JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20140224 13:40:49 (UTC)

20140224 13:40:49 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20140224050546.html
20140224 13:40:49 -NickServ- This nickname is registered. Please choose a different nickname, or identify via /msg NickServ identify <password>.
20140224 13:40:55 * jogamp (~jogamp@anon) has joined #jogamp
20140224 13:40:55 * Topic is 'http://jogamp.org | Hacking 3D Graphics, Multimedia and Processing across Devices'
20140224 13:40:55 * Set by rmk0 on 20130116 23:58:04
20140224 13:40:55 -NickServ- You are now identified for jogamp.
20140224 13:42:56 * magaio_ is now known as magaio
20140224 15:07:06 <xranby> #MWC14 is all about describing #ONONISM with positive words http://ononism.com
20140224 15:07:29 <xranby> #MWC14 #ONONISM (mind your vowels) is an extremely virulent disease, as yet unrecognised as such.
20140224 15:21:33 <xranby> Mozilla made some impressions on MWC14 by anouncing a 25$ Firefox OS smartphone http://www.phonearena.com/news/Mozilla-signs-a-deal-to-make-the-worlds-cheapest-smartphone-25-Firefox-OS-device-with-3.5-screen-and-HTML5-apps_id52963
20140224 15:22:20 <monsieur_max> oh, great, nice size too. However, FOS tends to be sluggish :(
20140224 15:23:06 <xranby> does it have a GPU?
20140224 15:23:28 <monsieur_max> xranby : straight to the point
20140224 16:57:20 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140224 17:00:09 * magaio (~magaio@anon) Quit (*.net *.split)
20140224 17:00:10 * hija (~hija@anon) Quit (*.net *.split)
20140224 17:00:11 * xranby-64 (~familjen@anon) Quit (*.net *.split)
20140224 17:00:12 * bbbruce (~bx@anon) Quit (*.net *.split)
20140224 17:00:12 * sgothel (~sgothel@anon) Quit (*.net *.split)
20140224 17:00:21 * kermyt_ (~kermyt@anon) Quit (*.net *.split)
20140224 17:00:23 * jvanek (jvanek@anon) Quit (*.net *.split)
20140224 17:06:07 * phao (~phao@anon) has joined #jogamp
20140224 17:06:07 * kermyt_ (~kermyt@anon) has joined #jogamp
20140224 17:14:51 * xranby-64 (~familjen@anon) has joined #jogamp
20140224 17:16:35 * magaio (~magaio@anon) has joined #jogamp
20140224 17:44:21 * sgothel (~sgothel@anon) has joined #jogamp
20140224 17:44:21 * hija (~hija@anon) has joined #jogamp
20140224 17:44:21 * bbbruce (~bx@anon) has joined #jogamp
20140224 17:54:25 * hija (~hija@anon) Quit (Quit: hija)
20140224 18:22:19 * phao (~phao@anon) Quit (Quit: Fui embora)
20140224 18:22:46 * phao (~phao@anon) has joined #jogamp
20140224 18:27:16 * monsieur_max (~maxime@anon) has joined #jogamp
20140224 19:39:53 * hija (~hija@anon) has joined #jogamp
20140224 19:48:33 * hija (~hija@anon) Quit (Quit: hija)
20140224 20:31:50 * kermyt_ is now known as kermyt
20140224 20:54:24 * kermyt (~kermyt@anon) Quit (Ping timeout: 240 seconds)
20140224 20:57:03 * kermyt (~kermyt@anon) has joined #jogamp
20140224 21:01:22 <rmk0> i'm trying to calculate the window-space bounds of an object
20140224 21:01:44 <rmk0> i'm transforming each of the eight vertices that make up the bounding box for the object, from object local space to clip space
20140224 21:02:00 <rmk0> and from there through normalized device coordinates and window-space coordinates
20140224 21:02:28 <rmk0> if the w component of all of the vertices in normalized device space is < 0, i assume the object is behind the viewer
20140224 21:02:54 <rmk0> i'm doing this in order to get a region that i can pass to glScissor
20140224 21:03:30 <rmk0> is there anything else i need to handle? i seem to be having issues when the camera is effectively perpendicular to the object (such that the object is completely off the screen)
20140224 21:04:10 <rmk0> as in, the object is still considered to be "not behind" the viewer, and the scissor region briefly covers the entire screen for some unknown reason
20140224 21:06:01 <rmk0> guess there could be a division by 0 in there, when w passes 0.0
20140224 21:06:03 * rmk0 checks
20140224 21:08:22 <rmk0> hm, no that doesn't seem to be it
20140224 21:09:06 <rmk0> it does seem to occur when the sign of the w components of some of the vertices are positive, and some are negative
20140224 21:09:26 <rmk0> that happens for a small range of angles when turning the view mostly perpendicular as described
20140224 21:09:37 <rmk0> not... exactly sure what the right way to deal with that is
20140224 21:10:59 <rmk0> when the observer is inside the object, it'll obviously be the case that the clip-space coordinates of some of the vertices will be negative
20140224 21:11:06 <rmk0> but this is happening when the observer is outside
20140224 21:13:44 <phao> rmk0, btw, I can't really help you, but the #opengl channel has lots more people.
20140224 21:14:14 <phao> Almost 30 times more, (300 people now)
20140224 21:14:16 <rmk0> phao: i've given up asking questions in there
20140224 21:14:22 <phao> why so?
20140224 21:14:33 <rmk0> have been insulted every time i've been in there... like it's illegal to ask questions or something
20140224 21:14:38 <rmk0> no patience for that
20140224 21:15:33 <phao> I see.
20140224 21:15:56 <rmk0> will try the opengl forums, generally get a good response from there
20140224 21:16:18 <phao> rmk0, window-space bounds of an object is the smallest rectangle which contains the object?
20140224 21:16:30 <phao> like a bounding box?
20140224 21:16:31 <rmk0> yep
20140224 21:17:20 <rmk0> the results i'm getting are mostly correct, there's just this anomaly when the object is off the screen and the view direction is roughly perpendicular to it
20140224 21:17:47 <rmk0> i mean the method is right... it's obviously just doing what opengl does but on the cpu
20140224 21:17:58 <phao> can't you write some sort of shader which after transforming the object to window-coordinates checks if its point has the last/max x, y, or
20140224 21:17:58 <rmk0> but presumably there's an edge case here i need to handle and don't know how to detect
20140224 21:17:59 <phao> z
20140224 21:18:15 <phao> and sends those as the output to another shader?
20140224 21:18:27 <rmk0> possibly
20140224 21:19:01 <phao> I know very little about shaders. But I'd try looking for something like a uniform-kind of variable to which I can write, with some kind of atomic operation
20140224 21:19:20 <rmk0> that's really a last resort for this sort of thing
20140224 21:19:24 <phao> I see.
20140224 21:19:33 <phao> No clue then =D
20140224 21:19:43 <rmk0> i mean... for something like this where it's a few simple math operations
20140224 21:20:05 <phao> Ahh, I'm sorry, you already have a way to do it?
20140224 21:20:13 <rmk0> mostly!
20140224 21:20:56 <phao> Well, good luck.
20140224 21:20:59 <rmk0> hehe
20140224 21:21:43 <phao> I'm not kidding. I know very little of opengl to help here.
20140224 21:21:50 <rmk0> it's not really an opengl problem
20140224 21:22:12 <rmk0> just happen to be doing the same math that opengl does to transform vertices to window space
20140224 21:23:03 <phao> the problem is that you mentioned glScissor, which I have no clue what it is. =)
20140224 21:23:24 <rmk0> ah, is a way of preventing writes to the framebuffer
20140224 21:23:35 <phao> What you could also do is to find the bounding box on the cpu during program start-up, and apply the same transformations to it as you apply to the original object.
20140224 21:23:37 <rmk0> you basically give opengl a rectangle in screen space and it'll prevent writes to anything outside of that
20140224 21:23:57 <rmk0> like a stencil
20140224 21:24:09 <phao> Because during start-up that object will likely to be in a convenient frame of reference.
20140224 21:24:24 <phao> I see.
20140224 21:24:26 <rmk0> that's... what i'm doing
20140224 21:24:32 <phao> Hehehhehe.
20140224 21:24:44 <rmk0> i have the eight vertices that form the bounding box of the object in object-local space
20140224 21:24:55 <rmk0> am transforming those to screen space on the cpu
20140224 21:25:16 <phao> I guess I just didn't understsand your original text.
20140224 21:25:25 * rmk0 is obscure \o/
20140224 21:27:20 <rmk0> i'm quite surprised that i'm having issues
20140224 21:27:23 * monsieur_max (~maxime@anon) has left #jogamp
20140224 21:27:28 <rmk0> is not like there's anything that isn't directly out of a textbook here
20140224 21:29:53 <phao> well, when you have perpendicular stuff, angles between vectors have their sin 0. You were worried about div by 0, ...
20140224 21:30:13 <phao> maybe you're doing sin of something, or cos of something. If the sine is 0, cross products also are.
20140224 21:30:30 <phao> Anyway, I'll be quiet. Gotta get back to studying.
20140224 21:31:00 <phao> By "perpendicular stuff", I mean what you said earlier.
20140224 21:31:15 <phao> "i seem to be having issues when the camera is effectively perpendicular to the object"
20140224 21:32:12 <rmk0> think i've narrowed it down
20140224 21:32:46 <phao> what is it?
20140224 21:32:56 <rmk0> it's essentially the case where some of the clip space coordinates w components are positive and some are negative
20140224 21:33:10 <rmk0> that'll happen when viewing at that angle, or when the observer is inside the object
20140224 21:34:40 <rmk0> if the object is a cube and i'm in the center of the cube looking out
20140224 21:34:49 <rmk0> four of the vertices are "behind" me, and so their w components are negative
20140224 21:34:55 <rmk0> four are in front and so are positive
20140224 21:35:28 <phao> btw, rmk0, what opengl do you consider good
20140224 21:35:28 <phao> ?
20140224 21:35:30 <phao> books*
20140224 21:35:56 <rmk0> sort of depends what you're looking for
20140224 21:36:45 <rmk0> i don't think there are any that actually walk through the construction of an engine
20140224 21:36:50 <phao> I am not sure. All I am "for sure" is that I want to be doing real time rendering related stuff.
20140224 21:37:09 <phao> I don't think that norrows much =D
20140224 21:37:14 <rmk0> mathematics for 3d game programming is excellent
20140224 21:37:31 <rmk0> is the most accessible and complete book i could find for that side of things
20140224 21:37:55 <rmk0> http://mathfor3dgameprogramming.com
20140224 21:38:36 <rmk0> plus http://arcsynthesis.org/gltut for basic lighting techniques
20140224 21:38:42 <phao> I see.
20140224 21:38:48 <phao> Thanks.
20140224 21:38:48 <rmk0> and the gpu gems books from nvidia (free) for specific techniques after that
20140224 21:39:13 <phao> aren't those techniques dated?
20140224 21:39:20 <rmk0> nope
20140224 21:39:35 <rmk0> most people are using deferred shading now, but the techniques stay the same
20140224 21:40:04 <phao> deferred shading?
20140224 21:40:34 <rmk0> er, well "forward" rendering has you render each object in the scene and apply lighting directly, per pixel
20140224 21:41:12 <rmk0> deferred rendering essentially renders the surface properties of the geometry in the scene to images
20140224 21:41:18 <rmk0> and then does lighting on those images
20140224 21:41:57 <rmk0> the lighting code is the same as you'd write in traditional rendering, it's just that the data comes from an image as opposed to being passed in to the shader, if you see the distinction
20140224 21:43:12 <rmk0> "forward" rendering being the traditional way
20140224 21:43:49 <rmk0> http://ogldev.atspace.co.uk
20140224 21:43:56 <rmk0> he covers both there
20140224 21:44:00 <phao> ok
20140224 21:44:09 <rmk0> his english is a bit rough, though
20140224 21:44:49 <phao> hehe, no problems. I can't be much picky myself.
20140224 21:45:28 <rmk0> there was a nice article written about the doom 3 engine a while back
20140224 21:45:54 <phao> are those tutorials good, btw?
20140224 21:45:56 <rmk0> http://fabiansanglard.net/doom3/index.php
20140224 21:46:02 <rmk0> they're ok, yes
20140224 21:46:08 <rmk0> few things in there i disagree with
20140224 21:46:22 <rmk0> ^^ the architecture of doom 3's renderer is still relevant now
20140224 21:47:08 <rmk0> everyone uses shadow mapping now though, rather than shadow volumes
20140224 22:25:07 * hija (~hija@anon) has joined #jogamp
20140224 22:43:43 * xranby-64 (~familjen@anon) Quit (Quit: Leaving.)
20140224 23:59:32 * kermyt (~kermyt@anon) Quit (Ping timeout: 246 seconds)
20140225 00:02:29 * bbbruce (~bx@anon) Quit (Ping timeout: 240 seconds)
20140225 00:04:53 * kermyt (~kermyt@anon) has joined #jogamp
20140225 00:23:06 * hija (~hija@anon) Quit (Quit: hija)
20140225 00:27:02 * bbbruce (~bx@anon) has joined #jogamp
20140225 00:43:26 * hija (~hija@anon) has joined #jogamp
20140225 00:45:39 * hija (~hija@anon) has left #jogamp
20140225 01:33:25 <phao> Hey... is there any guides on writing unit tests for code which uses opengl?
20140225 01:33:34 <phao> are*
20140225 01:35:12 * magaio (~magaio@anon) Quit (Ping timeout: 240 seconds)
20140225 01:36:56 * magaio (~magaio@anon) has joined #jogamp
20140225 02:34:41 * phao (~phao@anon) Quit (Disconnected by services)
20140225 02:34:43 * phao_ (~phao@anon) has joined #jogamp
20140225 02:34:55 * phao_ (~phao@anon) has left #jogamp
20140225 02:40:31 * phao (~phao@anon) has joined #jogamp
20140225 02:40:54 <phao> Ohhh dear.
20140225 02:41:49 <phao> Don't look this up... I'll regret forever of looking this disease ( Myiasis ) up. So disgusting...
20140225 03:04:45 * phao_ (~phao@anon) has joined #jogamp
20140225 03:07:46 * phao (~phao@anon) Quit (Ping timeout: 244 seconds)
20140225 03:10:38 * phao_ (~phao@anon) Quit (Read error: Connection reset by peer)
20140225 03:10:39 * phao__ (~phao@anon) has joined #jogamp
20140225 03:21:42 * phao__ (~phao@anon) Quit (Read error: Connection reset by peer)
20140225 03:21:43 * phao_ (~phao@anon) has joined #jogamp
20140225 03:22:28 * phao_ (~phao@anon) Quit (Client Quit)
20140225 03:37:16 * phao (~phao@anon) has joined #jogamp
20140225 03:44:57 * phao (~phao@anon) Quit (Read error: Network is unreachable)
20140225 03:52:04 * phao (~phao@anon) has joined #jogamp
20140225 04:11:38 * bbbruce (~bx@anon) Quit (Ping timeout: 264 seconds)
20140225 04:15:22 * phao (~phao@anon) Quit (Read error: Connection reset by peer)
20140225 04:32:53 * bbbruce (~bx@anon) has joined #jogamp
20140225 05:15:41 * hharrison (~chatzilla@anon) has joined #jogamp
20140225 05:16:26 <hharrison> sgothel: thanks for the CC on that webstart crash, things are crazy busy and I'll likely only be back to java3d stuff in another week or so
20140225 05:20:42 * phao (~phao@anon) has joined #jogamp
20140225 05:28:49 <sgothel> ay .. again, it's not a bug in java3d - but somewhat it triggers that bug in the JRE
20140225 05:29:23 <sgothel> so .. all 'good' .. hope things are fine. me hacking along on the 2.2-train
20140225 05:30:26 * phao (~phao@anon) Quit (Quit: Fui embora)
20140225 05:50:24 <hharrison> Yeah, I want to have a look over your binarystream stuff
20140225 05:50:49 <sgothel> thought it's good to have this issue tackled in one place
20140225 05:51:03 <sgothel> and soon .. I like to add some binary 3d formats .. etc
20140225 05:51:37 <sgothel> btw .. I put you also on the GlueGen maintainer list :)
20140225 05:56:36 <hharrison> I noticed
20140225 05:57:09 <hharrison> You're welcome to have a peak at the collada and obj loaders we've written for jaamsim
20140225 05:58:01 <hharrison> We even wrote a blender plugin to export those assets into our own mesh format
20140225 05:58:09 <sgothel> ah .. right, yes .. sounds great. would need a GPLv? license to use ?
20140225 05:58:17 <sgothel> now you got me interested
20140225 05:58:34 <sgothel> could you offer a BSD license for jogamp ?
20140225 05:58:48 <hharrison> If you were really interested, we could likely find a way to get it BSD or Apache2
20140225 05:58:58 <sgothel> sweet ..
20140225 05:59:03 <sgothel> yes very much for graph ui!
20140225 05:59:17 <sgothel> currently started to work on it
20140225 06:00:05 <sgothel> need to clean up the graph-curve packages incl. text rendering - then the scene-ui stuff
20140225 06:00:12 <sgothel> your importers would help a lot
20140225 06:00:26 <sgothel> i.e. not need to start from scratch
20140225 06:01:12 <sgothel> you have have the blender 'DNA' binary thing in ? Collada xml ?
20140225 06:01:29 <sgothel> which editor tool do you use for Collada ?
20140225 06:01:54 <hharrison> Blender + a python plugin to our own jsm format (collada subset)
20140225 06:02:13 <hharrison> exports very simple meshes w/ bone weights to use with some really simple shaders
20140225 06:02:30 <hharrison> also have a direct collada import and an OBJ import
20140225 06:02:36 <sgothel> ah .. so no binary blender parser itself
20140225 06:02:40 <hharrison> again...simple meshes w/ bone weights
20140225 06:02:50 <sgothel> yeah .. would need the CAD / curves ofc
20140225 06:03:02 <hharrison> Nope, ended up being easier to use the internal API from a plugin
20140225 06:03:16 <sgothel> but again .. it would help a lot to start w/ your code
20140225 06:03:20 <hharrison> Everything is a bit rough and ready, but quite functional
20140225 06:03:37 <sgothel> sure sure .. I can read code :)
20140225 06:03:57 <sgothel> working code == great documentation
20140225 06:04:15 <sgothel> at least semantics are clear
20140225 06:04:31 <hharrison> sure thing
20140225 06:11:44 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20140225 07:55:33 * monsieur_max (~maxime@anon) has joined #jogamp
20140225 08:43:07 * jvanek (jvanek@anon) has joined #jogamp
20140225 09:30:01 <xranby> https://www.youtube.com/watch?v=yBQoy-uFXZ8 - Jiri Vanek - IcedTea-Web slides: http://devconf.cz/filebrowser/download/442
20140225 09:46:05 <xranby> Intel releases their first #Android tutorial for #OpenCL: http://software.intel.com/sites/default/files/AndroidBasicOpenCL.pdf
20140225 09:51:25 <sgothel> Holy moly .. nice slides Jiri, I like that you always emphasize on the 'freedom', KUDOS
20140225 09:51:53 <sgothel> btw .. 'jweb' is not based on JOGL .. it's Plugin3 interface is completely toolkit agnostic
20140225 09:56:04 <sgothel> too bad .. the Intel paper doesn't yet push JOCL :)
20140225 10:02:19 * hija (~hija@anon) has joined #jogamp
20140225 10:04:28 <xranby> is the updated JOCL test apps from 2.1.4 uploaded to google play?
20140225 10:04:50 <sgothel> nope .. manual install, will do one for 2.1.5
20140225 10:07:11 <xranby> kudos to Wade to make the JOCL demos work on low memory devices!
20140225 10:09:00 <sgothel> yeah, we are doing a good job fixing things there ..
20140225 10:09:11 <sgothel> could you verify the crossplayer fix ?
20140225 10:13:57 <xranby> sgothel: all the locking warnings are gone, great sucess
20140225 10:16:09 <xranby> if you use 40+ parralel streams then the circular buffer in ALAudioSink can be "full" kind of expected
20140225 10:21:00 <xranby> thus the only "issues" i can create on my machine are related to out of memory when runnign too many streams.
20140225 10:23:08 <sgothel> well, we can't help the memory .. but you can adjust the buffer size AFAIK
20140225 10:24:09 <xranby> i can try create a junit test that tried to detect if we hold a lock if we trigger a OOM runtime exception while holding one
20140225 10:25:31 <xranby> the goal is to cleanup gracefully in case od out of memory
20140225 10:25:35 <xranby> case of
20140225 10:39:43 <hija> guys, where do I find the git-sha1 of my commit?
20140225 10:39:52 <hija> for this:
20140225 10:39:53 <hija> https://github.com/orange-vertex/jogl/commit/9a5b0c6e97b6e79287edf92361d765f7efe66c5e
20140225 10:40:04 <hija> its 9a5b0c6e97b6e79287edf92361d765f7efe66c5e ?
20140225 10:40:17 <hija> (the thing after the last /)
20140225 10:48:24 <xranby> hija: you may want to cleanup your merge request by first placing yourself at the current master
20140225 10:48:28 <xranby> and then git cherry-pick 9a5b0c6e97b6e79287edf92361d765f7efe66c5e
20140225 10:49:17 <hija> placing myself at the current master?
20140225 10:49:22 <xranby> git checkout master
20140225 10:49:30 <hija> meh using eclipse
20140225 10:49:35 <hija> let me see..
20140225 11:09:01 <hija> xranby: ok so I did this locally
20140225 11:09:20 <hija> up until the cherry picking
20140225 11:12:21 <hija> now?
20140225 11:17:42 <xranby> hija: i the history looks good then publish the branch
20140225 11:18:06 <xranby> i usually review the history using gitk before i git push
20140225 11:18:42 <hija> ok bear with the newbie here :P
20140225 11:19:09 <xranby> assume you local master is identical to the jogamp master + the cherry pick
20140225 11:19:10 <hija> how do I know if the history looks good? :p
20140225 11:19:34 <xranby> if the history only add the relevant chnages
20140225 11:19:40 <xranby> then the history looks giid
20140225 11:19:42 <xranby> good
20140225 11:25:05 <xranby> the reason i wanted you to cherry-pick your change is because this merge commit that you currently have ion your published branch makes https://github.com/orange-vertex/jogl/commit/cbcbafa139137f6a7143d5757c2e236e23613458 makes reviewing hard
20140225 11:26:39 <hija> yeah I guessed that would cause some problems...
20140225 11:26:58 <hija> the pull request back to my branch
20140225 11:26:59 <xranby> if you had done the merge from the other direction
20140225 11:27:35 <xranby> that is fist switching to the updated jogl master branch and then merge in your branch then it would be more easy to read
20140225 11:28:11 <xranby> because then the merge changeset would only contiain lines changed by your patch
20140225 11:28:36 <hija> so go to local master branch, checkout origin/master
20140225 11:28:46 <hija> create new branch
20140225 11:28:53 <hija> (bug 969)
20140225 11:28:57 <hija> do changes
20140225 11:29:15 <hija> then?
20140225 11:29:21 <hija> merge back to local master?
20140225 11:29:46 <xranby> i sugegst you do the following 1. checkout origin/master 2. git cherry-pick 9a5b0c6e97b6e79287edf92361d765f7efe66c5e 3. delete local master 4. git branch master
20140225 11:30:09 <xranby> thus the new local master will then be origin/master + the cherry pick
20140225 11:31:28 <xranby> 5. then replace the current master at github with your current local master git push github master --force
20140225 11:32:28 <xranby> you can delete your current master using: git branch -D master
20140225 11:35:33 <xranby> as long as you only rewrite your own history then its file :)
20140225 11:35:34 <xranby> fine
20140225 11:36:02 <xranby> thus the only thing to avoid is to rewrite someone elses history
20140225 11:36:20 <hija> i think i did it
20140225 11:36:20 <hija> i was missing --force
20140225 11:36:20 <hija> https://github.com/orange-vertex/jogl/commits/master
20140225 11:36:20 <hija> can I force the bug_969 branch to update?
20140225 11:36:20 <hija> rebase?
20140225 11:36:45 <hija> and if I do, will it remove the unnecessary commit from the pull request?
20140225 11:36:51 <xranby> you can update the bug_969 easy
20140225 11:37:30 <hija> how?
20140225 11:37:30 * phao (~phao@anon) has joined #jogamp
20140225 11:38:33 <xranby> do like this 1. git checkout master (that contains the nice history) 2. git branch -D bug_969 (remove the old bug_969 branch) 3. git branch bug_969 4. git push github bug_969 --force
20140225 11:39:01 <xranby> that will update the pull request
20140225 11:43:31 <hija> done
20140225 11:43:35 <xranby> yes the pull request looks much better now https://github.com/sgothel/jogl/pull/77
20140225 11:44:15 <hija> some times the non-terminal way is so convoluted it's not worth it
20140225 11:45:10 <hija> or it's just better to go through the terminal first to learn :P
20140225 11:45:30 <hija> in any case, thanks xr
20140225 11:46:33 <hija> so I guess the sha1 is: 78fcb8228d4a391054501aef16eb0462322ba39d ?
20140225 11:46:49 <xranby> yes
20140225 12:05:07 * pdurbin (~pdurbin@anon) has joined #jogamp
20140225 12:12:29 <sgothel> holy moly .. :)
20140225 12:12:36 <hija> sgothel: regarding http://forum.jogamp.org/NewtCanvasSWT-problem-tp4030973p4031719.html
20140225 12:12:46 <sgothel> yes ?
20140225 12:12:51 <hija> unit tests
20140225 12:13:04 <hija> well to test it I set up an eclipse project
20140225 12:13:08 <sgothel> sweet .. will merge later today ..
20140225 12:13:09 <hija> its not in SWT...
20140225 12:13:17 <sgothel> have you added a note about tested platforms ?
20140225 12:13:29 <hija> so I can't actually create a unit test..
20140225 12:13:32 <sgothel> pls prefer the bugzilla conversation ..
20140225 12:13:43 <sgothel> ?
20140225 12:14:09 <hija> its a different toolkit
20140225 12:14:28 <sgothel> the things you have fixed .. can they be tested ?
20140225 12:14:29 <hija> ViewPart does not belong to SWT
20140225 12:14:33 <sgothel> ah
20140225 12:14:36 <hija> yes, but not as a unit test
20140225 12:14:39 <sgothel> but other cases ..
20140225 12:14:53 <sgothel> license of 'ViewPart' ?
20140225 12:15:10 <hija> http://help.eclipse.org/indigo/nftopic/org.eclipse.platform.doc.isv/reference/api/org/eclipse/ui/part/ViewPart.html
20140225 12:15:23 <hija> its part of a ui plugin for eclipse
20140225 12:15:26 <sgothel> so we could add it to our repo ?
20140225 12:15:42 <sgothel> i.e. a small (if possible) jar w/ a license readme ?
20140225 12:15:50 <sgothel> like we drop SWT in jogl/make/lib
20140225 12:16:37 <hija> so, (i should) find the required libraries to build a unit test?
20140225 12:16:55 <sgothel> 'build a unit test' sounds dramatic
20140225 12:16:59 <hija> hehe
20140225 12:17:05 <sgothel> copy our existing ones .. and make it fit
20140225 12:17:17 <sgothel> need to edit the swt'ish classpath ..
20140225 12:17:25 <sgothel> in build-common.xml
20140225 12:17:55 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=969 <- tested platforms .. where it works (your patch) .. and the git-sha1 which I shall merge
20140225 12:18:11 <sgothel> i.e. I will fetch your repo ..
20140225 12:18:25 <sgothel> and then I will do a 'git merge lala/your-branch'
20140225 12:18:37 <hija> git sha1 is there
20140225 12:18:38 <sgothel> if no conflict .. all good
20140225 12:18:43 <hija> will add platforms
20140225 12:18:46 <sgothel> nice ..
20140225 12:19:34 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=980 <- for illustration how I comment w/ git-sha1
20140225 12:19:51 <sgothel> simply copy-paste of commit message :)
20140225 12:20:10 <sgothel> even though it never happens .. it gives you a chance to discuss it :)
20140225 12:20:25 <sgothel> I would see this to happen more frequently though :)
20140225 12:20:55 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=927 <- where Xerxes and I do discuss ..
20140225 12:22:22 <hija> hehe
20140225 12:22:36 <sgothel> last time I tried egit .. it crashed eclipse .. oh well
20140225 12:22:47 <sgothel> so .. still gitk, cola and qgit for UI
20140225 12:22:50 <hija> well I did type a lot of things in there, but no replies :p
20140225 12:23:04 <hija> https://jogamp.org/bugzilla/show_bug.cgi?id=969
20140225 12:23:10 <hija> added comment with systems
20140225 12:23:25 <sgothel> yes .. I read it ..
20140225 12:23:32 <sgothel> my fault ..
20140225 12:23:34 <sgothel> one sec :)
20140225 12:26:10 <hija> so for discussion bugzilla > irc > forum?
20140225 12:27:23 <sgothel> sort of ..
20140225 12:27:38 <sgothel> bugzilla is gold, i.e. it's preserved and searchable and owned by us
20140225 12:27:47 <sgothel> irc for transient ..
20140225 12:27:56 <sgothel> forum for some initial discussions ..
20140225 12:28:08 <sgothel> announcement etc ..
20140225 12:28:22 <sgothel> but for sure .. the hard stuff goes into bugzilla
20140225 12:28:32 <sgothel> if you like .. you sure can announce it in forum
20140225 12:29:01 <sgothel> for example .. I don't track these forum emails .. but do pick up bugreports even after month :)
20140225 12:29:50 <sgothel> so .. you are happy w/ your result ? great .. then it goes to 2.1.5
20140225 12:30:25 * sgothel (~sgothel@anon) Quit (Remote host closed the connection)
20140225 12:30:40 * sgothel (~sgothel@anon) has joined #jogamp
20140225 12:30:40 * sgothel (~sgothel@anon) Quit (Changing host)
20140225 12:30:40 * sgothel (~sgothel@anon) has joined #jogamp
20140225 12:30:40 * ChanServ sets mode +v sgothel
20140225 12:45:31 <xranby> added a manual way to reproduce a deadlock running 40 parallel players: https://jogamp.org/bugzilla/show_bug.cgi?id=927#c17
20140225 12:50:53 <sgothel> rossFadePlayer.java:81 is the initial play ..
20140225 12:51:09 <sgothel> and holds the player state-lock
20140225 12:51:51 <sgothel> is this a very short clip ?
20140225 12:52:40 <sgothel> the stream-worker already replies w/ pauseImpl .. hmm
20140225 12:52:49 <xranby> lewiscarroll.wav is a 46s clip
20140225 12:53:23 <sgothel> "Thread-121" prio=10 tid=0 (player)
20140225 12:53:28 <sgothel> "Thread-118-StreamWorker_34" da
20140225 12:55:15 <sgothel> an error must have occured .. ? hmm
20140225 12:57:48 <sgothel> very good .. player is in 'play' waiting for 'isActive' and streamworker does not notify it .. while already aborting
20140225 12:58:20 <sgothel> 'while( !isActive ) ' in doResume is a deadlock, i.e. also needs to check 'abortion'
20140225 12:58:38 <sgothel> i.e. 'isRunning'
20140225 12:59:03 <xranby> right... this is the first error exception seen in the log "mmap() failed: Kan inte allokera minne (can not allocate memory)
20140225 12:59:03 <xranby> AL lib: (EE) connect_context: pa_context_new() failed
20140225 12:59:03 <xranby> AL lib: (EE) alcOpenDevice: Option 'format' is deprecated, please use 'channels' and 'sample-type'
20140225 12:59:03 <xranby> Thread-51-StreamWorker_4: ALCError 0xa004 while makeCurrent. ALAudioSink[init true, playRequested true, device OpenAL Soft, ctx 0x1ab0d58, alSource 133, chosen AudioDataFormat[sampleRate 22050, sampleSize 8, channelCount 1, signed false, fixedP true, packed, little-endian], al[chan Mono, type u8, fmt 0x1100, soft true], playSpeed 0.0, buffers[total 32, avail 0, queued[32, apts 0, 5957 ms, 131072 bytes], queue[g 16, l 32]
20140225 12:59:03 <xranby> java.lang.Exception: Stack trace
20140225 12:59:03 <xranby> at java.lang.Thread.dumpStack(Thread.java:1364)
20140225 12:59:03 <xranby> at jogamp.opengl.openal.av.ALAudioSink.lockContext(ALAudioSink.java:232)
20140225 12:59:04 <xranby> at jogamp.opengl.openal.av.ALAudioSink.dequeueBuffer(ALAudioSink.java:549)
20140225 12:59:04 <xranby> at jogamp.opengl.openal.av.ALAudioSink.dequeueBuffer(ALAudioSink.java:622)
20140225 12:59:05 <xranby> at jogamp.opengl.openal.av.ALAudioSink.enqueueData(ALAudioSink.java:668)
20140225 12:59:05 <xranby> at jogamp.opengl.util.av.impl.FFMPEGMediaPlayer.pushSound(FFMPEGMediaPlayer.java:822)
20140225 12:59:06 <xranby> at jogamp.opengl.util.av.impl.FFMPEGv08Natives.readNextPacket0(Native Method)
20140225 12:59:22 <sgothel> :)
20140225 12:59:37 <sgothel> while we can't help that .. we surely can remove the deadlock
20140225 13:00:14 <sgothel> it's the 'audioEOS' case .. line 1265 .. 1275
20140225 13:00:23 <sgothel> last line it goes 'down'
20140225 13:05:08 <sgothel> hmm .. why isn't this error not propagated ?
20140225 13:05:49 <sgothel> throws a RuntimeException
20140225 13:07:11 <sgothel> so readNextTextureImpl .. call in line 1238 should have kicked out w/ above exception ..
20140225 13:07:51 <sgothel> or .. did the JNI code somewhat prohibit that .. ? not to my knowledge. i.e. here we: Java -> JNI -> Java -> Exception
20140225 13:14:24 <sgothel> @Xerxes: In doResume .. do a 'while( !isActive && !shallPause && isRunning ) {'
20140225 13:15:03 <sgothel> doPause: while( isActive && !shallPause && isRunning ) {
20140225 13:15:09 <sgothel> .. and maybe you can test again ..
20140225 13:15:09 <sgothel> should remove the deadlock
20140225 13:15:14 <sgothel> if so .. pls make a git commit
20140225 13:15:14 <sgothel> this should do the trick
20140225 13:15:33 <sgothel> still don't know 'who is catching the exception .. hmm'
20140225 13:18:14 <xranby> i am on it.. i have updated GLMediaPlayerImpl doPause and doResume now..
20140225 13:18:20 <sgothel> JVM/JNI bug ?
20140225 13:23:10 <xranby> according to http://www.ibm.com/developerworks/library/j-jtp0924/ - "RuntimeException is the leading cause of thread death" thus if i understand correctly a runtime exception only kills the thread that catches it
20140225 13:26:33 <sgothel> hu ?
20140225 13:26:51 <sgothel> me talking about why isn't StreamWorker catching the the exception ..
20140225 13:27:03 <sgothel> (A caught exception does not kill anybody here)
20140225 13:27:47 <sgothel> GLMEdiaPlayerImpl line 1282 .. should have caught it .. but it seems it's not propagated properly
20140225 13:28:48 <sgothel> hmm .. even in such case, we don't stop stream-worker .. quite tolerant :)
20140225 13:29:06 <sgothel> yes, user would receive the exception then ..
20140225 13:31:07 <sgothel> doPause only: while( isActive && isRunning ) {
20140225 13:32:36 <monsieur_max> not related to jogamp but somewhat interesting for reading at high speed : http://www.spritzinc.com/#
20140225 13:33:25 <sgothel> about ? (before we go through hoops and copy/paste .. :)
20140225 13:34:38 <monsieur_max> well, it's a website that show a kind of POC about displaying words at high speed and still allowing you to read
20140225 13:35:08 <monsieur_max> making it easier to read at 500words per minutes, faster that what is supposedly a mean of 210 wpm
20140225 13:36:18 <monsieur_max> ( totally not related to jogamp, but well, I find it somewhat interesting, sorry if that really does not belong here )
20140225 13:36:45 <sgothel> na .. I just got your text wrong .. i.e. 'read this in high speed ' :)
20140225 13:36:56 <sgothel> it's somewhat graphics/UI related ..
20140225 13:37:11 <sgothel> still I don't see it .. NoScript .. allow-all-this-page .. nada
20140225 13:38:58 <xranby> my build is acting strange on this machine.. the jogamp autobuilds can initialize openal while my own builds only get the nullaudiosink.. hmm i will test this https://github.com/xranby/jogl/commit/ac7de4a7b2cc939312cbeb88dd709a297c96c13c later at home
20140225 13:39:08 <sgothel> the last thing I read about conscious reading .. and memorizing it well, is to make it HARD to read (fonts .. etc) :)
20140225 13:39:35 <monsieur_max> sgothel: ah yes i read something like that too
20140225 13:39:37 <sgothel> @Xerxes: Always .. your git git :)
20140225 13:41:16 <sgothel> xorg naming tags: "Golden Gaytime" -> "Szechuan Hot Pot" :)
20140225 13:46:58 * monsieur_max1 (~maxime@anon) has joined #jogamp
20140225 13:47:09 * monsieur_max (~maxime@anon) Quit (Write error: Broken pipe)
20140225 13:47:34 * monsieur_max1 is now known as monsieur_max
20140225 13:48:06 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20140225 13:55:19 <hija> close to bug 1000...
20140225 13:55:24 <hija> what will it be??
20140225 13:57:41 <sgothel> must something .. err .. marvelous
20140225 13:58:42 * monsieur_max can't wait for the marvelous
20140225 15:25:39 * hija (~hija@anon) Quit (Ping timeout: 367 seconds)
20140225 16:00:20 * jvanek (jvanek@anon) has joined #jogamp
20140225 16:57:35 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140225 17:20:59 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20140225 17:48:48 * monsieur_max (~maxime@anon) has joined #jogamp
20140225 18:11:38 * magaio (~magaio@anon) Quit (Ping timeout: 264 seconds)
20140225 18:13:32 * magaio (~magaio@anon) has joined #jogamp
20140225 18:22:01 * sgothel (~sgothel@anon) Quit (Excess Flood)
20140225 18:22:10 * kermyt_ (~kermyt@anon) has joined #jogamp
20140225 18:22:22 * sgothel (~sgothel@anon) has joined #jogamp
20140225 18:22:22 * ChanServ sets mode +v sgothel
20140225 18:22:26 * kermyt (~kermyt@anon) Quit (Excess Flood)
20140225 18:22:35 * kermyt_ is now known as kermyt
20140225 21:59:20 * phao (~phao@anon) Quit (Read error: Network is unreachable)
20140225 22:02:56 * monsieur_max (~maxime@anon) has left #jogamp
20140226 05:05:49 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20140226050549.html