#jogamp @ irc.freenode.net - 20131212 05:05:57 (UTC)


20131212 05:05:57 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20131211050557.html
20131212 05:05:57 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20131212050557.html
20131212 07:03:22 * sgothel (~sgothel@anon) Quit (Quit: Leaving.)
20131212 07:11:20 * sgothel (~sgothel@anon) has joined #jogamp
20131212 07:11:20 * ChanServ sets mode +v sgothel
20131212 07:11:50 * sgothel (~sgothel@anon) Quit (Remote host closed the connection)
20131212 07:31:41 * sgothel (~sgothel@anon) has joined #jogamp
20131212 07:31:41 * ChanServ sets mode +v sgothel
20131212 08:03:16 <sgothel> http://jogamp.org/deployment/archive/master/gluegen_756-joal_506-jogl_1175-jocl_889/ <- will be 2.1.3 later today .. have to go now
20131212 08:06:42 * xranby (~xranby@anon) Quit (Ping timeout: 246 seconds)
20131212 08:11:12 * xranby (~xranby@anon) has joined #jogamp
20131212 08:17:33 * monsieur_max (~maxime@anon) has joined #jogamp
20131212 08:36:49 * hija (~hija@anon) has joined #jogamp
20131212 09:09:39 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131212 09:34:15 * hija (~hija@anon) Quit (Quit: hija)
20131212 09:54:03 * monsieur_max1 (~maxime@anon) has joined #jogamp
20131212 10:12:07 * hija (~hija@anon) has joined #jogamp
20131212 11:01:31 * monsieur_max1 is now known as monsieur_max
20131212 12:07:25 <rmk0> sgothel: i commented in 852 before i spoke here
20131212 12:07:37 <rmk0> and 'lo
20131212 12:28:41 <rmk0> having trawled through the ES3 spec, i'm reasonably convinced it's a mistake to make ES2 and ES3 contexts arbitrarily incompatible
20131212 12:28:50 <rmk0> i can't find any evidence that they're not compatible
20131212 12:29:15 <rmk0> the end result is obviously 915... we're rejecting both ES* contexts on mesa/amd for no good reason
20131212 13:34:52 * hija (~hija@anon) Quit (Quit: hija)
20131212 15:22:12 * rmk0 (~rmk0@anon) Quit (Quit: Lost terminal)
20131212 16:46:19 <hharrison> sgothel: we'll get 2.1.3 into production next week and see how it fares
20131212 16:55:40 * hija (~hija@anon) has joined #jogamp
20131212 16:55:52 * hija (~hija@anon) Quit (Client Quit)
20131212 16:57:20 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131212 17:48:12 * monsieur_max (~maxime@anon) has joined #jogamp
20131212 18:00:19 * rmk0 (~rmk0@anon) has joined #jogamp
20131212 18:00:19 * rmk0 (~rmk0@anon) Quit (Changing host)
20131212 18:00:19 * rmk0 (~rmk0@anon) has joined #jogamp
20131212 18:20:19 <sgothel> Back - @Mark: So you understand that 'CPU source not allowed in ES3 spec' right? -> Not compatible w/ ES2
20131212 18:21:03 <rmk0> sgothel: where does it say this?
20131212 18:21:08 <rmk0> i've tried to find that... can't
20131212 18:21:27 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=852
20131212 18:21:38 <sgothel> OpenGL 3.1 core spec 2.10 p41 Vertex Array Objects "An INVALID_OPERATION error is generated if any of the *Pointer commands specifying the location and organization of vertex array data are called while zero is bound to the ARRAY_BUFFER buffer object binding point, and the pointer argu- ment is not NULL."
20131212 18:21:50 <rmk0> that's opengl 3.1, not ES3
20131212 18:21:56 <rmk0> see my references to the ES3 spec
20131212 18:22:05 <rmk0> it seems to explicitly allow client side cpu sourced data
20131212 18:23:17 <sgothel> indices are diff .. let me double check w/ ES3 spec .. will do that later, IMHO ES3 is less than GL3-core ..
20131212 18:23:41 <rmk0> i don't really care either way, i'd never personally use the cpu sourced stuff
20131212 18:23:59 <rmk0> but if it turns out that ES3 *is* actually compatible, then we're rejecting compatible contexts
20131212 18:24:06 <rmk0> if it's not compatible, then mesa need to fix their side
20131212 18:24:13 <rmk0> i just need to find "proof" either way
20131212 18:26:38 <rmk0> i'm pretty sure that page 53 makes the case that client-side arrays are still supported
20131212 18:27:00 <rmk0> pdf page 53, that is... text page 41
20131212 18:27:37 <sgothel> your are perfectly right ..
20131212 18:27:44 <sgothel> and then we also have 2.1.4 :)
20131212 18:28:01 <sgothel> so let me read the spec .. it would simplify things ..
20131212 18:28:19 <rmk0> there's still the issue that mesa aren't setting the correct bits in their context, so i'll hassle them on that
20131212 18:28:31 <sgothel> pls .. that for sure is an issue
20131212 18:28:55 <sgothel> also .. assume 'cpu sourced' will be gone w/ ES4 .. than it will benefit to still not include it in the common profiles ..
20131212 18:29:08 <sgothel> i.e. GL4ES3 .. etc
20131212 18:29:09 <rmk0> let's hope
20131212 18:29:21 <rmk0> what sick people use client-side arrays these days...
20131212 18:29:29 <rmk0> sick!
20131212 18:29:47 <sgothel> it's a convenience .. impl. does the same ofc - create vbo, copy data .. blabla
20131212 18:30:02 <sgothel> so .. since core profiles shall be close to the metal .. it should be gone
20131212 18:30:08 <rmk0> sick, i say!
20131212 18:30:14 <sgothel> indicating: dude, this is not for free :)
20131212 18:30:18 <rmk0> yep
20131212 18:31:19 <sgothel> @Xerxes: Happy w/ GLMediaPlayer changes ?
20131212 18:31:30 <sgothel> .... and @Max ofc ..
20131212 18:32:21 <hharrison> sgothel: morbidly curious, did you intend people to use GL2GL3 in practice?
20131212 18:32:54 <hharrison> As it turns out, we liked that subset so much, we've essentially been using it everywhere as opposed to anything more specific
20131212 18:33:01 <sgothel> yes, a few years when I introduced the profiles .. one person was asking for it
20131212 18:33:11 <sgothel> around .. err .. 2009 or so
20131212 18:33:24 <hharrison> ....To much to ask to get a DebugGL2GL3 added?
20131212 18:33:37 <hharrison> Otherwise I've been doing:
20131212 18:33:50 * monsieur_max (~maxime@anon) Quit (Ping timeout: 240 seconds)
20131212 18:33:56 <hharrison> new DebugGL4bc((GL4bc)_sharedContext.getGL().getGL2GL3()
20131212 18:34:30 <sgothel> right .. that should do it ..
20131212 18:34:51 <hharrison> Just making sure that would wrk...the casting
20131212 18:35:08 <sgothel> the others are actually now doing the same, i.e. they are 'fakes' :)
20131212 18:35:09 <hharrison> I mean, it works in practice...just wondering if you saw something that would break eventually
20131212 18:35:13 <sgothel> so we can add it .. ofc
20131212 18:35:43 <hharrison> If it is safe...having a Debug4bc constructor that takes a GL2GL3 would be perfect for my purposes
20131212 18:36:11 <sgothel> oh ..I thought the other way around ..
20131212 18:36:12 <hharrison> Then if that ever becomes unsafe, you can kill the constructor which will let me notice something changed
20131212 18:36:36 <sgothel> DebugGL4bc a1;
20131212 18:36:47 <sgothel> GL2GL3 a2 = a1.getGL2GL3();
20131212 18:37:39 <sgothel> public class DebugGL3 extends DebugGL4bc {
20131212 18:37:39 <sgothel> public DebugGL3(GL3 downstream) {
20131212 18:37:39 <sgothel> super((GL4bc)downstream);
20131212 18:37:39 <sgothel> }
20131212 18:37:39 <sgothel> }
20131212 18:37:40 <sgothel> :)
20131212 18:38:16 <hharrison> Ahhh
20131212 18:38:27 <sgothel> so we only have the high DebugGL4bc .. in code -> expensive
20131212 18:38:36 <sgothel> so overall .. we save some bytes
20131212 18:38:41 * monsieur_max (~maxime@anon) has joined #jogamp
20131212 18:38:41 * monsieur_max (~maxime@anon) Quit (Client Quit)
20131212 18:38:48 <hharrison> That's a better way of providing the constructor I want
20131212 18:39:31 * hija (~hija@anon) has joined #jogamp
20131212 18:39:37 <sgothel> so you can add the other ones .. I pull and merge
20131212 18:40:05 <sgothel> (always add the Trace class then as well .. )
20131212 18:40:51 * monsieur_max (~maxime@anon) has joined #jogamp
20131212 18:42:36 <hharrison> Sure, I'm not super motivated if my casting above is safe, but you may see an add DebugGL2GL3 patch go by
20131212 18:43:05 <hharrison> Currently dealing with bugreports from all our new users that aren't in a north-american locale
20131212 18:43:35 <hharrison> Finding every single place that formats numbers dealing with ',' vs'.' as the numeric separator sure is fun
20131212 18:44:28 <rmk0> hharrison: have any of these string reduction patches made their way into a maven version yet?
20131212 18:44:52 <rmk0> i was testing on windows 7 a few days ago and noticed pretty heavy allocations of char[] in Debug contexts
20131212 18:45:11 <rmk0> am assuming that's what you're working to remove
20131212 18:46:04 <hharrison> Should already be in 2.1.2
20131212 18:46:08 <hharrison> Lemme check
20131212 18:46:24 <rmk0> i've been stuck on 2.1.0 for a while due to the context issues
20131212 18:47:56 <hharrison> It went in between 2.1.1, 2.1.2
20131212 18:48:22 <hharrison> Easliy cherry-picked onto 2.1.0 if you want to DIY
20131212 18:48:50 <rmk0> is no problem, i can wait
20131212 18:49:04 <rmk0> it's not so bad that it's noticeable outside of visualvm
20131212 18:49:08 <rmk0> was just... a bit alarming!
20131212 18:49:47 * hija (~hija@anon) Quit (Quit: hija)
20131212 18:50:06 <hharrison> Yeah, that's what got me onto it, we always run with the debug contexts now
20131212 18:50:38 <hharrison> Sure, you are doubling up the number of gl calls you make, but it's really helpful when stuff does go wrong
20131212 18:51:56 * [Mike] is now known as esemplare
20131212 18:52:04 * esemplare is now known as [Mike]
20131212 18:52:45 <rmk0> i hear that
20131212 18:56:31 <sgothel> (07:46:24 PM) rmk0: i've been stuck on 2.1.0 for a while due to the context issues <- If this is really a blocker .. we should have a workaround, didn't realize it is.
20131212 18:57:15 <rmk0> it's more that i like to make sure the library i'm writing works on ES2
20131212 18:57:24 <rmk0> so i couldn't upgrade to a version that denied access to it
20131212 18:57:32 <rmk0> i assumed it'd be fixed sooner or later, so it wasn't a huge problem
20131212 18:57:41 <sgothel> well, if you have ES3 .. you are fine
20131212 18:58:05 <rmk0> well with this mesa bug, i didn't have either of them
20131212 18:58:11 <sgothel> all we need is a workaround for mesa's es3 bit maybe
20131212 18:58:37 <rmk0> am currently talking them through it on their bug tracker
20131212 18:58:50 <rmk0> they seem to think the issue's in gallium's EGL
20131212 18:59:01 <rmk0> am trying to compile a version with it disabled now, to test
20131212 18:59:01 <sgothel> now I do the 2.1.3 duties .. maven, signing .. then I check the stuff
20131212 18:59:08 <sgothel> sounds reasonable, since intel works
20131212 18:59:19 <sgothel> but I thought mesa-soft also works .. ?
20131212 18:59:29 <rmk0> haven't tried that yet
20131212 19:01:17 <rmk0> i'm sure you know what's coming next... what's the correct way to make JOGL see my custom compiled libgl? (i don't want to replace the system one)
20131212 19:01:57 <sgothel> LD_LIBRARY_PATH
20131212 19:02:27 <rmk0> ah, right
20131212 19:02:28 <sgothel> jogl/make$ vi scripts/tests.sh
20131212 19:02:38 <rmk0> the infamous tests.sh
20131212 19:02:43 <sgothel> line 51 :)
20131212 19:02:54 * hija (~hija@anon) has joined #jogamp
20131212 19:03:04 <sgothel> you see .. it's all in there :)
20131212 19:03:13 <sgothel> #export LD_LIBRARY_PATH=`pwd`/lib/external/mesa_900/x86_64-linux-gnu:$LD_LIBRARY_PATH
20131212 19:03:13 <sgothel> #export LIBGL_DRIVERS_PATH=`pwd`/lib/external/mesa_900/x86_64-linux-gnu/dri
20131212 19:03:15 <sgothel> for example ..
20131212 19:03:18 <rmk0> yep
20131212 19:03:30 <sgothel> and below the env for mesa stuff
20131212 19:05:10 <sgothel> so .. thank you for triaging the Mesa bug and communicating it, I know .. time consuming stuff w/ lots of diplomacy involved
20131212 19:05:47 <rmk0> hehe
20131212 19:06:01 <rmk0> is nice to have the two bug trackers
20131212 19:06:04 <rmk0> can link between the two
20131212 19:06:28 <rmk0> https://bugs.freedesktop.org/show_bug.cgi?id=72600
20131212 19:06:42 <rmk0> ... will add that to the jogl tracker, if i haven't already
20131212 19:07:32 <rmk0> oh, actually linked there from 915
20131212 19:14:00 <sgothel> transfer from jogamp.org .. only ~50kB/s .. so many lechers ?
20131212 19:14:39 <sgothel> nope
20131212 19:15:36 <sgothel> can you try to download .. e.g. http://jogamp.org/deployment/archive/master/gluegen_756-joal_506-jogl_1175-jocl_889/archive/jogamp-all-platforms.7z
20131212 19:15:48 <sgothel> and dump the 'speed' ?
20131212 19:16:09 <rmk0> i only get 400kb/s down here max, so i'm probably not the best guide
20131212 19:16:15 <sgothel> oh
20131212 19:16:19 <sgothel> Harvey ?
20131212 19:16:43 <sgothel> should be more like 1-2 MB/s
20131212 19:16:58 <rmk0> i do get 400kb/s, at least
20131212 19:17:55 <sgothel> oh .. so it might be trouble w/ me .. hmm
20131212 19:18:21 <sgothel> I see a 41MB/s now ..
20131212 19:18:27 <sgothel> (from jogamp.org ..)
20131212 19:18:35 <sgothel> (on jogamp.org I mean)
20131212 19:18:41 <sgothel> so .. it's me :(
20131212 19:18:53 <hharrison> Yeah, it was plenty-fast for me
20131212 19:19:31 <sgothel> throttled by internet provider just to this address .. hmm .. or by server provider
20131212 19:21:26 * xranby (~xranby@anon) Quit (Ping timeout: 240 seconds)
20131212 19:22:10 * xranby (~xranby@anon) has joined #jogamp
20131212 19:22:16 * hija (~hija@anon) Quit (Quit: hija)
20131212 19:38:29 <sgothel> http://jogamp.org/deployment/archive/rc/v2.1.3/
20131212 19:39:40 * sgothel (~sgothel@anon) Quit (Remote host closed the connection)
20131212 19:51:50 * xranby (~xranby@anon) Quit (Ping timeout: 264 seconds)
20131212 19:58:06 * xranby (~xranby@anon) has joined #jogamp
20131212 20:03:23 * hija (~hija@anon) has joined #jogamp
20131212 20:08:46 <rmk0> sigh... can't get anything to behave
20131212 20:09:06 <rmk0> i build mesa without egl_gallium, i get no egl at all, apparently
20131212 20:40:51 * sgothel (~sgothel@anon) has joined #jogamp
20131212 20:40:51 * sgothel (~sgothel@anon) Quit (Changing host)
20131212 20:40:51 * sgothel (~sgothel@anon) has joined #jogamp
20131212 20:40:51 * ChanServ sets mode +v sgothel
20131212 20:41:13 <sgothel> (was cut off ..)
20131212 20:48:26 * hija (~hija@anon) Quit (Quit: hija)
20131212 20:49:00 * sgothel (~sgothel@anon) has left #jogamp
20131212 20:49:07 * sgothel (~sgothel@anon) has joined #jogamp
20131212 20:49:08 * ChanServ sets mode +v sgothel
20131212 20:50:40 * sgothel (~sgothel@anon) Quit (Quit: Leaving.)
20131212 20:50:56 * sgothel (~sgothel@anon) has joined #jogamp
20131212 20:50:56 * ChanServ sets mode +v sgothel
20131212 20:51:32 <sgothel> jogamp.org maybe analyzed .. and hence rebooted/rescue-system etc .. in a bit for a few minutes