#jogamp @ irc.freenode.net - 20130305 05:05:41 (UTC)


20130305 05:05:41 -CatOut- Previous @ http://jogamp.org/log/irc/jogamp_20130304050527.html
20130305 05:05:41 -CatOut- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130305050541.html
20130305 05:24:29 * hharrison (~chatzilla@anon) has joined #jogamp
20130305 06:53:04 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20130305 07:43:27 * [Mike] (~Mike]@anon) Quit ()
20130305 08:24:27 * ebrayet (~Adium@anon) has joined #jogamp
20130305 10:46:02 <sgothel> @Xerxes: Review of KeyEvent changes ?
20130305 10:47:37 <sgothel> http://jogamp.org/git/?p=jogl.git;a=commit;h=85338858f5c58694fa88e77df1386d0556887944
20130305 13:16:02 <xranby> sgothel: yes, all i find is that you have done a thorough job.
20130305 13:16:52 <sgothel> the TYPED event is synthesized .. so your Linux* input event drivers shall work
20130305 13:17:07 <xranby> i am trying to gasp why krishnak till have some issue with JMonkey engine
20130305 13:17:08 <sgothel> only auto-repeat is no more delivered as TYPED
20130305 13:17:44 <sgothel> they all shall stop using TYPED if possible now - since I will remove that soon
20130305 13:18:00 <sgothel> NEWT TYPED ... that is
20130305 13:18:13 <xranby> gouessej have switched JMonkeyEngine to only use the RELEASED events
20130305 13:18:20 <sgothel> great
20130305 13:19:43 <xranby> for what i can see is that you have changed the raspberry pi LinuxEventDeviceTracer to only send RELEASED events http://jogamp.org/git/?p=jogl.git;a=blobdiff;f=src/newt/classes/jogamp/newt/driver/linux/LinuxEventDeviceTracker.java;h=2c0e6d3dd5163b80b21de9e5fac15f8d499245cc;hp=5efce2524a9ac343d2a9e2c7ac4bab826252a149;hb=85338858f5c58694fa88e77df1386d0556887944;hpb=a40ee817a3bd537b0de7018772b0835f995a1bed
20130305 13:20:05 <xranby> which is ok
20130305 13:20:16 <sgothel> only PRESSED and RELEASED
20130305 13:20:22 <sgothel> all NEWT drivers do that now
20130305 13:20:38 <sgothel> i.e. we even through an exception if you try to deliver TYPED :)
20130305 13:20:39 <xranby> agree, before we also sent the TYPED
20130305 13:21:03 <xranby> for this work i say well done
20130305 13:21:03 <sgothel> -> WindowImpl.consumeKeyEvent(..)
20130305 14:22:30 <xranby> sgothel: there is two jogamp items jogamp tests and launcher from 2012 in the android "play" store that end users are unable to run because no gluegen or JOGL apk's gets installed.
20130305 14:23:36 <sgothel> the ones which cause problems contain native code .. and hence the google store may 'restrict' something .. which could be valid or not
20130305 14:23:46 <sgothel> pls try installing APK directly ...
20130305 14:23:46 <xranby> i tested this yesterday on a samsung BCM VC IV armv6 phone
20130305 14:24:08 <sgothel> e.g.http://jogamp.org/deployment/archive/master/gluegen_644-joal_405-jogl_924-jocl_753/apk/
20130305 14:24:52 <sgothel> will update the APK at a later time - very low prio
20130305 14:25:37 <sgothel> installing directly usually works pretty well, user just needs to say 'allow installation from unknown sources' somewhere in he settings (dev settings?)
20130305 14:26:10 <sgothel> I even could install them from the 7z file using a free 7z android tool :)
20130305 14:26:21 <xranby> hehe
20130305 14:26:27 <xranby> i am sure that work...
20130305 14:27:08 <xranby> odin_: ^ note that samsung ships broadcom video core iv based phones.
20130305 14:27:32 <xranby> odin_: thus hardware accelerated android is out there in the wild
20130305 14:27:50 <xranby> the main question is why the drivers are not available to the raspberry pi users
20130305 14:28:00 <sgothel> the discussion here .. yup
20130305 14:28:23 <sgothel> I tried to correct your statement .. i.e. the driver is wrapped by a google/android EGL/ES one
20130305 14:28:50 <sgothel> this wrapper mainly introduces the ANativeSurface semantics etc
20130305 14:29:15 <sgothel> you see this .. that the lib/EGL/vendor/libES.. thing exists loaded by lib/libES.so
20130305 14:29:29 <sgothel> or something similar .. you know it when you see it :)
20130305 14:29:42 <sgothel> this was our 1st trouble actually loading the native library
20130305 14:30:53 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130305 14:31:08 <sgothel> http://jogamp.org/git/?p=jogl.git;a=commitdiff;h=8c1007f28396956d5ef83fbbbea78e910c1e53bd <- FAILURE :)
20130305 14:31:19 <sgothel> /system/lib/egl/libEGL_POWERVR_SGX530_125.so
20130305 14:31:58 <xranby> sgothel: oh my
20130305 14:32:03 <sgothel> solution is to load the 'normal' libEGL.so 'globally'
20130305 14:32:24 <sgothel> i.e. you load the wrapper libEGL which itself will load the vendor libEGL
20130305 14:32:38 <sgothel> the dlopen must be performed w/ global flags
20130305 14:33:43 <xranby> ok, so if you have access to the broadcom wrapper lib.. load it and have hardware accelerated android on raspberry pi for free?
20130305 14:34:18 <xranby> i have to ask my work mate to borrow his phone a while
20130305 14:34:26 <sgothel> if you have this wrapper lib .. wrapping BCM handles to ANativeSurface .. yes
20130305 14:34:29 <xranby> and extract the required bits
20130305 14:34:45 <sgothel> the wrapper source is open source and available
20130305 14:34:58 <xranby> thats even better
20130305 14:35:10 <sgothel> don't know how much vendor magic is included, since there is one generic one
20130305 14:35:22 <sgothel> in the default/std android source tree
20130305 14:39:43 <sgothel> Magic EGL config on Android: /system/lib/egl/egl.cfg
20130305 14:40:20 <sgothel> i.e. to enable NVidia's perfhud:
20130305 14:40:42 <sgothel> echo 0 0 android > temp_perf.txt
20130305 14:40:42 <sgothel> echo 0 1 perfhud >> temp_perf.txt
20130305 14:40:42 <sgothel> adb push temp_perf.txt /system/lib/egl/egl.cfg
20130305 14:40:42 <sgothel> adb push libs/libEGL_perfhud.so /system/lib/egl/libEGL_perfhud.so
20130305 14:40:42 <sgothel> adb push libs/libGLESv2_perfhud.so /system/lib/egl/libGLESv2_perfhud.so
20130305 14:40:42 <sgothel> adb push libs/libGLESv1_CM_perfhud.so /system/lib/egl/libGLESv1_CM_perfhud.so
20130305 14:40:51 <sgothel> adb reboot
20130305 14:41:34 <sgothel> egl.cfg: defines how you load the implementation libs from the wrapper lib
20130305 15:05:16 <xranby> sgothel: ubuntu are dodging wayland, and instead they will use Mir https://wiki.ubuntu.com/MirSpec
20130305 15:06:17 <xranby> http://www.omgubuntu.co.uk/2013/03/ubuntus-new-display-server-mir-gets-demoed-video
20130305 15:06:33 <sgothel> sadly so - IMHO a disaster due to misunderstood architecture and lots of impatience .. see http://datenwolf.net/bl20110930-0001/
20130305 15:07:25 <sgothel> but - whatever, they still have an X11 client lib 'adapter' AFAIK
20130305 15:28:18 <sgothel> further: I don't know whether the term 'display server' applies to Wayland .. etc
20130305 15:28:43 <sgothel> I could simply throw up when I read something like this: 'However, users nowadays expect a more consistent and a more integrated user experience than what is possible to offer on top of the X window system.'
20130305 15:29:25 <sgothel> what an uneducated arrogant statement knowing nothing .. as it seems to be .. mama mia
20130305 15:31:24 <sgothel> all the pretty pictures (not really) .. don't show separation of: Compositor, WindowManager, Input System, Font Management , etc etc etc -
20130305 15:31:53 <sgothel> I remember KDE team said, even though they will offer Wayland binding, the KDE WM will still render decorations etc
20130305 15:32:13 <sgothel> -> own compositor/WM
20130305 15:33:45 <sgothel> If one reads the 'objectives' .. they feel like an insult to X11 - where one would need to ask, whether X11 was an attempt to sabotage a sane desktop environment or something like this :)
20130305 15:34:05 <sgothel> ok .. me better calming down :)
20130305 15:36:54 <sgothel> Mir Scope: I see modules similar to X11 - actually I (like many others) would have loved to see a more open X12 discussion - not a Ubuntu/Google/<BIG-COMPANY> driven something here. An X12 which may allow to get rid of something nor required anymore. However, identifying those redundancies .. is hard, since X.org's driver architecture almost got rid of most things, where the WM does render/composite everything.
20130305 15:37:53 * ebrayet (~Adium@anon) has joined #jogamp
20130305 15:38:21 <sgothel> In short: Mir looks like an attempt to own a technology by Ubuntu by creation and association (similar to what Google does) - to have a market power and veto right, instead of collaborating w/ X.org
20130305 15:39:19 <sgothel> IMHO this is very sad, since even the X.org and Linux-Kernel dependencies are already a huge mess especially for other free OS
20130305 15:40:59 <sgothel> You can see this w/ Mesa3D driver's today already and hence X.org, where most won't operate outside of the Linux kernel. Why? Failure in specs and licenses of kernel modules. X11 once was MIT licensed, where DRI/DRM drivers are GPLv2 and BSD folks cannot use them .. for example. etc etc
20130305 15:43:35 <sgothel> Besides power/politics, Mir is ofc driven by Ubuntu to follow their mobile goals. Well, this would have been also possible w/ a sort of X11 [or X12], or QT, however the latter would not give them enough unique flexibility to steer the ship :)
20130305 15:48:42 * ebrayet (~Adium@anon) Quit (Ping timeout: 264 seconds)
20130305 15:50:38 * ebrayet (~Adium@anon) has joined #jogamp
20130305 15:55:03 <sgothel> "developed by Ubuntu for Ubuntu" http://www.olli-ries.com/mir-unity-qml-unity-apis-unity/
20130305 16:01:58 <sgothel> .. so long story short: It would require us to have an Qt binding [Ubuntu goes Qt, finally] :)
20130305 16:06:56 <sgothel> http://phoronix.com/forums/showthread.php?78401-Ubuntu-Announces-Mir-A-X-Org-Wayland-Replacement#post315932 :)
20130305 16:10:29 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130305 16:15:13 * ebrayet (~Adium@anon) has joined #jogamp
20130305 16:16:32 * [Mike] (~Mike]@anon) has joined #jogamp
20130305 16:20:55 <sgothel> The Wayland folks discussion: https://plus.google.com/100409717163242445476/posts/jDq6BAgdpkG (Warning: Google+ :) .. "Mir is, I believe, the Russian word for peace. The irony." :)
20130305 16:21:54 <sgothel> Ofc .. Wayland is driven by GPU vendors (Intel, ..) to allow a most simplified EGL based desktop graphics driver
20130305 16:22:31 <sgothel> Same thing, IMHO - could be achieved by a simplified X12 GPU driver architecture as well .. but whatever
20130305 17:24:55 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130305 18:21:40 <odin_> yes hardware accelerated android is available from broadcom engineers, but I think they sitting on it, since they want RPi for "education" and don't want to be distracted by the android "circus"
20130305 18:22:42 <odin_> so it maybe a case that we need to release image for RPi first before broadcom engineers release theirs, to make sure they have best performance in any write-up/review that might damage image of broadcom's performance
20130305 18:22:47 <odin_> ^^^ my words
20130305 18:22:52 <sgothel> sounds similar to the RPi officials - babysitting a platforms user - but then again, the HW platform is as closed as any other ..
20130305 18:23:21 <odin_> my weekly football kick around now, time to exercise, laters
20130305 18:23:33 <sgothel> good kicking
20130305 19:45:53 * ebrayet (~Adium@anon) has joined #jogamp
20130305 23:09:49 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130306 02:53:27 * [Mike] (~Mike]@anon) Quit ()
20130306 03:19:39 * [Mike] (~Mike]@anon) has joined #jogamp
20130306 05:05:54 -CatOut- Continue @ http://jogamp.org/log/irc/jogamp_20130306050554.html