #minetest-dev
14:50 Hunterz joined #minetest-dev
14:51 MinetestForFun joined #minetest-dev
15:07 hmmmm joined #minetest-dev
15:16 proller joined #minetest-dev
15:38 Hunterz joined #minetest-dev
15:49 est31 joined #minetest-dev
17:12 est31 anybody time for "PR merging session"?
17:14 Wayward_One joined #minetest-dev
17:21 nore est31: I can help a bit, but I don't have much time now
17:23 est31 e.g. #2751 only needs a second +1
17:23 ShadowBot https://github.com/minetest/minetest/issues/2751 -- Add minetest.register_on_player_hpchange by TeTpaAka
17:24 est31 #2754 and #2742 to
17:24 est31 o
17:24 ShadowBot https://github.com/minetest/minetest/issues/2754 -- Fix bug when craft input isn't replaced by TeTpaAka
17:24 ShadowBot https://github.com/minetest/minetest/issues/2742 -- Add return list of individual counts to find_node_in_area by TeTpaAka
17:25 est31 I think I'll merge 2655
17:25 est31 #2655
17:25 ShadowBot https://github.com/minetest/minetest/issues/2655 -- Clean up mod profiler by ShadowNinja
17:30 est31 2754 has an open issue though, so perhaps we'd need to wait with it
17:30 proller joined #minetest-dev
17:36 kilbith nore ^
17:36 nore oops, sorry
17:36 Amaz joined #minetest-dev
17:37 nore est31, 2655 looks fine
17:37 est31 ah 2655 cant be merged yet
17:38 est31 not before sn's logging PR is merged
17:38 est31 #2628
17:38 nore 2742 is good too
17:38 ShadowBot https://github.com/minetest/minetest/issues/2628 -- Clean up logging by ShadowNinja
17:40 nore 2751 is fine as well
17:40 est31 pushed them
17:40 nore 2628 is not completed, so we will have to wait before 2655
17:42 est31 then there is #2771
17:42 ShadowBot https://github.com/minetest/minetest/issues/2771 -- UTF8 in formspecs by est31
17:47 nore est31: I want it :) (if it works well, of course :p)
17:48 est31 it works yes
17:49 est31 gonna merge it then
17:50 est31 oh trailing whitespace
17:51 est31 I need iconv for further work
17:51 nore sorry, gtg now :/
17:51 est31 bye
17:51 est31 already helped alot :)
17:52 Calinou a lot*
17:52 zat joined #minetest-dev
17:54 est31 argh dammit
17:55 est31 I totally forgot that we hard depend on iconv now.
17:55 est31 perhaps I should document that
17:56 est31 or is iconv unix??
17:58 est31 it is
17:58 est31 we need to do something for windows though
17:58 est31 pointing out that it isn't optional anymore
17:58 sfan5 gettext comes with iconv
17:59 sfan5 at least i have it in the windows libs
17:59 est31 gettext is optional
17:59 est31 iconv isn't
18:01 est31 perhaps the GETTEXT_ICONV_DLL cmake option name should be changed to to ICONV_DLL
18:02 sfan5 we also need ICONV_INCLUDE_DIR and ICONV_LIBRARY then
18:02 est31 iconv is usually part of glibc on unix
18:02 est31 https://en.wikipedia.org/wiki/Iconv
18:03 sfan5 yes
18:03 sfan5 but whata about the windows people that want to build without gettext
18:03 sfan5 -a
18:03 est31 for those we don't need ICONV_LIBRAR>
18:03 est31 Y*
18:04 est31 only ICONV_DLL and ICONV_INCLUDE_DIR
18:06 est31 All recent Linux distributions contain a free implementation of iconv() as part of the GNU C Library"
18:06 est31 "
18:06 est31 that part is in wikipedia for quite some time
18:06 Calinou musl users will whine :P
18:07 est31 they can send a PR
18:07 sfan5 est31: why no ICONV_LIBRARY?
18:08 est31 sfan5, I dont think its needed, but you can add it if you want.
18:08 est31 Calinou, it seems musl does include iconv
18:09 sfan5 est31: last time i check you need to link to libs to use them
18:09 sfan5 iconv is a seperate library on windoqws
18:09 sfan5 -q
18:09 est31 ah you need ICONV_LIBRARY for windows sfan5
18:09 est31 then add it
18:10 sfan5 <sfan5> but what about the windows people that want to build without gettext
18:10 sfan5 <est31> for those we don't need ICONV_LIBRARY
18:13 Calinou >2015 >building without gettext
18:16 sfan5 Calinou: that was the use case i could think of fastest
18:17 sfan5 iconv library at a different location if far more likely
18:28 OldCoder joined #minetest-dev
18:30 est31 somehow there is an issue with shaders
18:37 est31 https://github.com/minetest/minetest/issues/2780
18:40 paramat joined #minetest-dev
18:40 kilbith might be the same issue : http://irc.minetest.ru/minet​est-dev/2015-06-11#i_4284307
18:42 VanessaE est31: that happens to me, too
18:42 VanessaE (and I don't have to be flying to see it)
18:42 paramat hi sfan5, if you approve of this i can push it later https://github.com/minetest/minetest_game/pull/528 -- Bucket: Register river water and add texture
18:43 est31 VanessaE, somehow the bug shows twofold
18:43 VanessaE twofold?
18:43 est31 1. this spontaneous blinking shown in the gid
18:44 est31 gif*
18:44 VanessaE yes
18:44 est31 only when you have disabled full viewing range
18:44 VanessaE so it's an interaction with the fog then
18:44 est31 and 2. wrong rendering of areas when viewing range is full
18:45 est31 but then depending on pitch and yaw, and staying that way
18:45 VanessaE 2?  not sure I..
18:45 VanessaE oh wait
18:45 VanessaE yeah I have seen that second effect before as well (and it didn't depend on full view range)
18:52 rubenwardy joined #minetest-dev
18:55 Wayward_One joined #minetest-dev
18:57 sfan5 paramat: looks good
18:57 paramat thanks!
18:58 Krock joined #minetest-dev
18:58 sfan5 left #minetest-dev
18:58 sfan5 joined #minetest-dev
19:20 proller joined #minetest-dev
19:30 paramat left #minetest-dev
20:20 MinetestForFun joined #minetest-dev
20:29 troller joined #minetest-dev
20:38 proller joined #minetest-dev
20:44 Lunatrius joined #minetest-dev
20:45 guest__ joined #minetest-dev
22:07 Taoki Huh... just realized I never asked myself this: Does Minetest not yet support multi-threading? Or does it, and various tasks can be distributed to various CPU's?
22:08 Taoki **CPU cores
22:09 est31 Taoki, minetest mapgen supports it to some extent
22:09 Taoki Ah, nice
22:12 kahrl yeah, mapgen, meshgen and parts of the network stack are in background threads
22:12 kahrl pretty much everything else is not
22:13 est31 meshgen too? how does that work, irrlicht isnt multithreaded?
22:13 est31 thread safe*
22:15 kahrl you can create mesh buffers and meshes wherever you want
22:16 kahrl as long as you only add them to the scene in the main thread
22:17 est31 irrlicht isn't used to create those mesh buffers?
22:17 kahrl of course it is
22:17 est31 so in that regard, it is thread safe?
22:18 kahrl but something like "new scene::SMesh" does not access the gpu or global irrlicht state
22:18 est31 I see
22:18 est31 but when you dont draw on those buffers, what do you do then?
22:19 kahrl what you have to be careful about is loading textures/materials only on the main thread, which is what the queues in ITextureSource/IShaderSource are for
22:19 kahrl est31: you can add vertices and indices to a mesh on any thread, as long as you only do it on one at a time
22:20 kahrl it's just data, basically
22:20 kahrl it's only sent to the gpu once you add it to a scene
22:20 est31 so adding elements to a list is done in an extra thread?
22:20 kahrl (and draw that scene)
22:20 kahrl yeah
22:30 deltib joined #minetest-dev
22:58 paramat joined #minetest-dev
22:58 paramat will push https://github.com/minetest/minetest_game/pull/528 soon
23:05 Wayward_Tab joined #minetest-dev
23:22 prozacgod joined #minetest-dev
23:33 JZTech101 joined #minetest-dev

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