#jogamp @ irc.freenode.net - 20140817 05:05:29 (UTC)

20140817 05:05:29 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20140816050529.html
20140817 05:05:29 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20140817050529.html
20140817 06:21:35 * rmk0 (~rmk0@anon) Quit (Ping timeout: 244 seconds)
20140817 06:22:36 * rmk0 (~rmk0@anon) has joined #jogamp
20140817 07:15:40 * monsieur_max (~maxime@anon) has joined #jogamp
20140817 07:59:44 * doev (~doev@anon) has joined #jogamp
20140817 08:09:34 * doev (~doev@anon) Quit (Ping timeout: 244 seconds)
20140817 09:02:37 * doev (~doev@anon) has joined #jogamp
20140817 09:46:57 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 09:47:46 <xranby_ac100> gouessej: raspberry pi test results https://gist.github.com/xranby/04a30c83dcab9d784d82
20140817 09:52:39 * doev (~doev@anon) Quit (Ping timeout: 246 seconds)
20140817 10:30:15 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140817 10:31:18 * monsieur_max (~maxime@anon) has joined #jogamp
20140817 10:31:18 * monsieur_max (~maxime@anon) Quit (Client Quit)
20140817 11:03:21 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 246 seconds)
20140817 11:03:47 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 12:06:00 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 246 seconds)
20140817 12:07:19 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 13:24:46 * Eclesia (~eclesia@anon) has joined #jogamp
20140817 13:26:01 * Eclesia (~eclesia@anon) Quit (Client Quit)
20140817 13:27:07 * zzuegg| (zzuegg@anon) has joined #jogamp
20140817 13:28:32 * zzuegg (zzuegg@anon) Quit (Ping timeout: 245 seconds)
20140817 13:46:17 * zzuegg| (zzuegg@anon) Quit (Read error: Connection reset by peer)
20140817 13:49:06 * zzuegg (zzuegg@anon) has joined #jogamp
20140817 14:24:08 * monsieur_max (~maxime@anon) has joined #jogamp
20140817 15:04:36 * whiptail (~whiptail@anon) has joined #jogamp
20140817 15:04:56 <whiptail> Hey guys. Can anyone give me a hand getting gluegen-rt and jocl working in an Eclipse project?
20140817 15:19:03 <whiptail> Oh it's ok, figured out what was wrong. For some crazy reason jocl.jar is a zip file that contains the ACTUAL jar. Crazy.
20140817 15:21:19 <rmk0> whiptail: ?!
20140817 15:21:38 <rmk0> where'd this jocl jar come from?
20140817 15:22:16 * rmk0 engages full panic mode
20140817 15:22:54 <rmk0> the jocl jar on maven central is clean at least...
20140817 15:22:57 * rmk0 deflats
20140817 15:22:59 <rmk0> ...
20140817 15:22:59 <whiptail> http://jogamp.org/deployment/v2.2.0/jar/jocl.jar
20140817 15:23:00 * rmk0 deflates
20140817 15:23:17 * rmk0 grabs it
20140817 15:23:42 <rmk0> erm...
20140817 15:24:39 <rmk0> http://waste.io7m.com/2014/08/17/jocl.jar.txt
20140817 15:25:10 <whiptail> To get stuff to run, I need jocl.jar, gluegen-rt.jar, and all the jocl and gluegen natives for the architectures I want it to run on right?
20140817 15:25:20 <rmk0> yep
20140817 15:26:06 <rmk0> there is an archive that contains them all, if getting them all separately is annoying: http://jogamp.org/deployment/v2.2.0/archive/jogamp-all-platforms.7z
20140817 15:27:14 <whiptail> When I pop it in my Eclipse project and add to build path, I get: Archive for required library: 'lib/jocl.jar' in project 'WaveStudio' cannot be read or is not a valid ZIP file
20140817 15:27:30 <whiptail> And opening it up in xarchiver shows an embedded file called jocl.jar, and if I pull that out and use it instead, stuff works.
20140817 15:27:39 <rmk0> that is truly bizarre
20140817 15:28:11 <rmk0> the jocl.jar is definitely not in that state on the server, so it seems that it's being mangled on your end
20140817 15:28:21 <rmk0> how are you downloading the jar file?
20140817 15:28:23 <whiptail> The new jocl.jar is 227kb with the original being 201kb so perhaps you're using some hyperagressive compression?
20140817 15:29:53 <rmk0> the file on the server is 265K...
20140817 15:30:22 <rmk0> 233170 bytes, sha256sum is 88df286f7f80913b8ade85ba479a461d92a2d28ecd2289046b3bd2467f42bc13
20140817 15:32:07 <whiptail> ... da fuck. wget gets a file that's 233170 bytes... right-click save as in firefox, gets one that's 205982 bytes...
20140817 15:32:52 <rmk0> sounds like an issue with caching
20140817 15:33:44 <whiptail> I've never seen firefox ever mangle a file like that before :O
20140817 15:34:19 <rmk0> it does happen... i think they typically compound the problem by then caching the result of the failed transfer
20140817 15:34:28 <rmk0> so if you try downloading again, you get the same broken copy
20140817 15:34:35 <whiptail> It's not failed though, it gave me a valid file, just double compressed.
20140817 15:34:46 <rmk0> yeah.... not seen that!
20140817 15:34:56 <whiptail> Perhaps there's a proxy in the way, and that proxy deflated or 7zipped it or something :S
20140817 15:35:18 <rmk0> there isn't a proxy in front of jogamp.org, that i'm aware
20140817 15:35:26 <rmk0> but, could be one elsewhere, certainly
20140817 15:35:29 <whiptail> Mebbe my isp. Or the NSA.
20140817 15:36:01 <whiptail> Anyway, I may be back shortly, with questions about how to get buffers populated and working ;D
20140817 15:36:28 <whiptail> Not used NIO before and looks like that's key to passing stuff in and out of kernels
20140817 15:36:44 <rmk0> i've not used jocl, personally... don't have hardware that supports it
20140817 15:36:47 <rmk0> so i may not be much use
20140817 15:37:00 <rmk0> someone in here may have
20140817 15:37:38 <rmk0> a few people aren't back from SIGGRAPH, i think
20140817 15:38:13 <rmk0> if the jocl API is anything like jogl, you should be able to pass in kernels as direct FloatBuffers
20140817 15:58:52 <xranby_ac100> whiptail: its a known issue that firefox saves the file in gz compressed format it it have a choice of grabbing it using gz compression from the apacheserver
20140817 16:04:28 <xranby_ac100> whiptail: rmk0: https://bugzilla.mozilla.org/show_bug.cgi?id=610679
20140817 16:04:48 * rmk0 eyes it
20140817 16:05:26 <rmk0> heh, impressive
20140817 16:07:14 <xranby_ac100> firefox ask the jogamp server,, hi server do you support sending files in gz format for faster download times?, the jogamp server respond, sure
20140817 16:07:30 <xranby_ac100> and then firefox forgets to uncompress it
20140817 16:19:36 <whiptail> Thanks xrandby. First time I've ever had firefox mangle a download before :|
20140817 16:21:18 <xranby_ac100> whiptail: its a never ending argumentation, firefox developers consider it a configuration error to give offer compressed downloads of compressed files
20140817 16:21:18 <xranby_ac100> while the apache server developers consider it a client issue
20140817 16:22:21 <xranby_ac100> personally i side with the apache server developers
20140817 16:22:24 <rmk0> seems pretty clear that it's firefox's fault
20140817 16:22:59 <whiptail> If a server gives you a stream of bytes, indicates it's compressing it, then yeah, the client at the other end should uncompress that stream before handing it onto the user. Original file type is immaterial.
20140817 16:25:00 <xranby_ac100> whiptail: are you targeting opencl use on mobile? (jocl can do that)
20140817 16:26:07 <whiptail> Nah, desktop machines. Ooh! Any idea why this crashes the jvm:
20140817 16:26:14 <whiptail> http://pastebin.com/fewYZtHU
20140817 16:26:32 * xranby_ac100 eyes it
20140817 16:26:46 <whiptail> I'm pretty confused about the whole buffer aspect of things, and the use/copy/allocate options.
20140817 16:26:52 <xranby_ac100> can you make a complete example?
20140817 16:26:55 <whiptail> Trying to pass some floats into a kernel.
20140817 16:27:13 <xranby_ac100> that i can download compile and know what command line arguments you us
20140817 16:27:14 <xranby_ac100> es
20140817 16:28:39 <whiptail> I'm probably picking the wrong permutation of options for the buffer :<
20140817 16:30:07 <whiptail> Hunting down some examples on the site'll probably show me the light.
20140817 16:30:19 <xranby_ac100> if you can crash the jvm then please create a junit test that we can include in the automated testing of jocl
20140817 16:30:48 <whiptail> Haha ok ;D
20140817 16:30:57 <xranby_ac100> that is the best way of getting bugs fixed
20140817 16:36:33 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 246 seconds)
20140817 16:41:21 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 17:34:32 <whiptail> Aw no. The 32bit and 64bit libs are both called exactly the same thing :|
20140817 17:34:49 <whiptail> So when it builds a jar containing them all, they get overwritten :|
20140817 17:38:45 <whiptail> Really ought to call them jocl-x86.dll and jocl-x64.dll or something :V
20140817 17:46:17 <rmk0> whiptail: http://jogamp.org/log/irc/jogamp_20140816050529.html#l320
20140817 17:46:48 <whiptail> Ah cool, thanks ;D
20140817 17:47:35 <rmk0> did you see JOGL at SIGGRAPH, by any chance?
20140817 17:47:55 <rmk0> and/or is this thing you're working on part of a university course?
20140817 17:48:17 <whiptail> Nah, I been playing with opencl on java for a few weeks. Used another library called jocl that was pretty much just a wrapper for the C functions, but that kept crashing randomly.
20140817 17:48:29 <rmk0> ah, right
20140817 17:49:11 <rmk0> we had someone in here yesterday with similar deployment questiones
20140817 17:49:13 <rmk0> -e
20140817 17:49:22 <rmk0> was wondering if we were receiving graphics students one at a time
20140817 17:50:21 <whiptail> Haha nah I'm working on stuff that visualises wave interference patterns on surfaces, and opencl renders it a hundred times faster than cpu can.
20140817 17:50:39 <rmk0> nice
20140817 17:54:29 <whiptail> Do I need the invidual TAG classes in each native jar too, or is that just used for some reflection magic or something?
20140817 17:56:53 <rmk0> if you're arranging the native libraries in the natives/ directory inside the jar, you don't need the tags
20140817 17:57:00 <whiptail> Cool :D
20140817 17:57:09 <rmk0> they're used for finding natives if you were putting the individual natives jar files on the classpath (as maven would)
20140817 17:58:04 <rmk0> .. might be worth distributing a fat jogl/gluegen jar
20140817 17:58:11 <rmk0> seems like a lot of people want it
20140817 17:58:37 <rmk0> as in, one with the natives already arranged that could simply be merged with whatever jars the programmer's using
20140817 17:58:57 <rmk0> think people expect to be able to produce one-jars and are surprised when it doesn't work
20140817 18:00:22 <whiptail> Mine worked, but the 32 and 64 bit versions munged over each other, so it woulda broke if someone using a 32 bit machine tried running it
20140817 18:01:02 <whiptail> Or if the wind changed direction and eclipse decided to change the order it packed the jar.
20140817 18:01:35 * gouessej (52794fc6@anon) has joined #jogamp
20140817 18:02:00 <gouessej> whiptail, your web browser did that
20140817 18:02:07 <gouessej> for "security" concerns
20140817 18:03:22 <gouessej> rmk0: it has nothing to do with this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=610679
20140817 18:04:13 <gouessej> Several web browsers on some platforms wraps JARs into ZIP archives surreptitiously
20140817 18:04:22 <gouessej> Firefox, Chrome
20140817 18:06:37 <gouessej> Actually, it is a nice way of preventing "executable" JARs to work out of the box
20140817 18:06:47 <gouessej> "from working"
20140817 18:06:52 <gouessej> I'm tired
20140817 18:07:30 <gouessej> You know, this is always the same mantra: "use HTML5 or die"
20140817 18:08:44 <gouessej> xranby_ac100: I don't know yet why JogAmp's Ardor3D Continuation shows a black screen with ES 1 :(
20140817 18:09:33 * gouessej (52794fc6@anon) Quit (Quit: Page closed)
20140817 18:57:08 * zzuegg| (zzuegg@anon) has joined #jogamp
20140817 18:57:42 * zzuegg (zzuegg@anon) Quit (Ping timeout: 245 seconds)
20140817 19:02:08 * zzuegg| (zzuegg@anon) Quit ()
20140817 19:39:04 <xranby_ac100> gouessej: run ES1 with -Djogl.debug.DebugGL and it will likely find invalid parameters. also some drivers require square textures
20140817 19:44:46 * gouessej (52794fc6@anon) has joined #jogamp
20140817 19:46:05 <gouessej> xranby_ac100: I already do it. In my humble opinion, there is no texture bound and there is no vertex sent to the graphics card. It will work a bit better when the bugs 1045 and 1047 are fixed.
20140817 19:46:36 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 246 seconds)
20140817 19:48:03 <gouessej> It would be nice if someone knowing very well OpenGL 3 would review some of my code. I think that tons of features are build-in in OpenGL 3, JoglContextCapabilities should be deeply modified to take it into account
20140817 19:49:03 * gouessej (52794fc6@anon) Quit (Client Quit)
20140817 19:49:26 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 20:45:45 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 246 seconds)
20140817 20:46:44 * xranby_ac100 (~xranby@anon) has joined #jogamp
20140817 21:02:00 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140818 01:28:06 * whiptail (~whiptail@anon) Quit (Quit: Leaving)
20140818 05:05:29 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20140818050529.html