Time Nick Message 06:16 RealBadAngel ping pong 06:16 RealBadAngel anybody alive? 06:19 VanessaE hi 06:26 RealBadAngel hi VanessaE, whats up? 06:26 VanessaE not much 06:30 RealBadAngel https://github.com/minetest/minetest/issues/2748 06:30 RealBadAngel what do you think about that? 06:32 VanessaE no opinion really 06:32 RealBadAngel its damn easy to code 06:34 hmmmmmm paramat: i don't really have a problem with it 06:34 hmmmmmm RBA: what did you need before? 06:34 RealBadAngel hmmmmmm, got ya! 06:34 hmmmmmm you highlighted my name 06:34 RealBadAngel several times since 2 days :) 06:34 hmmmmmm yeah what's up 06:34 RealBadAngel see my PR, ive made it 06:35 hmmmmmm by the way how's the minimap been going 06:35 RealBadAngel waiting for this one to be merged 06:35 RealBadAngel for the minimap i need relief mapping 06:36 RealBadAngel to make terrain actually look 3d 06:36 hmmmmmm GLSL has an unsigned int type 06:36 hmmmmmm for counts of things or number of iterations, why not use 'uint'? 06:37 RealBadAngel if you count down, you can hit -1 06:37 hmmmmmm that would be iterating over a coordinate range though, would it not? 06:38 hmmmmmm typically speaking 06:38 hmmmmmm you normally wouldn't have a situation where you'd count down 06:38 RealBadAngel thats rather a manner 06:38 hmmmmmm it's not typical i mean to say 06:38 RealBadAngel i just used to rely on signed in such cases 06:38 hmmmmmm great job on the cleanup 06:39 RealBadAngel i did more than you wanted me to 06:39 hmmmmmm + if (fogDistance != 0.0) { 06:39 hmmmmmm is it safe to compare to a literal float value here? 06:39 hmmmmmm I don't know the answer, I'm just asking you to make sure that it's indeed correct here 06:39 RealBadAngel its so old code that for sure its safe ;) 06:40 hmmmmmm not sure if the rules for floats are different from C 06:41 hmmmmmm excellent 06:41 RealBadAngel very different 06:41 hmmmmmm I love how you were able to cut down on that if/else if chain for normal ranges 06:41 RealBadAngel you need to cast the value directly 06:41 hmmmmmm repetitive code is usually bad code 06:41 RealBadAngel so 1 is not the same as 1.0 06:41 RealBadAngel 1 will be in and 1.0 will be float 06:42 RealBadAngel but not for all compilers 06:43 RealBadAngel glsl is pain in the ass to debug and develop 06:43 hmmmmmm now wait 06:43 hmmmmmm explain to me why tangents are needed for shaders 06:43 RealBadAngel what works for me doesnt work elsewhere ;) 06:43 RealBadAngel sure 06:43 RealBadAngel tangent space is texture space 06:44 RealBadAngel every light, displacement, reflection etc calculations are done in that space 06:44 RealBadAngel not in the worlds one 06:45 RealBadAngel so each vertex has to be given all three vectors, normal, tangent and bitangent (binormal) 06:45 RealBadAngel those can be calculated out of 4 vertices 06:46 RealBadAngel shader knows at one time only 3 of them 06:46 hmmmmmm I mean how did it work before without explicit tangent values 06:46 RealBadAngel and the normal to the surface 06:47 RealBadAngel i simulated it 06:47 RealBadAngel it was absurdly costly and very faulty in certain cases 06:47 hmmmmmm oh i seee that 06:47 RealBadAngel thats why i wasnt able to make parallax effect more noticeable 06:48 hmmmmmm glMultiTexCoord1 is the tangent space 06:48 RealBadAngel 2 too 06:48 hmmmmmm well that's the binormal 06:48 hmmmmmm binormal is the thing orthogonal to normal and tangent, is it not? 06:48 RealBadAngel tangent space is 3 vectors 06:49 hmmmmmm right 06:49 hmmmmmm tangent space includes tangent as one of the vectors 06:49 RealBadAngel yes 06:49 hmmmmmm alright this all makes a lot of sense 06:49 RealBadAngel if you have 2 of them, you can get 3rd 06:49 RealBadAngel before we had just one 06:50 RealBadAngel now we got all 3 06:50 hmmmmmm does explicitly computing tangents outside of the shaders make a difference in the way the game looks normally, without parallax occlusion enabled or any of the heavy graphics things? 06:50 RealBadAngel no 06:51 RealBadAngel those are calculated in thread and only used when shaders enabled 06:51 RealBadAngel without shaders it doesnt change anything 06:51 RealBadAngel but 06:52 RealBadAngel if we have it calculated properly, we can start to use it 06:52 RealBadAngel we can implement proper lighting for all the objects 06:52 RealBadAngel even round 06:52 hmmmmmm right 06:53 hmmmmmm alrighty maybe i'm overestimating the memory usage impact for having those two extra floats 06:53 RealBadAngel im pretty sure they were used anyway 06:53 hmmmmmm see I originally was hesitant on bloating the size of a vector because there are like... millions of vectors on runtime 06:53 RealBadAngel but just not initialized 06:53 hmmmmmm S3DVertex doesn't have those fields 06:54 hmmmmmm by the way this is what I was saying about leveraging covariant types 06:54 RealBadAngel s3dvertex maybe not 06:54 RealBadAngel but gpu for sure uses those vectors 06:54 hmmmmmm S3DVertexTangents is a superclass of S3DVertex 06:55 hmmmmmm so you can do 06:55 hmmmmmm S3DVertex *verts = (S3DVertex *)(vertex_tangents); 06:55 hmmmmmm verts[i].Pos += whatever; regardless of actual vertex format 06:56 RealBadAngel with this version of irrlicht maybe 06:56 hmmmmmm so it seems what you did is change PreMeshBuffer to have two vectors 06:56 hmmmmmm this doesn't change. 06:56 RealBadAngel but if something changes we are still secure 06:56 hmmmmmm trust me. if this changes within the next 10 versions of irrlicht I'll eat my hat 06:56 RealBadAngel hehe 06:56 hmmmmmm it's part of the interface 06:56 hmmmmmm they can't break something like this 06:56 hmmmmmm so 06:57 hmmmmmm one thing you COULD do is make PreMeshBuffer a templated class 06:57 RealBadAngel ah 06:57 RealBadAngel i tried 06:57 hmmmmmm what happened 06:57 RealBadAngel our code 06:57 RealBadAngel we directly write that data runtime 06:58 RealBadAngel animations, lighting 06:58 hmmmmmm ?? 06:58 hmmmmmm rmmahh 06:58 RealBadAngel its not just grabbed, its modified on the fly 06:59 hmmmmmm i see a potential problem 06:59 hmmmmmm i have this problem too 06:59 hmmmmmm every time I say to myself, "Templates would work excellent here!" 06:59 RealBadAngel so premesh buffer is in fact the resulting mesh 06:59 hmmmmmm I try it and then I change my mind 06:59 hmmmmmm I can forsee a problem 07:00 hmmmmmm maybe you're right, just change all of them to S3DVertexTangents 07:00 RealBadAngel i spent 3 days coding the templates, then got pissed off and wrote what you can see now in 2 hours 07:00 hmmmmmm right 07:00 hmmmmmm sorry about that 07:00 hmmmmmm i shouldn'tve told you to do this 07:00 hmmmmmm templates never work well 07:01 RealBadAngel i did that to premesh buffers, ok 07:01 RealBadAngel then meshcollector wanted to be templated to 07:01 RealBadAngel then rest of the code happily announced that it wants direct access to it, not the templates 07:02 RealBadAngel only solution that could work could be to use original irrlicht meshbuffers 07:02 hmmmmmm I encountered the reason why we use PreMeshBuffer 07:03 hmmmmmm when I was writing my mini-minetest 07:03 RealBadAngel lighting and texture animations 07:03 RealBadAngel also append 07:03 hmmmmmm what I ended up doing was SMeshBuffer->Vertices.reallocate(4096); and then SMeshBuffer->Verticies.push_back(); manually instead of .append() 07:03 RealBadAngel original append does something with memory reallociation 07:03 hmmmmmm that avoids the copy altogether 07:04 hmmmmmm well 07:04 hmmmmmm that's something that would be cool to have in the future 07:04 hmmmmmm (removing PreMeshBuffer completely) 07:05 hmmmmmm alrighty then 07:05 RealBadAngel this time havent go wasted, ive learned a lot bout meshbuffers thx to that 07:05 hmmmmmm sorry about the whole template mess 07:05 RealBadAngel np 07:05 hmmmmmm change everything back to S3DVertexTangents :/ 07:05 hmmmmmm looks great otherwise 07:05 RealBadAngel actually im not sure now 07:05 hmmmmmm why 07:05 hmmmmmm this is so much messier and everybody uses shaders 07:06 hmmmmmm so it's not like it's making a huge difference 07:06 hmmmmmm if no shaders were the common case, then I could see 07:06 RealBadAngel i added some code to it to better handle premeshbuffers 07:06 hmmmmmm well in that case 07:06 hmmmmmm about the PreMeshBuffer modifications... 07:07 RealBadAngel only what worries me is that if (m_enable_shaders) 07:07 hmmmmmm why do you return u8 * for vertices? 07:07 RealBadAngel it doesnt really matter 07:07 RealBadAngel it could be void* 07:07 hmmmmmm void * is needed for a generic pointer 07:07 RealBadAngel its always cast to anything else 07:08 hmmmmmm there's no guarantee that pointer types are all the same size 07:08 hmmmmmm what if a u8 * is smaller than a S3DVertex * on a certain machine? 07:08 RealBadAngel irrlicht uses void* for meshbuffers 07:08 hmmmmmm void * is guaranteed to fit any pointer type. that's why we use it 07:08 RealBadAngel in fact that was kahrl suggestion 07:08 hmmmmmm then change it :) 07:09 RealBadAngel when we changed functions in mesh.cpp 07:09 RealBadAngel those switch things 07:09 hmmmmmm also you can make the code more expressive/compact by using ternary operators here 07:09 hmmmmmm and then the initializer list is supposed to be indented another level 07:09 hmmmmmm there's supposed to be a space between the constructor prototype and the : 07:10 hmmmmmm and there's not supposed to be a space between the function name and the argument list 07:11 RealBadAngel there can be some issues still 07:11 RealBadAngel im still working on this pr 07:12 RealBadAngel atm i wanted to show you what i did with premeshbuffer 07:12 RealBadAngel meanwhile ive added range limitation to parallax/relief effect 07:13 RealBadAngel it doesnt have to be applied to every rendered node - you simply could not see that out of 15-20 nodes range 07:14 RealBadAngel also i need to add in menu selection which displacement code to use, parallax occlusion or relief mapping 07:15 RealBadAngel first works ok with HD textures and is faster (but have some limitations, HD textures can cover them) 07:16 RealBadAngel relief is more accurate, can work with any data but is damn slower 07:16 hmmmmmm okay 07:17 kilbith relief and parallax mutually exclusive right ? 07:17 kilbith +are 07:17 RealBadAngel theyre basing on same math, almost 07:17 RealBadAngel but relief is more accurate 07:17 RealBadAngel hold on a sec 07:18 RealBadAngel http://www.cescg.org/CESCG-2006/papers/TUBudapest-Premecz-Matyas.pdf 07:18 RealBadAngel here you have all the methods explained 07:19 kilbith you may need to disable each other on checkboxes if one's selected 07:19 RealBadAngel i see it like off/parallax/relief 07:19 kilbith yes, dropdown better 07:19 RealBadAngel so no checkbox 07:20 RealBadAngel and definitely one day i will have to make my own tab for settings 07:20 kilbith btw this menu needs to be reworked, it's too confined in space 07:21 RealBadAngel number of shader options will grow enormously soon 07:22 RealBadAngel thx to tangent space finally in i can continue with all the projects i paused before 07:22 RealBadAngel lava/water shaders, reflections, post processing 07:22 RealBadAngel yikes! :) 07:23 hmmmmmm are you sure it's not a better idea to just make all of the s3dvertexes into s3dvertextangents? 07:24 OldCoder Attention. Server List stopped entirely for all of my worlds on Debian 8. Building a newer curl by hand seems to have corrected the problem. 07:24 OldCoder sfan5, ^ 07:26 RealBadAngel hmmmmmm, it could make code cleaner 07:26 hmmmmmm exactly 07:26 hmmmmmm it's up to you. 07:27 RealBadAngel lets get kahrls or c55 voice on that 07:28 hmmmmmm erm, well okay 07:28 hmmmmmm i highly doubt they're going to care :) 07:29 RealBadAngel est31 was complainin bout memory usage too 07:30 RealBadAngel its true but to some degree 07:31 RealBadAngel VanessaE did stress tests for it, with dreambuilder and hdx textures speed was almost the same (1 fps difference, can be just noise) 07:31 RealBadAngel and memory usage on the same level 07:31 RealBadAngel but with shaders, 30% fps gain 07:34 RealBadAngel hmmmmmm, btw, have you actually try the displacement effect in action? 07:34 hmmmmmm no 07:34 RealBadAngel i highly recommend that 07:34 RealBadAngel especially walk around cacti or diamond ore node :) 07:35 hmmmmmm i'll check it out when it's merged 07:35 RealBadAngel no screenshots can show that 07:36 RealBadAngel ok, so i will add the config options now and will wait for others votes if i should go back to tangent vertices 07:37 RealBadAngel i can think of just one reason why we would need them both 07:38 RealBadAngel android devices 07:39 RealBadAngel theyre not powerfull enough to use such advanced shaders and each byte here counts 07:44 Calinou shouldn't we disable shaders entirely on Android? they're likely to give a poor experience due to low FPS 07:47 RealBadAngel Calinou, we rather should prepare simplyfied version of them just for android devices 07:47 RealBadAngel only to handle finalColorBlend 07:47 RealBadAngel ie day/night transitions and nothing more 07:48 RealBadAngel and disable for android regular code 07:49 RealBadAngel those devices always support ES 07:49 RealBadAngel i bet ES shader will be faster than regular code 07:57 RealBadAngel brb 08:10 Calinou https://github.com/minetest/minetest/pull/2768 08:59 RealBadAngel Calinou, thx, i will have to manually rebase my PR then :) 09:00 RealBadAngel cant we have just one license file for christ sake? 09:00 RealBadAngel it starts to feel like pain in the ass to maintain 09:10 RealBadAngel just put into each file license header and update list of contributors in a file next to readme.txt 09:10 RealBadAngel without details, its a noise and most of us doesnt care of licenses even 09:12 RealBadAngel such shit just slows down development, you could use that time for something actually usefull 09:13 technomancy really? you could use the thirty seconds it takes to click "merge pull request" for something useful? 09:13 technomancy I agree having a list of names is a waste of time, but I hope "most of us don't care about licenses" isn't true =\ 09:14 technomancy I for one am glad that I have the freedom to legally copy and distribute minetest 09:15 jin_xi pls no. this is not the point. its more that this kind of pullrequest is not accepted ever, just like any other edit across the whole code base 09:17 technomancy fair enough 09:20 RealBadAngel if we have one central license file, the problem will be gone 09:20 RealBadAngel also updates to it will never break any pr 09:23 technomancy I agree; my understanding is it's only the Apache license that requires the header to be repeated in each file. no idea why you would do that in this codebase. 09:33 RealBadAngel technomancy, its not about the license at all. its about 300 copies of it 09:33 Calinou +RealBadAngel> cant we have just one license file for christ sake? 09:33 Calinou license headers are important in case someone pastes the code on eg. a pastebin 09:33 Calinou they are not entirely redundant 09:33 Calinou this is why the FSF recommends using them 09:34 Calinou (and in fact, you are supposed to have headers if you're using a GNU license) 09:34 RealBadAngel i hate lawyers. and the way they make our lives a hell :P 09:35 RealBadAngel lawyers should be shot down in a cradle. on a very first sign of a lawyer face ;) 09:36 Calinou this is not lawyering 09:37 Calinou this is common sense 09:37 Calinou if there is no header, no one has an idea of what the file does, who owns it, and what you can do with it, once the file is put on eg. a pastebin. 09:37 Calinou or distributed individuallyu 09:37 Calinou -u 09:42 RealBadAngel #license 09:42 RealBadAngel #put there general license 09:42 RealBadAngel #this file is part of mt project 09:43 RealBadAngel end of license 09:43 RealBadAngel erm #see license.txt in main folder for details 09:43 Calinou "see license.txt" doesn't work on pastebins. 09:43 Calinou nor does it work when the file is distributed individually 09:43 RealBadAngel put there link 09:43 RealBadAngel to that file 09:44 Calinou https://gnu.org/licenses/gpl-faq.html#NoticeInSourceFile 09:44 RealBadAngel pile of lawyers shit 09:44 RealBadAngel long live WTFPL! 09:45 RealBadAngel ;) 09:45 RealBadAngel anyway i do not care, if somebody HAVE to take care of it, im glad its you not me 09:45 Calinou even WTFPL is better with an header 09:46 Calinou my point is, you should always have an header, even a 1-line one :P 09:47 RealBadAngel and my point is that theres always a need to hire a lawyer 09:47 Calinou you don't have to hire a lawyer... 09:47 RealBadAngel or waste our time 09:47 Calinou except for enforcing the GPL :( 09:47 Calinou (unless you perform some sort of community enforcement; but you don't need copyleft for this) 09:47 RealBadAngel and our time is even more precious 09:48 RealBadAngel because we do things real 09:49 RealBadAngel and that rights or lefts doesnt add anything to the project 09:50 technomancy IMO it's the responsibility of the person putting it in the pastebin to make the license clear 09:50 RealBadAngel piece of code pasted elsewhere is worthless 09:51 RealBadAngel it doesnt mean anything without it 09:51 Calinou technomancy, but they can be lazy, or they can not know what licenses mean... 09:51 technomancy Calinou: so that's their problem 09:51 Calinou few pastebins have a "Description" field 09:51 RealBadAngel and pastebins are not for distributin the code 09:51 Calinou and even less a "License" field 09:51 Calinou RealBadAngel, you'd be surprised how often it happens in practice :P 09:52 RealBadAngel even if, so what? 09:52 RealBadAngel do we care bout it? 09:52 technomancy if you use pastebins for distributing code you likely have a lot bigger problems than determining its license 09:53 RealBadAngel we are doing it for free, not expecting anything 09:54 RealBadAngel if somebody knock my doors regarding license bout my code will earn punch in the face 09:54 RealBadAngel nothin less nothin more 09:55 RealBadAngel its free (tm) ;) 10:09 kilbith lol 14:46 est31 pushing in 5 minutes https://github.com/est31/minetest/commit/05f4e9ee08f93b17c3dbaf2b6a1022cc28a226ce 15:28 est31 pushed 16:50 Etzos What is the oldest/most oudated system minetest is intended to be compiled on? 16:53 est31 Ubuntu 12.04 16:53 est31 perhaps older ones too 16:53 est31 I think the very oldest is windows XP 16:59 Etzos So theoretically increasing the minimum cmake version to 2.8 should be okay then? 17:00 est31 what features of 2.8 would you want to use? 17:00 Etzos *theoretically, increasing 17:01 Etzos file(COPY ...) 17:02 est31 replacing what? 17:02 est31 and in which cmake file? 17:03 Etzos Not replacing, adding. I've been trying to get out-of-source builds functioning properly. I would like to have file(COPY ...) so that I can copy script files into the build directory for quick debugging behind a cmake option, naturally. 17:04 Calinou yes, we don't support older than Ubuntu 12.04 17:04 Calinou we will stop supporting it in April 2017 17:05 est31 perhaps 17:05 est31 perhaps earlier 17:08 Etzos I don't need cmake 2.8, but it certainly makes it easier. And if I can safely assume that it's available I don't see a reason to staying with 2.6. 17:09 est31 which script files? 17:09 est31 the builtin lua? 17:09 Etzos Yes. 17:10 Etzos Non-compiled sources. 17:10 Calinou <+est31> perhaps earlier 17:10 Calinou there will still be Ubuntu 12.04 (especially server) and Debian 7 (especially server) users in 2017 :( 17:10 Calinou this is the sad story of long term support 17:10 Calinou I wish Canonical didn't extend its support to 5 years, actually 17:10 Calinou it's of little purpose 17:11 est31 it is purposeful 17:11 est31 e.g. when you have a company and dont want to update every single year 17:11 Calinou that's an outdated culture IMO 17:12 est31 see how painful it was to leave windows XP 17:12 Calinou if you can't afford to upgrade every 2 years... 17:12 est31 yes perhaps 17:12 Calinou it's exactly how we land into Windows XP land 17:12 est31 but question is whether we should support an OS thats mostly interesting for the corporate world 17:12 est31 or big institutions 17:12 Etzos How many people are actually running minetest on servers that old? 17:13 est31 *shrug* 17:13 est31 or whether we should support users who are too afraid to click the "do-release-upgrade" button 17:13 Calinou we don't need to support XP anymore, but we still do for instance :P 17:13 Etzos Or compiling even. It could be compiled on a newer system and then moved to the old server if need be. 17:13 est31 like I think with ubuntu 12.04 support 17:13 Calinou many applications, especially modern games, stopped shipping 32-bit support 17:13 est31 Etzos, that wont work 17:13 Calinou we can't do that yet but maybe in 5 years 17:14 est31 thats only important when you have solely binary releases 17:14 est31 so only a matter for closed source software :) 17:15 Calinou Fedora plans to drop 32-bit soon, so will Ubuntu 17:16 est31 *shrug* 17:17 Calinou Debian will drop it in 2100 of course :p 17:17 Krock C:\Programme\cmake\bin>cmake --version 17:17 Krock cmake version 3.0.0 17:18 Krock est31 ^ 17:18 est31 Etzos, thats our XP user :) 17:18 Calinou 3.2.3 is easy to get on Windows 17:18 Krock Don't abuse my nick for linking it to Win 5.1 17:18 Calinou https://packages.debian.org/search?keywords=cmake 17:18 Calinou http://packages.ubuntu.com/search?keywords=cmake 17:19 Calinou you can see the versions shipped in both Debian and Ubuntu there ^ 17:19 Calinou we don't need to support CMake < 2.8.7 17:19 Calinou wheezy ships 2.8.9, precise ships 2.8.7 17:19 est31 good then 17:20 * est31 wonders whether cmake on precise supports c++11... 17:20 est31 err llvm* 17:20 Etzos est31: Why wouldn't it work? 17:21 est31 nice nice it supports c++11 17:21 est31 clang 3.0 has good c++11 adoption 17:21 est31 so people with older ubuntu, can use clang now to compile 17:21 Calinou we could default to Clang 17:21 Calinou AssaultCube does that already 17:22 est31 **we can upgrade to c++11 without losing users** 17:22 Krock *cough* 17:22 est31 precise users* 17:22 est31 Krock, you use msvc right? 17:22 Krock yes 17:24 Krock And if you're gonna use C++11, please put those things into #ifdefs, I'd like to be able to compile Minetest for the other win xp users 17:24 Krock if there are any 17:24 Calinou just use mingw :P 17:25 Krock Calinou, M$ windows works best with M$ compiled applications 17:25 Calinou it still works really well with mingw 17:25 Calinou and doesn't bother you to register 17:26 Krock blame profit 17:26 Krock register...? 17:26 est31 so that you get MSVC for free 17:26 Krock AFAIK, I didn't register and just used a key from somewhere 17:27 Krock ssht 17:28 Calinou telnett 17:28 est31 netcatt 17:28 Etzos Speaking of Windows, my cmake changes should also have the added benefit of showing the non-compiled source files in VisualStudio and QtCreator. 17:29 Krock What about outdated, compiled files? 17:29 est31 go on make a PR then. 17:29 Krock ^ great idea, realize your thoughts 17:30 Etzos est31: It's not done yet. I just wanted to see if I could safely increase the cmake version. 17:33 est31 k 18:56 Megaf folks, luacontrollers still active even if nobody is online 18:56 Megaf and they are keeping the region active 18:57 Megaf est31: 2015-06-09 14:43:10: ACTION[ServerThread]: Megaf leaves game. List of players: 18:57 Megaf 2015-06-09 14:43:16: ACTION[ServerThread]: sin takes stuff from chest at (45,12,-523) 18:57 Megaf est31: I did not highlight you, sorry, it was my client Oo 18:57 Megaf anyway, connected to that chest is some autocrafters and filters 18:58 est31 do you have technic installed on the server? 19:15 Megaf nope 19:15 Megaf mesecons and pipeworks 20:02 est31 pushing in 5 minutes https://github.com/est31/minetest_game/commit/f09ae11e8959e2d751bef3ee930bf678f7164e1d 20:31 est31 pushed 22:28 est31 why has intlGUIEditBox been removed? 22:49 kahrl est31: I guess #1166 22:49 ShadowBot https://github.com/minetest/minetest/issues/1166 -- Does not build with Irrlicht 1.9 (svn) and Freetype enabled 22:58 est31 kahrl, I guess we have to re-add it if we want to have proper entry of special characters in formspec? 22:59 est31 fixing the bug ofc 22:59 kahrl might be the only option, since there's no good way to subclass an irrlicht gui element 23:02 kahrl another way could be something with a global event handler, but that's probably incredibly hacky 23:02 est31 but 1.9 still isnt released 23:03 est31 they develop even slower than us 23:12 kahrl well, I think we should strive to be 1.9 compatible if at all possible 23:13 kahrl I had an annoying problem recently where I wanted to compile minetest 0.3.2 (for some map compatibility testing) 23:13 kahrl but I couldn't because it doesn't compile with irrlicht 1.8.1 23:14 kahrl (and my distro doesn't provide packages for irrlicht 1.7.* anymore) 23:14 kahrl I guess I could've compiled 1.7 myself but I didn't care enough 23:16 est31 perhaps somebody who has compiled 1.9 themselves can tell 23:16 est31 and give a fix 23:21 kahrl does intlGUIEditBox need to be a subclass of IGUIEditBox? could it be simply a subclass of IGUIElement? 23:47 paramat hi hmmmmm please could you review the implementation of noise defined biome blend if you have time later? https://github.com/minetest/minetest/pull/2764