JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20150811 05:05:28 (UTC)

20150811 05:05:28 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150810050528.html
20150811 05:05:28 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150811050528.html
20150811 06:38:14 * packet0 (~packet0@anon) has joined #jogamp
20150811 06:43:58 * elect_ (~elect@anon) has joined #jogamp
20150811 06:47:00 * packet0 is now known as p0_ack
20150811 06:47:51 * p0_ack is now known as packet0
20150811 07:06:42 <packet0> not a serious issue, but
20150811 07:39:41 * monsieur_max (~maxime@anon) has joined #jogamp
20150811 07:40:28 <packet0> ya, i had a question
20150811 07:41:23 <packet0> is there any particular reason why com.jogamp.nativewindow.swt.SWTAccessor has private variable DEBUG permanantly set to true?
20150811 07:42:21 <packet0> it causes com.jogamp.opengl.swt.GLCanvas to always print small debug info on init
20150811 07:47:48 <elect_> search on github who wrote it "blame" button (it's not a joke ^^)
20150811 07:53:42 * jvanek (jvanek@anon) has joined #jogamp
20150811 07:54:56 <packet0> seems like it was added when fixing GTK stuff...
20150811 07:55:16 <packet0> btw, didn't know about that botton, thanks
20150811 07:58:46 <elect_> I also recently knew it
20150811 08:05:39 * eclesia (~husky@anon) has joined #jogamp
20150811 08:05:56 <eclesia> hi
20150811 08:36:46 <monsieur_max> hi there
20150811 08:37:09 <monsieur_max> just saw the news, you guys may already know that : http://arstechnica.com/gadgets/2015/08/android-to-support-vulkan-graphics-api-the-open-answer-to-metal-and-dx12/
20150811 08:50:00 <xranby> monsieur_max: yes, happy times
20150811 08:50:40 <monsieur_max> :) not fan of android and google, but yes, indeed
20150811 08:52:04 <elect_> yep, glad google didnt make his own low level api
20150811 08:52:26 <elect_> I guess it was an fought decision, but at the end the good sense won
20150811 08:56:34 <packet0> talking about apis, wonder if apple's gonna keep following opengl
20150811 09:01:41 <xranby> packet0: hi, debug output should be off by default
20150811 09:02:07 <xranby> packet0: please change it to be false and prepare a git commit
20150811 09:02:56 <packet0> xranby: hi, would pull request on github be ok?
20150811 09:03:02 <xranby> packet0: yeas
20150811 09:03:03 <xranby> yes
20150811 09:03:57 <xranby> packet0: apple will support opengl es on ios because it is one of the reasons why ios is a success
20150811 09:04:15 <xranby> on the desktop ... it would be madness for them to stop supporting opengl
20150811 09:04:41 <xranby> but as we all have seen, apple like the idea to create apple specific graphics api's
20150811 09:05:25 <xranby> packet0: have you seen any news indicating that apple will drop opengl support?
20150811 09:07:08 <xranby> personally i do not think they will drop opengl support because apples own api's are using it
20150811 09:07:09 <packet0> xranby: no, just wondering since their impl have non-standard things, and is not fixed for years
20150811 09:07:48 <packet0> xranby: about the debug output, was wondering if jogamp.nativewindow.Debug should be used, instead of permanant false
20150811 09:08:24 <xranby> yes it is a good idea to re-use the nativewindow.Debug
20150811 09:10:06 <packet0> ok, thanks for the help
20150811 09:11:46 <xranby> you likely want to: import jogamp.nativewindow.Debug; and then use public static final boolean DEBUG = Debug.debug("SWT");
20150811 09:12:34 <elect_> apple suxs in GL
20150811 09:12:56 <elect_> they are always so late implementing stuff and have a lot of prop ext
20150811 09:13:05 <xranby> i think apple is accumulating technical debt in their GL implementation
20150811 09:13:17 <elect_> but they can't drop it given the importance and share, so they just act lazy
20150811 09:17:53 <xranby> so apple is "stuck" at opengl 4.1 for now?
20150811 09:20:20 <elect_> kind, plus relatively recent dev dont get update
20150811 09:21:23 <xranby> http://blog.imgtec.com/powervr/gnomes-per-second-in-vulkan-and-opengl-es - Gnomes per second in Vulkan and OpenGL ES
20150811 09:22:45 <elect_> yep
20150811 09:22:49 <elect_> it looks good
20150811 09:23:18 <elect_> but I afraid vulkan wont fit for us small dev
20150811 09:24:03 <elect_> I guess using modern GL/ES would be quite close to Vulkan
20150811 09:25:41 <elect_> when mantle came out Carmack said Nvidia ext could give similar benefits
20150811 09:25:42 <elect_> http://www.dsogaming.com/news/john-carmack-nvidias-opengl-extensions-rival-mantle-8gb-on-consoles-not-a-significant-change/
20150811 09:25:56 <elect_> he referred to bindless ext
20150811 09:43:06 <xranby> packet0: if you know which bug that added the permanent true then you may name your commit message as Bug XXX: Cleanup SWTAccessor disable debug messages
20150811 09:47:24 <packet0> don't think i was part of any bugfix, its there since Nov 2012 (v2.0 rc12)
20150811 09:48:01 <sgothel> git pull req. or email formatted patches are always welcome!
20150811 09:48:33 <sgothel> btw .. the whole SWT stuff needs a bit more love -> maintainer
20150811 09:48:47 <sgothel> quite some important bugs are still open AFAIK
20150811 09:49:08 <sgothel> @packet0 are you one of our SWT contribs?
20150811 09:49:14 <xranby> packet0: ok then name it: SWTAccessor: Cleanup disable debug messages
20150811 09:49:23 <xranby> or similar
20150811 09:49:40 <packet0> no, ok
20150811 09:50:49 <xranby> packet0: we use the bug numbers both for enhancements and bugs
20150811 09:51:06 <xranby> packet0: thus when you read the git log some bugs are new features
20150811 09:51:30 <xranby> you can consider our bugzilla as our work queue
20150811 09:51:30 <packet0> have made a pull req.; is it better if i go file a bug report too?
20150811 09:51:51 <xranby> fir this small change its easy for you to post the link to your commit here
20150811 09:51:58 <xranby> and we will review it
20150811 09:52:15 <packet0> https://github.com/sgothel/jogl/pull/90
20150811 09:52:39 <sgothel> great
20150811 09:52:55 <packet0> ya, I'm not used to fileing bugs and stuff, sorry...
20150811 09:53:06 <xranby> packet0: this change looks.. excellent :)
20150811 09:53:10 <xranby> ah no
20150811 09:53:17 <xranby> line 57
20150811 09:53:20 <xranby> have double ;;
20150811 09:53:28 <xranby> you can change it
20150811 09:53:44 <xranby> and then use git add and git --amend
20150811 09:53:51 <xranby> to add the change to you r last commit
20150811 09:54:39 <xranby> packet0: dont worry.. we are here to guide you
20150811 09:56:59 <xranby> packet0: best is if you use your real name as the contributor name
20150811 09:58:32 <packet0> oops, aa
20150811 10:02:37 <packet0> editted directly on github, hope its ok
20150811 10:04:12 <packet0> or was it better to ammend than commit?
20150811 10:04:33 <xranby> packet0: the changes to the files looks fine.
20150811 10:04:43 <xranby> i use git ammend to clean up my git log
20150811 10:04:59 <xranby> in your case i owuld consider using git rebase in interactive mode
20150811 10:05:23 <xranby> to squash the ;; change into the previous change
20150811 10:05:42 <xranby> that way i can send a clean looking change log for review
20150811 10:06:05 <xranby> personally i use git from the command line
20150811 10:06:14 <xranby> and find this "natural"
20150811 10:06:42 <xranby> i am not sure if github have any web interface tool that can do the same
20150811 10:06:51 <packet0> im not on my usual desktop and thought it would be faster than cloning the whole thing
20150811 10:07:37 <xranby> packet0: i leave it up to sgothel to merge it as is or wait for you if you want to clean up the git history and clean up your commit name
20150811 10:07:52 <packet0> ya, I'll try and clean it up
20150811 10:08:18 <xranby> ok we will wait to merge this untill you think it looks good
20150811 10:08:36 <packet0> thanks for the help
20150811 10:08:50 <xranby> packet0: thank you for collaboarting
20150811 10:10:09 <xranby> packet0: one question, which swt project are you working on?
20150811 10:11:44 <xranby> using git ammend and git rebase is always ok to use as long you only change your own history
20150811 10:13:07 <xranby> https://www.atlassian.com/git/tutorials/rewriting-history/
20150811 10:18:23 <xranby> http://streamcomputing.eu/blog/2015-08-06/mediateks-partners-deliver-opencl-on-their-phones/
20150811 10:55:22 <packet0> @xranby I think I've fixed it
20150811 10:57:22 <packet0> rebase -i is quite cool...
20150811 11:03:15 <packet0> as for your question, I'm only using swt personally, nothing big
20150811 11:07:10 <xranby> packet0: looking forward to see your project :)
20150811 11:08:38 <xranby> packet0: the git history looks good
20150811 11:10:07 <packet0> feeling it would have been easier for both if I just wrote it on bugzilla
20150811 11:18:57 <packet0> anyway, thanks for the help and the awesome lib
20150811 11:19:31 <xranby> sgothel: https://github.com/sgothel/jogl/pull/90
20150811 11:20:01 <xranby> packet0: tell teh world ! :)
20150811 11:20:04 <xranby> tell the world
20150811 11:55:22 * packet0 (~packet0@anon) Quit ()
20150811 11:55:34 * packet0 (~packet0@anon) has joined #jogamp
20150811 12:20:30 <xranby> we have many places where debug is permanent disabled.. hmm
20150811 12:20:40 <xranby> xranby@Xerxes-ESPRIMO:~/jogamp$ grep -r "DEBUG = false" jogl/src
20150811 12:20:41 <xranby> jogl/src/newt/classes/com/jogamp/newt/event/NEWTEvent.java: static final boolean DEBUG = false;
20150811 12:20:41 <xranby> jogl/src/jogl/classes/com/jogamp/gluegen/runtime/opengl/GLProcAddressResolver.java: public static final boolean DEBUG = false;
20150811 12:20:41 <xranby> jogl/src/jogl/classes/com/jogamp/gluegen/opengl/BuildStaticGLInfo.java: protected boolean DEBUG = false;
20150811 12:20:41 <xranby> jogl/src/jogl/classes/com/jogamp/opengl/util/texture/spi/SGIImage.java: final boolean DEBUG = false;
20150811 12:20:41 <xranby> jogl/src/jogl/classes/jogamp/opengl/util/pngj/PngHelperInternal.java: static final boolean DEBUG = false;
20150811 12:20:41 <xranby> jogl/src/jogl/classes/jogamp/graph/font/typecast/TypecastFont.java: static final boolean DEBUG = false;
20150811 12:20:42 <xranby> jogl/src/nativewindow/classes/com/jogamp/nativewindow/util/PixelFormatUtil.java: private static boolean DEBUG = false;
20150811 12:23:20 <xranby> and some places in gluegen where debug is permanently true and false.. hmm!
20150811 12:23:21 <xranby> xranby@Xerxes-ESPRIMO:~/jogamp$ grep -r "DEBUG = true" gluegen/src
20150811 12:23:21 <xranby> gluegen/src/java/com/jogamp/common/util/locks/SingletonInstance.java: protected static final boolean DEBUG = true;
20150811 12:23:21 <xranby> gluegen/src/java/jogamp/android/launcher/LauncherTempFileCache.java: private static final boolean DEBUG = true;
20150811 12:23:21 <xranby> xranby@Xerxes-ESPRIMO:~/jogamp$ grep -r "DEBUG = false" gluegen/src
20150811 12:23:21 <xranby> gluegen/src/java/com/jogamp/common/util/SecurityUtil.java: private static final boolean DEBUG = false;
20150811 12:23:22 <xranby> gluegen/src/java/jogamp/common/os/android/StaticContext.java: private static boolean DEBUG = false;
20150811 12:23:22 <xranby> gluegen/src/java/jogamp/common/os/android/AndroidUtilsImpl.java: private static boolean DEBUG = false;
20150811 12:23:23 <xranby> gluegen/src/java/jogamp/android/launcher/AssetDexClassLoader.java: private static final boolean DEBUG = false;
20150811 12:23:23 <xranby> gluegen/src/java/jogamp/android/launcher/TraceDexClassLoader.java: private static final boolean DEBUG = false;
20150811 12:23:24 <xranby> gluegen/src/junit/com/jogamp/common/util/locks/TestRecursiveLock01.java: static final boolean DEBUG = false;
20150811 12:23:24 <xranby> gluegen/src/junit/com/jogamp/common/util/locks/TestRecursiveThreadGroupLock01.java: static final boolean DEBUG = false;
20150811 12:23:25 <xranby> gluegen/src/junit/com/jogamp/common/util/RingBuffer01Base.java: private static boolean DEBUG = false;
20150811 12:23:51 * xranby is filing a bugreport to track this
20150811 12:31:32 * gouessej (5ee4b442@anon) has joined #jogamp
20150811 12:31:52 <xranby> packet0: i have filed a bugreport to track the permanently enabled or disabled debug messages: https://jogamp.org/bugzilla/show_bug.cgi?id=1191
20150811 12:32:58 <gouessej> Hi
20150811 12:34:40 <xranby> gouessej: he there!
20150811 12:34:42 <xranby> hi
20150811 12:35:23 <gouessej> packet0: If Apple drops OpenGL support, I drop OS X support :)
20150811 12:36:26 <gouessej> I just hope that Metal and Renderscript will be total failures.
20150811 12:46:28 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150811 12:51:21 <gouessej> sgothel: The JogAmp fat JAR seems to work :)
20150811 13:11:45 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20150811 14:08:35 <sgothel> jogamp-fat-test.jar: User unit-test runs incl. 'send back' test results, UI to run some demos
20150811 14:09:56 <sgothel> anybody know how to maximize (horz/vert) a window on Windows using the gdi stuff? looking for it now ..
20150811 14:10:20 <sgothel> .. oh and sticky/all-desktop for Windows's windows? :)
20150811 14:16:27 <sgothel> https://msdn.microsoft.com/en-us/library/windows/desktop/ms633548%28v=vs.85%29.aspx <- showWindow MAXIMIZE .. (horz+vert) .. ok, seems they don't have it for just horz and just vert
20150811 14:17:56 <sgothel> so I guess we calc that ourselves (as on OSX)
20150811 14:19:35 <sgothel> .. new Win10 has those 'virt. desktops' right? .. hmm, so a 'sticky' may be supported there
20150811 14:20:13 <gouessej> What is still missing for the Windows implementation?
20150811 14:20:36 <sgothel> max horz/vert and sticky
20150811 14:21:00 <sgothel> I guess I leave sticky then .. and use manual max horz/vert sizes hmm
20150811 14:21:07 <sgothel> lets see how it 'behaves' :)
20150811 14:21:18 <sgothel> (same on OSX)
20150811 14:22:19 <xranby> I will scrutinize the jni signatures and argument types passed for the X11 nativewindow DisplayDriver, i found one semireproducible odd crash seen using jamvm on armhf https://jogamp.org/bugzilla/show_bug.cgi?id=1192
20150811 14:23:35 <xranby> scrutinize jni signatures are my favourite activity this month
20150811 14:24:12 <eclesia> strange hobby
20150811 14:24:17 <sgothel> @Xerxes: the compiler validates them, .. i.e. the native C source includes the javah produces headers!
20150811 14:24:47 <sgothel> a mismatch should break compilation - thats the idea at least
20150811 14:24:48 <xranby> but as we have seen if we pass a jlong and the signature is jint it may truncate data on arm32
20150811 14:25:08 <xranby> something that goes undetected by the compilers
20150811 14:25:24 <sgothel> yes, so that is 'how we use/call some var-arg jni methods'
20150811 14:26:04 <gouessej> sgothel: I'm not sure that sticky windows are natively supported under Windows
20150811 14:26:13 <sgothel> funny question: then why wasn't it breaking on x32/i386 ?
20150811 14:26:24 <gouessej> sgothel: I'm still looking for the right method in GDI
20150811 14:28:02 <xranby> bug https://jogamp.org/bugzilla/show_bug.cgi?id=1166 was one such bug that only trigegred on arm32
20150811 14:28:16 <sgothel> yup .. the one I had in mind
20150811 14:28:31 <sgothel> i.e. question: how can we validate those automatically?
20150811 14:28:51 <xranby> i wish there was a way hmm
20150811 14:28:55 <sgothel> or at runtime (stack, ..)
20150811 14:29:23 <sgothel> so then check var-arg calls that is, i.e. our own 'native -> java' callbacks
20150811 14:30:33 <sgothel> but .. if all unit tests are passed .. it should be ok ..
20150811 14:30:59 <sgothel> don't they on arm32? (dropping experimental drivers etc etc)
20150811 14:31:17 <sgothel> they pass on i386
20150811 14:31:18 <xranby> i usually cross compile
20150811 14:31:25 <sgothel> ofc
20150811 14:31:30 <xranby> long time since i ran the full test suite
20150811 14:31:34 <xranby> on armhf
20150811 14:31:40 <sgothel> we have a remote unit test invocation in the ant build recipe
20150811 14:31:50 <sgothel> dunno whether they still work
20150811 14:32:10 <sgothel> w/ ssh, rsync .. and all that magic :)
20150811 14:32:33 <xranby> do the jogamp chuck buildbot run the full testsuite on armhf?
20150811 14:32:44 <sgothel> so an arm 32-bit test machine using all software rendering might be a desired test machine
20150811 14:32:53 <sgothel> nope
20150811 14:33:26 <xranby> https://jogamp.org/chuck/job/jogl_armv7_test/ last run 2012
20150811 14:33:37 <sgothel> 'all software' .. b/c we don't have 'good enough' drivers yet :-/
20150811 14:35:14 <xranby> its easy to make a raspbian system using mesa3d software rendering
20150811 14:35:21 <xranby> simply remove the /opt/vc filder
20150811 14:35:24 <xranby> folder
20150811 14:35:46 <xranby> and it will pick up the mesa3d drivers that are installed by default
20150811 14:51:44 <gouessej> sgothel: Why did you use a big sdhc card instead of a micro sdhc card with the Raspberry Pi 2?
20150811 14:52:46 <sgothel> duh .. raspi2 uses micro? then - my fault, sorry - sent the wrong one
20150811 14:53:37 <sgothel> at least the power adapter I checked properly (volts and ampers) :)
20150811 14:54:43 <sgothel> mac mini .. password .. well .. you can change that: 'julien' AFAIK, same as user name :)
20150811 14:54:59 <sgothel> w/ root access (sudo)
20150811 14:55:35 <sgothel> wonder when you would ask for that one :)
20150811 14:55:54 * gouessej (5ee4b442@anon) Quit (Ping timeout: 246 seconds)
20150811 14:56:58 * packet0 (~packet0@anon) Quit (Remote host closed the connection)
20150811 14:57:23 * packet0 (~packet0@anon) has joined #jogamp
20150811 14:57:41 <xranby> interesting: https://jogamp.org/bugzilla/show_bug.cgi?id=1192#c1 someone have any idea why final Map<?,?> _desktophints = (Map<?,?>)(Toolkit.getDefaultToolkit().getDesktopProperty("awt.font.desktophints")); can generate an ArrayIndexOutOfBoundsException ?
20150811 14:59:32 <sgothel> the problem w/ all those tests on arm here is: do you use a proven working software environment?
20150811 14:59:52 <sgothel> i.e. all those bug reports - where you don't know is it the glamour or gl/blob driver .. etc etc
20150811 14:59:54 <sgothel> :-/
20150811 15:00:12 <sgothel> well .. that message seems to be an AWT/native-X11 issue
20150811 15:00:15 <xranby> this time it is the "regular" raspbian system with the broadcom blob
20150811 15:00:39 <xranby> thus it is using the X11UnderlayTracker
20150811 15:00:56 <sgothel> please STATE that always on those reports .. i.e. platform+software, since it gets quite confusing :)
20150811 15:01:12 <xranby> i thought i did it :/ OS: Rasbian with /opt/vc GPU driver
20150811 15:01:29 <sgothel> so probably the BCM blob may not interfere .. but who knows?
20150811 15:01:44 <sgothel> framebuffer X11 and some WM (which one?)
20150811 15:01:57 <sgothel> best to make a few lines stating all those things explicitly
20150811 15:02:34 <xranby> WM = openbox
20150811 15:02:39 <sgothel> system: raspi2, cpu: armhf, xserver: framebuffer, WM:kde, opengl:BCM-blob
20150811 15:02:48 <sgothel> ^^ for example
20150811 15:03:15 <sgothel> java.lang.ArrayIndexOutOfBoundsException: 692 at sun.awt.XSettings$Update.getCARD8(XSettings.java:140) at sun.awt.XSettings$Update.updateOne(XSettings.java:249)
20150811 15:03:31 <sgothel> so they expect an answer to have some fields in an array - which it does not
20150811 15:03:42 <sgothel> answer from the xserver probably
20150811 15:03:53 * gouessej (5ee4b442@anon) has joined #jogamp
20150811 15:04:12 <sgothel> @Julien: macmini: pwd julien or julien0 .. hope it works :)
20150811 15:04:15 <xranby> ok let me dig up the exact xserver used
20150811 15:05:03 <sgothel> i.e. all those reports seem to be related to arm xserver and drivers - we need a new category :)
20150811 15:05:21 <sgothel> i.e. not our turf .. and poor me is getting a quite headache :)
20150811 15:05:33 <gouessej> sgothel: I still have to test my game under OS X even though I dislike this operating system but as the backward compatible context doesn't work, it won't run it anyway
20150811 15:06:35 <xranby> sgothel: i am trying to tackle them one by one, sorry for giving you an headache
20150811 15:06:41 <sgothel> yeah .. me dislike it also .. but since you accepted the device, well. supporting OSX and Windows is 'just' part of the JogAmp story .. well :)
20150811 15:07:02 <sgothel> @xerxes: no no .. just being emotionally honest :)
20150811 15:07:16 <sgothel> i.e. when those reports come in .. I jump on it .. and WTF? .. :)
20150811 15:07:30 <sgothel> i.e. the word CRASH always gets my attention ofc :)
20150811 15:08:00 <sgothel> but .. since I know you are on it - all is sailing smoothly - THANK YOU for testing those configs
20150811 15:09:37 <sgothel> @Xerxes: methodID=0x1718a58 <- so we might need to DBG_PRINT all method-ids ?
20150811 15:10:08 <xranby> hehhe.. i can tell you th XORG driver is no the one you expect.. the raspbian people are using
20150811 15:10:35 <xranby> /usr/lib/xorg/modules/libfbdevhw.so
20150811 15:10:48 <xranby> this module is called FBTURBO(0)
20150811 15:11:02 <xranby> FBTURBO(0): hardware: BCM2708 FB (video memory: 750kB)
20150811 15:11:48 <xranby> actually designed to be used with the "sunxi" Allwinner A10/A13 FBDEV
20150811 15:11:52 <sgothel> getDesktopProperty("awt.font.desktophints") <- if that fails .. that means AWT hasn't [yet] initialized the properties _or_ AWT is buggy
20150811 15:12:09 <xranby> but hey.. i guess they found it faster to use compared to the regular "framebuffer" driver
20150811 15:12:29 <sgothel> hmm
20150811 15:12:51 <sgothel> probably doesn't matter regarding this property ..
20150811 15:13:11 <sgothel> awt.font.desktophints is being documented to be available on AWT
20150811 15:13:29 <sgothel> but we can try to catch the exception .. for not so compatible AWT impl.
20150811 15:13:52 <xranby> the exception did not stop execution
20150811 15:14:03 <sgothel> i.e. simply add a catch clause .. ah .. good
20150811 15:14:34 <xranby> i was mostly concerned if we have already corrupted the heap somehow
20150811 15:14:49 <xranby> but i cant see how we can do that using this oneliner
20150811 15:15:18 <xranby> the reason i say we is because https://jogamp.org/bugzilla/show_bug.cgi?id=1166 had corrupted the heap
20150811 15:15:39 <xranby> anyway.. i will do the argument check round
20150811 15:15:49 <xranby> and file some patches to correct the comments
20150811 15:16:18 <xranby> as a way to make sure that https://jogamp.org/bugzilla/show_bug.cgi?id=1192 is not on our turf
20150811 15:18:43 <xranby> (17:12:16) sgothel: @Xerxes: methodID=0x1718a58 <- so we might need to DBG_PRINT all method-ids ? <- i can do that by compiling a debug version of jamvm
20150811 15:18:50 <xranby> i think
20150811 15:18:55 <xranby> good idea
20150811 15:20:23 <sgothel> added replies :)
20150811 15:20:55 <sgothel> yes you are right - it pays off to scrutinize those bugs .. i.e. not easily waving the 'not our turf' ..
20150811 15:21:15 <sgothel> hence big KUDOS for Bug 1166
20150811 15:27:03 <sgothel> assuming JamVM works correct, maybe its code is more sensitive to such var-arg 'mistakes' .. hmm
20150811 15:27:45 <sgothel> Bug 1166 was long for int .. lets see sizeof(..) diffs for i586/arm7
20150811 15:28:40 <sgothel> http://jogamp.org/deployment/jogamp-next/javadoc/gluegen/javadoc/overview-summary.html
20150811 15:29:12 <sgothel> so at least on Windows, 'long' is 32 bits wide, on Unix 64 bits wide
20150811 15:29:34 <sgothel> hence same Bug 1166 should have been caught on linux-32bit .. but it didn't .. hmm
20150811 15:31:02 <sgothel> ^^ squash that - its all 'long == 32bit' natively .. wrong column :)
20150811 15:31:31 <sgothel> ^^ on 32bit arch
20150811 15:34:57 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=1192#c5
20150811 15:55:33 * gouessej (5ee4b442@anon) Quit (Quit: Page closed)
20150811 16:00:18 * eclesia (~husky@anon) has left #jogamp
20150811 16:02:55 * elect_ (~elect@anon) Quit (Ping timeout: 256 seconds)
20150811 16:04:37 * i-make-robots (~i-make-ro@anon) has joined #jogamp
20150811 16:04:40 <i-make-robots> hi!
20150811 16:04:53 <xranby> i-make-robots: hi
20150811 16:05:05 <i-make-robots> how hard is it to run jogamp on a pi? I've got a mature project that people want to see ported.
20150811 16:05:17 <i-make-robots> https://github.com/MarginallyClever/Makelangelo/
20150811 16:06:23 <i-make-robots> i'm already using jogl for the 3d rendering on windows and OSX.
20150811 16:06:51 <sgothel> sounds fun .. great stuff, consider adding your project to our JOGL website?
20150811 16:07:01 <sgothel> you use AWT?
20150811 16:07:04 <i-make-robots> i'd be happy to.
20150811 16:07:10 <i-make-robots> erm, lemme check..
20150811 16:07:15 <i-make-robots> how do i confirm re: awt?
20150811 16:07:19 <xranby> i-make-robots: if you use NEWT and the GL2ES2 profile then it shall work out of the box
20150811 16:07:25 <xranby> for example
20150811 16:07:32 <xranby> if you use a GLWindow == NEWT
20150811 16:07:44 <xranby> and use setVisible()
20150811 16:07:53 <i-make-robots> loading my eclipse env to check, one moment...
20150811 16:07:54 <sgothel> then please provide a nice screenshot image (160x..) ... err and some text ad shown in http://jogamp.org/jogl/www/
20150811 16:08:18 <sgothel> ^^ 160x160 PNG
20150811 16:08:20 <xranby> loosk good :) http://www.makelangelo.com/
20150811 16:08:40 <i-make-robots> here's a screenshot: http://www.makelangelo.com/download/
20150811 16:09:49 <sgothel> .. and if you can make a short description, best similar to what we have ..
20150811 16:11:19 <sgothel> .. where is the '3d' in the picture?
20150811 16:11:26 <xranby> i-make-robots: the "default" OpenGL ES driver using teh broadcom binary blob on the raspberry pi works a bit different compared to other arm linux platforms, this driver can not output its rendering to a X11 window
20150811 16:11:46 <sgothel> the UI on the side is probably swing?
20150811 16:11:57 <i-make-robots> I just jogl for all the graphics rendering in the system. it's easier for me than most 2d drawing platforms, I was raised on opengl.
20150811 16:12:07 <i-make-robots> I have other projects that also just jogl with real 3d... for example,
20150811 16:12:10 <xranby> i-make-robots: there exist one new free software driver for the raspberry pi that is buuilt on top of mesa3d that implements full OpenGL 2 with some hardware acceleration
20150811 16:12:36 <xranby> i-make-robots: and there exist the mesa3d driver that implements OpenGL 2 using software rendering
20150811 16:12:59 <xranby> i belive that your project will work out of the box using any of the two mesa3d drivers
20150811 16:13:01 <sgothel> but GLES2 would be best, aka core profile w/o fixed function etc
20150811 16:13:28 <i-make-robots> https://github.com/MarginallyClever/Hexapod is an open source crab robot. i simulated the entire robot before I built it.
20150811 16:13:46 <sgothel> .. and then .. Xerxes is just about to finish X11 underlay .. and stuff to integrate hw-accel on raspi .. he knows best :)
20150811 16:14:13 <i-make-robots> https://github.com/MarginallyClever/Arm3 a 3 axis robot arm, also simulated. I copy & paste most of the code.
20150811 16:14:24 <xranby> wow, soo many cool robots! :D
20150811 16:14:25 <sgothel> holy moly :)
20150811 16:14:26 <i-make-robots> I used to try and make an engine to run all my robots, but that's proven way too hard.
20150811 16:14:27 <xranby> and simulators
20150811 16:14:42 <i-make-robots> xranby - thank you, that's very kind of you.
20150811 16:15:12 <i-make-robots> i really appreciate the hard work you guys & gals have put into jogamp. it's a wonderful tool.
20150811 16:15:50 <sgothel> looks surely like something to seriously play with - KUDOS
20150811 16:16:02 <i-make-robots> right now i'm building a simulator for this: https://instagram.com/p/6JII3nIfDz/?taken-by=imakerobots
20150811 16:16:04 <sgothel> yeah .. a robot framework ..
20150811 16:16:07 <i-make-robots> so i can program it offline.
20150811 16:16:20 <sgothel> I need a second life :)
20150811 16:16:26 <i-make-robots> i get lost in all kinds of engine design BS.
20150811 16:17:15 <i-make-robots> "do i make a world class, and then a factory to build different object types, one of which is robot? oh I guess I need a robot factory that returns the base class but makes each specific robot type. what with the UI look like? Argh."
20150811 16:17:18 <sgothel> please write 2 excerpts / descriptions w/ your choice of screenshots and write to http://jogamp.org/wiki/index.php/Maintainer_and_Contacts#Sven_Gothel (email)
20150811 16:17:32 <sgothel> then I like to add both projects (robots .. and 3d printer) ..
20150811 16:18:39 * monsieur_max (~maxime@anon) has joined #jogamp
20150811 16:18:40 <i-make-robots> lemme see what I can find. These will probably be a rush job...
20150811 16:18:44 <sgothel> then a fitness test (genetic algos) .. testing desired caps .. and resulting in the 3d printing plans :)
20150811 16:19:03 <xranby> you have a good laser cutter
20150811 16:19:35 <i-make-robots> it's a 60w. it's ok.
20150811 16:19:51 <i-make-robots> nobody else touches it so there's little chance someone borks it.
20150811 16:20:13 <i-make-robots> if you ever buy a laser cutter, get at least 80w. I struggle to cut 1/4".
20150811 16:20:32 <i-make-robots> so when I have 1/4" parts, I usually cut two 1/8" and glue them together.
20150811 16:20:34 <i-make-robots> not idea.
20150811 16:20:35 <i-make-robots> l
20150811 16:20:39 <sgothel> sounds like cool stuff I can do w/ my son (when he gets a bit older) .. hehe
20150811 16:22:52 <sgothel> btw .. I like your picture in 'About' very much .. puts it all together - does it look good in 160p?
20150811 16:24:00 <sgothel> you guys (robots.. and xerxes) surely have a lot to share I guess :)
20150811 16:24:19 <xranby> with I had time to realize all dreams :/
20150811 16:24:42 <xranby> aha.. tried to build the project.. it built
20150811 16:24:51 <xranby> but then i ran into the maven package bug
20150811 16:27:58 * elect (~elect@anon) has joined #jogamp
20150811 16:27:59 <xranby> https://jogamp.org/bugzilla/show_bug.cgi?id=1172
20150811 16:28:06 <i-make-robots> sent you three descriptions
20150811 16:28:35 <sgothel> thank you
20150811 16:29:32 <xranby> xranby@Xerxes-ESPRIMO:~/Makelangelo/java$ java -jar /home/xranby/Makelangelo/java/target/Makelangelo-v7.1.3-jar-with-dependencies.jar
20150811 16:29:32 <xranby> Exception in thread "AWT-EventQueue-0" java.lang.UnsatisfiedLinkError: /tmp/jogamp_0000/file_cache/jln8751896218628510696/jln7104418771196207347/libgluegen-rt.so: /tmp/jogamp_0000/file_cache/jln8751896218628510696/jln7104418771196207347/libgluegen-rt.so: kan inte öppna delad objektfil: Filen eller katalogen finns inte (Possible cause: can't load AARCH64-bit .so on a AMD 64-bit platform)
20150811 16:29:34 <sgothel> I was more in for a general description -> text .. similar length w/ URLs embedded as the other links .. if you don't mind .. :)
20150811 16:30:20 <xranby> the bove is caused by maven smasing all jogamp native jars on top of each other
20150811 16:30:25 <xranby> hmm
20150811 16:30:27 <xranby> above
20150811 16:30:43 <sgothel> yeah .. maybe @Mark, MARK, mark, .. can collab here?
20150811 16:30:47 <i-make-robots> Does anyone read those walls of text?
20150811 16:30:53 <xranby> my solution https://jogamp.org/bugzilla/show_bug.cgi?id=1172#c5
20150811 16:30:53 <sgothel> yes
20150811 16:30:57 <xranby> suggestion
20150811 16:31:02 <sgothel> @robot: yes ofc :)
20150811 16:31:16 <sgothel> the jogl main page has the highest hit number
20150811 16:31:40 <sgothel> *thousands*
20150811 16:32:31 <i-make-robots> xranby - I have been having major headaches with native jars. I still have to use a start.bat for windows users that detects the OS version. I just want to 2x click the jar and go! Minecraft does it right.
20150811 16:33:00 <xranby> i-make-robots: all your issues are caused by the maven maven-assembly-plugin
20150811 16:33:29 <sgothel> @robot: Unique Visitors: 161638 so far this year
20150811 16:33:55 <xranby> i-make-robots: in the next jogamp 2.3.2 release we will provile a "fat" jar
20150811 16:34:08 <xranby> that can be used in combination with the maven-assembly-plugin
20150811 16:34:32 <sgothel> 2014: 566928 unique visitors - hmm, I guess we are loosing :)
20150811 16:34:45 <xranby> all content of the jogamp fat jar can be unpacked into a new jar without any natives gets overwritten
20150811 16:34:56 <sgothel> yes
20150811 16:35:12 <sgothel> @xerxes: yes - I understood .. hence remodelling the maven stuff might be of interest
20150811 16:35:30 <sgothel> hence .. asking for @Mark, MARK, mark .. :)
20150811 16:36:01 <xranby> is tehere any technical issues to model the contents of the current multiple native jars to use the same in-jar fodler structure as sugegsted in comment https://jogamp.org/bugzilla/show_bug.cgi?id=1172#c5 ?
20150811 16:36:03 <sgothel> @rmk0 @rmk .. probably in deep sleep / holidays
20150811 16:37:07 <sgothel> I never really checked the maven stuff thoroughly .. was happy Mark did it all
20150811 16:37:26 <sgothel> i.e. maven-assembly-plugin: I dunno
20150811 16:38:05 <sgothel> offer a patch .. to the magic maven scripts in jogamp-scripting .. test .. etc etc
20150811 16:38:26 <xranby> ... yes good idea
20150811 16:38:31 <xranby> one more work item on my turf
20150811 16:38:56 <sgothel> hehe .. assign it to yourself then as well .. :)
20150811 16:39:23 <sgothel> I quite don't understand why the wrong arch is being picked .. but well .. what do I know about maven? not much ..
20150811 16:39:33 <sgothel> and I don't want to know :)
20150811 16:39:47 <sgothel> ok back to windows .. and bugs ..
20150811 16:40:40 <xranby> i understand why it breaks, i take it
20150811 16:40:49 <sgothel> sweet!
20150811 16:45:05 <i-make-robots> sgothel - better?
20150811 16:49:20 <i-make-robots> Jogamp a big project that takes a lot of hours. Surely you're getting compensated for your work. How do you make money with jogamp?
20150811 16:49:44 <i-make-robots> I struggle to make money with open source projects. People say I'm stupid for giving my work away.
20150811 16:50:30 <sgothel> wife holds company, by bugging 'users' .. i.e. in case they do it commercial .. and a company is visible, well, I remind them that commercial support is available.
20150811 16:50:49 <sgothel> http://jogamp.org/wiki/index.php/Maintainer_and_Contacts#Sven_Gothel <- as mentioned there .. low profile though :)
20150811 16:51:00 <sgothel> avoiding the big GIVE US YOUR MONEY button :)
20150811 16:51:46 <sgothel> .. and we are indeed grateful for CCT, a company supporting the project for 5 years already (they do the C3D application)
20150811 16:51:52 <sgothel> as listed on JOGL
20150811 16:52:30 <sgothel> however, indeed it is quite rare that companies decide to purchase services .. well
20150811 16:53:04 <sgothel> since the 'only' use case is attention .. usually, i.e. time to bug-fix and 'babysitting' .. you know what I mean
20150811 16:53:36 <sgothel> so we avoid this for normal free communication, here the users need to do their work as well ofc, as in all free projects
20150811 16:53:55 <sgothel> we thought about shipping something to sell .. but we are too lazy
20150811 16:54:16 <sgothel> how many days/hours do you have to use for shipping stuff? I assume quite some work .. hmm
20150811 16:55:48 <sgothel> hmm .. I read, so .. it must be a business case for companies IMHO ..
20150811 16:56:04 <sgothel> in your case (hobby/consumer) .. maybe hook up w/ some serious toy stores?
20150811 16:56:21 <sgothel> i.e. where they do RC cars/planes and stuff?
20150811 16:56:37 <sgothel> AFAIK .. families are keen to spend quite a buck ..
20150811 16:57:03 <sgothel> the other scenario is ofc professional .. for custom production in low amounts .. like what Xerxes does as well
20150811 16:57:27 <sgothel> demo devices .. maybe used later for bulk production
20150811 16:57:47 <sgothel> indeed .. I can imagine its hard to be found ..
20150811 16:57:57 <sgothel> in our case .. they come to us, usually b/c we have bugs :)
20150811 16:58:04 <sgothel> or features they need
20150811 16:58:34 <sgothel> but .. only 1-2 a year .. just surviving well .. so to speak
20150811 16:58:47 <sgothel> hope that helps in elaborating ..
20150811 16:59:52 <sgothel> pro CAD/CAM .. the latter w/ 3d printers and stuff .. there should be a market though .. ofc
20150811 17:00:21 <sgothel> IMHO .. open source doesn't hurt, contrary - it helps promoting. companies need service.
20150811 17:00:46 <xranby> i think of it as free/libre software
20150811 17:01:34 <sgothel> and if ethics are good, they will even pay for what is there - or they need support and sort of warranty for their big projects
20150811 17:01:35 <xranby> and if you run long time research and development you need the freedom to run it for any purpose and re-design it for your needs
20150811 17:01:42 <sgothel> the latter usually is the case, always
20150811 17:01:53 <sgothel> yes
20150811 17:02:14 <sgothel> companies may not use you [free] software, if they cannot purchase services!
20150811 17:02:31 <sgothel> it gives them peace of mind ofc ..
20150811 17:02:33 <sgothel> laters ..
20150811 17:02:41 <xranby> laters!
20150811 17:03:52 <i-make-robots> I'm glad you have a supportive family. I'm on my own, the gf just thinks what I do is wierd.
20150811 17:04:10 <i-make-robots> sgothel is that updated description I sent you good?
20150811 17:06:34 <xranby> i-make-robots: stay around this irc channel, we will repond eventually :)
20150811 17:07:43 <xranby> i-make-robots: if you have a raspberry pi at your location.. then you can try a fun demo
20150811 17:08:33 <xranby> wget http://labb.zafena.se/jogamp/vc4/libgdx-demo-pax-britannica-1.6.3-armhf.jar
20150811 17:08:40 <xranby> java -jar libgdx-demo-pax-britannica-1.6.3-armhf.jar
20150811 17:09:12 <xranby> it shall work from both xorg and from console
20150811 17:10:17 <xranby> let me think
20150811 17:10:41 <i-make-robots> I have many Pis. none connected.
20150811 17:10:51 <xranby> there was something with libopenal and the pi's soundcard
20150811 17:11:51 <xranby> If OpenAL fail to initialize
20150811 17:11:51 <xranby> "AL lib: (EE) ALCplaybackOSS_open: Could not open /dev/dsp: No such file or directory"
20150811 17:11:51 <xranby> this originate from that the libopenal sound library, bundled with JOAL, failed to find your raspberry pi soundcard
20150811 17:11:51 <xranby> then you have two options:
20150811 17:11:51 <xranby> A) load the linux kernel driver
20150811 17:11:52 <xranby> sudo modprobe snd_pcm_oss
20150811 17:11:52 <xranby> B) disable the OpenAL soft oss driver:
20150811 17:11:53 <xranby> edit /etc/openal/alsoft.conf
20150811 17:11:53 <xranby> and make sure that the drivers list contains -oss (it disables the OSS backend that tries to use the non existant dsp file)
20150811 17:12:46 <xranby> hard to make things work out of the box.. have to think on all details :/
20150811 17:13:12 <xranby> can probably be fixed by improving openal-soft
20150811 17:50:44 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150811 17:58:10 * monsieur_max (~maxime@anon) has joined #jogamp
20150811 18:00:13 * monsieur_max (~maxime@anon) Quit (Client Quit)
20150811 18:12:07 * packet0 (~packet0@anon) Quit ()
20150811 18:12:35 * gouessej (52794dec@anon) has joined #jogamp
20150811 18:12:56 <gouessej> i-make-robots: What's wrong with the native libraries?
20150811 18:14:00 <gouessej> sgothel: I can ship some bread :p
20150811 18:14:36 <sgothel> @robot: desc is fine - thx
20150811 18:14:56 <sgothel> @julien: french croissants .. hmm :)
20150811 18:15:42 <sgothel> but they must be fresh .. dripping of butter .. yummy
20150811 18:15:56 <sgothel> guess we have to visit you then .. to get those :)
20150811 18:16:15 <gouessej> sgothel: French croissants are better than American croissants: https://gouessej.files.wordpress.com/2012/08/photo20120806_105232.jpg
20150811 18:16:26 <sgothel> haha .. I remember :)
20150811 18:17:21 <sgothel> I lived in Hildesheim .. small city .. and there was a bakery .. starting doing the doe around 1am .. or earlier, picking up the croissants around 5:30am .. that was the real deal
20150811 18:18:05 <gouessej> I make my own bread and my own "fallue" (Normandy style brioche)
20150811 18:19:00 <gouessej> i-make-robots; Actually, just double-click on a JAR has some severe limitations
20150811 18:19:31 <gouessej> i-make-robots: You can't pass any option to the JVM or you have to create a second JVM :s
20150811 18:20:17 <gouessej> i-make-robots: I use my own packaging tool, it produces a ZIP with a BAT file for Windows. In the future, it will rather create a MSI file for Windows
20150811 18:20:41 <gouessej> i-make-robots: It's a lot better than what is done for Minecraft
20150811 18:21:07 <gouessej> i-make-robots: You can easily run out of native or Java memory when using the default values
20150811 18:22:13 <gouessej> sgothel: I can postpone the start of the program so that I obtain a very fresh brioche when I wake up :)
20150811 19:05:45 <sgothel> :)
20150811 19:11:03 * elect (~elect@anon) Quit (Ping timeout: 256 seconds)
20150811 19:13:29 <sgothel> https://www.marginallyclever.com/blog/2014/03/automate-your-phone-app-testing-with-a-delta-robot/
20150811 20:29:55 * gouessej (52794dec@anon) Quit (Quit: Page closed)
20150811 20:33:21 <i-make-robots> you like that one? :)
20150811 20:34:08 <sgothel> yes.. especially assuming the accuracy, otherwise .. well, poor devices being 'touched' :)
20150811 20:34:35 <sgothel> the video was impressive .. you might want to show videos for all products?
20150811 20:34:40 <i-make-robots> i'm working on adding encoders so all my robots will be compliant.
20150811 20:34:58 <i-make-robots> Agreed. I don't have enough video skills.
20150811 20:35:11 <i-make-robots> makelangelo needs a video, but the software is always changing.
20150811 20:35:54 <sgothel> prospective users need visuals .. to see what it really means - an inspiration
20150811 20:36:25 <i-make-robots> yeah. I've got a video guy helping me now. I'll put him on that.
20150811 20:36:31 <sgothel> makelangelo: can it do layer by layer to test 3d printing (fast) ?
20150811 20:36:56 <sgothel> I was curious .. since you mentioned '3d' ..
20150811 20:37:16 <i-make-robots> would something like this do? https://www.youtube.com/watch?v=dHGkYwhfWgI
20150811 20:38:03 <sgothel> quite professional cut ..
20150811 20:38:06 <sgothel> sweet
20150811 20:38:09 <sgothel> slow-mo ..
20150811 20:38:24 <sgothel> I was also looking into the LED strips .. a month ago :)
20150811 20:39:04 <sgothel> sweet
20150811 20:40:18 <sgothel> now showing how you do custom robots for all kinds of things .. well, there is an industry ofc. maybe even architecture and building construction to demo a process? hmm
20150811 20:44:56 <sgothel> https://www.youtube.com/watch?v=2TiCZmJC9aU .. also ..
20150811 20:45:15 <sgothel> he points to one use-case business case: SCALE of the picture
20150811 20:45:22 <sgothel> now .. how about BUILDINGS?
20150811 20:45:55 <sgothel> paintings on interior walls .. and even exterior walls?
20150811 20:46:44 <sgothel> drone + spraying ? :)
20150811 20:46:58 <bruce-> there are some projects with paint guns
20150811 20:49:23 <bruce-> http://sonicedevelopment.com/thewild/
20150811 20:49:25 <sgothel> if the pen could compensate the inaccuracy of the wall .. well, the inexpensive 2 motor thingy may be just fine?
20150811 20:49:39 <sgothel> ^^ depth .. and 'obstacles' .. hmm
20150811 20:50:23 <sgothel> ^^ hehe .. that might be a bit too fuzzy :)
20150811 20:50:24 <bruce-> Jurg Lehni's Hector was also used at quite large scales, but is also less precise
20150811 20:50:35 <sgothel> ah
20150811 20:50:55 <sgothel> 'Hector' is also 'makelangelo' source of inspiration .. right
20150811 20:51:08 <sgothel> nice nice ..
20150811 20:52:01 <bruce-> yeah, we actually had a piece in the same show as Hector, the curators told me it was an 'interesting' piece to set up :)
20150811 20:54:02 <i-make-robots> there are people using it for interior wall painting already. Had one guy approach me about using it for safer exterior building inspection.
20150811 20:55:27 <sgothel> .. combined w/ a laser scanner .. hmm
20150811 20:55:29 <i-make-robots> these are great suggestions. it's so nice to get out of my usual circle and get some feedback from people who make things.
20150811 20:55:48 <i-make-robots> nah, they just want a webcam and some position information.
20150811 20:55:59 <i-make-robots> so they can review at a later date.
20150811 20:57:18 <sgothel> when Rami returns .. I will give him a little summary of these ideas. great stuff
20150811 20:57:55 <sgothel> (they were testing laser scanner etc .. for a similar validation process .. construction and the like)
20150811 20:59:05 <sgothel> now, if one could create a miniature process of the real construction process .. in 'no time' .. to validate the construction process itself .. physically .. hmm :)
20150811 20:59:32 <sgothel> i.e. what architects do in old days .. building those little models for 'show'
20150811 20:59:51 <sgothel> but here .. the construction process itself would be validated
20150812 01:38:13 * packet0 (~packet0@anon) has joined #jogamp
20150812 05:05:29 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150812050529.html