#jogamp @ irc.freenode.net - 20130913 01:00:12 (UTC)


20130913 01:00:12 -jogamp- Previous @ http://jogamp.org/log/irc/jogamp_20130912050611.html
20130913 01:00:12 -NickServ- This nickname is registered. Please choose a different nickname, or identify via /msg NickServ identify <password>.
20130913 01:00:17 * jogamp (~jogamp@anon) has joined #jogamp
20130913 01:00:17 * Topic is 'http://jogamp.org | Hacking 3D Graphics, Multimedia and Processing across Devices'
20130913 01:00:17 * Set by rmk0 on 20130116 23:58:04
20130913 01:00:18 -NickServ- You are now identified for jogamp.
20130913 05:38:22 * monsieur_max (~maxime@anon) has joined #jogamp
20130913 06:07:55 * monsieur_max (~maxime@anon) Quit (Quit: Leaving.)
20130913 07:51:52 <xranby> Jerry Jongerius: Have a quite good think list regarding java plug-in browser security http://www.duckware.com/tech/index.html
20130913 07:52:45 <xranby> http://www.duckware.com/tech/javacodesigningfailure.html
20130913 08:01:07 <sgothel> good morning :)
20130913 08:01:40 <sgothel> _seem_ like Oracle/TX doesn't 'give a flying fuck' :)
20130913 08:03:48 <xranby> i will forward the test page to icedtea-web, since their plugin is vulnable to this name change attack as well
20130913 08:03:55 <xranby> http://www.duckware.com/tech/javafailure.html
20130913 08:44:24 <xranby> sgothel: godo morning as well :)
20130913 08:44:26 <xranby> good
20130913 08:45:37 <xranby> Jiri Vanek who runs the icedtea-web team is now well informed
20130913 08:46:29 <sgothel> hehe, good stuff - sure he is, you are the messenger!
20130913 08:46:54 <sgothel> great job
20130913 09:05:16 <sgothel> http://www.youtube.com/watch?v=z-sdO6pwVHQ <- Help Kickstart World War III! gogogoggo :)
20130913 09:17:54 <xranby> haha
20130913 11:30:35 <xranby> This is actually a good idea: USBCondoms prevents accidental data excange while charging your toy http://int3.cc/collections/frontpage/products/usbcondoms
20130913 11:31:33 <sgothel> right .. read an article (german) about those usb/firewire root-kits ..
20130913 11:32:24 <sgothel> .. infection proxies in the network (replacing or modding download/pages on the fly .. omg)
20130913 11:42:13 <xranby> replacing firmware on harddrives
20130913 11:42:26 <xranby> etc
20130913 11:43:46 <sgothel> what I liked at the Star Trek 'design' of computers as shown in 'Deep Space 9', is that all items have a signature and all operations are logged :)
20130913 11:44:03 <xranby> did they have source?
20130913 11:44:32 <sgothel> hehe .. dunno
20130913 11:44:37 <xranby> how else can you fix your starship
20130913 11:44:50 <sgothel> sure .. 'O' Brian' fixes everything :)
20130913 11:45:03 <xranby> ah he must have source in the brain
20130913 11:45:16 <sgothel> .. and Scotty ofc
20130913 11:45:47 <sgothel> so all thing peripheral .. should use such signatures .. and logging .. hmm
20130913 11:46:07 <sgothel> firewire was difficult, since it gave you access to the hardware bus itself
20130913 11:46:25 <sgothel> probably same w/ the new Intel hw bus .. thunder* whatever :)
20130913 11:47:12 <sgothel> usb depends on OS vulnerabilities AFAIK
20130913 12:03:05 <xranby> i am sure you have seen the usb rubber ducky www.usbrubberducky.com
20130913 12:05:31 <sgothel> hmm .. not that one :)
20130913 12:05:36 <sgothel> cute name
20130913 12:06:17 <sgothel> you may add HID identification to 'your' NEWT USB input tasks/bugs .. hmm :)
20130913 12:06:52 <sgothel> where did the world go .. you even cannot trust assumed humans anymore :)
20130913 12:09:41 <xranby> should we stop users who wants to use rubber duck "keyboards"? :)
20130913 12:10:34 <sgothel> hmm .. the UI/WM may only want to allow the HID used to login the user ?
20130913 12:11:05 <xranby> can we post pone this to when we aim to design a secure os?
20130913 12:13:45 <sgothel> right .. right after adding our brainwave scanner :)
20130913 12:19:34 <dfj> xranby: actually, I'm totally cool with breaking support for sufficiently crappy keyboards, or those in egregious taste.
20130913 12:20:22 <dfj> luckily I don't contribute. :)
20130913 12:20:50 <sgothel> :)
20130913 12:26:39 <dfj> I am a little peeved about the recent blog complaingin about oracles' warnings vs sandboxing cruft at present... if I want to run a sandboxed applet, with access to jogamp stuff, say... does this issue bite us? will my oracle vm users be getting big red warnings?
20130913 12:27:19 <sgothel> sadly yes .. and soon maybe not possible, so you would need to sign your jars ..
20130913 12:27:52 <sgothel> I saw the big red warning w/ our applet demos and 7u40
20130913 12:28:55 <dfj> ugh, so, outside the 'ease of install/keeping users uptodate' compared to making folks dl and run an app, this is not very awesome. :(
20130913 12:29:19 <sgothel> yes
20130913 12:29:32 <xranby> dfj: oracle is working hard to remove the java security sandbox
20130913 12:29:45 <sgothel> well, we have to test how it behaves .. signing your jar with your key and using our signed jars ..
20130913 12:29:54 <sgothel> i.e. from a user perspective ..
20130913 12:30:08 <dfj> I have few users at present, and they can be taught how to heed or ignore a warning, but... this is not goign to be good in teh future.
20130913 12:30:10 <sgothel> what we were talking about is the security aspect :)
20130913 12:30:24 <xranby> dfj: this all calls for a plugin replacement implemented using a free software license
20130913 12:30:32 <sgothel> so practically you have to sign your stuff
20130913 12:30:48 <sgothel> and maybe use our plugin when ready :)
20130913 12:31:08 <sgothel> -> Ji Gong -> JWeb -> SCC (wiki)
20130913 12:31:29 <dfj> yes - I would prefer that you folks folks got your holes patched, auditted, and signed libraries out there so that unsigned applets could use them...
20130913 12:31:41 <dfj> like... we just had working.
20130913 12:32:08 <sgothel> that is what all folks thing .. see links in our SCC wiki page
20130913 12:32:11 <xranby> i am confused why people put their trust in oracle
20130913 12:32:30 <sgothel> that is what Ji Gong is about - a choice
20130913 12:32:48 <sgothel> folks simply want to use an applet/app .. and hence use Oracle - the only provider
20130913 12:33:17 <xranby> alternatives exist like non browser applet viewer
20130913 12:33:18 <dfj> <grumpy old man voice> I've been in this industry for blah years, I've never seen oracle act as anything other than an adversary to anyone.
20130913 12:33:28 <xranby> netx for jnlp
20130913 12:34:12 <sgothel> @Xerxes: when all is build for all platforms -> Ji Gong's goal -> availability
20130913 12:34:30 <sgothel> as it is now, openjdk+icedtea is _not_ available on all platforms
20130913 12:35:00 <dfj> the sandbox is a hard problem, and a continual support problem, bring bad PR to the mothership... I'm not shocked, just sad.
20130913 12:35:01 <sgothel> so maybe that one dude is doing/helping w/ it .. or we have to do it alone :(
20130913 12:35:06 <xranby> netx that icedtea-web is based on worked on all platforms standalone http://jnlp.sourceforge.net/netx/pics.html
20130913 12:35:24 <xranby> the applet viewer is included in openjdk for all platforms as well
20130913 12:35:32 <xranby> openjdk/bin/appletviewer
20130913 12:35:32 <sgothel> @xerxes: you know what I mean ? Like 'download this' .. done!
20130913 12:35:36 <sgothel> tsts
20130913 12:35:36 <xranby> sure
20130913 12:36:04 <xranby> i know.. deployment sell
20130913 12:36:11 <sgothel> yes sir
20130913 12:36:20 <dfj> my users can handle a download... currently. But I would like to support less techy folks in the future.
20130913 12:36:33 <sgothel> so 'djf' maybe you like to jump in .. or somebody else ..
20130913 12:36:48 <sgothel> it's tedious hard work .. which must be done, we do the rest (plugin, SCC, ..)
20130913 12:37:09 <xranby> its like day of wonder say.. we want a deployment winner.. as long as somebody wins
20130913 12:37:10 <sgothel> 'dfj' :)
20130913 12:37:16 <dfj> I'm working on it - hanging out in a few dev chans to try and get my brain back in gear - I've been out of the industry for a few years. :/
20130913 12:37:17 <sgothel> hehe
20130913 12:37:37 <sgothel> read our wiki entries in this regard ..
20130913 12:38:08 <sgothel> oh well, so we were able to build manuallly for GNU/Linux .. now need the other OS ..
20130913 12:38:21 <dfj> could you relink that - I tagged my bookmark daftly last time and can't find it. :)
20130913 12:38:42 <sgothel> if we have GNU/Linux, Windows and OSX .. I guess we can start test deployment
20130913 12:39:02 <sgothel> https://jogamp.org/wiki/index.php/Main_Page
20130913 12:39:10 <sgothel> -> Ji Gong, SCC
20130913 12:41:00 * rmk0 (~rmk0@anon) Quit (Quit: leaving)
20130913 12:41:02 <dfj> oh - teh build kinda works for linux, but still needs to handle the others? I just poked my head into the end of a gradle migration last few weeks - my testcles are still sore. :/
20130913 12:41:52 <sgothel> yeah, and I like to see this work to be delegated as well - since I cannot do it all by myself, or by Xerxes ..
20130913 12:42:00 <sgothel> so a Ji Gong maintainer would be AWESOME!
20130913 12:43:21 <dfj> wish I knew good folks for you, I'll poke around, try to get folks interested.
20130913 12:43:37 <sgothel> .. even though I think, Xerxes _IS_ an ideal person doing it, he is so close to all the stuff (icedtea, openjdk, ..) :)
20130913 12:44:52 <sgothel> http://jogamp.org/wiki/index.php/Maintainer_and_Contacts#Maintainer <- Just add yourself to the Ji Gong position Xerxes, if you dare :)
20130913 12:44:53 <xranby> Martijn Verburg (karianna) do a good job as well working on OpenJDK deployment with the London JUG
20130913 12:45:02 <sgothel> sweet ..
20130913 12:45:32 <sgothel> well .. they are all welcome ofc .. and if we can trust them .. great
20130913 12:45:38 <xranby> Alex Kasko is also a champion who do package IcedTea 6 & 7 today for Windows Mac and Linux
20130913 12:45:47 <sgothel> trust: access to server and stuff .. to build the beast
20130913 12:46:00 <xranby> https://github.com/alexkasko/openjdk-unofficial-builds
20130913 12:46:05 <dfj> I personally can't do decent maintenance, I can only code when my sleep is working, which is about 1 month in three, randomly sprayed over the year. I envy those who can produce either consistent work, and those who can consistently produce work. :)
20130913 12:46:25 <dfj> either/and, whee, ugh.
20130913 12:46:27 <xranby> right now I, Gnu Andrew and Alex Kasko is testing what will become the next icedtea 7 2.4 release
20130913 12:46:31 <sgothel> so Alex has it all ..
20130913 12:46:35 <xranby> (that matches OpenJDK 7u40)
20130913 12:46:45 <sgothel> just need to add openjdk8 w/ profiles ..
20130913 12:46:57 <xranby> + getting it to build using free software
20130913 12:47:03 <sgothel> ofc
20130913 12:47:04 <xranby> Alex uses windows toolchain for windows
20130913 12:47:08 <xranby> currently
20130913 12:47:08 <sgothel> jikes
20130913 12:47:21 <dfj> jikes still is?
20130913 12:47:31 <sgothel> mingw64 preferable ..
20130913 12:47:40 <sgothel> :)
20130913 12:47:47 <sgothel> yikes ? :)
20130913 12:47:57 <dfj> I'm still on cygwin...
20130913 12:48:17 <dfj> jikes - used to be one of IBM's javac alternatives.
20130913 12:48:32 <sgothel> sure .. we use cygwin as the dev environment, also for sshd to our jenkins nodes
20130913 12:48:37 <sgothel> right
20130913 12:48:55 <dfj> got behind for a while, so I stopped tracking it - has it come back to life?
20130913 12:49:03 <sgothel> but native compiler is ming64 by now (32 and 64 bit)
20130913 12:49:55 <sgothel> @Xerxes: if we can have a chat session w/ Alex .. that would be great .. or email thread
20130913 12:50:17 <sgothel> i.e. introduce him to the idea .. and whether he likes to help and even has time
20130913 12:50:41 <sgothel> minimum benefit: share some experience
20130913 12:50:49 <dfj> so - did you laugh tghe first time you dug through the vm, and noticed that out at the edge, the C++ was so java -like, then deeper became more 'large program C++' ish, and finally in code-gen, it went all the way back to large, deep C functions with irregular, near mystical comments?
20130913 12:51:33 <dfj> - or cry?
20130913 12:51:56 <xranby> and in the middle is a scattered assembler and machine code core :)
20130913 12:52:02 <sgothel> oh .. the C++ <-> C thing etc ? well, that is quite normal today :)
20130913 12:52:29 <sgothel> yes .. asm rocks :)
20130913 12:52:54 <xranby> how else can the jit implement the native calls :)
20130913 12:52:59 <xranby> ?
20130913 12:53:10 <sgothel> sure sure
20130913 12:53:16 <xranby> some projects have stashed all the assembler into libffi
20130913 12:53:24 <xranby> to perform the calls
20130913 12:53:51 <dfj> no - not the transition, the different coders began to be visible with their styles, out at the edge teh conventions dominated - so without knowing them personally it was hard to see them in the code. Deeper it was clearly written by a tiny group of interesting people. I haven't dug around in there in 8 years, so it might be much cleaner now - or more insane.
20130913 12:55:51 <dfj> so - I thought I was clever replacing bytecode inserted code that would consistently compile->optimize to the same binary - and then i could safely rewrite from below, but then I noticed that sometiems I wsa not the first to have danced over the binary - there were other point patches in there...
20130913 12:56:22 <dfj> what really threw me was the first time I spotted teh 5-byte nop :)
20130913 12:57:03 <dfj> which - was a thing of beauty, once I realized that it let one patch consistently through line transitions on running code.
20130913 12:58:14 <sgothel> I guess you looked into it in good detail, me not so much
20130913 12:58:36 <dfj> I've alwyas been performance side - so tricks that native debuggers and such might have known about forever were pretty novel to me.
20130913 13:00:50 <dfj> well - I liked my work - I was writing minimal profilers to profile our other profilers... a bit recursive, but crazy fun - and the end goal was faster code for everyone, which is spiritually the closest thing I have to go on. :)
20130913 13:03:07 <sgothel> hmm, my last low level perf. enhancement was about realizing how expensive memcpy was for small chunks in a GL driver (attributes, uniforms)
20130913 13:03:36 <sgothel> then simply replacing it w/ loops .. some inline optimizations and gaining like 200% :)
20130913 13:03:59 <dfj> so - deciding when to make the call, and when to just copy by hand?
20130913 13:04:00 <sgothel> while doing so .. I saw that the next gcc in works does it already implicit :)
20130913 13:04:06 <sgothel> yes
20130913 13:04:40 <dfj> yeah - I whined about that exact same issue - and my friend linux-side was all - buts - that will be working so soon - this was over a year ago, though..
20130913 13:05:06 <sgothel> 2005/6 ..
20130913 13:06:29 <dfj> hmm - that would explain more - that winter I kinda could bare sleep at all - my memories from '05 to '08 are kinda... loose.
20130913 13:07:42 <dfj> I'm a little spooked byt eh stuff teh 'dose gl drivers do when 'linking' to your code... a bit aggressive in the thunking.
20130913 13:08:29 <dfj> I guess convincing users to make less tiny calls is and always will be, a lost cause.
20130913 13:09:36 <dfj> How is the overhead these days crossing from the vm to gl - does it still require setting up and passing a full JNI context or similar?
20130913 13:10:19 * dfj should really be testing this himself... easier to ask, though.
20130913 13:10:27 <sgothel> don't know really .. AFAIK it's like 1:1, no expensive switch
20130913 13:10:45 <sgothel> what may be expensive is to register/attach a thread to a JVM instance
20130913 13:11:23 <sgothel> but we do this rarely .. and iff, only once in a long term threads lifecycle, like a UI thread
20130913 13:11:32 <sgothel> gone for a while .. laters
20130913 13:11:36 <dfj> well, I like to fancy that would be uncommon, so latency there could be handled.
20130913 13:12:15 <dfj> have fun - thanks for entertaining me while I sip my coffee. :)
20130913 13:21:17 <xranby> (14:50:13) sgothel: @Xerxes: if we can have a chat session w/ Alex .. that would be great .. or email thread <-- email thread will most likely work, he is active on the distro-pkg-dev mailinglist http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2013-September/024640.html
20130913 13:48:53 <xranby> Henri Gomez is also a good candidate for Ji Gong maintance, he built the OBuildFactory project https://github.com/hgomez/obuildfactory have provided mac os x builds in the past , and linux builds now
20130913 13:51:55 <xranby> one idea is to write a nice email and send it out to Martijn Verburg, Alex Kasko and Henri Gomez
20130913 14:03:24 <xranby> Henri got jenkins scripts that we may try
20130913 14:09:32 <xranby> Me and Mark Wielaard have tested to perform builds using buildbot for linux http://builder.classpath.org/icedtea/buildbot/waterfall we use these to test icedtea builds
20130913 14:11:03 <xranby> and icedtea-web
20130913 16:57:07 * monsieur_max (~maxime@anon) has joined #jogamp
20130913 17:46:26 * hharrison (~chatzilla@anon) Quit (Remote host closed the connection)
20130913 21:02:47 * monsieur_max (~maxime@anon) has left #jogamp
20130913 21:16:45 * rmk0 (~rmk0@anon) has joined #jogamp
20130913 21:16:45 * rmk0 (~rmk0@anon) Quit (Changing host)
20130913 21:16:45 * rmk0 (~rmk0@anon) has joined #jogamp
20130913 21:21:28 * jk4 (~jk4@anon) Quit (Remote host closed the connection)
20130913 22:32:14 * void256 (~chatzilla@anon) has joined #jogamp
20130913 22:41:41 * rmk0 (~rmk0@anon) Quit (Quit: leaving)
20130913 22:42:03 * rmk0 (~rmk0@anon) has joined #jogamp
20130914 00:15:36 * void256 (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.90.1 [Firefox 23.0.1/20130814063812])
20130914 00:58:41 * jk4 (~jk4@anon) has joined #jogamp
20130914 05:06:12 -jogamp- Continue @ http://jogamp.org/log/irc/jogamp_20130914050612.html