#jogamp @ irc.freenode.net - 20130902 05:06:10 (UTC)

20130902 05:06:10 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20130901050610.html
20130902 05:06:10 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130902050610.html
20130902 07:20:48 * hrenovka (hrenovka@anon) has joined #jogamp
20130902 08:28:22 * xranby (~xranby@anon) Quit (Read error: Operation timed out)
20130902 08:44:06 * xranby (~xranby@anon) has joined #jogamp
20130902 09:04:09 <xranby> good morning
20130902 09:37:43 <hrenovka> :) :P
20130902 09:38:36 <xranby> hrenovka: welcome
20130902 10:38:42 * chakie (~chakie@anon) has joined #jogamp
20130902 10:44:27 <xranby> sgothel: is VBO glBufferData garbage collector safe? if so then we may start allow transfer of data using non jni memory pinned arrays
20130902 10:44:52 <xranby> that is if GL always copy the data
20130902 10:45:07 <sgothel> good morning :)
20130902 10:45:48 <sgothel> I will be very carefully with such a comment now .. since our last failure :)
20130902 10:46:03 <xranby> the last failure was with the immediate mode
20130902 10:46:10 <xranby> AFAIK
20130902 10:46:11 <sgothel> yup
20130902 10:46:49 <sgothel> so the 'data source' of VBO is not required to stay alive after pushing the data down to the VBO ..
20130902 10:47:03 <sgothel> cpu/gpu ..
20130902 10:53:41 <sgothel> maybe we can 'enhance' GLArrayHandler and the corresponding GLArrayData utils w/ java arrays for 'animation' purposes, while also dropping the cpu side data (optional)
20130902 10:53:58 <sgothel> having perf. tests then would be nice
20130902 10:54:49 <sgothel> indeed, it boils down to the glBufferData(..) calls once in a VBO lifetime (static) .. or 'BufferSubData(..)' (?) if 'animated' data shall be updated
20130902 10:55:30 <sgothel> currently still fighting w/ builds - note to myself: be careful w/ driver updates, especially AMD/fglrx :|
20130902 10:56:31 <sgothel> so .. for static data it doesn't really matter whether we use NIO, but if it's animated stuff .. java arrays would be better suited, similar to PMVMatrix
20130902 10:58:07 <sgothel> not to fglrx users: last and only good driver is amd-driver-installer-catalyst-13-6-beta-x86.x86_64
20130902 10:58:25 <xranby> right now i have the happy feeling that by enforcing the switch from imm mode code to VBO will fix all GC issues like situations that the GC thinks the memory is elligable for collection. with imm mode it was an issue even with using JNI pinned memory
20130902 10:59:24 <sgothel> using VBO .. there should be no issue, since we do not pass a data pointer or the like for being required available after the 'push data' call
20130902 10:59:37 <sgothel> but hey .. thats the spec .. we have to see how reality is :)
20130902 11:00:08 <sgothel> spec wise: { byte[] data; ..; glBufferData(,.. data,..) }
20130902 11:00:34 <sgothel> relieve .. amd node didn't crash
20130902 11:00:56 <xranby> yes yes.. i have to experiment with the reality of glMapBuffer and if it complicates things
20130902 11:01:32 <sgothel> MapBuffer is another thing .. :)
20130902 11:01:52 <sgothel> you need NIO for this for sure ..
20130902 11:02:14 <sgothel> and we do track this .. actually need more testing of our memory tracking and performance for MapBuffer
20130902 11:02:31 <sgothel> since it's included in ES3 .. it sure becomes an issue
20130902 11:03:31 <sgothel> old goals for GLArrayData: Drop the cpu memory to be easy on the java heap - additionally using java-arrays is a great idea
20130902 11:03:47 <sgothel> iff you have time .. perf tests are in jogl-demos
20130902 11:04:30 <sgothel> http://jogamp.org/git/?p=jogl-demos.git;a=tree;f=src/demos/es2/perftst;h=de811b065af92cd860de5284334e0e12a8fa8dfa;hb=HEAD
20130902 11:04:33 <xranby> so android launchers wound be nice to have for the perf tests
20130902 11:04:39 <sgothel> quite ole stuff .. needs to be migrated to jogl
20130902 11:05:31 <xranby> btw pushed one more fix for the raw GL2ES2 demo made the code work on the AMD OpenGL 4 drivers as well https://github.com/xranby/jogl-demos/commit/594cba9b34faf0ef2f0f7bf7801c2b228984b02a
20130902 11:05:54 <xranby> that required the explicit #version 130
20130902 11:06:09 <xranby> and made some commercials for the jogamp utility classes
20130902 11:06:21 <sgothel> why don't you use the advertised GLContext / ShaderCode stuff (forum post) ?
20130902 11:06:35 <xranby> because i wanted this code to only use the raw api
20130902 11:06:45 <sgothel> whatever that is :)
20130902 11:06:52 <xranby> without using utility classes
20130902 11:07:04 <sgothel> it's in GLContext .. pretty much raw
20130902 11:07:21 <sgothel> i.e. get the GLSL version string matching current context
20130902 11:07:41 <xranby> i had issues using com.jogamp.opengl.util.glsl.ShaderCode to simply update the shaders
20130902 11:07:57 <xranby> since i dont load the shaders from a file
20130902 11:08:00 <xranby> or a resource
20130902 11:08:17 <sgothel> pls .. tell me about it .. so we can enhance them
20130902 11:09:14 <sgothel> they sure can edit the code in memory (simple string stuff), but even w/o ShaderCode, GLContext gives you the GLSL version for your shader code. But whatever works .. sure.
20130902 11:09:22 <xranby> in its simlpest form i would like to pass a String with the shader, gl, and the shader type VERTEX / fragment hader and have it return an updated shader
20130902 11:10:19 <xranby> in form of a String
20130902 11:10:52 <xranby> input GLSL 1.1 shader -> output GLSL matching the current context
20130902 11:10:57 <sgothel> right .. totally file based .. sure .. can add that
20130902 11:11:34 <sgothel> well, adding a basic shader 'stub' .. then using the customize methods .. users choice
20130902 11:11:38 <sgothel> similar to all our demos
20130902 11:11:45 <sgothel> just not file based
20130902 11:12:10 <sgothel> public ShaderCode(int type, int count, CharSequence[][] source) ?
20130902 11:12:25 <xranby> hmm
20130902 11:12:47 <xranby> i guess i did miss that one if it existed
20130902 11:13:03 <sgothel> haha .. I haven't squeezed it in just now :)
20130902 11:13:24 <sgothel> make sure you use a StringBuffer .. allowing to edit it .. hmm let me see
20130902 11:13:26 <xranby> what is count?
20130902 11:13:33 <sgothel> StringBuilder
20130902 11:13:53 <sgothel> ah .. old stuff, orig GL API was an array of strings ..
20130902 11:13:57 <sgothel> so just one
20130902 11:14:37 <sgothel> API doc .. number of shaders
20130902 11:14:54 <sgothel> and per shaders (1st array order) .. number of strings
20130902 11:15:33 <sgothel> yes, we can add a simple ctor w/ only one CharSequence if so desired
20130902 11:19:41 <sgothel> can you resend me the link to the OMX player demo in ANSI-c ?
20130902 11:19:52 <sgothel> for raspb.
20130902 11:20:32 <sgothel> very good .. < 10 failures b 1072
20130902 11:21:30 <xranby> sgothel: https://github.com/raspberrypi/userland/tree/master/host_applications/linux/apps/hello_pi/hello_video
20130902 11:21:44 <sgothel> thank you
20130902 11:21:54 <xranby> https://github.com/raspberrypi/userland/tree/master/host_applications/linux/apps/hello_pi/hello_teapot
20130902 11:22:05 <xranby> https://github.com/raspberrypi/userland/tree/master/host_applications/linux/apps/hello_pi/hello_videocube
20130902 11:22:27 <xranby> https://github.com/raspberrypi/userland/tree/master/host_applications/linux/apps/hello_pi/hello_encode
20130902 11:22:30 <xranby> .. and so on
20130902 11:22:35 <sgothel> hoho
20130902 11:23:21 <xranby> https://github.com/raspberrypi/userland/tree/master/host_applications/linux/apps/raspicam
20130902 11:23:38 <sgothel> yes .. you have to test that one later .. when 1st steps are made :)
20130902 11:24:19 <sgothel> http://jogamp.org/deployment/archive/master/gluegen_711-joal_472-jogl_1072-jocl_836/
20130902 11:24:32 <xranby> hopefully all the camera functionality is available using openmax il
20130902 11:24:47 <sgothel> as far as I remember, sure is
20130902 11:24:48 <xranby> broadcom have also added a parralel mmal api that is broadcom specific
20130902 11:24:59 <sgothel> oh ..
20130902 11:25:02 <xranby> as mentioned in the raspicam demo
20130902 11:25:14 <sgothel> yeah, the non-spec stuff will be interesting, i.e. demuxer etc
20130902 11:25:32 <sgothel> or maybe they have an ffmpeg/stagefright impl ? would be best :)
20130902 11:26:00 <sgothel> impl -> backend
20130902 11:26:32 <sgothel> i.e. demuxing in OMX was the NV specific part I had to remove from the ole OMX code ..
20130902 11:27:09 <xranby> that is for sure not in
20130902 11:27:21 <xranby> the NV specifix part
20130902 11:27:26 <xranby> in broadcom source
20130902 11:30:06 <xranby> (13:25:09) sgothel: http://jogamp.org/deployment/archive/master/gluegen_711-joal_472-jogl_1072-jocl_836/ <- cool the first test version since 2.0.2!
20130902 11:31:51 <sgothel> OMX/BCOM: ilclient.c <- the ugly part .. platform dependent
20130902 11:32:45 <xranby> a wrapper or simply ugly?
20130902 11:33:01 <sgothel> well, it wraps the BCOM dependencies ..
20130902 11:33:17 <sgothel> IL/OMX_Broadcom.h
20130902 11:33:35 <sgothel> so if we seriously like to do OMX .. we need to abstract this as well
20130902 11:33:44 <xranby> they mention ilclient.h // Video deocode demo using OpenMAX IL though the ilcient helper library
20130902 11:34:11 <sgothel> sure .. I got it from the lib folder in you repo
20130902 11:34:20 <sgothel> from *your* mentioned repo
20130902 11:35:04 <sgothel> looks like we can feed the data from java side -> NIO - good
20130902 11:35:27 <sgothel> plus we can run the OMX stuff one-threaded due to our StreamWorker - good
20130902 11:35:50 <sgothel> however, will need to see when I have time for this
20130902 11:36:12 <sgothel> in EU .. at least, we could distribute ffmpeg/libav ? I guess so ?
20130902 11:36:35 <sgothel> if so - we may simply offer a native jar file for folks who feel legally entitled :)
20130902 11:36:44 <sgothel> for windows and OSX .. I mean
20130902 11:37:42 <sgothel> we could drop this jar file .. and mention it's existence in a movie demo with such remarks
20130902 11:37:52 * xranby (~xranby@anon) Quit (Ping timeout: 246 seconds)
20130902 11:54:37 * xranby (~xranby@anon) has joined #jogamp
20130902 12:22:54 <sgothel> testing memory / thread behavior of GLMediaPlayer w/ visualvm .. looks good, no leak .. StreamWorker busy LFRingbuffer get/put is cheap
20130902 12:41:41 * sgothel (~sven@anon) Quit (Quit: Leaving.)
20130902 13:15:04 * xranby (~xranby@anon) Quit (Read error: Operation timed out)
20130902 13:29:41 * xranby (~xranby@anon) has joined #jogamp
20130902 14:04:28 <xranby> sgothel: http://www.hfrmovies.com/high-frame-rate-example-videos/
20130902 14:07:22 <xranby> makes sense to test the media palayer with high fps movies!
20130902 14:07:27 <xranby> player
20130902 14:26:18 <xranby> JiGong: http://mail.openjdk.java.net/pipermail/jigsaw-dev/2013-August/003328.html <- Jigsaw prototype, take 2
20130902 14:26:38 <xranby> why did they name it jake?! :D
20130902 14:26:43 <xranby> http://hg.openjdk.java.net/jigsaw/jake
20130902 14:28:11 <xranby> i guess this means jigsaw (1) is dead
20130902 15:29:26 * xranby (~xranby@anon) Quit (Quit: Leaving.)
20130902 19:18:47 * sgothel (~sven@anon) has joined #jogamp
20130902 19:18:47 * ChanServ sets mode +v sgothel
20130902 19:19:12 <sgothel> https://jogamp.org/wiki/index.php/SW_Tracking_Report_Objectives_for_the_release_2.1.0 <- emphasizing features ..
20130902 19:27:57 <hrenovka> ok i don't get it
20130902 19:31:23 <sgothel> added features subsections referencing our bugzilla entries ... thats all
20130902 19:32:23 <hrenovka> not that
20130902 19:32:25 <hrenovka> lol
20130902 19:32:34 <hrenovka> i'll take a lok
20130902 19:32:35 <hrenovka> x-D
20130902 19:34:20 <hrenovka> xranby ?
20130902 19:34:22 <hrenovka> :)
20130902 19:39:38 <hrenovka> can anbody tell me whats that with share one context ( openal ) with classes.i'm new in java and jogamp. need to play soun in diff. classes. realy dont get it
20130902 19:39:51 <hrenovka> can anbody give me a brief intro. please
20130902 19:42:58 <sgothel> pls refine your question
20130902 19:44:22 <hrenovka> so...with current context i can replace or delete Alut.alutInit
20130902 19:44:28 <hrenovka> with alc class
20130902 19:45:22 <sgothel> like GL, you need a valid AL ctx bound for most operations
20130902 19:45:26 <hrenovka> that i'm redaing in practical java game programming
20130902 19:45:37 <sgothel> have a look at ALAudioSink
20130902 19:53:06 <hrenovka> ok...i se i need to make my own initialization routines
20130902 19:53:20 <hrenovka> for openal .... whic means alc
20130902 19:53:24 <hrenovka> class
20130902 19:55:00 <hrenovka> eh.. lol x-D
20130902 19:55:05 <hrenovka> that sad
20130902 19:55:07 <hrenovka> sand
20130902 19:55:08 <hrenovka> :-P
20130902 19:57:43 <hrenovka> a little yoke
20130902 19:57:51 <hrenovka> no regards please
20130902 19:57:54 <hrenovka> :)
20130902 19:58:19 <hrenovka> is it correct vocabulary ?
20130903 03:25:02 * jk4 (~jk4@anon) Quit (Remote host closed the connection)
20130903 04:34:15 * jk4 (~jk4@anon) has joined #jogamp
20130903 05:06:10 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20130903050610.html