JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20131023 14:10:49 (UTC)

20131023 14:10:49 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20131023120140.html
20131023 14:10:49 -NickServ- This nickname is registered. Please choose a different nickname, or identify via /msg NickServ identify <password>.
20131023 14:10:50 * jogamp (~jogamp@anon) has joined #jogamp
20131023 14:10:50 * Topic is 'http://jogamp.org | Hacking 3D Graphics, Multimedia and Processing across Devices'
20131023 14:10:50 * Set by rmk0 on 20130116 23:58:04
20131023 14:11:00 -NickServ- You are now identified for jogamp.
20131023 14:39:41 <eclesia> pff. not fast enough this computer ...
20131023 14:40:01 <monsieur_max> eclesia: time to buy a $3.000 mac i think
20131023 14:40:27 <eclesia> I don't think any mac would be faster then the one I have here ^^
20131023 14:42:06 <monsieur_max> ok so next step is to build your own own cloud computing system :)
20131023 14:43:20 <eclesia> just joking , takes 5minutes to unpack/read world openstreet map datas (~30Gb)
20131023 14:43:39 <eclesia> I'm so used to this computer, waiting a few minutes seems long
20131023 14:45:41 <monsieur_max> what is the project you're working on ?
20131023 14:46:45 <eclesia> some kind of navy
20131023 14:46:52 <eclesia> sea maps
20131023 14:47:31 <monsieur_max> using jogl ?
20131023 14:47:52 <eclesia> yes
20131023 14:48:00 <monsieur_max> neat :)
20131023 14:48:00 <eclesia> but not my lib :(
20131023 14:51:58 <sgothel> @Brice: Bug 866 pull req avail ? Otherwise I patch and attribute you (thx for finding and fixing)
20131023 14:52:08 <sgothel> btw - https://jogamp.org/wiki/index.php/SW_Tracking_Report_Objectives_for_the_release_2.1.2
20131023 14:55:52 <sgothel> @Brice: Me pushing it .. attributing you ..
20131023 15:09:29 <ebrayet> @sven Thanks
20131023 15:27:05 <masterzen> sgothel: thanks!
20131023 15:34:18 * hharrison (~chatzilla@anon) has joined #jogamp
20131023 15:55:53 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131023 15:57:53 <sgothel> https://jogamp.org/wiki/index.php/SW_Tracking_Report_Objectives_for_the_release_2.1.2 <- fixing some OSX issues .. then looking @Harvey 's changes
20131023 15:58:22 <sgothel> @Harvey - if your changes cover more than beauty .. maybe a bug entry would be nice for reference
20131023 16:01:10 <hharrison> Hmm, the only thing I've really done is add a fast-path out to the debug piplines, so they don't build a large error string, even in the success cases
20131023 16:01:31 <hharrison> but sure, I will open a bug report
20131023 16:02:59 <hharrison> Just doing a bit more profiling at the moment.....GL4bcImpl has a ton of interned strings for building error messages we could save a lot of size on
20131023 16:03:41 <hharrison> anyways...openign bug
20131023 16:08:31 * eclesia (jsorel@anon) has left #jogamp
20131023 16:10:21 <hharrison> https://jogamp.org/bugzilla/show_bug.cgi?id=868
20131023 16:26:30 <hharrison> Let me know if you want more detail
20131023 16:27:03 <sgothel> if you can remove those duplicates .. even better .. sure
20131023 16:27:07 <sgothel> (strings)
20131023 16:28:29 <sgothel> the Bug entry doesn't match the description
20131023 16:28:29 * ebrayet (~Adium@anon) Quit (Ping timeout: 272 seconds)
20131023 16:28:48 <sgothel> i.e. not about 'garbage' but detailed error report - which is good
20131023 16:29:00 <sgothel> or am i missing something ?
20131023 16:29:26 <sgothel> also: 'int err= ..' -> "final int err=.." allowing easier optimization
20131023 16:29:33 * monsieur_max (~maxime@anon) has joined #jogamp
20131023 16:31:00 <sgothel> also: to compare both code path .. the new code template in description shall be downstreamGL4bc.wrappedFunction(arg0,arg1,arg2...); .. as well .. (now I am getting annoying - SORRY :)
20131023 16:31:53 <sgothel> this ofc will explode the Debug* class quite a bit .. however, since it's pipelined - GREAT
20131023 16:32:26 <hharrison> Sorry, I don't follow, the emitted messages are exactly the same, we just only build them when there is an error
20131023 16:32:28 <sgothel> Bug 868 is added ..
20131023 16:32:42 <hharrison> Avoiding generating a pile of garbage-collected Strings
20131023 16:32:49 <sgothel> oh .. haven't compared in detail
20131023 16:33:02 <sgothel> ah ..
20131023 16:33:17 <sgothel> yup - get it .. missed the ole 'String txt=.." perfect
20131023 16:33:44 <hharrison> I tried to stage the changes in order, the commit messages hopefully tell a good tale
20131023 16:34:08 <hharrison> I
20131023 16:34:21 <sgothel> will look at it after fixing OSX .. sure do
20131023 16:34:38 <hharrison> 'll do the final int err thing, sure, the indentation also needs a fixup in the generator (pre-existing problem)
20131023 16:35:04 <sgothel> indentation .. well, I don't care :)
20131023 16:35:25 <sgothel> (for generated code .. ofc)
20131023 16:35:56 * monsieur_max (~maxime@anon) has left #jogamp
20131023 16:36:26 <hharrison> Yeah, last steps, while I'm touching this code, may as well do it while the details are fresh
20131023 16:37:10 <sgothel> where is it ?
20131023 16:37:19 <sgothel> ah .. jogl .. duh !
20131023 16:37:22 <hharrison> ;-)
20131023 17:15:20 * [Mike] (~Mike]@anon) has joined #jogamp
20131023 17:21:39 <hharrison> Just doing final testing of the indent/final stuff in case you haven't already pulled
20131023 17:33:59 <sgothel> the graph clone mods .. I assume you tested them .. :)
20131023 17:35:06 <sgothel> output.format(" @Override%n public %s %s(%s)%n",
20131023 17:35:06 <sgothel> + JavaType.createForClass(m.getReturnType()).getName(), ??
20131023 17:35:24 <sgothel> @Override%n ?? '%n' ?
20131023 17:35:41 <hharrison> oh...that
20131023 17:35:52 <hharrison> I'm outputting two lines in one call
20131023 17:36:07 <sgothel> %n == \n ?
20131023 17:36:10 <sgothel> oh
20131023 17:36:12 <hharrison> %n gets replaced with platform-specific line ending....same as what println("") would output
20131023 17:36:20 <sgothel> black magic :)
20131023 17:37:53 <hharrison> the graph clone stuff - I did test it...but the thing is, that method can _only_ throw excpetions as-weitten
20131023 17:38:00 <hharrison> It can never complete successfully
20131023 17:38:11 <sgothel> yup .. must have been a looong night :)
20131023 17:38:13 <hharrison> Which is why I felt comfortable killing it...it cannot ever work
20131023 17:38:41 <hharrison> So unless someone is relying on catching an InternalErrorException, nobody is relying on that code
20131023 17:38:58 <hharrison> Sorry, InternalError()
20131023 17:39:18 <sgothel> nobody clones OnlyShape I hope :)
20131023 17:39:48 <hharrison> I'm pretty sure they'd notice the Exception
20131023 17:40:10 <hharrison> I only touched the ones that I could prove were totally broken
20131023 17:40:30 <hharrison> There's another set of maybe broken I haven't looked at yet
20131023 17:50:03 <hharrison> crap, away from my ssh key, you can pull the indentation fix (1 more commit) from:
20131023 17:50:05 <hharrison> https://github.com/AusencoSimulation/jogl.git master
20131023 17:51:07 <hharrison> I just noticed the indentation code for the Trace pipelines never got implemented, bug has been there for a long time
20131023 17:51:38 <sgothel> beauty bugs have low prio since I was the only code dweller :)
20131023 17:52:15 <hharrison> This is functional...the trace pipelines try to format the traces by indenting the messages based on context, but that code appears to be broken
20131023 17:52:25 <sgothel> so you say (7fd9174afe73ae2a61af9a6d101afb7f21bb608b) identical duplicated strings gets optimized away ?
20131023 17:52:37 <sgothel> great
20131023 17:53:10 <hharrison> indent -= 2 on every glEnd...but no indent += 2 anywhere in site
20131023 17:53:10 <sgothel> oh .. :)
20131023 17:53:10 <sgothel> well ..
20131023 17:53:12 <sgothel> will merge you fixes .. sure
20131023 17:53:15 <hharrison> I'm guessing that should be += 2 on glBegin
20131023 17:57:49 * monsieur_max (~maxime@anon) has joined #jogamp
20131023 17:57:56 <hharrison> Yes
20131023 17:58:22 <hharrison> The identical string thing, it's required as part of the Java language spec
20131023 17:58:31 <sgothel> sweet
20131023 17:58:41 <hharrison> It gets resolved at class-load time
20131023 17:58:50 <hharrison> Which is why:
20131023 17:59:02 <hharrison> String s1 = "a string";
20131023 17:59:11 <hharrison> String s2 = "a string";
20131023 17:59:38 <hharrison> String s3 = new String("a string");
20131023 18:01:27 <hharrison> s1 == s2 => true
20131023 18:01:27 <sgothel> s1==s2, s1!=s3 .. nice
20131023 18:01:27 <hharrison> yep
20131023 18:01:27 <hharrison> even better...that works across class boundaries, if s1 and s2 are in different classes
20131023 18:01:27 <sgothel> so the GLProfile strings can also be compared w/ a local "GL3" .. hoho
20131023 18:01:27 <hharrison> So...in the code gen, that's why the gl method name is an argument to the format, not part of the format string, I plan to pull this into the implementation class itself and share all these strings
20131023 18:01:27 <sgothel> yup .. great optimization
20131023 18:01:27 <hharrison> :-)
20131023 18:01:27 <hharrison> It's the C-programmer in me...I can't help it
20131023 18:02:10 <hharrison> ALso, all those Strings get laoded to perm-gen and the garbage collector no longer even considers them until a class gets unloaded
20131023 18:03:27 <hharrison> (can you tell I've been profiling this week)
20131023 18:04:28 <hharrison> trivia question: what can be paired with glEnd/glEndList other than glBegin?
20131023 18:05:24 <hharrison> just taking a crack at fixing the trace pipeline formattig
20131023 18:08:09 <hharrison> sgothel: need a roadmap...what generates the implementation classes like GL4bcImpl?
20131023 18:08:48 <sgothel> glEnd/glEndList other than glBegin: thats it AFAIK
20131023 18:09:08 <hharrison> OK, patch in progress
20131023 18:13:10 <sgothel> one of the '*Emission' classes .. overriden in an JOGL pecialization
20131023 18:13:15 <sgothel> *specialization
20131023 18:13:31 <sgothel> JavaEmittor .. similar
20131023 18:14:17 <hharrison> Sure, I'll figure it out, was hoping you had it off the top of your head
20131023 18:15:05 <sgothel> sorry .. a bit distracted .. friend's computer maintenance .. fixing OSX .. review your code :)
20131023 18:15:21 <sgothel> will help later ..
20131023 18:15:46 <hharrison> Don't worry about it, it might be better for me to figure it out by myself, will help me get more familiar with the codebase
20131023 18:15:50 <sgothel> GLEmitter
20131023 19:53:00 * ebrayet (~Adium@anon) has joined #jogamp
20131023 20:25:04 <hharrison> I think this rabbit-hole goes a bit too deep for me today, ended up in gluegen methodEmitters that I'd likely have to change
20131023 20:26:15 <sgothel> me grokking new kCGLOGLPVersion_GL4_Core ..
20131023 20:27:04 <sgothel> returns 4.1 APPLE SOFT, but req. GL3_Core .. gives me 4.1 NV HW :)
20131023 20:27:20 <sgothel> looks like a bug in OSX
20131023 20:34:35 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20131023 20:46:12 * xranby-64 (~familjen@anon) has joined #jogamp
20131023 21:32:21 <xranby-64> http://shadow-file.blogspot.ch/2013/10/complete-persistent-compromise-of.html - Complete, Persistent Compromise of Netgear Wireless Routers
20131023 21:35:26 <monsieur_max> another one ... great
20131023 21:41:15 * Schrostfutz (~who@anon) has left #jogamp
20131023 21:44:41 <xranby-64> monsieur_max: 攻撃力 blog explained it quite nicely: "Congratulations on your device purchase! Would you like: A. A US Backdoor. B. A Chinese Backdoor. C. A Vendor Backdoor."
20131023 21:45:54 <hharrison> D. All of the above
20131023 21:46:00 <xranby-64> 攻撃力 blog == kougekiryoku.blogspot.se https://twitter.com/_wirepair
20131023 21:49:09 <xranby-64> hharrison: yes.. home fabricated chips will take a while... before its of any use
20131023 21:53:03 <monsieur_max> :)
20131023 21:53:33 * monsieur_max (~maxime@anon) has left #jogamp
20131023 22:34:05 <hharrison> sgothel: post in the Java3d forum has a good looking test case attached that may allow debugging the OSX applet canvas problems....in your copius spare time, of course
20131023 22:35:08 <hharrison> Has anyone else had to work around this:
20131023 22:35:10 <hharrison> http://stackoverflow.com/questions/17223304/appcontext-is-null-from-rmi-thread-with-java-7-update-25
20131023 22:36:02 <hharrison> Trying to decide the best way to add a quirk for that java7 u25 on OSX problem in Java3d...or if I should just tell people to avoid that version
20131023 22:54:29 * xranby-64 (~familjen@anon) Quit (Ping timeout: 248 seconds)
20131023 23:19:49 <sgothel> @Harvey: avoid - since 7u45 is a security update
20131023 23:20:36 <sgothel> @Harvey: pls attach/link test case to [[re]open] bug report .. otherwise .. I will loose it probably
20131023 23:20:46 <sgothel> you may also add it to 2.1.2 roadmap .. wiki
20131024 00:10:20 * masterzen (~masterzen@anon) Quit (Ping timeout: 260 seconds)
20131024 00:19:50 * masterzen (~masterzen@anon) has joined #jogamp
20131024 00:24:33 * xranby (~xranby@anon) Quit (Read error: Operation timed out)
20131024 00:41:06 * xranby (~xranby@anon) has joined #jogamp
20131024 01:46:28 * hharrison (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 25.0/20131017174213])
20131024 02:30:51 * xranby (~xranby@anon) Quit (Ping timeout: 245 seconds)
20131024 02:32:48 * xranby (~xranby@anon) has joined #jogamp
20131024 03:13:30 * xranby (~xranby@anon) Quit (Ping timeout: 256 seconds)
20131024 03:23:31 * xranby (~xranby@anon) has joined #jogamp
20131024 03:36:10 * hharrison (~chatzilla@anon) has joined #jogamp
20131024 05:58:36 <hharrison> Fun trivia:
20131024 05:58:53 <hharrison> byte[] data;
20131024 05:59:19 <hharrison> int val = (data[0] << 8) | data[1];
20131024 05:59:50 <hharrison> what's wrong with val
20131024 06:17:06 * xranby (~xranby@anon) Quit (Ping timeout: 256 seconds)
20131024 06:17:43 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20131024 06:18:49 * xranby (~xranby@anon) has joined #jogamp
20131024 06:57:59 * ebrayet (~Adium@anon) has joined #jogamp
20131024 07:24:19 * monsieur_max (~maxime@anon) has joined #jogamp
20131024 09:16:55 * xranby (~xranby@anon) Quit (Read error: Operation timed out)
20131024 09:31:18 * xranby (~xranby@anon) has joined #jogamp
20131024 09:34:14 * eclesia (jsorel@anon) has joined #jogamp
20131024 09:35:17 <eclesia> hi
20131024 09:36:15 <eclesia> xranby: Jmescher : Sorry, I'm not interested in relicensing JMescher or distributing it without copyright, if
20131024 09:36:15 <eclesia> that is the question. You're free to create a derivative work, but it would have to be
20131024 09:36:15 <eclesia> under the BSD license. :(
20131024 11:00:11 <masterzen> sgothel: would it be possible to release a 2.1.2-rc or something on maven so that we can try the fix for bug 866?
20131024 14:08:13 <monsieur_max> xranby: are you available for a quick question ?
20131024 14:08:30 <xranby> monsieur_max: sure
20131024 14:10:54 <monsieur_max> i'm trying to play a bit with GLMediaPlayer . The example/test works fine but whenever i try to make it work inside my own game, i'm facing an issue while trying to call initGL on it ( http://pastebin.com/X1Lz5dSc )
20131024 14:11:49 <monsieur_max> I tried to follow the same "flow" as you did but as it's supposed to be inside my game, it's not quite the same. Maybe i'm missing something
20131024 14:12:22 <monsieur_max> have you a rough idea ? if not, i'll investigate by my own, but just wanted to know if this was a common mistake
20131024 14:12:27 <xranby> hmm.. the code that failed to initialize is the texture cache
20131024 14:12:33 <xranby> Caused by: java.lang.RuntimeException: TextureNames creation failed (num: 4): err 0x500
20131024 14:12:33 <xranby> at jogamp.opengl.util.av.GLMediaPlayerImpl.createTexFrames(GLMediaPlayerImpl.java:615)
20131024 14:14:03 <monsieur_max> is it something i have to initialize before ?
20131024 14:14:21 <monsieur_max> ( i don't use such a thing in my game )
20131024 14:16:32 <xranby> is gl initialized?
20131024 14:16:49 <monsieur_max> yep no doubt at this point, i even had other textures loaded
20131024 14:17:10 <xranby> there is quite little going on in this function
20131024 14:17:26 <monsieur_max> yep, that's my concern, i can't understand what i'm missing
20131024 14:17:30 <xranby> protected TextureFrame[] createTexFrames(GL gl, final int count) {
20131024 14:17:30 <xranby> final int[] texNames = new int[count];
20131024 14:17:30 <xranby> gl.glGenTextures(count, texNames, 0);
20131024 14:17:44 <xranby> and after that you caught a gl error
20131024 14:18:31 <xranby> thus your opengl driver is upset
20131024 14:18:48 <monsieur_max> he's not the only one :)
20131024 14:18:55 <monsieur_max> i've got a question that might sound silly
20131024 14:20:17 <monsieur_max> does this init have to be called in init(GLAutoDrawable glDrawable) or can it be called after, as long as gl is initialized ?
20131024 14:22:36 <xranby> its most safe to call it in init(GLAutoDrawable glDrawable) gl is guaranteed to be initialized inside this function
20131024 14:23:40 <xranby> The message "GL_INVALID_ENUM (0x500) from glGenTextures" is puzzling.
20131024 14:23:45 <monsieur_max> yep
20131024 14:23:46 <xranby> glGenTextures can't generate that particular error code
20131024 14:24:09 <xranby> thus i belive the error originate from some code before
20131024 14:24:36 <monsieur_max> xranby: if the code is callad from display(GLAutoDrawable glDrawable); it could be ok too no ?
20131024 14:24:58 <xranby> and the error was simply thrown by some gl call done before this point and then later got caught here..
20131024 14:25:06 <xranby> that would be ok as well
20131024 14:25:20 <monsieur_max> ... strange
20131024 14:25:39 <xranby> http://www.opengl.org/sdk/docs/man/xhtml/glGenTextures.xml
20131024 14:26:06 <xranby> GL_INVALID_VALUE is the only error code this particular gl call can return
20131024 14:26:31 <monsieur_max> note that along my test i sometimes manage to make a native error thrown somewhere around the same call
20131024 14:27:01 <monsieur_max> xranby: ok, don't bother about it, i'll kepp on looking
20131024 14:27:20 <xranby> try use the GLTrace pipeline
20131024 14:27:37 <monsieur_max> ermmm .... the what ? ;)
20131024 14:29:04 <xranby> jogamp have got two wrapper GL classes GLTrace and GLDebug... that can be used to monitor all opengl calls inside your application
20131024 14:29:16 <xranby> let me find one example
20131024 14:30:16 <monsieur_max> oh great
20131024 14:31:12 <monsieur_max> speaking about debugging, did one of you managed to make jogamp project debugged with gdebugger ? ( http://www.gremedy.com/ )
20131024 14:31:25 <eclesia> I tryed but failed
20131024 14:31:48 <monsieur_max> eclesia: ok, i'm not the only one
20131024 14:31:51 <eclesia> didn't search very long anyway...
20131024 14:31:57 <monsieur_max> neither did i
20131024 14:32:32 <monsieur_max> eclesia: found someone who did it : http://stackoverflow.com/questions/15960786/profiling-jogl-with-java-visualvm
20131024 14:32:40 <eclesia> I even thinked about making a virtual GL implementation for debugging :D
20131024 14:33:46 <monsieur_max> but when i asked him, he just answered that he succeeded without much trouble
20131024 14:40:52 <monsieur_max> xranby: about the GlMediaPlayer, i got one question, is it possible to use a same instance to play multiple resources ( not at the same time ), is it a correct design ?
20131024 14:41:53 <sgothel> @Max: Yes, you can [re]set the stream
20131024 14:42:30 <monsieur_max> sgothel: great :) thanks Sven
20131024 14:42:53 <sgothel> look at the API doc .. state diagram -> initStream(..)
20131024 14:44:48 <eclesia> sgothel: I was searching for a use case of the Dummy GL context, aside from GLMediaPlayer I didn't find any. think it could be possible to have a small example of loading something in a different thread using this dummy gl ?
20131024 14:45:29 <xranby> monsieur_max: GLDebug GLTrace... use GLPipelineFactory
20131024 14:45:39 <monsieur_max> sgothel: found it, thanks
20131024 14:45:55 <xranby> gl = drawable.setGL( GLPipelineFactory.create("javax.media.opengl.Debug", null, gl, null) );
20131024 14:45:55 <xranby> gl = drawable.setGL( GLPipelineFactory.create("javax.media.opengl.Trace", null, gl, new Object[] { System.err } ) );
20131024 14:46:06 <xranby> import javax.media.opengl.GLPipelineFactory
20131024 14:46:25 <sgothel> @Eclesia: Well .. GLMediaPlayer was 1st use case .. and hence is an example :) - Answer: yes, it is just like that
20131024 14:46:27 * [Mike] (~Mike]@anon) Quit (Ping timeout: 265 seconds)
20131024 14:46:39 <sgothel> i.e. for loading/editing textures 'on other thread'
20131024 14:47:37 <eclesia> I tryed on my current project. to take advantage of my 16 cores. but always have a error :/
20131024 14:48:05 <sgothel> hmm
20131024 14:48:19 <monsieur_max> xranby: ok, thanks 'ill look at that
20131024 14:48:45 <eclesia> I used a ExecutorService, and synchronize on the dummy GL instance each time I use it
20131024 14:49:29 <eclesia> but when I use the texture in the main GL it says index is not valid (or something like that)
20131024 14:49:39 <sgothel> @Exclesia: Maybe you can provide a little unit test ..
20131024 14:49:48 <sgothel> then we can fix it together
20131024 14:50:15 <eclesia> ok
20131024 14:51:11 <sgothel> pls copy one of our unit tests .. and mod it .. so we can add it later)
20131024 14:51:25 <sgothel> your 1st JOGL contribution under BSD .. :)
20131024 14:51:31 <sgothel> (couldn't help it)
20131024 14:52:22 <monsieur_max> :)
20131024 14:52:50 <eclesia> nooooooooooo, not bsd !! argh :D
20131024 15:25:02 * hharrison (~chatzilla@anon) has joined #jogamp
20131024 15:30:27 <hharrison> sgothel: any more changes you'd like from my side on the debugpipeline stuff?
20131024 15:47:15 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131024 15:53:42 <eclesia> sgothel: here is a demo, I tryed to make it as minimal as possible. http://pastebin.com/Ysihh1iB
20131024 15:53:57 <eclesia> error is : javax.media.opengl.GLException: Create texture ID invalid: texID 0, glerr 0x0
20131024 15:54:16 <eclesia> maybe it's an obvious error.
20131024 15:57:11 <eclesia> I have to go, coming back in ~30minutes
20131024 15:57:31 * eclesia (jsorel@anon) has left #jogamp
20131024 16:21:00 * monsieur_max (~maxime@anon) has joined #jogamp
20131024 16:24:39 * Eclesia (~eclesia@anon) has joined #jogamp
20131024 16:27:51 * ebrayet (~Adium@anon) Quit (Ping timeout: 268 seconds)
20131024 16:43:36 * xranby (~xranby@anon) Quit (Ping timeout: 252 seconds)
20131024 16:50:41 * xranby (~xranby@anon) has joined #jogamp
20131024 17:56:07 * xranby-64 (~familjen@anon) has joined #jogamp
20131024 17:57:48 <xranby-64> http://streamcomputing.eu/knowledge/sdks/arm-mali/ - Linux drivers now available for Chromebook. OpenCL
20131024 18:07:18 * [Mike] (~Mike]@anon) has joined #jogamp
20131024 18:18:01 <sgothel> great - if I get your README .. hmm .. I will turn-on the device and make it GNU/Linux (next week or so) :)
20131024 18:34:27 <Eclesia> and one more 3d format support, STL, ascii and binary. http://unlicense.developpez.com/gallery/3D_format_stl_l.png
20131024 18:52:21 * ebrayet (~Adium@anon) has joined #jogamp
20131024 18:52:30 * ebrayet (~Adium@anon) Quit (Client Quit)
20131024 19:31:22 <monsieur_max> Eclesia: congratulations , mon cher
20131024 19:31:55 <Eclesia> it was the easiest 3d format I've seen
20131024 19:39:37 <monsieur_max> fear our french technology ! http://arstechnica.com/business/2013/10/move-over-google-glass-here-comes-300-competition-from-france
20131024 20:32:21 * monsieur_max (~maxime@anon) has left #jogamp
20131024 21:14:57 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20131024 21:24:04 * void256 (~void@anon) has joined #jogamp
20131024 21:39:49 * xranby (~xranby@anon) Quit (Ping timeout: 248 seconds)
20131024 21:40:44 * xranby (~xranby@anon) has joined #jogamp
20131024 21:54:04 <[Mike]> Hey guys quick question, is there a way to prevent a NewtWindow/GLWindow from resizing?
20131024 22:12:52 * kermyt (~kermyt@anon) Quit (Ping timeout: 272 seconds)
20131024 22:15:28 <sgothel> @Mike: Quick Answer: No [not implementing fixed size native windowing toolkit flag on any platform] - Long answer: your patches in this regard would be welcome
20131024 22:15:52 <sgothel> (you could listen to the resize and bring back the size thought ..)
20131024 22:15:57 <sgothel> *though
20131024 22:16:19 * kermyt (~kermyt@anon) has joined #jogamp
20131024 23:14:59 * xranby (~xranby@anon) Quit (Ping timeout: 260 seconds)
20131024 23:21:11 * xranby (~xranby@anon) has joined #jogamp
20131024 23:24:40 <[Mike]> sgothel: Thanks sven, yeah thats the intermediate fix. As far as a patch goes, I can look into it. Was reading that the fixed size flag has to be instantiated when creating the native window, correct?
20131024 23:45:08 <sgothel> @Harvey: Can you add the 'Bug xxx' to your future commits ?
20131024 23:45:31 <sgothel> I am ready to merge now (puhhh ..) so we need to map the commits to Bug .. :)
20131024 23:46:05 <sgothel> (The 'SCM Refs' field .. you might have noted ..)
20131024 23:46:16 <sgothel> Sure .. minor stuff can go undetected :)
20131024 23:47:27 <sgothel> b77bd2e9de054124ef81413aa338c635dbdc5a54 -> Bug 870 ?
20131024 23:47:36 <sgothel> ah ..
20131024 23:48:05 <sgothel> I see, you have added it in the bug report descr. .. yeah, will copy it in the field as well .. for search queries :)
20131024 23:50:25 <sgothel> @Harvey: Excellent patches - THX
20131024 23:53:47 <hharrison> sure, can do
20131024 23:53:56 <sgothel> merged and pushed ..
20131024 23:54:08 <sgothel> doing some local tests .. then kick off jenkins
20131024 23:54:27 <sgothel> 3 bugs left .. for 2.1.2 .. hmm
20131024 23:54:36 <sgothel> excl. fonts: easy
20131024 23:54:47 <sgothel> the OSX VBO sharing .. tricky
20131024 23:54:56 <sgothel> VAO .. known .. I guess
20131024 23:55:11 <sgothel> anything else we shall have in 2.1.2 ?
20131024 23:56:48 <sgothel> DebugGL4bc: Pre 568170 / Post: 477812 (bytes of bytecode)
20131024 23:56:53 <sgothel> KUDOS
20131024 23:57:27 <sgothel> Trace.. almost no diff ..
20131024 23:57:33 <hharrison> I tried a few approaches there...have something similar in the works for GL4bcImpl
20131024 23:57:42 <sgothel> great!
20131024 23:58:08 <hharrison> My local build saw 30kb of the jogl-all.jar size
20131024 23:58:14 <sgothel> even though .. this needs a good review since it's risky :)
20131024 23:58:16 <sgothel> great
20131024 23:58:49 <sgothel> kicking out the fonts will reduce footprint as well - at least the jar size - not really the RAM footprint
20131024 23:59:01 <hharrison> Yeah, when it comes to GL4bcimpl...it is building some very large stinrgs in the slow-path error message, would do much better to have a format string
20131024 23:59:27 <hharrison> with the name of the api entry as the one replacement.....it's slow error path anyways, the extra format won't hurt
20131024 23:59:37 <sgothel> slow can be slow :)
20131024 23:59:43 <hharrison> exactly!
20131024 23:59:52 <hharrison> But at least it should be small
20131025 00:00:11 <sgothel> yup .. me kicking out fonts next .. then
20131025 00:00:21 <hharrison> ...and, we'll share all the API function name strings with the trace and debug pipelines, leading to double-savings
20131025 00:00:43 <sgothel> within JVM .. not to be measured by class/jar size - gotcha
20131025 00:00:43 <hharrison> because String constants get interned for all classes at load time, as we discussed
20131025 00:01:21 <sgothel> won't hurt .. if you add your numbers to git commit and bug report - if you have them .. i.e. the JVM heap size
20131025 00:01:31 <sgothel> show off when you can :)
20131025 00:02:06 <sgothel> will help GLES3Impl as well .. embedded will be thankful !
20131025 00:02:11 <hharrison> I go one better, you can actually run sql on jvisualvm hep dumps to count Strings
20131025 00:16:24 <sgothel> masterzen: sgothel: would it be possible to release a 2.1.2-rc or something on maven so that we can try the fix for bug 866? <- 2.1.2 is around the corner ..
20131025 00:17:19 <sgothel> will see .. if next jenkins builds are find, will push a maven
20131025 01:11:06 * void256 (~void@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 24.0/20130910160258])
20131025 02:45:40 <sgothel> @Mark: I guess you mentioned this (ES3 GLSL ..) https://jogamp.org/bugzilla/show_bug.cgi?id=872
20131025 02:48:41 <sgothel> tested against OpenGL ES 3.0 Mesa 9.2.1 [GL 3.0.0, vendor 9.2.1 (Mesa 9.2.1)]
20131025 03:01:19 <sgothel> http://www4.mercedes-benz.com/manual-cars/ba/foss/content/en/assets/FOSS_licences.pdf <- TONS of FOSS in Mercedes .. :)
20131025 03:05:50 <sgothel> SGI .. BSD .. Sun .. almost looks like an old JOGL version is in :)
20131025 03:27:46 * hharrison_ (~chatzilla@anon) has joined #jogamp
20131025 03:29:55 <hharrison_> thanks for merging those patches...please let me know what else you'd like to see in future patces to make it easier on you
20131025 03:30:30 <sgothel> they are perfect .. (better partitioned than mine)
20131025 03:30:51 <hharrison_> i need baby steps for now ;-)
20131025 03:31:20 <sgothel> before I push an impacting commit, I may create the bug report on-the-fly .. so it's in the commit message :)
20131025 03:31:39 <hharrison_> yeah..I will do to help tracking
20131025 03:31:47 <sgothel> feel free to add those to the current release wiki ..
20131025 03:32:06 <sgothel> or if you [or anybody else] feels they should be included
20131025 03:33:39 <hharrison_> what's next on your roadmap....inspired by anything?
20131025 03:34:03 <sgothel> have to check 776 and 692
20131025 03:34:31 <sgothel> since you like to get into the details .. hmm .. maybe you even dare to add GL 4.4 stuff ?
20131025 03:34:43 <sgothel> the ctx is already recognized ..
20131025 03:35:07 <sgothel> so .. the additional functions must be added to the headers .. and some thinking about spec details .. maybe
20131025 03:35:30 <sgothel> I usually update all GL headers when bumping the versions ..
20131025 03:35:38 <sgothel> i.e. I use kdiff3 .. and merge manually
20131025 03:36:12 <sgothel> or you like to play w/ Android and JOCL ?
20131025 03:36:27 <sgothel> or SWT ? :)
20131025 03:37:01 <hharrison_> jocl seems pretty interesting
20131025 03:37:16 * xranby-64 (~familjen@anon) Quit (Ping timeout: 246 seconds)
20131025 03:37:45 <sgothel> the most disturbing part for me is - it still uses the NB crap (ant files)
20131025 03:37:52 <hharrison_> in a previus life I did a lot of traffic modelling I always wanted to try porting the models to a gpu
20131025 03:38:06 <sgothel> i.e. I would love to have it streamlines w/ JOGL/JOAL style .. of manual ant files
20131025 03:38:29 <sgothel> hehe .. closing eyes .. pressing thumbs .. JOCL maintainer ?
20131025 03:38:38 <hharrison_> that I may be able to tackle...I like to understand how to build my prereqs
20131025 03:38:48 <sgothel> cool
20131025 03:39:02 <hharrison_> ...learn jogl/gluegen first...then transfer to jocl
20131025 03:39:08 <sgothel> then .. Android/JOCL .. well, have to add the demo to test Bug 863
20131025 03:39:32 <sgothel> JOAL make is most simple .. and similar complexity .. JOGL is .. err complex :)
20131025 03:39:55 <hharrison_> hah!
20131025 03:39:57 <sgothel> JOCL also doesn't recognize when things do not need to be rebuild .. so it's all ugly
20131025 03:40:27 <sgothel> JOGL .. you know, are 3 modules .. nativewindow, jogl and newt ..
20131025 03:40:38 <sgothel> then .. it's simple, seen as such
20131025 03:41:00 <sgothel> but JOAL would be best to start w/ .. straight fwd
20131025 03:41:24 <sgothel> but they are all the same .. in style/logic
20131025 03:41:47 <sgothel> most magic is hidden in gluegen
20131025 03:43:02 <hharrison_> indeed, the patches I'm working on for gl4bc are in gluegen too
20131025 03:43:06 <sgothel> you might have noticed that ant-cpptasks and ant-contrib do have their own repo now. cpptasks for sure is ugly. I started to remove some redundancies while adding things (xcode/clang)
20131025 03:43:18 <hharrison_> I need to adjust the callbacks available for methodemitters
20131025 03:43:59 <sgothel> ay - sure we have to run the cmp new/old script .. and check carefully :)
20131025 03:44:03 <hharrison_> (I am _deep_ down the rabbit hole at this point...but it's fun!)
20131025 03:44:31 <sgothel> really ? oh .. nice, touching these deep recursive OO things .. is meditation :)
20131025 03:44:35 <hharrison_> Yeah, it's a bit of latency running the code gen, and doing the diffs after, but it is easily repeatable
20131025 03:44:47 <hharrison_> Don't worry, IO
20131025 03:44:57 <hharrison_> I'll be pretty confident before sending stuff
20131025 03:45:32 <sgothel> I know - hence merging yours is a pleasure .. and something for me to learn as well
20131025 03:46:43 <sgothel> VBO names / objects [all objects] are shared between ctx .. but IMHO not their bound state .. hmm
20131025 03:47:09 <sgothel> like texture objects .. VBO .. etc .. sounds right ?
20131025 03:48:00 <hharrison_> really? I thought ctx's were their own islands....that's the way I've always treated them
20131025 03:48:24 <hharrison_> Other than explicitly shared ctxs
20131025 03:48:34 <sgothel> the specs like to differ .. depending on GL version (-> headaches :)
20131025 03:48:44 <sgothel> yes, shared ctx ..
20131025 03:49:04 <sgothel> ES3 has funny wording .. regarding what gets shared
20131025 03:49:34 <sgothel> but as I remember .. the memory objects are shared only .. not their bound state .. hmm
20131025 03:53:16 <sgothel> GL spec 4.3 core - 5 Shared .. p47 .. reading ..
20131025 03:54:47 <sgothel> VAO not shared (i.e. all container objects == client state == redundant convenience stuff)
20131025 05:13:30 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=776 <- oh dear .. a truly buggy test case - result: no issue w/ us :)
20131025 05:15:12 <sgothel> so yes .. the 'binding state' of buffers is _not_ shared!
20131025 05:18:43 * hharrison_ (~chatzilla@anon) Quit (Remote host closed the connection)
20131025 06:44:17 * xranby-64 (~familjen@anon) has joined #jogamp
20131025 07:11:22 * eclesia (jsorel@anon) has joined #jogamp
20131025 07:11:56 <eclesia> good morning
20131025 07:34:07 * monsieur_max (~maxime@anon) has joined #jogamp
20131025 07:35:26 * xranby (~xranby@anon) Quit (Ping timeout: 240 seconds)
20131025 07:40:27 * xranby (~xranby@anon) has joined #jogamp
20131025 09:24:52 * xranby-64 (~familjen@anon) Quit (Ping timeout: 260 seconds)
20131025 12:33:55 * [Mike] (~Mike]@anon) Quit (Ping timeout: 260 seconds)
20131025 14:31:56 <rmk0> hharrison: 'lo. any chance you could spare a few minutes on a math question?
20131025 14:39:26 <hharrison> sure...I have my coffee now
20131025 14:41:26 <rmk0> i'm wondering if my problem is just numerical inaccuracy
20131025 14:41:32 * rmk0 locates images
20131025 14:42:05 <rmk0> i'm using the following to generate projection matrices:
20131025 14:42:11 <rmk0> is just glFrustum, obviously
20131025 14:42:17 <rmk0> http://waste.io7m.com/2013/10/25/frustum.txt
20131025 14:42:37 <rmk0> now, er
20131025 14:43:07 <rmk0> *ahem*, sorry, host is taking ages to respond
20131025 14:43:23 <rmk0> http://waste.io7m.com/2013/10/25/03.png
20131025 14:43:36 <rmk0> so that's an image using a projection produced from the parameters shown
20131025 14:43:53 <rmk0> the "camera" is at (0.0, 1.0, 5.0)
20131025 14:44:02 <rmk0> looking towards the origin
20131025 14:44:19 <rmk0> red axis is positive X running right, green is positive Y running up, etc
20131025 14:44:50 <rmk0> my problem is that if i set the near plane to 5.0 (which would place it directly on the origin)...
20131025 14:45:05 <rmk0> http://waste.io7m.com/2013/10/25/04.png
20131025 14:45:27 <rmk0> you can see that the projection actually seems to be some way beyond the origin
20131025 14:45:41 <rmk0> if i slowly reduce the value i give for the near plane:
20131025 14:45:51 <rmk0> http://waste.io7m.com/2013/10/25/02.png
20131025 14:45:58 <rmk0> that's where the near clipping plane effectively crosses the origin
20131025 14:46:11 <rmk0> http://waste.io7m.com/2013/10/25/01.png
20131025 14:46:49 <rmk0> so... even though i specify a near clipping plane distance of 5, i actually get something more like 4.76190
20131025 14:47:20 <hharrison> hmmm
20131025 14:47:36 <rmk0> as i'm currently down a rabbit hole of my own, working on shadow mapping, the discrepancy between the vertices i generate for frustums and what's actually rendered through a given projection is... distressing
20131025 14:47:58 <rmk0> i'm wondering if it's just an accuracy problem and if there's anything i can do about it
20131025 14:48:02 <hharrison> I tend to use something more like gluePerspective, but I'll have a look
20131025 14:51:10 <rmk0> it's that third row, fourth column value that's bothering me
20131025 14:51:36 <rmk0> the near clipping plane obviously crosses the origin when that value reaches -10.0
20131025 14:52:07 <rmk0> i feel like it might be numerical errors accumulating that result in a "wrong" value there
20131025 14:52:17 <rmk0> but... i'd not expect errors to accumulate that quickly
20131025 14:52:37 <rmk0> a negation, two multiplies, a subtraction, and a divide
20131025 14:53:02 <hharrison> (2.0 * f * n) / (n - f)
20131025 14:53:06 <rmk0> yeah
20131025 14:53:12 <hharrison> not that it saves anything
20131025 14:53:25 <rmk0> ah, didn't realise you'd rewritten it
20131025 14:53:26 <hharrison> sorry, what'
20131025 14:53:30 <sgothel> (not that have followed the above .. but looks like a cumulative / number-range issue .. i.e. use epsilon (precision) when comparing ..)
20131025 14:53:39 <hharrison> s your near/far plane distances
20131025 14:54:21 <hharrison> Also, c2r2 = -1
20131025 14:54:50 <rmk0> hharrison: er, for this example, the near plane is 5.0, the far plane is 100.0
20131025 14:58:37 <hharrison> wait a second c2r2...shouldn't that be - (f + n) / (f - n)
20131025 14:58:56 <hharrison> sign error?
20131025 14:59:29 <rmk0> you may be right... i'm actually just picking up this code again after a year
20131025 14:59:51 <rmk0> ah!
20131025 15:00:31 <rmk0> yes, i've modified that line as you said, and giving a value of 5 for the near plane results in:
20131025 15:01:08 <rmk0> https://outland.arc7.info/2013/10/25/05.png
20131025 15:01:19 <rmk0> you can see the tiny strip of red at the bottom indicating the X axis
20131025 15:01:26 <rmk0> thanks very much, i'd never have spotted that one
20131025 15:01:44 <hharrison> 'given enough eyes, all bugs are shallow'
20131025 15:01:49 <rmk0> scary how you can be using code for years with a bug like that...
20131025 15:01:52 <hharrison> you're welcome
20131025 15:02:38 <rmk0> am also surprised that an error in signs would have such an apparently innoccuous effect
20131025 15:02:47 <rmk0> would expect something a bit more explosive...
20131025 15:03:36 <rmk0> i think you've saved me a week of suffering
20131025 15:03:41 <rmk0> \o/
20131025 15:04:17 <rmk0> heh, i owe you something... let me know if there's something i can do in future
20131025 15:04:33 <rmk0> been banging my head against subtle problems like this for over a week now
20131025 15:04:49 <rmk0> shadow mapping is evil stuff... can't wait until we can all have real time global illumination
20131025 15:04:56 <hharrison> I imagine that is actually the worst term to have an error in as all it would do is bias the clip distance
20131025 15:05:38 <hharrison> Or maybe the least-obvious term to have a failure in
20131025 15:07:41 <hharrison> We had/have a prototype shadow-mapping path in or in house-stuff, ended up being of dubious value for engineering assets
20131025 15:08:04 <rmk0> hehe, yes
20131025 15:08:52 <rmk0> is lighting taken into consideration much with those sorts of projects?
20131025 15:10:09 <hharrison> For day to day work....everyone wants it turned off...for a video at the end for the client 'MOAR SHINY!!'
20131025 15:12:41 <rmk0> hehe
20131025 15:28:46 <hharrison> sgothel: how do you feel about the format string + args pattern for the error paths patch?
20131025 15:29:14 <hharrison> I'm actually tempted to add a constructor to GLException that takes those args
20131025 15:29:30 <hharrison> This is all related to the GL4bcImpl stuff
20131025 15:30:31 <hharrison> Every entry point currently has an error path with this in the error-block:
20131025 15:30:32 <hharrison> throw new GLException("Method \"glActiveStencilFaceEXT\" not available");
20131025 15:30:42 <hharrison> Which would become:
20131025 15:31:09 <hharrison> throw new GLException("Method \"%s\" not available", "glActiveStencilFaceEXT");
20131025 15:31:57 <hharrison> Which would save a significant amount of space
20131025 15:52:35 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131025 16:04:07 * eclesia (jsorel@anon) Quit (Quit: Leaving.)
20131025 16:34:51 * monsieur_max (~maxime@anon) has joined #jogamp
20131025 17:20:47 <sgothel> @Harvey: ofc - np
20131025 17:21:30 <sgothel> @H&M: Great that you could solve - sometimes you just need fresh eyes
20131025 17:22:04 <sgothel> updating kernel after a loong time, any problems w/ 3.11.6 ?
20131025 17:22:25 <sgothel> (Linux kernel 3.11.6 that is ..) - i.e. NV proprietary driver .. etc
20131025 17:22:40 <sgothel> yeah, disabled intel RNG :)
20131025 17:23:47 * [Mike] (~Mike]@anon) has joined #jogamp
20131025 17:26:37 <sgothel> @Mark: 'final float C=-1.0f*(zFar+zNear)/dz;' PMVMatrix.glFrustumf(..) :)
20131025 17:27:39 <sgothel> always easy to find references when you know where to look at it :)
20131025 17:27:50 <sgothel> (i.e. thx Harvey)
20131025 17:28:11 <sgothel> experiences w/ 'write later' RAID1 for SDD + HDD ?
20131025 17:29:28 <sgothel> (found this one .. http://www.tansi.org/hybrid/ )
20131025 17:29:59 <sgothel> (yes, no zfs, since I use loop-AES encryption .. for mobile, etc)
20131025 17:32:14 <sgothel> and again .. grrr .. solaris node's NV driver crashed ..
20131025 17:35:52 * xranby (~xranby@anon) Quit (Ping timeout: 256 seconds)
20131025 17:39:15 <sgothel> bump jenkins to 1.536 (2013/10/20) .. pressing thumbs
20131025 17:46:38 * xranby (~xranby@anon) has joined #jogamp
20131025 18:26:11 <rmk0> sgothel: \o/
20131025 18:48:00 <sgothel> crying for mama - amd machine crashed .. :(
20131025 18:53:09 <hharrison> Oh no! The gremlins are loose at Sven's house
20131025 18:53:26 <sgothel> 4th build attempt now .. 'I love it'
20131025 18:54:02 <hharrison> On another note...anyone had any weird issues with Win8 having long latency in either the mouse event handling or the actual display?
20131025 18:54:25 <sgothel> w/ latest ?
20131025 18:54:37 <hharrison> No, 2.0.2
20131025 18:54:47 <sgothel> I just added touch support .. mouse/touch works fine w/ 2.1.1
20131025 18:55:04 <sgothel> haven't tested 2.0.2 on win8 .. but I guess not too much diff.
20131025 18:55:17 <sgothel> however .. I overhauled the whole Pointer
20131025 18:55:23 <sgothel> Pointer/Mouse event handleing
20131025 18:55:31 <sgothel> *keep my typos*
20131025 18:55:39 <hharrison> OK, I'll try that first....I am just about to start debug, not sure if it's the mouse handling or the actual getting pixels on screen....something sure is 'spongy'
20131025 18:55:52 <sgothel> NEWT ?
20131025 18:56:05 <hharrison> NewtCanvasAWT
20131025 18:56:10 <hharrison> :-(
20131025 18:56:11 <sgothel> (i.e. otherwise we .. ah .. ok so NEWT)
20131025 18:57:11 <hharrison> Anyways, I'll update to later jogl first, and then figure out what side the latency is coming from
20131025 18:57:21 <sgothel> I have an unpacked box here for Win8 desktop .. will check there as well at some point in time
20131025 18:57:27 <sgothel> yes pls.
20131025 18:57:53 <monsieur_max> xranby: remember the issue i had with glmediaplayer ?
20131025 18:57:59 <hharrison> This is the first time we've tried on Win8...everything works, just updates are lagging mouse movement significantly
20131025 18:58:05 <sgothel> and the -Dnewt.debug.Window.MouseEvent should tell about latency .. (timestamp in 'when')
20131025 18:58:15 <hharrison> Thought I'd just ask if it was a known issue
20131025 18:58:28 <sgothel> nope sorry
20131025 18:58:49 <hharrison> no problem, I'll report back what I find out
20131025 18:58:56 <sgothel> but what we know is .. that some magic mouse driver may cause havoc .. i.e. steal focus .. to draw a picture etc ..
20131025 18:59:09 <sgothel> thx .. bug report always welcome
20131025 18:59:17 <hharrison> There's also the possibility I'm doing something stupid in my code...we'll see
20131025 19:00:08 <sgothel> pls try our infamous .. com.jogamp.opengl.test.junit.jogl.demos.es2.newt.TestGearsES2NEWT com.jogamp.opengl.test.junit.jogl.demos.es2.newt.TestLandscapeES2NewtCanvasAWT
20131025 19:00:17 <sgothel> i.e. if lag doesn't happen there ..
20131025 19:00:22 <sgothel> it might be you :)
20131025 19:02:30 <hharrison> Hmm, may be a few days before the laptop makes its way back to the office
20131025 19:24:03 <sgothel> reboot ..
20131025 19:24:15 * sgothel (~sven@anon) Quit (Quit: Leaving.)
20131025 19:42:51 * sgothel (~sven@anon) has joined #jogamp
20131025 19:42:51 * ChanServ sets mode +v sgothel
20131025 19:43:27 <sgothel> good .. 3.11.6 w/ patch for nv kernel module works
20131025 19:43:44 <sgothel> and loop-AES-v3.7a
20131025 20:06:29 <sgothel> finally .. a build went through
20131025 20:08:09 <sgothel> anybody knows whether Debian 8 (jessie testing) will have Mesa 9.2 (which is still in experimental)
20131025 20:08:10 <monsieur_max> sgothel: found something "interesting" with glmediaplayer ... lost a few hours on this one :-/
20131025 20:08:23 <sgothel> oh .. shoot
20131025 20:08:34 <sgothel> (make it hours well spent)
20131025 20:09:06 <monsieur_max> well, :) fortunately i found out before the week end ;)
20131025 20:09:20 <sgothel> details .. ?
20131025 20:09:26 <sgothel> lack of API doc ?
20131025 20:09:50 <monsieur_max> so the trick is that if i call GLMEdiaPlayer.initGL(..) with gl.glEnable(GL2.GL_ALPHA_TEST) , it fails
20131025 20:10:26 <sgothel> hmm .. tbh me only tested it in a GL core env.
20131025 20:10:38 <sgothel> do you know why ?
20131025 20:11:04 <monsieur_max> no, not really, apparently fail o init the texture cache
20131025 20:11:14 <sgothel> so you are saying glmp.initGL(..) shall not be called why GL2.GL_ALPHA_TEST is enabled ?
20131025 20:11:27 <monsieur_max> xranby: took a look, but .. this was quite puzzling
20131025 20:11:30 <sgothel> can you make a git patch .. and add this info ?
20131025 20:11:36 <monsieur_max> well, i do, your honor
20131025 20:11:42 <monsieur_max> ermmm git ..
20131025 20:11:55 <sgothel> i.e. "Note: .."
20131025 20:12:00 <monsieur_max> ermmmmm how to say that... i don't know ( yet ) how to git
20131025 20:12:06 <sgothel> emphasizing GL2 ..
20131025 20:12:19 <sgothel> so a great scenario to start contributing ..
20131025 20:12:29 <monsieur_max> aha, was about to wirte that
20131025 20:12:31 <monsieur_max> write
20131025 20:12:36 <sgothel> we have some git links on our page wiki, and our git url
20131025 20:12:44 <sgothel> thx
20131025 20:12:58 <monsieur_max> yeah, i'll do that :)
20131025 20:13:14 <sgothel> this GL states are killing me .. similar issue w/ GLJPanel ..
20131025 20:13:47 <monsieur_max> well, i totally understand that now :)
20131025 20:13:52 <sgothel> so 1st use case of GLMP w/ a compatibility profile
20131025 20:14:02 <monsieur_max> i was stripping down the code since days to understand
20131025 20:14:23 <monsieur_max> well not full days, lets say evening hours
20131025 20:14:55 <sgothel> could be that the ALPHA test state went into the shared ctx .. somewhat .. or textures .. and rendered it all black ? ..
20131025 20:15:20 <sgothel> me usually doing all manually w/o GL states .. (shaders, uniforms ..) so never an issue
20131025 20:16:23 <monsieur_max> well it throws a real clean error
20131025 20:16:52 <sgothel> oh .. thats great, yes and property jogl.debug.DebugGL shall trigger those
20131025 20:16:55 <monsieur_max> it's not even a rendering issue, it throws a 0x500 error trying to generate textures
20131025 20:17:09 <sgothel> very good .. in those cases -> read spec
20131025 20:17:21 <monsieur_max> haha ... :)
20131025 20:17:23 <sgothel> or maybe 'just' a driver issue :)
20131025 20:17:47 <monsieur_max> well got the same issue on AMD desktop windows and nvidia laptop linux
20131025 20:18:02 <sgothel> try DebugGL .. should be more verbose .. w/ DEBUG context
20131025 20:18:08 <sgothel> especially w/ AMD
20131025 20:18:30 <sgothel> http://jogamp.org/deployment/archive/master/gluegen_737-joal_492-jogl_1120-jocl_867/
20131025 21:00:39 <sgothel> http://jogamp.org/deployment/archive/master/gluegen_737-joal_492-jogl_1120-jocl_867-signed/
20131025 21:02:35 <sgothel> maven 2.1.2-rc-20131025 .. done
20131025 22:23:43 <hharrison> sgothel: what kind of space savings are we talking about with unbundlng the ubuntu fonts?
20131025 22:23:54 <sgothel> ~1M
20131025 22:23:57 <sgothel> huge
20131025 22:24:22 <hharrison> 1M of jar contents, or 1M of final jar size?
20131025 22:24:40 <sgothel> will check now ..
20131025 22:24:48 <sgothel> and tackle it ..
20131025 22:27:28 <sgothel> compression ~50%
20131025 22:29:15 <sgothel> 3 MB uncompressed .. du -hsc src/jogl/classes/jogamp/graph/font/fonts/ubuntu/*
20131025 22:29:25 <sgothel> ~1.5M in jar
20131025 22:29:40 <hharrison> Holy shit
20131025 22:30:05 <hharrison> That's like 30% of jogl-all.jar
20131025 22:30:06 <sgothel> it's like one of those .. for loops to degrade performance :)
20131025 22:30:21 <sgothel> so .. big savior .. hahaha
20131025 22:30:49 <sgothel> in future .. user has to either take it (atom) or load own single ttf file
20131025 22:31:00 <hharrison> Definitely looking forward to that....and I'm sure your bandwidth bills will thank you
20131025 22:31:21 <hharrison> What functionality requires it?
20131025 22:31:26 <hharrison> graph?
20131025 22:31:41 <sgothel> bandwidth bills ? dude .. we have true flatrate (server as well)
20131025 22:31:47 <sgothel> graph .. yup
20131025 22:32:30 <sgothel> well .. if you have your 100mbit line .. +3M is faster than a 2nd jar to process :)
20131025 22:32:48 <sgothel> and if cached already .. who cares
20131025 22:32:54 <sgothel> but yes, will do so
20131025 22:34:48 <hharrison> Yeah, but I statically link jogl-all.jar inside a launcher
20131025 22:36:53 <hharrison> which runs by writing jar files out to the temp folder...so I save 1.5M of I/O every time the exe starts
20131025 22:52:18 <sgothel> bedtime .. good night & laters
20131025 22:56:27 * xranby (~xranby@anon) Quit (Ping timeout: 248 seconds)
20131025 22:57:50 * monsieur_max (~maxime@anon) has left #jogamp
20131025 22:59:23 * xranby (~xranby@anon) has joined #jogamp
20131025 23:04:06 * xranby (~xranby@anon) Quit (Ping timeout: 252 seconds)
20131025 23:07:54 <rmk0> heh, speaking of numerical precision...
20131025 23:08:01 <rmk0> https://outland.arc7.info/2013/10/25/banding_bad2.png
20131025 23:08:24 <rmk0> the light at the top right is actually four very low intensity lights in the same place
20131025 23:08:37 <rmk0> guess the lack of precision soon adds up when you repeatedly blend
20131025 23:08:46 <rmk0> maybe i should use an RGBA32F framebuffer...
20131025 23:13:37 <rmk0> as you can see, the effect is cumulative... a single light set to a colour and intensity that results in the same area of effect doesn't have the same banding
20131025 23:14:23 <rmk0> well, you can't actually see that, you'll have to take my word for it
20131025 23:14:25 <rmk0> /o/
20131025 23:36:39 * xranby (~xranby@anon) has joined #jogamp
20131025 23:51:24 * xranby (~xranby@anon) Quit (Ping timeout: 252 seconds)
20131025 23:54:30 * xranby (~xranby@anon) has joined #jogamp
20131026 01:16:24 * hharrison (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 25.0/20131021191948])
20131026 03:03:53 * [Mike] (~Mike]@anon) Quit ()
20131026 03:52:18 * xranby (~xranby@anon) Quit (Ping timeout: 252 seconds)
20131026 03:54:36 * xranby (~xranby@anon) has joined #jogamp
20131026 04:13:14 * [Mike] (~Mike]@anon) has joined #jogamp
20131026 04:42:08 * hharrison (~chatzilla@anon) has joined #jogamp
20131026 07:50:50 * xranby (~xranby@anon) Quit (Read error: Operation timed out)
20131026 08:07:22 * xranby (~xranby@anon) has joined #jogamp
20131026 08:16:14 * monsieur_max (~maxime@anon) has joined #jogamp
20131026 09:24:48 * xranby (~xranby@anon) Quit (Ping timeout: 240 seconds)
20131026 09:33:01 * xranby (~xranby@anon) has joined #jogamp
20131026 10:08:19 <monsieur_max> pfewww, adapting UI for Gamestick console, real pain in the ... back
20131026 10:45:12 * [Mike] (~Mike]@anon) Quit ()
20131026 12:27:25 * xranby (~xranby@anon) Quit (Ping timeout: 272 seconds)
20131026 12:31:54 * xranby (~xranby@anon) has joined #jogamp
20131026 13:29:59 * xranby (~xranby@anon) Quit (Ping timeout: 260 seconds)
20131026 13:35:47 * xranby (~xranby@anon) has joined #jogamp
20131026 17:38:35 * monsieur_max (~maxime@anon) has left #jogamp
20131026 21:04:14 * kermyt (~kermyt@anon) Quit (Ping timeout: 240 seconds)
20131026 21:07:05 * kermyt (~kermyt@anon) has joined #jogamp
20131026 22:01:50 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20131026 22:09:40 <rmk0> hm... anyone got any reference code for conversion to/from 16 bit floating point numbers
20131026 22:09:47 <rmk0> as defined in the OpenGL standard?
20131026 22:10:20 <sgothel> Good morning :) .. I remember something like this .. in GL4 ?
20131026 22:10:24 <rmk0> 'lo
20131026 22:10:38 <rmk0> oh, they exist as far back as 3.1, actually
20131026 22:10:44 <rmk0> is there something in the jogl tree?
20131026 22:10:47 <sgothel> must be some source .. if you find it, pls post it - I gave up on it when doing color lookup
20131026 22:10:51 <sgothel> no sorry
20131026 22:10:55 <rmk0> will do
20131026 22:11:20 <sgothel> i.e. an int lookup table .. guess I used the build in conversion .. SHORT -> F32
20131026 22:11:33 <sgothel> w/ RED .. etc
20131026 22:11:47 <sgothel> in the end .. it all ends up to be float on GPU
20131026 22:12:04 <rmk0> yeah, am just after something to pack values into textures
20131026 22:12:31 <sgothel> piggy bag .. :) don't rely on the value preservation there
20131026 22:13:02 * monsieur_max (~maxime@anon) has joined #jogamp
20131026 22:18:03 <rmk0> seen a few examples online... the code to do the conversions is repulsive in all cases
20131026 22:18:09 <rmk0> think i'll... be leaving this one out
20131026 22:18:14 <sgothel> Obama 2008: "allies who will listen to each other, who will learn from each other, who will, above all, trust each other." hahaha
20131026 22:18:42 <sgothel> w/ GL3 texture fetch cmds .. you may be lucky to avoid interpolation ..
20131026 22:18:48 <sgothel> i.e. int coords
20131026 22:23:18 * monsieur_max (~maxime@anon) has left #jogamp
20131026 22:23:26 <sgothel> IMHO .. if you can pick the right value of the sampler .. it's easy to make it an int (spec, i.e. a/65536 .. etc)
20131027 01:05:15 <sgothel> ha! DST ended .. and me wondering why the heck files are omitted at update (ant jar task ..) - duh
20131027 01:05:58 <sgothel> don't get it though .. system did change time .. hmm
20131027 01:07:27 * kermyt_ (~kermyt@anon) has joined #jogamp
20131027 01:08:43 * kermyt (~kermyt@anon) Quit (*.net *.split)
20131027 01:08:56 * kermyt_ is now known as kermyt
20131027 01:29:34 <sgothel> May need to add a convenience method to GLAutoDrawable to attach a shared GLContext / GLAutoDrawable .. hmm
20131027 01:29:52 <sgothel> now I am bound to our no API change ..
20131027 01:30:25 <sgothel> so this would break folks who impl. their own GLAutoDrawable .. which probably is something nobody does ?
20131027 01:30:52 <sgothel> .. or adding another interface .. hmm
20131027 01:34:16 * kermyt (~kermyt@anon) Quit (Ping timeout: 260 seconds)
20131027 01:35:17 * kermyt (~kermyt@anon) has joined #jogamp
20131027 01:35:45 <sgothel> .. will sneak this in w/o interface change .. maybe refined later
20131027 05:05:50 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20131027050550.html