Time Nick Message 00:00 VanessaE UGH, this half-double-click shit has to GO. 00:01 sapier atm doubleclick is issued if you click twice within 500ms 00:03 VanessaE sapier: it happens even with a single click sometimes. 00:03 VanessaE sapier: also, cursor keys no longer function in the list 00:03 VanessaE (used to be you could move the highlight bar up/down) 00:03 sapier yes I don't think they will work anytime soon 00:05 sapier I don't think it's a showstopper issue 00:05 VanessaE no, but with that mis-handling of click/double-click, it is. 00:05 VanessaE the menu is less usable now than before because of it. 00:06 sapier two possibilitys disable doubleclick completely or force irrlicht guys to add that doubleclick patch someone wrote years ago 00:06 VanessaE then disable it completely. 00:06 VanessaE especially if "enter" is sufficient to get into a world. 00:07 sapier it's working for most ppl you know? 00:07 VanessaE consistency. If a basic function like a double click doesn't work properly for someone, the problem needs to be fixed or the function needs to be disabled. 00:08 VanessaE I can't be the only one who has this trouble - I'm just the only one who's reported it. 00:08 PilzAdam chat doesnt work on sfan5's phone, lets disable it until its fixed 00:08 VanessaE PilzAdam: non sequitur. 00:08 VanessaE chat isn't a "basic function". a double-click is. 00:09 VanessaE I'm talking about from a user interface standpoint, not something application-specific 00:09 sapier vanessa normaly ppl complaining are overrepresentet due to those not having problems dont bother to report ;-) 00:10 VanessaE sapier: standard rule of communication: for every one person who has a complaint, there are 10 more who have the same problem but won't bother to say anything. 00:18 sapier and 100 who don't have a problem ... but I don't wanna discuss the problem away I just don't understand why it's that worse for you 00:20 VanessaE I wish I knew 00:21 VanessaE it's always been glitchy like that for me, but without those cursor keys, it's that much worse 02:13 hmmmm aghk 02:13 hmmmm i never thought placing some nodes on top of some other nodes could be so freaking difficult 02:14 hmmmm then again, mapgen v6 had two distinct advantages that made this incredibly easy: 02:14 hmmmm 1). no terrain overlapped, and 2). the terrain never went above 48 nodes high 02:17 VanessaE in plants_lib I had to resort to manually scanning the whole block for targets, then extracting the surface from that 02:17 VanessaE s/block/chunk/ 02:18 VanessaE (both were done in 1 step, but in a somewhat naive way) 02:20 VanessaE once I have the surface map, then I just blindly assume ;) that the next node above each is the real target (except in a certain mode) and the object gets placed there. 03:09 hmmmm i can't assume things or else it'll look wrong 03:09 VanessaE well you've got a lot more data to work with than I had at the time, so you'll surely figure it out 03:09 hmmmm yeah i know what to do... 03:12 VanessaE well have fun :) 03:12 VanessaE I'm gonna try to get some sleep. 03:12 hmmmm i have to do a check against the noise for that height 03:12 hmmmm that point* 03:13 hmmmm i have to preserve the heightmap that i get from generating the terrain, not the one i recalculate 04:09 ShadowNinja Does this look good? http://ix.io/6wP 04:10 ShadowNinja (Untested) 04:16 kaeza ShadowNinja, can you add metamethods to it? 04:17 ShadowNinja kaeza: What do you mean? v:add(v2)? 04:17 kaeza no 04:17 kaeza like, if v1 == v2 then ... instead of if v1.compare(v2) 04:18 kaeza (or v1 = v1 + v2) 04:18 ShadowNinja kaeza: Oh, right, Lua can do that. I will have to look into it. 04:20 kaeza you'll need to create a Vector "class" (which would be basically a table plus metamethods) 04:21 ShadowNinja Yes, that means the regular pos tables won't work... 04:22 * kaeza wonders if that could be added to tables returned by core or would be an overkill 04:23 kaeza anyway, the code looks good as-is :) 04:23 kahrl ShadowNinja: just an idea, but you could use a (numerically) better algorithm for vector.distance 04:23 hmmmm kaeza, i was thinking about doing this for about 5 seconds before i realized how many things use v3fs 04:24 hmmmm you'd have to keep doing things like local vec = v3f:new{myplainvec} 04:25 hmmmm but it would be cool to use metamethods 04:25 kaeza hmmmm, I mean directly in core 04:25 kaeza isn't there a Lua helper to create v3f tables? 04:25 hmmmm push_v3f? 04:25 kaeza yep 04:25 kaeza you could just assign a metatable inside that function 04:26 hmmmm yeah but the problem is that you'd have to make that a usertype 04:26 hmmmm userdata* type rather 04:26 kaeza uhh no 04:26 ShadowNinja kahrl: Like? What is wrong with it? 04:26 hmmmm no? 04:26 kaeza tables can have metatables 04:26 kahrl ShadowNinja: http://en.wikipedia.org/wiki/Hypot#Implementation 04:27 kahrl to extend that to 3 dimensions use hypot(x,y,z) = hypot(x,hypot(y,z)) 04:40 ShadowNinja So should I use metatables and add a seperate vector table? 04:40 kaeza hmmmm, http://pastebin.com/dPWRXE2M 04:40 kaeza (untested) 04:40 hmmmm why would you have the add function in C 04:41 kaeza so for example player:getpos() already has the metamethods 04:41 hmmmm what'd be ideal would be to have the arithmetic things in pure Lua so there's no switching back and forth 04:45 kaeza hrm... OTOH, this kind of thing leads to ugly side effects (either in Lua or in C++) 04:45 kaeza so I guess the functions are better 04:57 ShadowNinja hmmmm: I have a few of these: http://pastebin.ubuntu.com/5848854/ 04:58 hmmmm yup 05:05 kahrl I think those should be fixed 05:06 kahrl the byte order of these literals is implementation-defined and I think on big endian architectures the compilers interpret them differently 05:07 hmmmm nonsense 05:07 hmmmm the only issue would be if there was a platform we supported that used 16 bit ints 05:07 kahrl explain why big endian is not an issue 05:07 hmmmm if byte order did matter, then 'a' wouldn't be 'a' on different architectures 05:08 kahrl 'a' is one byte 05:08 hmmmm actually 05:08 hmmmm the expression 'a' is of type int 05:08 kahrl and yes it can be a different value on different architectures, but we only support ASCII ones 05:09 kahrl 'a' is 97 on big and little endian architectures (if they use ASCII) 05:09 kahrl 'MTSM' is not the same 05:09 hmmmm i don't believe you at all. 05:10 kahrl it's hard to find a reliable source for this (I can't test it myself) but for example http://www.mail-archive.com/palm-dev-forum@news.palmos.com/msg40918.html 05:11 hmmmm i'm firing up my sunblade for this 05:15 kahrl https://gist.github.com/kahrl/5938740 (untested) 05:16 hmmmm erm, if we're really going to change it, i'd rather just change it to an integer constant 05:16 hmmmm #define MT_SCHEM_FILE_SIGNATURE 0x5035904whatevr 05:16 kahrl your choice 05:21 ShadowNinja kaeza: Metatables aren't working. http://ix.io/6wS 05:22 hmmmm kahrl, http://pastebin.com/hTgpkM1f 05:25 kahrl prints little endian for me 05:25 hmmmm 'abcd' prints out 0x64636261? 05:25 kahrl 0x61626364, &foobar: 0x7ffffe647e00 05:25 hmmmm yeah 05:26 hmmmm it's 0x61626364 everywhere 05:26 hmmmm even on big endian architectures 05:26 hmmmm just tested it 05:26 kahrl well some might do it 05:27 kahrl but it's not guaranteed by anything 05:27 hmmmm well we don't know anything until we actually consult the standard 05:27 hmmmm which i'll do right now 05:27 kahrl it says implementation defined 05:27 ShadowNinja kaeza: Do you know why they aren 05:27 ShadowNinja 't working? 05:28 kahrl C99 says anyway, will check the C++ standard 05:29 kahrl A multicharacter literal has type int and implementation-defined value. 05:30 kahrl 2.13.2 05:30 hmmmm welp 05:30 hmmmm alright, i'll change it 05:30 hmmmm and i like multicharacter constants 05:32 hmmmm this really sucks 05:32 hmmmm in my past projects i have mutlicharacter constants everywhere 05:33 kahrl it does, I sometimes wish the standard committee wouldn't be so implementation-defined trigger happy 05:33 hmmmm if they weren't like that you'd be forced to do unoptimal things on some weirder architectures because of this arbitrary restriction 05:34 hmmmm the odd floating point rounding behavior of C reminds me of this 05:34 sfan5 hello 05:35 hmmmm there's almost always an instruction for round to 0, but C's implementation specifically says round to -inf, so that was really slow for the longest time until an instruction was specifically added for this on x86 05:35 hmmmm so it could be that they're afraid of making bad standardization decisions like that 05:39 kaeza ShadowNinja, what's the error? 05:40 kaeza ah 05:40 kaeza ShadowNinja, you are assigning the fields to nil 05:40 ShadowNinja Oh, I see. 05:41 kaeza move the vector_meta to after the function defs 05:41 kaeza do just "local vector_meta" at the beginning 05:41 kaeza and vector_meta = { ... } at the end 05:42 kaeza (the second one without the 'local') 05:44 ShadowNinja There, at least eq works... 05:48 kaeza it's interesting that this way you can get the lenght of a vector with #v 05:50 ShadowNinja Yes, I thought of that but it doesn't seem to work. Also add/sub... don't work with ints. Is there any way to fix that kaeza? 05:54 kaeza seems to work here 05:54 kaeza local v = vector.new({x=2,y=2,z=2}); v = v + 1; print(("x=%d,y=%d,z=%d"):format(v.x, v.y, v.z)) --> x=3,y=3,z=3 05:56 kaeza ShadowNinja, also, small issue, but is hypot() intended to be global? 05:56 ShadowNinja kaeza: Not sure. Should it? 05:57 ShadowNinja hmmmm, kahrl: ^ 05:57 kaeza ShadowNinja, it is now, that's why I ask 05:58 kaeza I don't think it is needed as global, since you have vector.lenght() 05:58 kahrl it should be called either minetest.hypot or math.hypot if global 05:58 kahrl but yeah it probably doesn't have to be 05:58 hmmmm huh? 05:59 hmmmm i don't get why you're actually computing the hypotnuse of the angle when you could just compare it against the length squared 05:59 hmmmm ||v||^2 06:00 hmmmm call it 'norm_sq' 06:02 ShadowNinja init.lua:14: attempt to perform arithmetic on local 'v1' (a table value) 06:03 ShadowNinja Why doesn't it work for me? v1 = v1 + v2 works. 06:03 hmmmm because it doesn't have the correct metatable 06:04 kaeza you sure you're calling v1 = vector.new(pos) ? 06:04 kaeza v2 does not need metatable AFAIK 06:05 ShadowNinja Yes. 06:05 kaeza can you paste your init.lua? 06:06 kaeza (and if possible, the new version of vector.lua) 06:07 ShadowNinja http://pastebin.ubuntu.com/5848983/ and http://ix.io/6wV 06:07 ShadowNinja kaeza: ^ 06:08 kaeza ah! 06:08 kahrl small thing: use hypot in vector.length too 06:08 kaeza ShadowNinja, the table you return by add/sub/etc does not have a meta 06:09 ShadowNinja Ah, ok. 06:10 kaeza ShadowNinja, http://pastebin.com/AVHgtD2i 06:11 kahrl shouldn't vec * vec be a dot product? 06:11 ShadowNinja kaeza: Or return vector.new(...) 06:11 kaeza yup 06:12 sfan5 could someone tell me how to propely delete an std::istringstream? 06:12 kahrl sfan5: let it go out of scope 06:13 sfan5 ok 06:14 sfan5 thexyz: I made a patch 06:14 sfan5 damnit 06:14 sfan5 thexyz: I finished the patch that changes from colorkeys to RRGGBB (hex) 06:16 ShadowNinja kaeza: There, but #v always returns 0. 06:18 kaeza local v = vector.new({x=2,y=2,z=2}); print(#v) --> 3.4641016151378 06:19 kahrl ShadowNinja: is it intentional that vector.new(table) doesn't copy the table? 06:21 kaeza ShadowNinja, maybe you are using # on a regular table (possibly by not setting the metatable before returning in some metamethod) 06:21 sfan5 thexyz: https://gist.github.com/sfan5/5938884 06:21 ShadowNinja kahrl: Nope, although it probably leaves one less table for the garbage collector. 06:22 hmmmm welp 06:22 hmmmm https://github.com/kwolekr/minetest/commit/18d7bc7fa1f2621eb593969b7bcccfeb4918c05f 06:23 hmmmm there are some parts that are simply not possible to fix or would be way too slow to do correctly 06:24 ShadowNinja sfan5: Instead of ## use \#. 06:25 kahrl ShadowNinja: might be confusing that when someone changes the original table it changes the vector, and vice versa 06:25 sfan5 ShadowNinja: why? thats complicated because in lua you would need to do "\\#" 06:26 ShadowNinja Because '\' is a universal escape character. 06:26 ShadowNinja kahrl: Good point. 06:28 ShadowNinja kaeza: It definitely has the metatable set. 06:33 kaeza ShadowNinja, http://pastebin.com/qkGgRvtS 06:35 ShadowNinja http://ix.io/6wV -> 0, 0, false, ..... 06:36 ShadowNinja Oops 06:36 ShadowNinja http://ix.io/6wW 06:37 ShadowNinja kaeza: ^ 06:38 kaeza ShadowNinja, crap 06:38 kaeza __len does not work with Lua 5.1 06:38 kahrl just leave it out then 06:38 kahrl vector:length() is fine 06:39 kaeza * vector.lenght(v) 06:39 ShadowNinja Why? If/when we upgrade it will work. 06:40 kahrl kaeza: maybe __index should be set to vector? 06:40 kaeza could be 06:40 kahrl ShadowNinja: it's kind of weird (to me) to overload # 06:41 kaeza hmm... actually no 06:41 kahrl I would assume it would return 3 06:41 kaeza nope 06:41 ShadowNinja It isn't very usefull like that. 06:41 kaeza the # operator only works for arrays 06:41 kahrl I know but that's I would think when I first saw code like that 06:42 kahrl that's what* 06:42 kaeza there's a small issue here 06:42 ShadowNinja What will setting __index = vector do? 06:43 kaeza if we go with metamethods, things like v1:add(1) won't work 06:43 kaeza if we go with regular methods, things like v1 + 1 won't work 06:44 kaeza (unless you make separate metamethods instead of making them point to vector.add, ...) 06:45 kaeza that's because one would expect v1:add(x) to modify the table in-place, while the metamethod would return a copied modified table 06:46 ShadowNinja You could probably do if self then... 06:46 ShadowNinja But that would be ugly. 06:46 kaeza It's more complicated than that I guess 06:47 ShadowNinja I have a idea... 06:48 kaeza if getmetatable(self) == vector_meta then ... 06:49 kaeza I'd suggest going the easy way and supporting only one way (either metamethods or regular methods) 06:50 ShadowNinja metamethods are cooler. ;-) 06:51 kahrl perhaps rename vector.add to vector.sum and vector.subtract to vector.difference 06:51 kahrl and make vector.add and vector.subtract the in-place versions 06:51 kaeza that could work 06:52 kaeza I think you would only need 2 functions 06:53 ShadowNinja kahrl: And what about mul/div that is pretty ugly. 06:53 kaeza vector.translate() (for + and -) and vector.scale() (for * and /) 06:53 ShadowNinja +? 06:53 kaeza err... whatever 06:54 ShadowNinja Night. 06:54 kahrl that sounds good (__unm would have to be added to the metatable too) 07:01 kaeza offtopic: if this helps at all, I also have the red screen problem, but only when both enable_shaders and smooth_lighting are on 07:01 kaeza if either are off, the prob goes away 07:21 thexyz sfan5: sorry, was sleeping 07:22 thexyz sfan5: how about optional alpha-channel? 07:22 sfan5 thexyz: everyone has to sleep ;) 07:22 sfan5 thexyz: not a bad idea 07:22 thexyz sfan5: so our color is either rrggbbaa or rrggbb 07:22 sfan5 that wouldn't work 07:22 thexyz why? 07:22 sfan5 since it needs to know how long the color string is 07:23 sfan5 currently: if string.beginswith("#") and string[1] != "#" then set_color(getColor(s.substr(1, 6), foobar) end 07:24 sfan5 .. 07:25 thexyz oh, they can be prepended with color 07:25 sfan5 yep 07:25 thexyz what a great idea 07:25 thexyz whatever, then just forget about alphas 07:26 sfan5 may I push? 07:26 sfan5 (compiles and works fine, tested it) 07:26 thexyz sfan5: https://gist.github.com/sfan5/5938884#file-gistfile1-diff-L173 why don't you like loops? 07:27 * sfan5 likes loops, but a loop wouldn't make the code more compact by that much 07:27 sfan5 s/loop/loop there/ 07:28 sfan5 should I put a loop there or not? 07:29 thexyz no, I mean, even if it "wouldn't make the code more compact by that much"; what was the point of using copypasta instead of it? 07:29 sfan5 hm 07:29 sfan5 theres no point actually 07:31 thexyz right 07:31 thexyz also, shouldn't irr::video::SColor GUIFormSpecMenu::getColor(std::string color,bool& valid_color) be something like bool GUIFormSpecMenu::getColor(std::string color,irr::video::SColor &color) ? 07:31 sfan5 yeah, probably 07:32 sfan5 take a look at the same gist again and tell me if thats better 07:32 thexyz it shouldn't be called getColor then, something like parseColor 07:33 thexyz sfan5: actually, what's your experience with C/++? 07:33 sfan5 thexyz: I work more with C.. 07:34 sfan5 (haven't used c++ for some time) 07:34 thexyz well, I'd like to stress out that we ain't C89 and you can declare variables inside the loop like `for (int i = 0; i < 6; ++i)` 07:35 kahrl sfan5: about the loop, you could use string_allowed 07:36 sfan5 why didn't I think of that?... 07:38 kahrl on that note, why does string_allowed use a O(n*m) algorithm when it could be O(n+m) 07:45 sfan5 thexyz: I renamed the function to parseColor ; -> same gist 07:57 thexyz sfan5: I wasn't talking about just renaming.. 07:58 sfan5 the arguments too? 07:58 sfan5 one sec 07:59 thexyz that was just an idea 08:00 sfan5 its more clear to return bool 08:03 sfan5 thexyz: gist updated 08:23 thexyz sfan5: it's fine now, I think 08:32 sfan5 thexyz: pushing it now 09:16 sfan5 what about merging #799? 09:26 kahrl the new mapgen v7 crashes when I start a new world 09:26 kahrl I've traced it to the contents of the default biome http://paste.dy.fi/LGw 09:27 kahrl see those content_t values above 0xfff? 09:27 kahrl they cause assertion failures later (when collecting nodes for liquid transform, in my case) 10:37 Taoki sfan5: Back. Any news on that pull? 10:51 sfan5 Taoki: n 10:51 sfan5 *no 10:52 Taoki Probably gonna wait 100 more years again till someone will confirm it :< 11:28 kahrl I tried it earlier and I'm not a fan of it 11:28 kahrl it makes it seem like I'm standing in a giant cardboard box 11:32 kahrl then again I can't be bothered to care about fog and sky colors, so I'm not going to cast any sort of negative vote 11:35 Jordach overall, it's a nice feature graphics wise, but unless we can dynamically change the sky's colour to any colour and at any time of day -- it might get merged (this also includes the overall colour of the sky too) 11:36 Taoki Jordach: That is changed dynamically and the fog effect is only meant for the horrizon 11:36 Taoki I could easily modify the main sky color with a few lines of code too, but that would look bad 11:42 Taoki Well, I'm trying it now.Seems slightly modifying main sky color too looks indeed better 11:43 Jordach Taoki, you know how the player's movement is a defined variables 11:44 Jordach (well. ther keyframes..) 11:44 Jordach do that with time of day, as well as colour 11:44 Jordach sky.lua 11:45 Taoki I don't know about that nor can do any extreme changes to the code now 11:45 Jordach minetest is about modding. 11:45 Jordach not to mention custom skies will make paramat faint. 11:46 Taoki Custom skies are what I want too, but it must be done later on 11:46 Taoki And this system will be adapted then 12:36 Taoki Jordach: Force pull from my branch again. I amended the comit and made sky color also influenced a bit 12:36 Jordach <- not using *nix 12:37 Jordach you might wanna ask PilzAdam or sfan5 12:37 sfan5 Jordach: delivering 12:38 Taoki No point in asking PilzAdam, he'll find reasons to ask me to do more no matter what else I'll change 12:38 Taoki sfan5 yes, curious what you think 12:39 Taoki sfan5: Note that I only amended the squashed comit (no new one), so you might have to do a force pull to update it 12:46 ironzorg https://0fps.wordpress.com/2013/07/03/ambient-occlusion-for-minecraft-like-worlds/ 12:46 Calinou \o/ 12:46 Calinou this guy used a gif from me 12:46 * Calinou sues them for commercial use 12:47 Jordach OMFG 12:48 Taoki ironzorg: Nice! We need that as well, totally :D 12:48 Taoki But first hardware lighting 12:49 Calinou there is already ambient occlusion, but no SSAO 12:49 Calinou some guy made a SSAO shader for minecraft a while ago 12:49 Taoki oh, nice 12:49 Calinou Taoki: you've been playing without it all time :P 12:49 Taoki Calinou: In Minetest? Without which? 12:50 Calinou in most of the screenshots you do of minetest, you have smooth lighting off 12:50 Taoki And yes, looking now I can see there is ambient occlusion in Minetest too. Though apparently not as strong and correct 12:50 Calinou there is since march 2011 :P 12:50 Taoki No, smooth lighting is enabled 12:50 Calinou smooth lighting == ambient occlusion 12:51 Calinou mojang went silly and renamed it 12:59 Taoki Amended the gof color comit again to fix some junk (adding empty spaces and the like, no functionality or code change) 12:59 Taoki **fog color 13:08 ironzorg In thread 7fdc4d442700: 13:08 ironzorg /home/lenorm_f/games/minetest/src/nodedef.cpp:434: virtual const ContentFeatures& CNodeDefManager::get(content_t) const: Assertion 'c <= 0xfff' failed 13:08 ironzorg yer boi 13:08 PilzAdam ironzorg, your mods register too many items 13:08 ironzorg I don't have mods 13:12 kahrl ironzorg: using mgv7? 13:13 ironzorg yes kahrl 13:14 kahrl these assertion failures are a new feature added in the latest mgv7 commit 13:14 ironzorg «feature» 13:14 PilzAdam I got something else in latest v7: https://gist.github.com/PilzAdam/5939792 13:20 Taoki sfan5: Any news? 13:20 sfan5 Taoki: nope 13:22 Taoki I love how the new mapgen v7 is looking so far 13:29 kahrl ironzorg: this should fix it: https://gist.github.com/kahrl/5939885 13:33 ironzorg kahrl: it's not crashing anymore, and I don't have any warning, but the world is now made of stone and ores :) 13:33 Exio you need to register biomes 13:34 ironzorg err 13:54 kaeza ironzorg, http://pastebin.com/qQ2uw3Nm 13:55 kaeza thank hmmmm about those :) 13:56 ironzorg I would have expected the game to do all this by itself 13:56 ironzorg but thanks hmmm 13:56 ironzorg wherever you are 14:00 PilzAdam ironzorg, the biome definitions would be in minetest_game if v7 will be officialy "released" 14:01 ironzorg ok, makes sense 14:01 ironzorg I like v7 14:01 PilzAdam here is a WIP game from me that uses v7: https://github.com/PilzAdam/pilztest 18:56 hmmmm schematic file signature, vmanip updateliquid for non-mapgen, schematic rotation facedirs, sokomine's thing, remove mapgen selection from menu 18:56 hmmmm alright this is today's agenda 18:56 hmmmm after this i think i'm going to slow down a bit 19:08 hmmmm erm sfan5, you *removed* the three-character color specifiers? 19:08 sfan5 hmmmm: yes, thexyz confirmed it was fine 19:08 hmmmm were you supposed to do that? doesn't that break other previous formspec things that used them? 19:08 hmmmm if nothing other than the main menu used them, then it'd be fine 19:09 sfan5 [yesterday 13:16] no one uses it now, just make it #rrggbb before anyone notices 19:09 sfan5 *23 19:09 hmmmm alright, that's good 19:09 PilzAdam hmmmm, I got something for you: https://gist.github.com/PilzAdam/5939792 19:09 PilzAdam this happens after generating some terrain with v7 19:09 hmmmm sorta wish you could do #define BLK .. "#000000" .. in lua 19:10 kahrl btw there's still stuff in the main menu that uses the color constants 19:10 * sfan5 also wants #define in LUA 19:10 kahrl credits tab for example 19:10 sfan5 kahrl: must've missed that 19:10 sfan5 I grepped for "box[" 19:10 hmmmm yes, and is it intentional to have the game selection area at the bottom green? 19:11 sfan5 its grey 19:11 hmmmm seems pretty green to me 19:11 kahrl it's green for me too 19:11 thexyz depends on irrlicht version IIRC 19:11 PilzAdam its _sometimes_ green for me 19:11 hmmmm yes, this qualifies as a "bug" 19:11 thexyz it should always be green 19:11 hmmmm ....:/ 19:11 PilzAdam or rather, it was green once 19:11 thexyz right? right! 19:11 PilzAdam thexyz, we are not talking about textboxes 19:12 thexyz oh, well, okay 19:15 hmmmm oh shoot, thanks pilzadam 19:17 xrogaan heh 19:17 xrogaan don't you think that this one looks a lot like minetest? https://play.google.com/store/apps/details?id=com.br4mmie.minebuilder&feature=apps_topselling_paid_game 19:17 kahrl hmmmm: also if no biomes are defined, mgv7 sets bogus node IDs because c_water and other members of the default biome are uninitialized 19:18 kahrl this causes assertion failures later 19:18 thexyz wait, why does Minetest use left handed coordinate system? 19:18 xrogaan (i've got this window open for month, but didn't notice that i was kicked) 19:18 hmmmm oh shooot 19:18 kahrl my attempt to fix it: https://gist.github.com/kahrl/5939885 19:18 hmmmm yeah, i added all those new node parameters and i forgot to add them to the default 19:20 hmmmm but the default biome node names don't need to be set, because they're skipped over in resolveNodeNames() 19:21 hmmmm maybe i shouldn't do that and let them resolve, just setting the names and letting them fall back to air on failure 19:21 hmmmm this way i can use water for biome water content 19:21 kahrl I did it for completeness's sake, to be on the safe side 19:22 hmmmm that's what i'll do 19:23 thexyz well, whatever, I guess there's a good reason 19:24 mikolalysenko hi 19:24 mikolalysenko saw this channel and wondered if it had anything to do with minecraft/voxel games 19:25 sfan5 mikolalysenko: this channel is for Minetest(a voxel game), not Minecraft 19:25 mikolalysenko ah, cool 19:25 mikolalysenko I have a blog, and I've written a bit about voxel games in the past as kind of a hobby. 19:25 kahrl thexyz: why does *irrlicht use left handed coordinate system 19:26 sfan5 mikolalysenko: "Minetest core development and maintenance. Chit-chat goes to #minetest. " --Topic 19:26 PilzAdam mikolalysenko, this is the development channel, please go to #minetest 19:26 mikolalysenko ah, ok 19:27 PilzAdam kahrl, now the game selection background is blue for me... 19:28 sfan5 PilzAdam: it seems to be random 19:28 PilzAdam apparently 19:30 hmmmm by the way, guys, you can't use the old biomes with mapgen v7 anymore, i changed a bit 19:30 thexyz kahrl: oh, right 19:30 hmmmm right now there's a problem with grass placement in snow biomes but i need to work on something else in order to get it working the correct way 19:32 hmmmm here are the updated biome and decoration definitions: http://pastebin.com/1kjU5tN7 19:44 PilzAdam hmmmm, so dirt, dirt_with_grass and stone are default nodes now? 19:45 hmmmm yes 19:46 hmmmm to get a "normal" biome, with dirt and grass and stone, you don't need to do anything 19:46 PilzAdam and how would I set a biome with e.g. dirt2 and dirt_with_grass2 ? 19:46 PilzAdam (and stone under it) 19:46 hmmmm set those to node_top and filler_top 19:47 hmmmm seee beaches, deserts, etc. 19:47 PilzAdam so its nto possible to use stone2 under that 19:47 PilzAdam *not 19:47 hmmmm two layers of stones? 19:47 hmmmm i guess not 19:48 hmmmm i really don't think i'm going to do that either, if i start taking peoples' feature quests, where will it end? 19:48 hmmmm i did a bad enough thing by taking the suggestion to let biomes define their own water, no more requests 19:48 PilzAdam cant we just pass a table with the different layers? like {"dirt_with_grass", "dirt", "dirt", "dirt", "stone1", "stone1", "stone2"} 19:49 hmmmm no, you can't just do that 20:09 hmmmm hey guys, what do you think about that? about the pause game idea in #minetest. i think it's doable and would be benefiical; minetest is quite heavy to leave in the background. what if we had it stop rendering and wait on the server/connection threads if in singleplayer with a "halt game on pause" option? 20:10 Exio i really would like that idea 20:11 PilzAdam but dont pause it when just pressing Esc, there was a pull request that limited the FPS to 3 or so when the Esc menu was open, and it looked horrbile 20:11 hmmmm noo, what i'm talking about would completely halt the game. 20:12 hmmmm it'd be equivalent to ctrl+Z but in a more controlled manner 20:22 kahrl sounds good, just need to make sure it's still possible to leave the game running in the background if one desires to 20:32 hmmmm hey kahrl, i was talking about this a couple nights ago but i never got any sort of response: 20:32 hmmmm are we able to solve the transparency sorting problem on our own (or work around it by hiding transparent things in back of it), or would we need to modify irrlicht? 20:37 kahrl I probably didn't answer because I don't know the answer 20:37 kahrl give me a bit of time 20:46 celeron55 pausing the whole thing in the singleplayer pause menu would be fine, as long as ui responsivity is preserved in some way 20:47 celeron55 (when running client-only that doesn't make much sense because one could just shut down the client) 20:52 kahrl hmmmm: these are the options that I think would be possible to implement with irrlicht: http://paste.dy.fi/irl 20:53 Jordach IRL, is that a IRL paste 20:54 kahrl IRL I have to go to bed :P 20:54 Jordach kahrl, when the PC tells you to sleep; do it 20:59 PilzAdam http://mg.viewskew.com/mgoblin_media/media_entries/64/MapgenV7map1.png 20:59 PilzAdam the mountains are cut-off I think, so they are grey 23:43 hmmmm i still need to find better tree perlin noise params 23:44 PilzAdam I just use an extra forest biome and fill it with trees with fill_ratio = 0.01 23:46 hmmmm i should make the cave-popping-out-of-ground amount configurable 23:46 hmmmm right now it's a 1/10 chance 23:46 hmmmm (well, probably not 1/10, the LCG used doesn't seem very random)