#jogamp @ irc.freenode.net - 20150805 05:05:27 (UTC)


20150805 05:05:27 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150804050527.html
20150805 05:05:27 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150805050527.html
20150805 05:40:24 * xranby_ (~familjen@anon) Quit (Ping timeout: 250 seconds)
20150805 06:30:28 * elect_ (~elect@anon) has joined #jogamp
20150805 06:58:03 <elect_> hi
20150805 07:40:12 <elect_> I finished the jogl hello triangle
20150805 07:40:53 <elect_> https://github.com/elect86/jogl-samples/blob/master/jogl-samples/src/helloTriangle/HelloTriangle.java
20150805 07:43:12 <elect_> I implemented pure newt, animator, element drawing, dinamic attribute and frag data binding, vao, vbo (with normalization), ibo (with shorts), glsl program and matrix jogl util
20150805 07:43:27 <elect_> plus key listener
20150805 07:43:32 <elect_> and right way to dispose
20150805 07:43:38 * eclesia (~husky@anon) has joined #jogamp
20150805 07:43:43 <eclesia> hi
20150805 07:43:57 <elect_> on the "esc" animator.stop + glWindow.destroy and System.exit on dispose()
20150805 07:44:00 <elect_> hi
20150805 07:50:40 <elect_> I added a small entry in the wiki, https://jogamp.org/wiki/index.php/Jogl_Tutorial#Hello_Triangle
20150805 08:25:52 * xranby_ (~familjen@anon) has joined #jogamp
20150805 09:09:39 <xranby_> sgothel: this simple program is "enough" to recreate the 1183 bug https://jogamp.org/bugzilla/attachment.cgi?id=715
20150805 09:09:49 <xranby_> https://jogamp.org/bugzilla/show_bug.cgi?id=1183#c9
20150805 11:13:11 <sgothel> @Xerxes: 1st of all thx! Is this issue confined to raspi+alpha-driver ?
20150805 11:14:18 <sgothel> I would say: yes :)
20150805 11:14:26 <sgothel> (looking at it)
20150805 11:15:46 <sgothel> so confined to raspi .. hmm
20150805 11:21:31 <sgothel> I wonder .. why it is not reproducible on other platforms then .. hmm
20150805 11:27:41 <sgothel> so RandR13 simply returns null .. 'occasionally' .. due to ... (buggy driver?)
20150805 11:28:17 <sgothel> ok .. handling NULL .. but I am afraid thats it then
20150805 11:46:20 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=1183#c12
20150805 12:19:28 * jvanek (jvanek@anon) has joined #jogamp
20150805 12:58:17 <xranby_> sgothel: I can confirm that I no longer can see the exception with your patch for bug 1183 applied
20150805 13:01:09 <xranby_> sgothel: changed the bug to verified fixed
20150805 13:01:25 <sgothel> thank you!
20150805 13:01:41 <sgothel> so root cause still remains .. where randr13 is supposed to work .. ofc
20150805 13:01:57 <sgothel> i.e. occasionally not working -> bug in 'driver'
20150805 13:02:43 <xranby_> i agree, i am not entirely satisfied not knowing why it fails sometimes
20150805 13:02:54 <sgothel> working on Bug 1062 .. weird stuff ..
20150805 13:03:27 <xranby_> unfortunally i do not have a vivante board here to test on :/
20150805 13:03:55 <sgothel> Bug 1062 is restricted to vivante? oh ..
20150805 13:04:26 <sgothel> Bug 1185 .. I like to mark this as invalid .. hmm see https://jogamp.org/bugzilla/show_bug.cgi?id=1185#c8
20150805 13:05:23 <xranby_> Bug 1062 -> https://jogamp.org/bugzilla/attachment.cgi?id=630 -> EGL vendor Vivante Corporation, version 1.4
20150805 13:08:19 <xranby_> i have no idea how to reproduce 1185
20150805 13:09:07 <sgothel> 'i.mx6' -> vivante ? hmm
20150805 13:09:28 <xranby_> sgothel: do you have any vivante board at your location?
20150805 13:09:29 <eclesia> alive*
20150805 13:09:38 <sgothel> surely an issue w/ libGL/libGLES* issues
20150805 13:09:49 <sgothel> no vivante :-/
20150805 13:10:37 <sgothel> guess a workaround alike 'gles-only' flag would be useful here .. hmm
20150805 13:11:19 <xranby_> http://forum.jogamp.org/OneTriangleAWT-crashes-with-quot-glGetString-quot-message-td4035015.html -> glxinfo.log ->
20150805 13:11:19 <xranby_> OpenGL vendor string: Vivante Corporation
20150805 13:11:19 <xranby_> OpenGL renderer string: Vivante GC2000
20150805 13:11:19 <xranby_> OpenGL version string: 2.1 2.0.1
20150805 13:11:39 <xranby_> yup definately vivante
20150805 13:11:56 <sgothel> and that is a vendor blob driver?
20150805 13:11:58 <sgothel> hmm :)
20150805 13:13:02 <xranby_> GC2000 used in i.MX6 Dual and Quad
20150805 13:13:20 <xranby_> according to https://en.wikipedia.org/wiki/Vivante_Corporation
20150805 13:14:03 <xranby_> "There are no plans on writing a new DRM/KMS driver kernel driver for the Vivante hardware, since Vivante previously put out their Linux kernel component under the GNU General Public License (GPL), instead of maintaining it as a proprietary blob. The free Gallium3D-style device driver etna_viv has surpassed Vivante's own proprietary user-space driver in some benchmarks.[citation needed] It supports Vivante's product line of GC400
20150805 13:14:42 <xranby_> maybe a solution is to ask the user to install etna_viv ?
20150805 13:15:52 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=1062#c14 - thx!!
20150805 13:16:39 <sgothel> excellent
20150805 13:17:18 <sgothel> so I guess I can add the workaround, we already have a 'skip GLES' flag - so adding a 'skip GL-desktop' flag seems to be appropriate
20150805 13:19:25 <sgothel> jogl.disable.opengldesktop
20150805 13:19:30 <sgothel> ^^ we have that already :)
20150805 13:20:00 <xranby_> excellent :)
20150805 13:20:20 <xranby_> you can skip desktop by using jogl-all-mobila.jar as well
20150805 13:20:24 <xranby_> mobile
20150805 13:20:42 <sgothel> right .. things get .. err .. big :)
20150805 13:20:53 <sgothel> I see, you adopted the fat-jar already :)
20150805 13:21:19 <sgothel> maybe it speeds up loading your app a bit, since it requires to only load this one jar + your app
20150805 13:21:32 <sgothel> then .. you can merge your app to the fat-jar .. -> one file only
20150805 13:21:44 <xranby_> loading times are still quite heavy using jar's
20150805 13:22:02 <sgothel> assuming jar is memory mapped and not inflated completely ..
20150805 13:22:04 <xranby_> the fastest way to launch jogl on my arm systems is to first unpack the ajr
20150805 13:22:10 <sgothel> ah
20150805 13:22:38 <xranby_> too much time is spent in java's lib zip
20150805 13:22:55 <sgothel> hmm
20150805 13:23:12 <sgothel> maybe an alternative jar w/ zero compression? :)
20150805 13:23:25 <eclesia> or pack200 compression
20150805 13:23:32 <xranby_> is the ubuntu fonts still included?
20150805 13:23:33 <sgothel> however .. unpacked seems to be fine
20150805 13:23:40 <xranby_> i recall they being 4mb something
20150805 13:24:02 <sgothel> nope .. atomic for quite a while
20150805 13:24:44 <xranby_> i will start experimenting again so see how to deploy an jogl app on mobile the quickest
20150805 13:25:08 <xranby_> starting jogl on android is fast... they must have done something right
20150805 13:25:13 <sgothel> maybe resulting in a new mobile jar ..
20150805 13:25:49 <sgothel> @eclesia: pack200 .. yeah, we do that when signing/deploying for network .. probably quite slow on ARM :)
20150805 13:26:28 <eclesia> it implies far less reading operations, so it might faster even if it's more compressed
20150805 13:26:52 <sgothel> good hint .. lets see what xerxes perfect results will be
20150805 13:27:21 <sgothel> atomic/jogl-fonts-p0.jar
20150805 13:27:34 <eclesia> another compression I use is XZ. produces pretty good performance and compression ratio. and there is a pure java decoder available
20150805 13:28:29 <sgothel> xz/7x .. excellent, yup, pack200 is similar if I remember correctly
20150805 13:28:57 <sgothel> however .. we need the signature entries in the jar file when loading via GlueGen (natives .. etc)
20150805 13:29:01 <eclesia> pack200 is a bit more tricky to decompress, need to call a tool from the jvm
20150805 13:29:26 <xranby_> let me try pack200 on the jogamp-fat.jar
20150805 13:30:24 <sgothel> http://jogamp.org/git/?p=jogamp-scripting.git;a=blob;f=deployment/funcs_jars_pack_sign.sh;h=979f57ceaf9e7778db53ecc3a76b54d99bec201b;hb=HEAD#l34
20150805 13:30:29 <sgothel> [1] repack
20150805 13:30:48 <sgothel> http://jogamp.org/git/?p=jogamp-scripting.git;a=blob;f=deployment/funcs_jars_pack_sign.sh;h=979f57ceaf9e7778db53ecc3a76b54d99bec201b;hb=HEAD#l81
20150805 13:30:50 <sgothel> [2] pack
20150805 13:31:00 <sgothel> w/ PACK200_OPTIONS="--segment-limit=-1"
20150805 13:31:31 <sgothel> if it helps .. well, we can add [re]packing the fat-jars in that script as well ..
20150805 13:32:05 <sgothel> I just don't like to sign it for network usage .. until proven faster via network (unlikely due to download size)
20150805 13:32:21 <sgothel> well .. probably depends on location :)
20150805 13:38:38 <xranby_> Brainstorming on, bug 1062 : can jogl be engineered to restrict the LD_LIBRARY_PATH in order to ensure that we use libGL and libEGL from the same location? the goal is to prevent a driver from loading parts using the shared library linker that belongs to a differnt driver
20150805 13:39:10 <sgothel> the key here is lookup-global ..
20150805 13:39:36 <sgothel> however, for glGetString symbol we use the tool lookup, i.e. eglGetProcAddress etc
20150805 13:39:54 <sgothel> and I validated that the right tool is being used .. as well as the right symbol :-/
20150805 13:39:55 <xranby_> glxgears always "work" because it has the paths to one driver hardcoded since compiletime ?
20150805 13:40:04 <sgothel> yes
20150805 13:40:23 <sgothel> in bundle-info we say how libs are being loaded ..
20150805 13:40:29 <sgothel> global .. or not
20150805 13:41:24 <sgothel> we seems to have issues here, current setting is the 'works in most cases' compromise .. where 'local only' is ofc desired
20150805 13:42:07 <sgothel> DynamicLibraryBundleInfo shallLinkGlobal() / shallLookupGlobal() <- this
20150805 13:42:57 <xranby_> ... meh pack200 failed on my raspberry pi -> out of bounds exception
20150805 13:43:05 <xranby_> probably out of ram
20150805 13:43:27 <xranby_> quicksort failed
20150805 13:43:30 <zubzub> that's an array index out of bound
20150805 13:43:40 <zubzub> don't see how that is related to out of ram?
20150805 13:43:59 <sgothel> a = new int[max]; ? :)
20150805 13:44:15 <sgothel> heap .. whatever
20150805 13:44:25 * jvanek (jvanek@anon) Quit (Ping timeout: 240 seconds)
20150805 13:45:03 <zubzub> out of heap or ram gives different exceptions
20150805 13:45:20 <xranby_> i guess. in any cause.. it didnt work
20150805 13:45:23 <xranby_> :/
20150805 13:45:31 <xranby_> zubzub: you are right.
20150805 13:45:36 <sgothel> and a jar w/ zero compression ?
20150805 13:45:58 <xranby_> i will postpone these experiments to a day i am back to work
20150805 13:46:04 <xranby_> i am on vacation
20150805 13:46:13 <sgothel> ah :)
20150805 13:46:43 <sgothel> happy vacation .. traveling the world .. or in your sweet home, as there is no place alike? :)
20150805 13:48:05 <xranby_> traveling the world... some days doing something completely different
20150805 13:48:43 <sgothel> our loss is somebody else gain .. I am afraid then :-/
20150805 13:49:17 <xranby_> i am sorry i cant help with the last tail of bugs before the release
20150805 13:49:40 <sgothel> all good .. not the end of the [raspi] world :)
20150805 13:50:03 <sgothel> you already enhanced all use cases - KUDOS
20150805 13:51:12 <xranby_> i think the raspi is starting to become quite nifty my only big complain is loadingtimes :/ unfortunally
20150805 13:51:50 <xranby_> thus no bugs bugging me :)
20150805 13:52:11 <xranby_> only finetuning of startup times and it is stellar
20150805 13:52:39 <xranby_> surely there is some cornercases that can be polished but all in all... it work!
20150805 13:52:45 <sgothel> then again: suspend to ram/sdcard .. <- my earlier recommendation
20150805 13:52:47 <sgothel> great
20150805 13:53:26 <sgothel> or .. if one could 'suspend' just one process-tree .. hmm, like a virtual machine :) .. but then again, it needs to be loaded as well
20150805 13:53:38 <xranby_> ok have a good evening and weekend
20150805 13:53:48 <sgothel> you too
20150805 13:54:06 <xranby_> thank you, and superkudos for all work you do sven!
20150805 13:54:29 <xranby_> send cheers and greetings to all of your family
20150805 13:54:43 <eclesia> have a nice trip :)
20150805 13:55:02 <sgothel> thank you .. happy days & cheers to your family as well, ofc
20150805 13:55:18 <xranby_> eclesia: thank you, and well done having a release!
20150805 13:56:56 * xranby_ (~familjen@anon) has left #jogamp
20150805 14:02:37 * jvanek (jvanek@anon) has joined #jogamp
20150805 14:03:20 * jvanek (jvanek@anon) Quit (Client Quit)
20150805 14:03:36 * jvanek (jvanek@anon) has joined #jogamp
20150805 14:04:50 * elect_ (~elect@anon) Quit (Ping timeout: 240 seconds)
20150805 15:01:28 * jvanek (jvanek@anon) Quit (Ping timeout: 246 seconds)
20150805 15:18:10 * jvanek (jvanek@anon) has joined #jogamp
20150805 15:30:45 * elect (~elect@anon) has joined #jogamp
20150805 16:00:35 * eclesia (~husky@anon) has left #jogamp
20150805 16:50:43 * elect (~elect@anon) Quit (Ping timeout: 252 seconds)
20150805 18:23:31 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20150805 22:40:02 * trigger_ (5ce09c82@anon) has joined #jogamp
20150805 22:42:26 * trigger_ (5ce09c82@anon) Quit (Client Quit)
20150806 05:05:28 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150806050528.html