#jogamp @ irc.freenode.net - 20140304 05:05:51 (UTC)


20140304 05:05:51 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20140303050550.html
20140304 05:05:51 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20140304050551.html
20140304 07:24:33 * phao (~phao@anon) has joined #jogamp
20140304 08:21:30 * monsieur_max (~maxime@anon) has joined #jogamp
20140304 10:20:19 * hija (~hija@anon) has joined #jogamp
20140304 11:24:31 * jvanek (jvanek@anon) has joined #jogamp
20140304 15:37:55 * phao (~phao@anon) Quit (Quit: Fui embora)
20140304 16:59:17 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20140304 17:20:35 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20140304 18:51:08 * hija (~hija@anon) Quit (Quit: hija)
20140304 18:58:01 * zubzub (~zubzub@anon) has joined #jogamp
20140304 18:58:31 <zubzub> is there a working maven repo for newt?
20140304 18:58:46 <zubzub> all the jars from the official maven repo and the jogamp repo only contain the manifest
20140304 18:58:56 <zubzub> for all the versions of nativewindow at least
20140304 18:59:07 <rmk0> ...
20140304 18:59:15 <rmk0> if that's the case, that shouldn't be happening
20140304 18:59:17 * rmk0 eyes it
20140304 19:00:55 <rmk0> zubzub: ah, d'you mean "newt.jar"?
20140304 19:01:22 <zubzub> well the nativewindow jar actually
20140304 19:01:39 <zubzub> but it seems the newt.jar seems to suffer from the same problem afaik
20140304 19:01:46 <rmk0> yes, right
20140304 19:01:50 <zubzub> judging from its size
20140304 19:02:00 <rmk0> what's going on there is that the jogamp project doesn't actually distribute a newt.jar or nativewindow.jar
20140304 19:02:12 * monsieur_max (~maxime@anon) has joined #jogamp
20140304 19:02:15 <rmk0> there's newt-core.jar, newt-event.jar, etc
20140304 19:02:17 <zubzub> :(
20140304 19:02:23 <zubzub> ah
20140304 19:02:27 <rmk0> but we have to include those empty jars to satisfy maven's naming schemes
20140304 19:02:40 <zubzub> ah so I have to add a classifier thingy in my maven dep?
20140304 19:02:45 <rmk0> yep, was just typing that
20140304 19:02:53 <zubzub> ok will do that
20140304 19:02:55 <zubzub> thx!
20140304 19:02:58 <rmk0> no problem
20140304 19:03:14 <zubzub> I see odin_ is idling this channel as well
20140304 19:03:16 <rmk0> i don't think the "atomic" jars see a lot of use, as people tend to grab the all-inclusive things
20140304 19:03:20 <zubzub> hi odin_ *waves*
20140304 19:03:26 <zubzub> yeah
20140304 19:03:26 <rmk0> at least from maven... if there are issues, let me know
20140304 19:03:29 <rmk0> i handle that side
20140304 19:04:06 <zubzub> are there any plans to support wayland?
20140304 19:04:17 <rmk0> seem to remember there being talk of it
20140304 19:04:23 <zubzub> (there are maven java bindings now)
20140304 19:04:27 <zubzub> *wayland
20140304 19:04:35 <rmk0> nice
20140304 19:11:05 <zubzub> hmmm
20140304 19:11:22 <zubzub> are those classifier jars supposed to be visible in the maven repo?
20140304 19:14:03 <rmk0> visible?
20140304 19:15:30 <rmk0> as in visible from search.maven.org?
20140304 19:17:28 <rmk0> hm, looks like they're being missed by the deployment scripts
20140304 19:17:34 * rmk0 files bug
20140304 19:24:09 <rmk0> https://jogamp.org/bugzilla/show_bug.cgi?id=991
20140304 19:36:15 <zubzub> aight thx!
20140304 19:54:45 * hija (~hija@anon) has joined #jogamp
20140304 20:32:08 <rmk0> zubzub: thanks for bringing this to my attention, there are quite a few files missing
20140304 20:32:13 <rmk0> shows how much they get used, i suppose...
20140304 20:32:18 <rmk0> the deployment scripts are getting a beating
20140304 20:39:29 * hija (~hija@anon) Quit (Ping timeout: 240 seconds)
20140304 20:40:15 * hija (~hija@anon) has joined #jogamp
20140304 21:02:57 <rmk0> sgothel: it'd be a good idea to push a new maven build to the jogamp repository using the new build scripts
20140304 21:03:11 <rmk0> seems that a lot of the atomics were missing
20140304 21:08:10 <rmk0> zubzub: http://jogamp.org/wiki/index.php/Maven#The_jogamp.org_test_repository_.28optional.29 <- that'll most likely have the missing files whenever sgothel awakes
20140304 21:08:21 <rmk0> if i've not screwed up anywhere
20140304 21:09:01 <sgothel1> sorry I was missing the conversation ..
20140304 21:09:07 <rmk0> 'lo
20140304 21:09:20 <sgothel1> deep inside font AA :)
20140304 21:09:30 <rmk0> ow
20140304 21:09:56 <sgothel1> 7f28cc15004e143f77f7b90f8e543c16f84dcff0 <- me merging ?
20140304 21:10:17 <sgothel1> I wait for one confirmation regarding 2.1.5 ..
20140304 21:10:21 <rmk0> should already be there
20140304 21:10:26 <sgothel1> probably will do a pre 2.1.5 then .. hmm
20140304 21:10:28 <sgothel1> ah .. great
20140304 21:10:39 <sgothel1> so next aggregated build will do it then ..
20140304 21:10:42 <rmk0> right
20140304 21:10:59 <sgothel1> you may also close the bug then and add it to 2.1.5 wiki page
20140304 21:11:14 <rmk0> will wait and see if everything actually made it into the repos, first!
20140304 21:11:26 <rmk0> pretty sure there's nothing missing, but someone should test it
20140304 21:11:32 <rmk0> i don't have anything that uses atomics
20140304 21:12:03 <rmk0> is just a case of someone wanting a specific atomic and not being able to get it from the repos
20140304 21:12:11 <rmk0> at this point, only jogl-cg* is missing
20140304 21:12:16 <sgothel1> ah .. so the fonts will be offered now as well, good
20140304 21:12:38 <sgothel1> they are atomic exclusive .. for a while
20140304 21:15:27 <sgothel1> pulled and pushed commit to other repos ..
20140304 21:15:46 <sgothel1> rebuilding jogl .. and make a test later
20140304 21:16:34 <sgothel1> 'fonts-p0' .. yup
20140304 21:16:48 <rmk0> what is jogl-cg?
20140304 21:17:02 <sgothel1> that horrible nvidia cg binding
20140304 21:17:18 <sgothel1> the nv proprietary glsl counterpart
20140304 21:17:21 <rmk0> ugh
20140304 21:17:43 <rmk0> i was considering it as a separate top-level project (at least in maven terms) because it had its own set of natives
20140304 21:17:48 <rmk0> is why it's still missing
20140304 21:18:05 <rmk0> can add it later if anyone demands it
20140304 21:18:08 <sgothel1> don't worry .. I don't think anybody cares
20140304 21:18:25 <rmk0> hehe
20140304 21:18:32 <sgothel1> I also fail to see a reason to need it, since you can access nv's cg via glsl interfaces
20140304 21:58:27 * phao (~phao@anon) has joined #jogamp
20140304 21:59:26 <sgothel1> @zubzub: *wayland* .. yes, sure there are plans .. also a bug report already
20140304 21:59:46 <sgothel1> you like to participate in this effort ?
20140304 22:11:38 <zubzub> sure
20140304 22:11:52 <zubzub> but currently my wayland knowledge is pretty limited
20140304 22:12:04 <zubzub> but I have good contacts with the guy that created the java wayland bindings
20140304 22:12:36 <zubzub> also my gl knowledge is close to 0 but I plan on changing that :)
20140304 22:12:55 <sgothel1> ah .. we will only have a small native chunk, like the X11 code for NEWT
20140304 22:13:27 <sgothel1> sure, Wayland is new .. I have to use some demo-code and spec as well
20140304 22:14:47 <zubzub> http://ppaalanen.blogspot.be/2012/03/what-does-egl-do-in-wayland-stack.html
20140304 22:14:56 <zubzub> this is also an interesting read
20140304 22:15:37 <sgothel1> I have to assume that somehow EGL, and maybe even GLX will digest the Wayland 'window/drawable' handle
20140304 22:16:00 <zubzub> sorta
20140304 22:16:03 <zubzub> but not glx
20140304 22:16:09 <sgothel1> ah ..
20140304 22:16:19 <zubzub> there are 0 deps on X11 in wayland
20140304 22:16:23 <rmk0> wonder if it'll improve mesa's performance...
20140304 22:16:28 <sgothel1> EGL in theory can also be used for desktop GL profiles ..
20140304 22:16:41 <rmk0> doing a few full-screen passes with blending is still far slower on mesa than the same code running on amd's blob, for example
20140304 22:16:46 <zubzub> https://github.com/jekstrand/wayland-java
20140304 22:16:50 <rmk0> can't imagine what mesa can possibly be doing that would slow down blending of all things
20140304 22:17:16 <sgothel1> you mean mesa-amd vs amd-blob ?
20140304 22:17:21 <rmk0> yeah
20140304 22:17:42 <sgothel1> hmm .. just testing text rendering modes: blending, alpha :)
20140304 22:18:06 <sgothel1> blending: just framebuffer 'combiners'
20140304 22:19:40 <sgothel1> http://jogamp.org/git/?p=jogl.git;a=commit;h=8bab15934fe63e44084294e315492420e07a902b <- review ? Especially the shader vbaa.fp (AA, fragment position) ;
20140304 22:20:41 <sgothel1> http://jogamp.org/git/?p=jogl.git;a=blob;f=src/jogl/classes/jogamp/graph/curve/opengl/VBORegion2PMSAAES2.java;h=9b15ba3192c64b86411dc5b6634c2f2ed70cb140;hb=8bab15934fe63e44084294e315492420e07a902b#l189 <- also the orthogonal window box calculation .. hmm
20140304 22:21:13 <sgothel1> works nice if not too much rotated (trapezoid projection) .. but iff .. pixel volume is too low
20140304 22:22:13 <sgothel1> hope the git commit message is comprehensive ..
20140304 22:22:25 * monsieur_max (~maxime@anon) has left #jogamp
20140304 22:22:57 <rmk0> i have to admit to not really understanding a word
20140304 22:23:16 <sgothel1> and I was so happy w/ the write-up :(
20140304 22:23:27 <rmk0> is more than i've still not even read the original graph paper /o\
20140304 22:23:42 <sgothel1> ah .. it's pretty generic
20140304 22:23:45 <sgothel1> this code
20140304 22:23:55 <sgothel1> so you have a bbox in 3d space
20140304 22:24:06 <sgothel1> 2d bounding box .. for the text
20140304 22:24:43 <sgothel1> here, we need to get the window-box as it is projected onscreen (gluProject)
20140304 22:25:07 <sgothel1> multiply that one w/ sampleCount (width and height)
20140304 22:25:08 <rmk0> right
20140304 22:25:27 <sgothel1> now .. we have a delta .. due to int vs float (the fraction)
20140304 22:25:49 <sgothel1> need to add-up that fraction to vertices for 2nd pass (FBO -> screen)
20140304 22:25:57 <sgothel1> otherwise distortion
20140304 22:26:13 <sgothel1> and so on .. and so forth
20140304 22:26:19 <rmk0> i understand \o/
20140304 22:26:32 <sgothel1> great .. hope message makes sense now ..
20140304 22:26:54 <sgothel1> so if we have rotated text like the last letters are huge .. and first tiny (trapezoid)
20140304 22:27:07 <sgothel1> then the overall pixel volume is very small .. and the AA sucks ..
20140304 22:27:26 <sgothel1> either a bug .. or wrong assumption
20140304 22:27:33 <sgothel1> need to earmark that ..
20140304 22:28:16 <sgothel1> review of whether the fragment position thingy (AA part) is properly .. and the fraction stuff is properly would be nice
20140304 22:28:33 <sgothel1> will produce a few screenshots later as well as an update of the demos to test
20140304 22:28:48 <sgothel1> so far .. looks even great w/ font-size 8 and 10
20140304 22:29:43 <sgothel1> I added a few remarks to text rendering in the Font class .. it's a science for itself :) .. so using this manual 'scene AA' is legit
20140304 22:30:09 <sgothel1> will try FXAA as well, since it seems to be able to produce good results for sample-count 2
20140304 22:30:15 <sgothel1> (i.e. 4 fragments)
20140304 22:30:53 <sgothel1> (actually they use 4+1 .. i.e. the center piece as well)
20140304 22:31:55 <sgothel1> was also playing w/ pre-multiplied-alpha (in *msaa.fp) .. however this doesn't work w/ blending
20140304 22:38:27 * kermyt (~kermyt@anon) Quit (Read error: Operation timed out)
20140304 22:44:56 * kermyt (~kermyt@anon) has joined #jogamp
20140304 22:44:56 * kermyt (~kermyt@anon) Quit (Changing host)
20140304 22:44:56 * kermyt (~kermyt@anon) has joined #jogamp
20140304 23:17:17 <phao> Using lambdas is so nice. It makes me remind me somewhat of a while back when I programmed in scheme and sml just for kicks.
20140304 23:44:45 * odin_ (~Odin@anon) Quit (Quit: Leaving)
20140304 23:59:46 * phao (~phao@anon) Quit (Read error: Connection reset by peer)
20140305 00:34:52 * phao (~phao@anon) has joined #jogamp
20140305 01:46:15 * phao (~phao@anon) Quit (Ping timeout: 244 seconds)
20140305 05:05:51 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20140305050551.html