Time Nick Message 01:09 RealBadAngel kahrl, hmmmm (and others ofc) https://www.youtube.com/watch?v=CM2b_vmM5rg 06:14 hmmmm grr. 06:15 hmmmm I wonder if it'd be worth it to start adding real unit tests 06:16 hmmmm like with every submission of a new feature or whatever, it needs to have unit tests 06:38 Zeno` hmmm? 06:40 Zeno` hmmmm, https://github.com/Zeno-/minetest/commit/4bc3a4d2392cf11ff4ea0d024dcdae79c5aec324 06:41 Zeno` lines 192-194, do they look right to you? 06:41 Zeno` or possibly a mistake? 06:41 Zeno` oh dear 06:41 Zeno` I'd better remove changes to test.cpp heh 06:43 Zeno` https://github.com/Zeno-/minetest/commit/e7e920d85f8156170536082389f7ff96a8bb0600 06:43 Zeno` anyway those bitwise ands look wrong to me, but I'll leave them for now 06:44 hmmmm yeah they should be logical ands 06:44 hmmmm the way it is right now doesn't hurt anything, but it prevents the compiler from short-circuiting the expression 06:47 Zeno` ok 06:48 Zeno` Changed them to logical 06:52 Zeno` valgrind now clean 06:52 Zeno` well apart from those 2 unfixable irrlicht issues 07:00 hmmmm sometimes I wonder if the changes I made actually broke things 07:01 hmmmm it's usually hard to tell if it's mapgen-related :< 07:34 Zeno` unit test: http://codepad.org/gtFjMRF2 07:34 Zeno` not doing that again 11:05 kilbith sfan5: I don't understand at all; when I cmake with -DENABLE_OGLES=1, it fetch the wrong lib : 11:05 kilbith -- Found OpenGL: /usr/lib/arm-linux-gnueabihf/libGL.so 11:05 kilbith yet I have have the good one : libglib-2.0.so 11:06 kilbith so why it don't fetch this one ? 11:06 kilbith sorry, this one rather : libGLESv2.so.2 12:11 Krock Why does a player on multiplayer (or myself) decrease the FPS more than "expanding"/flowing water? 13:48 Krock !doge 16:51 gravgun Hi everybody! sfan5, anything new about the custom model VM thingy? 16:52 sfan5 anything new 16:52 sfan5 no 16:52 sfan5 I thought about it 16:52 sfan5 but nothing more 16:52 gravgun m'kay :) 17:00 gravgun sfan5: Thought about something though: if we use a register & address-based memory model, we must implement some kind of RAM pages for the VM program... 17:01 sfan5 yes 17:01 gravgun Which I have no idea how much we should be able to give 17:03 gravgun I started writing the Triangulator VM spec a bit blindly, I even barely know assembly in general 17:04 gravgun Me be back when done reading http://en.wikibooks.org/wiki/Creating_a_Virtual_Machine/Register_VM_in_C 17:10 kilbith sfan5: http://irc.minetest.ru/minetest-dev/2014-10-07#i_3968961 17:10 sfan5 kilbith: Found OpenGL is for opengl not opengl es 17:11 kilbith sfan5: there is a "FindOpenGLES2.cmake" in the modules 17:11 sfan5 -- Found system opengles2 library /usr/lib64/libGLESv2.so;/usr/lib64/libGLESv2.so;/usr/lib64/libSM.so;/usr/lib64/libICE.so;/usr/lib64/libX11.so;/usr/lib64/libXext.so 17:11 sfan5 thats how it looks for me 17:12 kilbith so why it fetch the wrong lib so ? 17:12 sfan5 no no 17:12 sfan5 -- Found OpenGL: 17:12 sfan5 -- Found system opengles2 library 17:12 sfan5 it's different 17:12 sfan5 you are looking at the wrong line 17:13 kilbith hmm... 17:14 kilbith ok I checked again, it don't show at all : "-- Found system opengles2 library" 17:14 kilbith but only : -- Found OpenGL: /usr/lib/arm-linux-gnueabihf/libGL.so 17:15 kilbith and I have the GLES libs in /usr/libs 17:15 sfan5 it should show -- Found system opengles2 library somewhere below the opengl one 17:23 kilbith_ sfan5: http://paste.debian.net/125024/ 17:24 kilbith_ no trace of the GLES libs... 17:24 gravgun sfan5: any specific coding style convention in MT's source? (Highly ressembles my style, but question specifically applies to types: can I use uint8_t? ) 17:29 sfan5 gravgun: depending on what you want to do just use int http://dev.minetest.net/Code_style_guidelines 17:30 sfan5 kilbith_: do "rm -Rf CMakeFiles CMakeCache.txt" and then run cmake again 17:31 gravgun "rm -Rf ..." ... "-Rf"... "f" nawpe. 17:32 kilbith_ sfan5: exactly the same result :( 17:33 sfan5 kilbith_: then something else is wrong 17:33 kilbith_ I'm wondering if the opengl libs are picked in priority, instead of the gles ones 17:34 kilbith_ my raspbian is newly installed yet... 17:34 kilbith_ I don't understand at all 17:35 kilbith_ Megaf got the same issue iirc 17:37 sfan5 kilbith_: is libGLESv2.so installed? 17:38 kilbith_ /usr/lib/arm-linux-gnueabihf/libGLESv2.so.2 17:38 kilbith_ /usr/lib/arm-linux-gnueabihf/libGLESv2.so.2.0.0 17:38 kilbith_ /usr/lib/arm-linux-gnueabihf/libglib-2.0.soo 17:38 Krock soo? 17:38 kilbith_ here my files for the GLES one ^ 17:38 sfan5 libglib-2.0.so is some GTK thing 17:38 sfan5 has nothing to do with gles 17:39 kilbith_ but the GLES libs are pre-installed in Raspbia 17:39 kilbith_ n 17:40 kilbith_ sorry for libglib-2.0, wrong pasting 17:41 sfan5 kilbith_: set OPENGLES2_INCLUDE_DIR to /usr/include, OPENGLES2_gl_LIBRARY to /usr/lib/arm-linux-gnueabihf/libGLESv2.so.2, EGL_INCLUDE_DIR to /usr/include and EGL_egl_LIBRARY to and run cmake again 17:41 sfan5 you can set them like this 17:41 sfan5 -DOPENGLES2_INCLUDE_DIR=/usr/include/ 17:42 kilbith_ after cmake ? 17:43 kilbith_ eh, sorry, misread 17:51 kilbith_ well, same result 17:52 kilbith_ it shall have a conflict with OpenGL or I don't understood the set INCLUDE_DIR things 17:53 kilbith_ I'm being mad... 17:58 sfan5 the FindGLESv2.cmake module could use a rewrite 18:00 kilbith_ i'll try to replace directly the settings you gave me in the this file 18:01 kilbith_ after 'IF (NOT BUILD_ANDROID)' I guess... 18:10 kilbith_ damn, I give up until a ready-to-play MT raspi edition 18:18 Krock kilbith_ += " releases" 18:52 kilbith_ Krock: there is no armhf/opengles release 18:52 Krock damn, I give up until a ready-to-play MT raspi edition + " releases" 18:54 kilbith_ sfan5: if you already have a compiled MT for raspi, could you upload it please ? 18:57 Krock kilbith_, Megaf_ did it too IIRC 18:57 kilbith nope, same issue that me 18:59 Megaf_ Krock: I didnt manage to compile the minetest client, only the server 18:59 Megaf_ I gave up already 19:00 Krock ah 19:00 Krock but the RPi doesn't have a graphics card 19:00 Megaf_ Actually I managed to compile it using OpenGL, that would be done via software, now, you can imagine how slow it is... 19:00 Megaf_ Krock: You are totally wrong, it has and quite a good one 19:01 Krock ah 19:01 kilbith but you said I got not success to compile it with opengles 19:01 kilbith you* 19:01 Megaf_ Krock: as I said a couple of days ago, it is able to run Quake III at 1920x1080 will all setting maxed out 19:02 Megaf_ including anti aliasing 19:02 Krock Megaf_, so that tiny thing is better than my PC 19:02 Megaf_ I think that is by far more taxing on GPU than minetest is 19:02 Krock \o/ BlockMen 19:02 Megaf_ Krock: the Raspberry Pi COU/GPU is actually intended for media intensive tasks 19:03 BlockMen o/ Krock 19:03 kilbith but does MT is better optimized than Quake III ? ;) 19:03 Megaf_ if you use a raspberry pi for a headless server or desktop you are underusing it 19:04 Krock And I really would like to know why flowing water gives a lower fps decrease than digging a node 19:04 BlockMen oh, seems someone has its period again: http://irc.minetest.ru/minetest/2014-10-07#i_3969478 19:04 Megaf_ Krock: 24 GFLOPS, that's the power of the Raspberry Pi GPU 19:04 Krock such a flop :3 19:04 Megaf_ when running OpenGL ES 2.0 19:06 Megaf_ Krock: thats more powerful than 2 x Xeon 5310 19:07 * Krock searchs a FLOPS benchmark program 19:07 Megaf_ IF and WHEN using optimized software that can use that power 19:07 kilbith plus his arm support very well the overclocking 19:07 kilbith reached 1 Ghz lately 19:07 Megaf_ its ARM* 19:07 kilbith sorry 19:07 Megaf_ RPis ARM* 19:08 Megaf_ not my arm :P 19:08 Megaf_ my arm can do like 2 operations per second 19:08 Megaf_ lol 19:08 kilbith it depends of the operation...... 19:09 Krock Doctor: my arm can do one operation per hour 19:09 Calinou but does MT is better optimized than Quake III ? ;) 19:09 Calinou yes, definitely 19:09 Calinou but more demanding 19:10 Megaf_ Calinou: do you really think that MT is more demanding? (GPUwise) 19:10 Megaf_ Krock: http://www.techpowerup.com/forums/threads/processor-gflops-compilation.94721/ 19:12 Megaf_ i7 975 D0 @ 4.3ghz = 13.96 GFLOPS 19:13 Megaf_ why arent we running software on GPUs? 19:14 Megaf_ my Phenom II x4 955 BE @ 3,2 GHz, 12 GFlops 19:15 Calinou yes, even without shaders 19:15 Calinou both games are heavily CPU-bottlenecked 19:15 Calinou why arent we running software on GPUs? 19:16 Calinou GPUs are suitable for heavily parallel operations only 19:16 Calinou it is harder to program for them 19:16 Calinou due to their parallelism 19:16 Megaf_ Yep, we arent even utilizing SMP power properly... 19:17 Megaf_ Calinou: and my the way, MT runs pretty well on my EeePC 701 netbook, with a Celeron at 630 MHz, it slows down because of its terrible GPU 19:17 Megaf_ but I think its a drivers fault 19:18 Megaf_ I used to play Need For Speed Porsche on Windows 2013 Enterprise on it, that was a fantastic perfomance 19:24 Calinou Yep, we arent even utilizing SMP power properly... 19:24 Calinou requires a whole engine rewrite 19:25 Calinou much harder than you think 19:25 Calinou Megaf_, on PCs with good CPU and dedicated graphics card, the game is CPU-bottlenecked 19:26 PenguinDad Megaf_: SMP is also slower sometimes 19:26 Megaf_ Calinou: That's exactely what I propose, a whole rewrite of minetest 19:26 Megaf_ and that you drop irrlicht 19:27 Megaf_ several bugs and several limitations of minetest comes from irrlicht, is not even a minetest fault 19:27 Megaf_ I propose dropping irrlicht and swiching to python 19:27 Megaf_ or better, adding support to python and keeping LUA 19:28 Megaf_ and if I was good in C++ I would do it myself for sure 19:29 Megaf_ Minetest is a fantastic game, I woul change almost nothing on the gameplay and visuals 19:29 Megaf_ would* 19:29 Megaf_ Its perfect the way it is 19:30 Calinou WTF 19:30 Calinou can you stop suggesting stupid things 19:30 sfan5 Megaf_: rpi designed for media intensive tasks? 19:30 sfan5 pff 19:31 Calinou Python is likely slower 19:31 Calinou stop being one of these Raspberry Pi fanatics… 19:31 Megaf_ sfan5: nope, its "processor" 19:31 Calinou or better, adding support to python and keeping LUA 19:31 Calinou it is possible, but not worth it for a game 19:31 sfan5 Megaf_: does it encode H.264 faster than my cpu? 19:31 Calinou you are free to make your own rewrite of Minetest 19:31 sfan5 ^ 19:31 Calinou but don't expect a rewrite… 19:31 Calinou Irrlicht is fine 19:32 Calinou it's a simple engine 19:32 Calinou better than using OGRE ;) 19:32 Megaf_ sfan5: for sure it does, via hardware, its GPU can do that via hardware 19:32 Megaf_ sfan5: I watch 1080p videos on it 19:32 sfan5 Megaf_: encoding not decoding 19:32 Megaf_ and 3D wise, it has the same GPU as this phone for example, http://en.wikipedia.org/wiki/Samsung_Galaxy_Fame 19:32 sfan5 well 19:32 PenguinDad Megaf_: you are talking about decoding 19:33 Megaf_ if minetest can run on that phone, it can run on the RPi 19:33 sfan5 it actually does h.264 encoding 19:33 Megaf_ PenguinDad: ok, let me research about encoding 19:33 sfan5 I even have code for it 19:33 sfan5 but the libraries are a PITA to get working 19:33 Megaf_ The VideoCore IV BCM2763 processor improves on the VideoCore III with support for 1080p encode and decode, along with higher resolution camera support and faster 2D and 3D graphics, all at very low power. It is used e.g. in the Nokia 808 PureView.[4] 19:33 sfan5 I'm aware 19:34 Megaf_ so yes, it is a purpose built chip, built for media 19:34 sfan5 but there is literally no software that supports the H.264 encoding interface 19:34 sfan5 lol 19:34 Megaf_ And I do know that is really hard to use that power 19:35 sfan5 I don't think it could software decode H.264 1080p 19:35 sfan5 so much for "media" 19:35 Megaf_ sfan5: of course it can't software decode anything 19:35 Megaf_ not fast... 19:35 sfan5 H.264 is media 19:35 sfan5 got you 19:35 Megaf_ it has a dedicated hardware to do that, using 3 watts... 19:35 sfan5 what about 10-bit 1080p H.264? 19:36 sfan5 it doesn't decode that 19:36 sfan5 and thats not "some format nobody uses" 19:36 sfan5 almost all anime releases are in H.264 10-bit since a few years 19:36 shadowzone !seen myself 19:37 shadowzone oops 19:37 shadowzone Wrong channel. 19:37 Megaf_ sfan5: you are deviating from my point 19:37 Megaf_ and 1 Gigapixel per second fill rate is a considerable perfomance 19:37 sfan5 am I? 19:37 sfan5 it's media too 19:38 Megaf_ ok, my original point is, we are having trouple to compile minetest with OpenGL ES 19:38 Megaf_ this discussion we are having now will lead to nowhere 19:39 sfan5 what was the problem again? 19:39 Megaf_ let me turn it on again, wait a second please 19:43 Megaf_ sfan5: ok,my cmake looks like this cmake ../../ -DVERSION_EXTRA=MinetestPi -DRUN_IN_PLACE=1 -DENABLE_GLES=1 -DIRRLICHT_SOURCE_DIR=~/ogl-es/ -DENABLE_SOUND=0 -DENABLE_LEVELDB=0 -DENABLE_CURL=0 -DENABLE_REDIS=0 19:43 Megaf_ I will run make again 19:43 Megaf_ a clean make 19:43 sfan5 nono 19:44 sfan5 just paste the output of cmake 19:45 Megaf_ sfan5: https://gist.github.com/bcae0c623ca46ca29773.git 19:45 Megaf_ cmake runs fine, in a couple of hours I will paste the output of make 19:46 Megaf_ sfan5: anyway, theres no GL ES on that output, 19:46 sfan5 -- Found system opengles2 library /usr/lib/arm-linux-gnueabihf/libGLESv2.so;/usr/lib/arm-linux-gnueabihf/libGLESv2.so;/usr/lib/arm-linux-gnueabihf/libX11.so;/usr/lib/arm-linux-gnueabihf/libXext.so 19:46 sfan5 did you ignore that? 19:47 sfan5 thats at least what I see in the cmake output 19:47 kilbith sfan5: could you upload your compilated MT with OGLES, pretty please ? 19:47 Megaf_ sfan5: isnt there a way to disable OpenGL? Found OpenGL: /usr/lib/arm-linux-gnueabihf/libGL.so 19:48 sfan5 I didn't succeed in compiling Minetest on my rpi 19:48 sfan5 Megaf_: why would you want to disable opengl if it works 19:48 sfan5 Megaf_: also if you use the ogl-es irrlicht you can't use opengl anyway 19:48 kilbith ah :( 19:51 Megaf_ sfan5: because OpenGL use useless for RPi 19:51 sfan5 Megaf_: why? 19:52 Megaf_ sfan5: the RPi GPU doesnt do OpenGL 19:52 Megaf_ OpenGL is done via software 19:55 Megaf_ "This is because the Videocore GPU runs a RTOS that handles the real processing; all of the actual video acceleration is done using this RTOS firmware coded for its proprietary GPU" 19:55 Megaf_ lol, there's an operating system running o the RPis GPU 19:56 Megaf_ http://en.wikipedia.org/wiki/Free_and_open-source_graphics_device_driver#Broadcom 20:05 ShadowNinja kahrl: Are you O.K. with me merging that coroutime patch now? I don't now if the SCRIPTAPI_PRECHECKHEADER functions are called from Lua API functions, but it doesn't seem like it, and this at least fixes a lot of functions. 20:05 ShadowNinja (Even itf it doesn't fix everything) 20:37 kahrl ShadowNinja: yeah, do it 20:38 ShadowNinja kahrl: Done, how about https://github.com/minetest/minetest/pull/1721 (with my commented modification, better commit messages, and squash). 20:38 ShadowNinja +? 20:40 kahrl seems good 20:41 kahrl merge it 20:55 BlockMen is there a reason why https://github.com/minetest/minetest/pull/1699 is not merged yet? 20:59 Calinou https://github.com/minetest/minetest/pull/1636 → :'( probably dead 20:59 Calinou (“rebase needed” should generally be closed, to me…) 21:00 Calinou add label, close, author reopens if they rebased 21:00 ShadowNinja BlockMen: Yes. Nobody's done it. :-P Otherwise, not as far as I can see. 21:01 ShadowNinja Calinou: A core dev can probably rebase it, if it's agreed to be merged. 21:02 ShadowNinja I think that core.colorize should add a reset token instead of the color white after the text. We might provide a text color option later. 21:03 Calinou there could be some automatic colours 21:03 Calinou error messages in slight red, warnings in slight yellow, infos in slight blue (in chat) 21:03 Calinou and coloured join/leave messages of players 21:03 Calinou but that's just additional features 21:05 ShadowNinja Calinou: With chat command return values error/success would be trivial, but there's no other classifications. 21:06 ShadowNinja Third time in two days someone's left a second too early. 21:13 ShadowNinja kahrl / BlockMen: What about https://github.com/minetest/minetest/pull/1673? 21:14 BlockMen ShadowNinja, not tested but seems ok 21:14 * ShadowNinja has tested it. 21:17 BlockMen then merge it 21:39 Megaf sfan5: kilbith: That's the error I have when compilining minetest on my raspberry pi 21:39 Megaf https://gist.github.com/bcae0c623ca46ca29773.git 21:39 Megaf https://gist.github.com/Megaf/bcae0c623ca46ca29773#file-rpicmake-L219 21:39 Megaf ^ 21:42 PenguinDad Megaf: looks you are missing these egl symbols 21:44 Megaf PenguinDad: allright, and how do I fix that> 21:44 Megaf ? 21:45 kilbith well then, the opengl es integration needs to be re-based 21:46 Megaf kilbith: weird thing is, the Android build works, and it uses GL ES 21:46 Megaf and it doesnt uses X 21:46 Megaf I wish sapier were here, he would be able to help us out 21:46 kilbith Megaf: I know 21:47 kilbith sapier == proller ? 21:47 proller WAT 21:47 proller NOOO!!!! 21:47 kilbith just wondered 21:47 proller NOOOOO 21:47 proller also, who can compile irrlicht demos under pi ? 21:47 PenguinDad Megaf: you have to somehow point cmake to your libEGL but I don't know how 21:47 proller and run them 21:56 Megaf BlockMen: any idea? 21:56 Megaf kahrl: ? 21:56 Megaf ShadowNinja: ^ 21:56 Megaf we are trying to port Minetest to Raspberry Pi 21:58 Megaf proller: what error do you get when you try to compile irrlicht demos? 21:58 Megaf I got this when trying to compile hello world /usr/bin/ld: cannot find -lGLES_CM 21:58 Megaf /usr/bin/ld: cannot find -lXcursor 21:59 proller my error is dark screen after running 22:00 Megaf proller: at least you can compile them 22:03 proller and i can compile freeminer on pi ;) 22:03 Megaf almost a year later and we still have the same problems... 22:03 Megaf thats so frustrating and disapointing 23:53 paramat hmmmm, could you take a quick look at this voxelmanip lighting issue? https://github.com/minetest/minetest/issues/1727 If you have advice on this i'll code it, test and make a pull request.