#jogamp @ irc.freenode.net - 20150906 05:05:09 (UTC)
20150906 05:05:09 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20150905050508.html
20150906 05:05:09 -jogamp- This channel is logged @ http://jogamp.org/log/irc/jogamp_20150906050509.html
20150906 08:49:01 * elect (~elect@anon) has joined #jogamp
20150906 09:13:56 * monsieur_max (~maxime@anon) has joined #jogamp
20150906 10:14:49 * elect (~elect@anon) Quit (Ping timeout: 252 seconds)
20150906 14:33:47 * elect (~elect@anon) has joined #jogamp
20150906 14:46:45 <elect> jgo is down?
20150906 18:26:09 * elect (~elect@anon) Quit (Ping timeout: 252 seconds)
20150906 19:33:05 * xranby_f22 (~familjen@anon) has joined #jogamp
20150906 19:33:14 <xranby_f22> event at Oracle. All of the Java evangelists have been let go. Sad this happened to a great dedicated enthusiastic group
20150906 19:33:23 <xranby_f22> happening on twitter right now
20150906 19:33:41 <xranby_f22> thus java fx is ńow officially dead i think
20150906 19:34:04 <xranby_f22> http://www.infoworld.com/article/2980686/java/sources-oracle-releases-java-evangelists.html
20150906 19:34:27 <xranby_f22> Java Evangelists: Meet the Team: Evangelists for hire http://www.oracle.com/technetwork/java/community/javaevangelists-325324.html
20150906 20:09:02 <xranby_f22> http://www.theregister.co.uk/2015/09/04/oracle_reportedly_lays_off_java_evangelists/
20150906 20:22:31 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20150906 20:39:54 <guillaum1> sad :-(
20150906 20:49:50 <xranby_f22> some of the people has already got a new jog at google and redhat
20150906 20:49:53 <xranby_f22> job
20150906 20:51:13 <xranby_f22> John Clingan -> new job at Red Hat after 18.5 years at Sun/Oracle
20150906 20:58:21 <xranby_f22> i thought oracel made money on java ee.. thus seeing people like John off the paycheck is interesting
20150906 20:59:10 <xranby_f22> John was the "GlassFish Product Manager"
20150906 21:44:55 <sgothel> tbh .. I never got the idea of this 'tech evangelists' .. well, sort of advertisement I guess
20150906 21:46:00 <sgothel> i.e. in the little group where I was .. we had no such evangelists - just 'workers' :)
20150906 21:46:13 <sgothel> no silk suits :)
20150906 21:47:04 <sgothel> ofc Oracle is shrinking down the java group .. due to economy overall- and java- conraction
20150906 21:47:08 <sgothel> *contraction*
20150906 21:48:09 <sgothel> Red Hat eh? .. well, I wish them luck. Sadly we got no offer for JogAmp support from them yet :)
20150906 21:50:00 <sgothel> ORCL -8%, RHT -12%
20150906 21:50:33 <sgothel> annually RHT performs better though ..
20150906 21:50:56 <sgothel> better than many .. not bad
20150906 21:50:59 <xranby_f22> https://jogamp.org/bugzilla/show_bug.cgi?id=1214 <- me adding the macosx newt issues reported from the processing team
20150906 21:51:18 <sgothel> RHT -1.46%, IBM -2.13%, ORCL -1.71%
20150906 21:52:31 <sgothel> ok - removing the OSX from the title .. generalizing .. depending on Bug 1188 maybe?
20150906 21:53:26 <xranby_f22> sgothel: i think you where blessed to be in a pretty awesome group, maybe too radical, not even the evangelists mention on their when jogl evolve
20150906 21:53:36 <xranby_f22> on their blogs
20150906 21:54:04 <xranby_f22> sgothel: the bug has one macos x newt specific part , that the contect is rendered outside the newt window after auto-resize
20150906 21:54:18 <sgothel> the ORCL layoffs were known to all employees AFAIK .. for .. around 3 years, at least my ex-colleagues knew the schedule for the UI cancellation
20150906 21:54:25 <sgothel> ok
20150906 21:54:49 <sgothel> yeah well, we didn't have 'press interviews' w/ other groups :)
20150906 21:55:21 <sgothel> and tbh .. we also were a assumed a hostile group, since we disliked many 'solutions'
20150906 21:55:56 <xranby_f22> i assume: its good that both sides can speeak freely now
20150906 21:56:09 <sgothel> sort of .. sort of
20150906 21:56:11 <xranby_f22> and maybe work on stuff for real
20150906 21:56:36 <sgothel> yes, I am glad .. to be in my current role - somewhat free
20150906 21:57:24 * xranby_f22 is running: xrandr --fb 16384x16384 --output DVI-I-2 --mode 1920x1080 --panning 16384x16384
20150906 21:57:25 <sgothel> Johnny once called me 'brave', but I said, well, you must be brave, b/c you live in the US and have a family to feed while being in an employers market :-/
20150906 21:58:03 * xranby_f22 will try create some.. XXL large GLWindows
20150906 21:58:09 <sgothel> hehe
20150906 21:58:23 <sgothel> w/ current GlueGen/JOGL .. does the IRQ still happen?
20150906 21:58:46 <sgothel> so I will cont. hacking all the cond-wait code sections .. I guess
20150906 21:58:53 <xranby_f22> right now i am on the machine that i only manage to crash gnome shell with
20150906 21:59:01 <sgothel> ah
20150906 21:59:25 <xranby_f22> thus on this machine jogl seems stable
20150906 21:59:46 <sgothel> the new test case is able to detect the source of the IRQ call .. pls use it - when you are back on your lucky machine :)
20150906 22:00:20 <xranby_f22> i have my lucky machine at work at the office, it can reproduce it within 1 minute
20150906 22:00:43 <sgothel> so consensus is .. AFAIK .. for dedicated threads like EDT, we shall be non-cancelable but reporting it.
20150906 22:00:58 <xranby_f22> usually in combination trying to browse the web using firefox while GLWindows are opening and closing
20150906 22:01:11 <sgothel> for simple invoke-wait, we shall be cancelable (sic)
20150906 22:01:37 <sgothel> yeah, the 2nd test of new test case does it .. while being covered w/ own Thread calss
20150906 22:01:39 <sgothel> class
20150906 22:02:18 <xranby_f22> let me run the new test on this machine.. to test all the things
20150906 22:06:59 <sgothel> -time01 100000 <- i.e. push up that 2nd sub-test01 .. opening and closing .. like GLWindow startup ..
20150906 22:07:25 <sgothel> it could be .. that it was the classloader/readBytes .. I mentioned there
20150906 22:07:47 <sgothel> it could be .. that this was a side effect of my GlueGen group-locking unlock bug
20150906 22:09:10 <sgothel> "GlassFish Product Manager" is not an 'evangelist' !
20150906 22:10:24 <xranby_f22> i have one question about jogl's role on application termination. If we have a GPU driver that is running background threads like the nvidia gpu driver do, shoud we ensure that all these threads has joined and finished during the jogl shutdown hooks?
20150906 22:10:49 <sgothel> good question - I am glad you asked! :)
20150906 22:11:06 <xranby_f22> at the moment this is handled by the JVM
20150906 22:11:16 <sgothel> to deal w/ such driver bugs (and yes, there were many) - we keep the SharedResourceRunner alive
20150906 22:11:22 <sgothel> (for X11 and WGL)
20150906 22:11:22 <xranby_f22> but some JVM do not wait for threads before running the shutdown hooks
20150906 22:11:52 <sgothel> so this thread queries capabilities and keeps one context alive on X11 and WGL
20150906 22:12:09 <sgothel> at shutdown .. we shut them down .. release and end thread
20150906 22:12:30 <sgothel> "some JVM do not wait for threads before running the shutdown hooks"
20150906 22:12:54 <sgothel> well, the shutdown hooks are being processed .. and here we release the 'last resource'
20150906 22:12:57 <xranby_f22> https://github.com/robovm/robovm/issues/85 <- this one in particular
20150906 22:13:28 <sgothel> the JVM must not wait for other threads befor starting the hook ..
20150906 22:13:43 <sgothel> some threads even might be killed ..
20150906 22:14:03 <sgothel> but the hooks shall be executed .. and the JVM shall wait for some 'reasonable amount of time'
20150906 22:14:20 <sgothel> experience is: its enough for us - no locking involved here
20150906 22:14:35 <sgothel> also: we don't release the shared libraries - here we experienced crashes indeed
20150906 22:14:58 <sgothel> the OS exit code must release the open handlers .. libraries etc
20150906 22:15:38 <sgothel> but we take care to release all windowing and GL resources, since some drivers simply don't act on the SIGNAL EXIT (OS exit() call)
20150906 22:16:09 <sgothel> hence .. user code shall also take care of issuing 'destroy()' .. i.e. to be nice
20150906 22:18:14 <xranby_f22> i will ventilate my thoughts.. on one of my friends nvidia GTX laptop running windows 10 i get a windows notification that the gpu driver has crashed after a newt window terminate.. no java crash log etc. thus i suspect something similar to this robovm crash have happened but using hotspot
20150906 22:18:17 <sgothel> hard to associate the stack trace of that RoboVM issue
20150906 22:18:55 <xranby_f22> the windows popup dissapears automatically and the os continue humming thus the user did not care at all :/
20150906 22:19:29 <xranby_f22> but i had no clue how to debug it
20150906 22:19:47 <sgothel> it is hard .. w/ those NV threading issues
20150906 22:20:02 <sgothel> we have some workarounds for it already, i.e. the initialization
20150906 22:20:06 <xranby_f22> yes, all recent pain has been with nvidia drivers recently
20150906 22:20:09 <sgothel> but all this was .. err .. years ago
20150906 22:20:18 <sgothel> so it happens again? hmm
20150906 22:20:45 <xranby_f22> maybe the latest drivers are stable hmm?
20150906 22:20:48 <sgothel> windows + NV ?
20150906 22:20:54 <sgothel> linux as well ?
20150906 22:21:13 <xranby_f22> windows + NV <- i have to investigate the driver version on my friends machine
20150906 22:22:32 <sgothel> WindowsWGLDrawableFactory.SharedResourceImplementation.releaseSharedResource(..) line 386
20150906 22:22:42 <xranby_f22> linux + NV <- old driverNVIDIA 331.113 used in combination with robovm
20150906 22:22:48 <sgothel> similar for X11GLX ...
20150906 22:23:28 <sgothel> you see the comment: // may cause JVM SIGSEGV: sharedContext.destroy();
20150906 22:23:43 <sgothel> in the past .. it was causing a crash .. (-> driver issue)
20150906 22:23:52 <sgothel> since a while - all was/is stable
20150906 22:24:23 <sgothel> if you are in doubt .. uncomment the context destroy .. and maybe even the drawable destruction via setRealize(false)
20150906 22:24:41 <xranby_f22> i tried to reproduce the robovm issue using a new nvidia driver such as 355.00.05 on my fedora 22 machine.. but it runs stable.. cant reproduce it using a new driver
20150906 22:24:46 <sgothel> maybe all .. alltogether :)
20150906 22:25:18 <sgothel> yeah I am running NVIDIA-Linux-x86_64-355.11.run now .. all seems to be fine
20150906 22:27:38 * xranby_f22 files the issues seen under old nvidia driver blob bugs that are all fixed in new versions
20150906 22:28:22 <xranby_f22> in any case... i will run the tests such as 1211 on all drivers i can find
20150906 22:28:41 <xranby_f22> to get a better picture of what is going on
20150906 22:39:46 <xranby_f22> good night, 1211 testing tomorrow on my lucky machine
20150906 22:42:12 <sgothel> good night
20150907 05:05:09 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20150907050509.html