#jogamp @ irc.freenode.net - 20151030 05:05:19 (UTC)
20151030 05:05:19 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20151029050519.html
20151030 05:05:19 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20151030050519.html
20151030 08:03:48 * elect (~GBarbieri@anon) has joined #jogamp
20151030 08:23:57 * monsieur_max (~maxime@anon) has joined #jogamp
20151030 08:39:52 * doev (~doev@anon) has joined #jogamp
20151030 08:56:14 * gouessej (5ee4b442@anon) has joined #jogamp
20151030 08:56:29 <gouessej> Hi
20151030 08:57:04 <gouessej> bruce-: Look at JoglImageLoader in JogAmp's Ardor3D Continuation. I have some code that does what you want
20151030 08:57:12 * gouessej (5ee4b442@anon) Quit (Client Quit)
20151030 11:02:16 * jvanek (jvanek@anon) has joined #jogamp
20151030 11:06:55 * doev (~doev@anon) Quit (Ping timeout: 260 seconds)
20151030 12:55:15 * doev (~doev@anon) has joined #jogamp
20151030 12:55:39 * doev (~doev@anon) Quit (Client Quit)
20151030 13:33:32 * elect (~GBarbieri@anon) Quit (Ping timeout: 272 seconds)
20151030 13:37:35 * jvanek (jvanek@anon) Quit (Quit: Leaving)
20151030 15:19:08 <rmk0> am i required to "release" a context before destroying it?
20151030 15:19:17 <rmk0> i'm seeing an error on windows using the old proprietary AMD drivers
20151030 15:19:39 <rmk0> "Caught NativeWindowException: didn't release surface Handle: GDISurface[...]"
20151030 15:19:52 <rmk0> it's happening when calling destroy() on an offscreen drawable
20151030 15:32:28 * elect (~GBarbieri@anon) has joined #jogamp
20151030 15:48:41 <rmk0> yes... i can reliably reproduce
20151030 15:48:58 <rmk0> if you don't release() the context inside the drawable before attempting to destroy the drawable, you'll get the above exception
20151030 15:49:03 <rmk0> should i file a bug on this?
20151030 15:49:07 <rmk0> they're pretty old drivers...
20151030 16:00:00 <rmk0> https://jogamp.org/bugzilla/show_bug.cgi?id=1259
20151030 16:01:52 <rmk0> this could potentially be a documentation issue... "don't destroy that unless you've released it"
20151030 16:02:01 <rmk0> but it doesn't occur on other platforms, so there's at least an inconsistency
20151030 16:10:19 * elect (~GBarbieri@anon) Quit (Ping timeout: 240 seconds)
20151030 16:13:55 * Eclesia (~eclesia@anon) has joined #jogamp
20151030 16:17:53 <rmk0> additionally, i'm discovering that with shared contexts on windows, you'll get the same exception if you attempt to destroy a released context that is shared with a context that is current
20151030 16:18:15 <rmk0> are these platform-specific semantics documented somewhere, or am i the first person hitting this stuff?
20151030 16:18:16 <Eclesia> hi
20151030 16:18:18 <rmk0> lo
20151030 19:37:07 <Eclesia> rmk0: small question, when loading a DXT texture, what format/internalformat value do you use ? I found GL_COMPRESSED_RGBA_S3TC_DXT5_EXT but the 'EXT' is annoying me, as if this was somekind of old constant value
20151030 19:48:49 <rmk0> haven't used DXT textures
20151030 19:56:12 <Eclesia> rmk0: no compressed textures ?
20151030 20:00:04 <rmk0> no compressed textures
20151030 20:11:32 <rmk0> it'll be EXT forever, unfortunately
20151030 20:11:40 <rmk0> the algorithm is patented, so it'll never make it into the spec
20151030 20:20:10 <Eclesia> damn, I wasn't aware of that ...
20151030 20:21:43 <Eclesia> Patent: Fixed-rate block-based image compression with inferred pixel values Estimated expiry: Oct 2, 2017
20151030 20:22:16 <rmk0> it's been replaced by ASTC, but that's not implemented widely yet
20151030 20:22:39 <Eclesia> yes I've started working on this one
20151030 20:23:03 <Eclesia> but it's really a nasty bit by bit compression :/
20151030 20:54:08 <rmk0> as i understand it, JOGL provides a default VAO
20151030 20:54:13 <rmk0> is there some way to turn it off?
20151030 20:54:41 <rmk0> by providing a default one, it's helpfully adding silent failure, because you'd normally get an error if you tried to render without a VAO bound
20151030 20:54:48 <rmk0> with JOGL, there's always one, so you get nonsense results instead
20151030 21:03:12 <Eclesia> you could work directly on the GL interfaces without the utility classes
20151030 21:03:36 <rmk0> i don't use any of the utility classes
20151030 21:03:58 <rmk0> the default VAO is created per context, and there doesn't seem to be any way to stop it
20151030 21:04:49 * Eclesia is clueless
20151030 21:05:22 <rmk0> i know why it was done
20151030 21:06:04 <rmk0> 3.2 suddenly mandated that a VAO be bound with certain operations, and because JOGL only exposes versions at the granularity of GL3, GL2, etc, it meant that code was valid with some GL3 instances and not others
20151030 21:06:27 <rmk0> by introducing a default VAO, it meant that code would work the same on all 3.* versions
20151030 21:06:40 <rmk0> but it also introduces some silent failure, as i mentioned
20151030 21:07:57 <rmk0> the opengl API is just bad
20151030 21:20:30 * Eclesia small victory, no more dxt decompression when loading in opengl
20151030 21:22:53 <rmk0> ... i badly want vulkan
20151030 21:23:20 <rmk0> i'd almost forgotten what a deeply shitty experience it is to program with this joke of an API
20151030 21:23:28 <Eclesia> ^^
20151030 21:24:06 <rmk0> it would be slightly better if GPUs actually worked this way, but instead i'm fighting an abstraction dreamt up by the ARB
20151030 21:25:03 <Eclesia> nothing's perfect the first 20 times
20151030 21:25:10 <rmk0> hehe
20151030 21:55:33 <rmk0> http://jogamp.org/deployment/v2.2.1/javadoc/jogl/javadoc/javax/media/opengl/GLContext.html#getDefaultVAO%28%29
20151030 21:55:37 <rmk0> X(
20151030 22:07:44 <Eclesia> rmk0: good luck & good night ++
20151030 22:08:04 * Eclesia (~eclesia@anon) Quit (Quit: Leaving.)
20151030 23:27:16 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20151031 02:05:54 * bigpet (uid25664@anon) Quit (Quit: Connection closed for inactivity)
20151031 03:54:16 * monsieur_max (~maxime@anon) has joined #jogamp
20151031 03:54:17 * monsieur_max (~maxime@anon) Quit (Client Quit)
20151031 05:05:20 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20151031050520.html