#jogamp @ irc.freenode.net - 20130718 05:06:03 (UTC)
20130718 05:06:03 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20130717050603.html
20130718 05:06:03 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130718050603.html
20130718 08:20:15 <sgothel> good morning
20130718 08:20:55 <sgothel> @Mark: Well .. ES3 _is_ a 'subset' of 'things' .. i.e. GL3+ES3_compat or 4.3
20130718 11:15:09 <sgothel> http://jogamp.org/jogl/doc/Overview-OpenGL-Evolution-And-JOGL.html <- updated UML .. and text!
20130718 12:03:56 <rmk0> lo
20130718 12:04:00 <rmk0> down to 41 errors
20130718 12:04:26 <rmk0> not sure if any particular version of GLSL ES is mandated for 3.0, but mesa only supports GLSL ES 1.0
20130718 12:05:46 <rmk0> also... ES 3.0 doesn't have glMapBuffer, but it does appear in the ES3 interface in jogl
20130718 12:05:49 <rmk0> not sure if that's intended
20130718 12:05:59 <rmk0> it has glMapBufferRange
20130718 12:07:35 <sgothel> hmm .. thx .. ES2 and ES3 do not have MapBuffer ?
20130718 12:07:48 <sgothel> then it's an interface bug of ours .. hmm
20130718 12:08:24 <sgothel> Mesa / Intel does have glsl '#version 300 es' AFAIK
20130718 12:08:36 <sgothel> i.e. we do push that version and it works
20130718 12:08:45 <rmk0> ES2 doesn't have any buffer mapping at all, ES3 has MapBufferRange
20130718 12:08:54 <rmk0> right, have only checked mesa on AMD at the moment
20130718 12:08:56 <sgothel> otherwise it would not be ES3 ..
20130718 12:08:58 <rmk0> will try intel next
20130718 12:09:15 <sgothel> thank you - will fix the MapBuffer issues .. even if it breaks the world :)
20130718 12:09:26 <sgothel> if you find anything .. shoot at me :)
20130718 12:09:26 <rmk0> hehe
20130718 12:09:41 <rmk0> yep, still just working through the test suite
20130718 12:09:48 <rmk0> the mesa people certainly are ... focused
20130718 12:09:56 <rmk0> it really doesn't suppport anything that isn't required by the spec
20130718 12:10:05 <rmk0> i know this hardware supports RG88 textures, for example
20130718 12:10:15 <rmk0> but they aren't required by the ES 3 spec, so the ES 3 context doesn't have them
20130718 12:10:33 <rmk0> the 3.1 context does
20130718 12:11:21 <sgothel> RG88 in headers ?
20130718 12:11:36 <rmk0> haven't read the headers
20130718 12:11:52 <rmk0> am just going by the list of "required texture formats" in the ES3 spec
20130718 12:12:03 <sgothel> grep -RI RG88 stub_includes/opengl empty
20130718 12:12:17 <rmk0> yeah, it's probably not in there
20130718 12:12:20 <sgothel> stub_includes/opengl/GLES3/gl3.h:#define GL_RG 0x8227
20130718 12:12:25 <sgothel> required!
20130718 12:12:38 <sgothel> stub_includes/opengl/GLES3/gl3.h:#define GL_RG8 0x822B
20130718 12:12:42 <rmk0> i'm referring to the specifically sized RG88 format
20130718 12:13:00 <sgothel> right .. seems not to be in _any_ header!
20130718 12:13:13 <rmk0> hm
20130718 12:13:43 <rmk0> heh, weird... wonder where that came from ...
20130718 12:16:21 <sgothel> http://jogamp.org/jogl/doc/Overview-OpenGL-Evolution-And-JOGL.html <- me happy now
20130718 12:16:34 <sgothel> fixed uml layout ..
20130718 12:17:27 <sgothel> @Mark: stub_includes/opengl/GLES2/gl2ext.h:GL_APICALL void* GL_APIENTRY glMapBufferOES
20130718 12:17:40 <sgothel> so it may exist as an extension :)
20130718 12:17:53 <sgothel> (in all profiles .. )
20130718 12:18:03 <sgothel> so it is mandatory to check the API doc .. a bit :)
20130718 12:18:17 <rmk0> looks good
20130718 12:18:22 <rmk0> the UML i mean
20130718 12:18:39 <sgothel> i.e. the API doc (mouse over) shows 'where it is coming from' ..
20130718 12:18:50 <rmk0> would need to check the behaviour of that extension
20130718 12:19:05 <sgothel> well .. :)
20130718 12:19:11 <rmk0> i seem to have GL_OES_mapbuffer here, but the glMapBuffer function raises errors, so the behaviour may be subtly different
20130718 12:19:26 <rmk0> am still going through the tests, haven't investigated that one yet
20130718 12:19:36 <sgothel> AFAIK semantics have only changed on FBOs .. a bit (shared context .. etc)
20130718 12:20:20 <sgothel> we could take it out .. i.e. that extension (not subsume it) .. hmm
20130718 12:20:30 <rmk0> ah, the GL_OES_mapbuffer accepts some weird values for 'access'
20130718 12:21:02 <sgothel> I was also quite cocky adding all [vendor] extensions of ES2 to ES3 .. seems naturally to me :)
20130718 12:21:03 <rmk0> in 3.0, glMapBuffer takes GL_READ_ONLY or GL_WRITE_ONLY, in that extension it's GL_WRITE_ONLY_OES only
20130718 12:21:28 <sgothel> hope values are same ?
20130718 12:21:40 <sgothel> GL_WRITE_ONLY == GL_WRITE_ONLY_OES
20130718 12:22:01 <rmk0> yeah, seems it
20130718 12:22:03 <rmk0> 0x88B9
20130718 12:22:07 <sgothel> so same semantics, but restricted caps ..
20130718 12:22:15 <sgothel> thats quite normal
20130718 12:22:32 <sgothel> good .. I will leave it as is then
20130718 12:22:51 <rmk0> the ES2 pipeline really doesn't have any concept of reading back results does it...
20130718 12:23:11 <sgothel> readPixels :)
20130718 12:23:15 <rmk0> heh
20130718 12:23:48 <sgothel> adding PBO now to our utils .. is a next task .. :)
20130718 12:50:02 <sgothel> no ES3 on Nexus 10 :(
20130718 15:05:13 * armurray (~andrew@anon) has joined #jogamp
20130718 18:42:55 <sgothel> I need a good method name ..
20130718 18:43:04 <sgothel> you have a buffer / list .. etc
20130718 18:43:14 <sgothel> size() -> number of elements in ..
20130718 18:43:30 <sgothel> XXX() -> number of elements available to write in ..
20130718 18:43:55 <sgothel> available/remaining ... etc .. are ambiguous ..
20130718 18:43:57 <sgothel> Mark ?
20130718 18:44:46 <sgothel> getFreeSlots() ?
20130718 18:44:47 <rmk0> would use "availableForWriting" personally
20130718 18:45:11 <sgothel> good name - but a bit loong ?
20130718 18:45:35 <rmk0> being shorter might be ambiguous, as you said
20130718 18:45:38 <rmk0> not sure on the context
20130718 18:45:54 <sgothel> one is ringbuffer .. new class for a/v resource handling
20130718 18:46:13 <rmk0> maybe getWritableSlots
20130718 18:46:37 <rmk0> if it's important that it be clear they're "available for writing"
20130718 18:46:39 <sgothel> one is in AudioSink .. i.e. get byte count of either free bytes .. or buffered bytes
20130718 18:47:15 <sgothel> ok .. will use FreeSlots for ring buffer .. will use the other for AudioSink
20130718 18:47:41 <sgothel> i.e. I remove the data copy stuff .. which exist currently
20130718 18:53:25 <rmk0> love being able to type "required texture formats" into a spec and actually get a list of them
20130718 18:53:41 <sgothel> :)
20130718 19:05:12 <rmk0> ugh, mesa may not provide two of the required texture formats on ES 3.0
20130718 19:05:45 <rmk0> page 126 of the spec... required texture formats
20130718 19:05:55 <rmk0> RG8 and R8, amongst others
20130718 19:06:15 <rmk0> seems like i get an error if i ask for either of those
20130718 19:06:29 <rmk0> trying to reproduce it in C, and of course i can't seem to get an ES 3.0 context open in C
20130718 19:07:59 <sgothel> we currently simply create ctx .. w/o KHR extension ..
20130718 19:08:11 <sgothel> parsing the GL version ..
20130718 19:08:23 <sgothel> (all w/ EGL ofc)
20130718 19:08:33 <sgothel> - or - you can use normal
20130718 19:08:51 <sgothel> .. X11/GLX and use the ES3_compat* .. assuming it's ES3
20130718 19:09:00 <sgothel> for mesa .. should be same thing
20130718 19:12:14 <rmk0> do we have a simple example of opening a context in C in the tree?
20130718 19:12:36 <rmk0> trying to get something together to file a bug report for mesa... would ideally prefer something that isn't hundreds of lines
20130718 19:12:52 <sgothel> yes .. the old one you used .. once
20130718 19:13:05 <rmk0> i ... forget what that was
20130718 19:13:22 * rmk0 eyes logs
20130718 19:13:25 <sgothel> jogl/src/test-native/
20130718 19:13:39 * rmk0 eyes test-native
20130718 19:14:27 <sgothel> ../src/test-native/mesa-demos-patched/ w/ egl .. as well
20130718 19:14:42 <rmk0> thanks
20130718 21:54:22 * armurray (~andrew@anon) Quit (Quit: armurray)
20130718 22:20:09 <rmk0> phew, ES3 integration almost complete
20130718 22:20:13 <rmk0> no real issues so far
20130718 22:20:32 <rmk0> am satisfied with the organization of the various interfaces
20130718 22:20:46 <sgothel> thank you .. happy :)
20130718 22:20:48 <rmk0> GL2ES3, GL3ES3, etc
20130718 22:20:54 <sgothel> no dups
20130718 22:21:16 <rmk0> i've not come across any, but i may've missed some
20130718 22:21:30 <sgothel> my sort -d scripts didn't find any :)
20130718 22:21:34 <rmk0> hehe
20130718 22:21:45 <sgothel> (yes a few GL2 / something .. but that was unavoidable)
20130718 22:21:49 <rmk0> i've started writing small test cases using reflection to check API uniformity in my own libraries
20130719 00:17:23 * rmk0-alt (~rmk0-alt@anon) Quit (Ping timeout: 240 seconds)
20130719 00:30:37 * rmk0-alt (~rmk0-alt@anon) has joined #jogamp
20130719 00:30:37 * rmk0-alt (~rmk0-alt@anon) Quit (Changing host)
20130719 00:30:37 * rmk0-alt (~rmk0-alt@anon) has joined #jogamp
20130719 02:42:50 * xranby_ac100 (~xranby@anon) has joined #jogamp
20130719 02:43:20 <xranby_ac100> I have arrived in Anaheim LA
20130719 02:43:29 <xranby_ac100> using hotel wifi
20130719 02:54:32 <xranby_ac100> safe travels to all of you heading to LA
20130719 02:59:31 <sgothel> hehe ..
20130719 02:59:46 <sgothel> I finally played w/ audio and so ..
20130719 02:59:49 <sgothel> will push in a sec ..
20130719 03:00:16 <sgothel> threadin .. well didn't work properly - but .. ok, removed data copy at least .. and some type cleanups .. *chaos* :)
20130719 03:00:23 <sgothel> btw
20130719 03:00:28 <sgothel> .. KUDOS - GREAT JOB
20130719 03:01:08 <xranby_ac100> thank you, kudos for testing!
20130719 03:02:12 <xranby_ac100> yes, i aimed for functionality first. i must admit i did not have much experience in a/v work to make it better on the first try :/
20130719 03:02:36 <xranby_ac100> so KUDOS for looking into reducing data copy!
20130719 03:03:33 <sgothel> I could not find a way to preserver the NIO buffer mapping - to the AV data frame .. well .. we discussed this once .. hidden magic
20130719 03:05:12 <xranby_ac100> sgothel: when do your flight leave?
20130719 03:05:27 <xranby_ac100> i know when it arrive .. tomorrow
20130719 03:05:32 <sgothel> 8 am .. so me hurry :)
20130719 03:06:30 <xranby_ac100> OK, i hope you live next door to your airport :D
20130719 03:06:47 <sgothel> 30 min :)
20130719 03:09:18 <sgothel> very short commit message .. sorry .. pushed
20130719 03:09:18 <xranby_ac100> it is still the 18th 4h more in LA
20130719 03:09:25 <xranby_ac100> good i will read the code
20130719 03:09:31 <xranby_ac100> *holding thumbs for jenkins*
20130719 03:10:21 <sgothel> Adroid still works :)
20130719 03:11:31 <xranby_ac100> thank you androi
20130719 03:11:33 <xranby_ac100> d
20130719 03:12:07 <xranby_ac100> please scale to desktop as well, and liberate the JVM
20130719 03:13:40 <sgothel> https://jogamp.org/chuck/job/gluegen/708/
20130719 03:13:49 <sgothel> almost forgot to .. do the version commit :)
20130719 03:23:06 <xranby_ac100> \o/ for making the audio sinks support AudioDataFormat!
20130719 03:24:04 <sgothel> not really - yet :)
20130719 03:24:11 <sgothel> but .. it's a blueprint
20130719 03:29:54 <xranby_ac100> sgothel: and \o/ 2.0.2
20130719 03:30:01 <xranby_ac100> sgothel: enjoy your flight!
20130719 04:45:39 * sgothel (~sven@anon) Quit (Quit: Leaving.)
20130719 05:06:03 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20130719050603.html