JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20131126 05:05:54 (UTC)

20131126 05:05:54 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20131125050554.html
20131126 05:05:54 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20131126050554.html
20131126 05:07:07 * xranby (~xranby@anon) Quit (Ping timeout: 272 seconds)
20131126 05:08:10 * xranby (~xranby@anon) has joined #jogamp
20131126 08:48:47 * monsieur_max (~maxime@anon) has joined #jogamp
20131126 09:07:37 <xranby> sgothel: your workaround work... and at the moment i thought everything was fine... this landed in icedtea-web that broke all the jogl applets https://jogamp.org/bugzilla/show_bug.cgi?id=912
20131126 09:09:14 <xranby> icedtea-web now get upset and throw a runtime exception: Multiple main JARs specified, refusing to continue."
20131126 09:10:21 <xranby> triggers for the version applet as well
20131126 09:20:08 <xranby> i am chatting with jvanek to see what icedtea-web complain about
20131126 09:31:39 <xranby> (10:11:47) xranby: jvanek: good morning.. tested the latest icedtea-web 1.5+ hg tip and all jogl applets broke https://jogamp.org/bugzilla/show_bug.cgi?id=912
20131126 09:31:39 <xranby> (10:12:42) jvanek: xranby, hello!
20131126 09:31:39 <xranby> (10:12:57) xranby: it would be good to know what we do wrong in our jnlp or object tags to trigger this odd exception
20131126 09:31:39 <xranby> (10:13:09) jvanek: xranby, hm, I was digging recently in multiple main jars appelts specified
20131126 09:31:39 <xranby> (10:14:15) xranby: jvanek: we have several applets and all of them now trigger this multiple main jars http://jogamp.org/deployment/archive/master/gluegen_751-joal_502-jogl_1161-jocl_882-signed/jogl-test-applets.html
20131126 09:31:39 <xranby> (10:14:35) jvanek: xranby, What is done in 1.5, is , that when more them one "main" attribute is specified, or when more then one jar have main-class specified in manifest, itw will stop work
20131126 09:31:40 <xranby> (10:15:32) xranby: for applets main is specified by the jnlp or the object tag no?
20131126 09:31:40 <xranby> (10:15:53) jvanek: yes
20131126 09:31:41 <xranby> (10:16:05) jvanek: but jnlp can have (by mistake) more "main" specified
20131126 09:31:41 <xranby> (10:16:19) jvanek: and if no main is specified, we are scaning individual maifest
20131126 09:31:41 <xranby> (10:16:50) jvanek: so where we were ok in 1.4, we stopped now
20131126 09:31:42 <xranby> (10:16:53) xranby: we use the object tag and use the code=" parameter to specify main
20131126 09:31:42 <xranby> (10:17:00) jvanek: that should work
20131126 09:31:43 <xranby> (10:17:13) xranby: and we opnly specify one main as fair as i am aware
20131126 10:34:12 * hija (~hija@anon) has joined #jogamp
20131126 10:49:34 <xranby> (11:40:33) jvanek: for jogl it is clear itw bug
20131126 10:49:34 <xranby> (11:40:46) jvanek: (maybe we recieve the params, but we should hande it correctly
20131126 10:49:34 <xranby> (11:46:26) jvanek: xranby, just fdor curiosity, is <comment> specified somewhere?
20131126 10:49:34 <xranby> (11:47:16) xranby: jvanek: http://docs.oracle.com/javase/7/docs/technotes/guides/plugin/developer_guide/using_tags.html#object
20131126 10:49:34 <xranby> (11:47:42) jvanek: xranby, I see, thanx :-/
20131126 10:49:34 <xranby> (11:47:48) jvanek: yup we ignore it....
20131126 10:49:34 <xranby> (11:48:05) jvanek: I mean, we dont ignre the content of it :")
20131126 10:49:35 <xranby> (11:48:12) jvanek: just now it flowed up
20131126 10:49:35 <xranby> (11:48:25) xranby: ok great.. problem understood
20131126 11:03:04 <sgothel> ok - so there is another bug ..
20131126 11:03:26 <sgothel> you can close 912 then .. "assigned to IcedTea Web"
20131126 11:03:40 <sgothel> It would be great, if you can show them that 910 is also their bug!
20131126 11:03:56 <sgothel> i.e. don't remove stuff before Applet.destroy() has been called
20131126 11:04:07 <xranby> (10:56:22) xranby: jvanek: jogamp added some workarounds for icedtea-web yesterday https://jogamp.org/bugzilla/show_bug.cgi?id=910 - IcedTea Web removes Components from the Applet Container before Applet.stop() causing a Crash. happens for icedtea-web 1.3 1.4 and 1.5
20131126 11:04:07 <xranby> (10:58:20) jvanek: mmmmmnt :)
20131126 11:04:36 <sgothel> .. and it should not be done before Applet.destroy() !!
20131126 11:04:49 <sgothel> _has_been_called <- just to make sure :)
20131126 11:05:14 <sgothel> KUDOS from me for fixing the AWT-EDT issue though ..
20131126 11:05:24 <xranby> KUDOS!
20131126 11:06:06 <sgothel> goody .. at least NApplets work fine now it seems
20131126 11:06:29 <sgothel> thank you for communicating all this!
20131126 11:11:23 <sgothel> @Xerxes: 'Multiple Main' is proper design - not by accident :)
20131126 11:11:48 <sgothel> i.e. it allows an applet to use JNLP if supported - or fall back to traditional Applet
20131126 11:12:10 <xranby> yes it is described in http://docs.oracle.com/javase/7/docs/technotes/guides/plugin/developer_guide/using_tags.html#html
20131126 11:12:11 <sgothel> i.e. either pick up the jnlp-applet .. or use traditional applets
20131126 11:12:37 <sgothel> Ken and I took care those days that this is supported
20131126 11:13:18 <xranby> 912 triggers for NApplets as well http://jogamp.org/deployment/archive/master/gluegen_751-joal_502-jogl_1161-jocl_882-signed/jogl-applet-version-napplet.html where no jnlp is in use
20131126 11:14:42 <sgothel> then I may misunderstood - where is the 'double main' ?
20131126 11:15:44 <sgothel> <param name="code" value="jogamp.opengl.awt.VersionApplet">
20131126 11:15:44 <sgothel> <param name="archive" value="jar/gluegen-rt.jar,
20131126 11:15:44 <sgothel> jar/jogl-all.jar">
20131126 11:15:44 <xranby> the icedtea-web parser somehow adds one main from the <object <param name="code" value=" and another main from <embed code=" at the same time
20131126 11:15:56 <sgothel> omg ..
20131126 11:16:21 <xranby> because they did not implement the <comment> tag
20131126 11:16:26 <sgothel> so .. if they have a proper bug in their tracker .. pls close 912 as invalid w/ a reference
20131126 11:16:30 <sgothel> same for 910 ..
20131126 11:16:37 <sgothel> (add reference ..)
20131126 11:16:44 <xranby> i am opening an icedtea-web bug as we speak
20131126 11:16:49 <sgothel> AWESOME!
20131126 11:19:22 <xranby> (12:18:12) xranby: jvanek: so if i understand it correctly the icedtea-web parser somehow adds one main from the <object <param name="code" value=" and another main from <embed code=" at the same time
20131126 11:19:22 <xranby> (12:18:25) jvanek: I'm still on it
20131126 11:19:23 <xranby> (12:18:32) jvanek: but it stinks :)
20131126 11:19:23 <xranby> (12:18:56) xranby: jvanek: are you working on a bugreport as well or should i add one ?
20131126 11:19:23 <xranby> (12:19:06) jvanek: I'm working on fix...
20131126 12:07:13 <xranby> when i run http://jogamp.org/deployment/archive/master/gluegen_751-joal_502-jogl_1161-jocl_882-signed/jogl-application-version.jnlp using icedtea-web javaws then there is a black border at the bottom below the bottom scrollbar.. if i run it inside the browser then this border is filled with uninitialized garbage data.. this is using pure awt thus must be an openjdk awt layout bug...
20131126 12:07:43 <sgothel> not a bug .. a feature :)
20131126 12:07:50 <sgothel> it's a GLCanvas ..
20131126 12:07:55 <xranby> oh
20131126 12:07:57 <sgothel> GL FBO not cleared :)
20131126 12:08:02 <sgothel> *FB*
20131126 12:08:26 <sgothel> the one used to get the GLCOntext .. and information :)
20131126 12:08:51 <sgothel> -> "Sven's crazy tech demos" :)
20131126 12:08:59 <xranby> odd then that it gets cleared to gray if i cover the region inside the browser while it gets painted black using javaws
20131126 12:09:15 <xranby> is it transparent?
20131126 12:09:34 <sgothel> could be .. not req. it - but info will tell you
20131126 12:09:48 <sgothel> the caps .. -> alpha and opaq
20131126 12:12:09 <xranby> it have listed 100+ caps some with and some without in the bottom scroll window.. not sure which one is in use
20131126 12:12:19 <xranby> i guess the last one
20131126 12:12:33 <xranby> GLCaps[glx vid 0xe4, fbc 0x1c8: rgba 8/8/8/8, trans-rgba
20131126 12:13:45 <sgothel> ah sorry .. we don't print it in the 1st text area ..
20131126 12:14:08 <sgothel> will do that ..
20131126 12:15:08 <sgothel> black here as well (IcedTea NApplet
20131126 12:15:32 <sgothel> chosen GLCaps[glx vid 0x27, fbc 0x10d: rgba 8/8/8/0, opaque, accum-rgba 16/16/16/16, dp/st/ms 24/0/0, dbl, mono , hw, GLProfile[GL4bc/GL4bc.hw], on-scr[.]]]]],
20131126 12:15:42 <sgothel> (from debug log)
20131126 13:30:10 <xranby> http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2013-November/025360.html - [rfc][icedtea-web] removed recent check for multiple main jar
20131126 13:30:48 <sgothel> nice
20131126 14:08:46 <xranby> sgothel: are your simple test.net/a.html applet used to produce the logs in 910 available someplace?
20131126 14:12:08 * [Mike] (~Mike]@anon) has joined #jogamp
20131126 14:13:48 <sgothel> I just shortend the URL
20131126 14:14:25 <sgothel> but any simple applet should do it .. overload removeNotify() of an added AWT component .. will show it's called before destory()
20131126 14:14:32 <sgothel> destroy()
20131126 14:15:05 * hija (~hija@anon) Quit (Quit: hija)
20131126 14:27:23 <xranby> lets see.. if this static test applet is of use http://docs.oracle.com/javase/tutorial/deployment/applet/lifeCycle.html
20131126 14:29:00 <sgothel> they can simply use ours .. and if that removeNotify() appears before destroy() .. gotcha .. should be piece of cake to code this
20131126 14:29:42 <xranby> what i am looking for is some sort of spec that explicit say that removeNotify cant appear before destroy
20131126 14:29:54 <sgothel> in init .. just call add(new MyComponent())
20131126 14:30:07 <sgothel> MyComponent overrides removeNotify() ..
20131126 14:30:52 <sgothel> well .. before you applet is stopped or destroyed .. it shall not be secretly touched/content-removed
20131126 14:31:14 <sgothel> wording may not say this, but it says what stop means .. and destroy
20131126 14:31:23 <sgothel> stop means, it can be start'ed again
20131126 14:31:40 <sgothel> destroy means - _after_ calling destroy -> EOL
20131126 14:31:53 <sgothel> i.e. you have your chance to cleanup here
20131126 14:32:14 <xranby> well icedtea-web passes the lifeCycle outlined by the oracle documentation
20131126 14:32:26 <sgothel> so the test is buggy!
20131126 14:32:50 <sgothel> since 610 proves that the applet container removes an AWT component
20131126 14:33:15 <sgothel> _before_ stop() and before destroy()
20131126 14:33:22 <sgothel> so we have no chance to cleanup
20131126 14:33:37 <xranby> right!
20131126 14:34:01 <xranby> ok i am on track
20131126 14:35:10 <sgothel> Quitting the Browser When the user quits the browser, the applet has the opportunity to stop itself and perform a final cleanup before the browser exits.
20131126 14:35:23 <sgothel> final cleanup -> destroy()
20131126 14:35:26 <xranby> icedtea-web got a testsuite
20131126 14:35:26 <xranby> so if we can give them a test then its gold
20131126 14:35:43 <sgothel> ok .. I will attach one :)
20131126 14:55:56 <xranby> How about this summary:
20131126 14:56:11 <xranby> https://jogamp.org/bugzilla/show_bug.cgi?id=910 - IcedTea Web removes Components from the Applet Container before Applet.stop() causing a Crash.
20131126 14:56:11 <xranby> Assigned bug to IcedTea Web.
20131126 14:56:11 <xranby> This bug is observed using IcedTea-web 1.3, 1.4 and 1.5pre hg tip.
20131126 14:56:11 <xranby> IcedTea-web contains code that call AWT Component.removeNotify(), too early when the user leave/reload an applet page, before Applet.stop() and Applet.destroy() has been called.
20131126 14:56:11 <xranby> Removing components before Applet.stop() and Applet.destroy() has been called violates the applet lifeCycle "spec":
20131126 14:56:11 <xranby> http://docs.oracle.com/javase/tutorial/deployment/applet/lifeCycle.html
20131126 14:57:44 <sgothel> Yes .. pls reference https://jogamp.org/bugzilla/show_bug.cgi?id=910#c17 and the upcoming standalone test -> hence wait a bit
20131126 14:58:12 <sgothel> also add a 'confirmed by xranby + sgothel' :)
20131126 15:05:36 <xranby> sgothel: will you publish the applet lifecycle test under http://jogamp.org/deployment/test/ ?
20131126 15:11:59 <sgothel> will do ..
20131126 15:12:07 <sgothel> testing locally now ..
20131126 15:12:32 <sgothel> will add it to usual jogl-test-applet location, but also preview it as you mentioned ..
20131126 15:12:56 <sgothel> code is standalone ofc
20131126 15:15:06 <xranby> Are people who want to be part of the FOSDEM 25min jogamp talk available tomorrow? We need to have the email sent in before the 1st of dec with our talk proposal.
20131126 15:15:28 <sgothel> sven: yes
20131126 15:16:31 <xranby> https://lists.fosdem.org/pipermail/java-devroom/2013-October/000042.html - Free Java DevRoom FOSDEM 2014 Call For Participation now open!
20131126 15:16:35 <xranby> xerxes: yes
20131126 15:16:50 <xranby> there is currently only 4 talk proposals sent in to that list
20131126 15:17:27 <xranby> https://lists.fosdem.org/pipermail/java-devroom/2013-November/thread.html
20131126 15:22:01 <xranby> thus it may be possible to have one 25min Ji Gong talk and one 25min JogAmp talk
20131126 15:22:26 <xranby> all depending on how many talk proposals we send in
20131126 15:30:33 <xranby> sgothel: i will depart for today , i can file the icedtea-web bug later this evening when your test is ready. KUDOS for all your support to find and fix these kind of issues!
20131126 15:48:15 <sgothel> KUDOS to us. Sounds great
20131126 16:56:08 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131126 16:57:26 <rmk0> there's no way to distinguish between a uniform that was never declared, and a uniform that has been optimized out, is there...
20131126 16:57:30 <rmk0> without parsing GLSL, that is
20131126 17:34:28 * xranby (~xranby@anon) Quit (Ping timeout: 272 seconds)
20131126 17:36:42 * xranby (~xranby@anon) has joined #jogamp
20131126 18:27:20 * monsieur_max (~maxime@anon) has joined #jogamp
20131126 18:34:44 * monsieur_max (~maxime@anon) has left #jogamp
20131126 18:35:42 * monsieur_max (~maxime@anon) has joined #jogamp
20131126 18:43:34 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20131126 18:44:26 * monsieur_max (~maxime@anon) has joined #jogamp
20131126 19:09:45 <sgothel> http://jogamp.org/git/?p=jogl.git;a=commit;h=9310b11b2b6e1e89fa5ed9b8de26e56ff6a6b262
20131126 19:16:46 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=910#c23
20131126 22:21:26 <sgothel> @Xerxes: http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=1607 is fixed ?
20131126 22:24:23 <sgothel> ^^ sorry, I had javascript disabled
20131126 22:34:09 * void256 (~void@anon) has joined #jogamp
20131126 22:40:35 * xranby-64 (~familjen@anon) has joined #jogamp
20131126 22:41:44 <xranby-64> http://icedtea.classpath.org/bugzilla/show_bug.cgi?id=1607 still needs to be patched inside icedtea-web
20131126 22:42:36 <sgothel> yup .. as I said, I had javascript disabled .. and then bugzilla shows all the other states (i.e. NEW + FIXED :)
20131126 22:43:04 <xranby-64> interesting side-effect :)
20131126 22:56:46 <sgothel> political mode [again] - interesting development lately :)
20131126 22:58:20 <sgothel> China was probed a few times US/Al-Quaida .. Silence w/ Iran while Israel is jumping on the spot (good silence?) .. err .. and our governments in Europe seem all to agree that we need an even more powerful 'lawful' NSA aquivalent
20131126 22:59:27 * monsieur_max (~maxime@anon) has left #jogamp
20131126 23:03:41 * xranby-641 (~familjen@anon) has joined #jogamp
20131126 23:05:47 * xranby-64 (~familjen@anon) Quit (Ping timeout: 272 seconds)
20131126 23:25:02 * xranby-641 (~familjen@anon) Quit (Ping timeout: 252 seconds)
20131126 23:32:24 * void256_ (~void@anon) has joined #jogamp
20131126 23:34:27 * void256 (~void@anon) Quit (Ping timeout: 252 seconds)
20131126 23:34:33 * void256_ is now known as void256
20131127 00:30:33 * void256 (~void@anon) Quit (Ping timeout: 252 seconds)
20131127 01:39:18 * sgothel (~sgothel@anon) Quit (Ping timeout: 246 seconds)
20131127 01:40:04 * sgothel (~sgothel@anon) has joined #jogamp
20131127 01:40:04 * sgothel (~sgothel@anon) Quit (Changing host)
20131127 01:40:04 * sgothel (~sgothel@anon) has joined #jogamp
20131127 01:40:04 * ChanServ sets mode +v sgothel
20131127 04:33:13 * xranby (~xranby@anon) Quit (Ping timeout: 272 seconds)
20131127 04:36:21 * xranby (~xranby@anon) has joined #jogamp
20131127 05:05:54 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20131127050554.html