JogAmp IRC Chat Logs

#jogamp @ irc.freenode.net - 20130206 05:06:16 (UTC)

20130206 05:06:16 -CatOut- Previous @ http://jogamp.org/log/irc/jogamp_20130205050602.html
20130206 05:06:16 -CatOut- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130206050616.html
20130206 07:15:39 * DemoscenePassiv (~Lutsche@anon) has joined #jogamp
20130206 07:19:38 * xranby (~xranby@anon) Quit (Ping timeout: 272 seconds)
20130206 07:21:47 * xranby (~xranby@anon) has joined #jogamp
20130206 07:45:36 * DemoscenePassiv (~Lutsche@anon) Quit (Ping timeout: 272 seconds)
20130206 09:13:01 * ebrayet (~Adium@anon) has joined #jogamp
20130206 09:21:31 * xranby (~xranby@anon) Quit (Ping timeout: 244 seconds)
20130206 09:24:03 * xranby (~xranby@anon) has joined #jogamp
20130206 09:53:17 <xranby> sgothel: the video editing is *awesome* !
20130206 09:53:47 <xranby> http://jogamp.org/doc/fosdem2013/
20130206 09:57:38 * xranby (~xranby@anon) Quit (Quit: Leaving.)
20130206 10:46:06 * rmk0 (~rmk0@anon) Quit (Read error: Operation timed out)
20130206 11:43:07 * xranby (~xranby@anon) has joined #jogamp
20130206 12:36:00 * rmk0 (~rmk0@anon) has joined #jogamp
20130206 12:36:00 * rmk0 (~rmk0@anon) Quit (Changing host)
20130206 12:36:00 * rmk0 (~rmk0@anon) has joined #jogamp
20130206 12:50:08 <sgothel> good morning
20130206 12:50:17 <sgothel> @Xerxes: thx ..
20130206 12:52:40 <sgothel> tip for kdenlive users .. if the source video sync is in question (here MTS/AVCHD files), import the files directly and transcode in kdenlive, i.e. -> DNxHD 720p 23.97fps 90Mb/s, only then audio/video is in sync. I had to re-edit the whole full session again :)
20130206 13:40:53 <sgothel> @Xerxes: You haven't tagged your blog entry properly to http://labb.zafena.se/?category_name=jogamp
20130206 13:44:36 <xranby> i will fix it now
20130206 13:45:15 <xranby> done!
20130206 13:49:46 <sgothel> thx
20130206 13:50:09 <sgothel> I like the details about the JVMs .. very very important!
20130206 13:50:49 <sgothel> I actually have ignored your obsession about the variety of JVMs in the past, but considering the 'free' attribute .. I honor it now very much
20130206 13:51:24 <sgothel> Also read your RoboVM .. very interesting alternative to GCJ
20130206 13:51:57 <xranby> sgothel: kudos
20130206 13:53:49 <sgothel> maybe .. we should set up a node for a few other VMs .. at least earmark it ..
20130206 13:54:02 <xranby> good point
20130206 13:54:07 <xranby> at least run the tests
20130206 13:54:44 <sgothel> yup .. even though this may explode the space of machines, i.e. jamvm+openjdk, jamvm+classpath, ..
20130206 13:55:05 <sgothel> but .. well, thats the price for freedom :)
20130206 13:55:50 <xranby> sure
20130206 13:55:57 <xranby> freedom got a high price
20130206 13:56:30 <sgothel> Bug 680: 10th Anniversary @ 2013-06-06 07:30:12 (UTC?) - haha
20130206 13:56:43 <sgothel> Maybe we can make this a theme on SIGGRAPH 2013 ?
20130206 13:57:01 <sgothel> .. even though a bit late, but always a reason to have a beer
20130206 13:57:08 <xranby> :)
20130206 13:57:29 <xranby> SIGRAPH will be a rocking party
20130206 13:58:13 <sgothel> note: while editing the full session .. we got around ~10min of being IDLE, which I cut-off
20130206 13:58:23 <xranby> amazing
20130206 13:58:32 <xranby> so much free time!
20130206 13:58:52 <sgothel> this can be a lesson for next time - better demo preparations, which gave us IDLE time
20130206 13:59:04 <sgothel> most -> monitor/projector setup I was fighting with :(
20130206 13:59:42 <sgothel> so maybe .. next time we rehearse the demos completly - I know you did, hence you were very fast!
20130206 14:00:24 <sgothel> then .. we could make it in one hour w/ lots of free time, depending on other material to show
20130206 14:02:55 <xranby> we should use a chess clock
20130206 14:03:02 <xranby> haha
20130206 14:03:29 <sgothel> idle/active - giving us active time 40min :)
20130206 14:03:38 <sgothel> old man need a rest ..
20130206 14:03:47 <xranby> it was an amazing presentation
20130206 14:04:09 <xranby> i ran it for non technical biology experts here and they gasped the content!
20130206 14:04:19 <sgothel> good back/forth talk .. I agree, sure I am nagging about my 'errr ..' and 'so ..' :)
20130206 14:04:46 <sgothel> I could cut off some of our 'moments' though :)
20130206 14:05:08 <xranby> why that was an eye opener
20130206 14:05:24 <xranby> made the whole thing memorable
20130206 14:05:51 <sgothel> yes .. and I heard your entry questions now .. indeed, nobody really knew about the hardware thingy
20130206 14:06:11 <xranby> it was soo nice that the person in front of the camera did not raise the hand on the third question
20130206 14:10:22 <sgothel> So .. to make it official: Also a great KUDOS for Qun, our dear camera-woman!
20130206 14:12:23 <xranby> i have added that to the post now :)
20130206 14:14:10 <sgothel> nothing about good journalism - and it's nice to have my w/ t-shirt like a bum next to all the dress man on stage :)
20130206 14:14:41 <sgothel> our love for diversity :)
20130206 14:15:37 <sgothel> @ my 1st talk @ JavaOne .. Ken was asking, 'so you want to wear this?' .. me having a t-shirt on, 'ok, well then' :)
20130206 14:16:47 <xranby> i dont remember, did you wear a suit then?
20130206 14:16:53 <xranby> i have tried to locate the vide
20130206 14:16:57 <sgothel> hell no - never .. ever
20130206 14:17:00 <xranby> but after oracle re-org its hard to locate
20130206 14:17:16 <sgothel> yeah, since they seem to not love me anymore, they removed my content :)
20130206 14:17:17 <xranby> good to stay true to oneself
20130206 14:17:33 <xranby> since thats the only one you can be sure of will love you in the end
20130206 14:21:45 <sgothel> oho .. sad but sure, always working on a better finale :)
20130206 16:17:20 <sgothel> Bug 681 - https://jogamp.org/bugzilla/show_bug.cgi?id=681 - working on it now .. hope they can re-use it ..
20130206 16:21:29 <xranby> nice, i didnt think about using the gluegen struct accessor! :D
20130206 16:22:40 <sgothel> so we make a 'static' pre-generated impl. of it - like NativeLib stuff .. solving chicken-egg problem and allowing others to reuse it
20130206 16:24:22 <xranby> lovely @neugens the freejava devroom organiser sent the oracle community manager this http://t.co/LDmaGPvZ
20130206 16:26:28 * xranby (~xranby@anon) Quit (Quit: Leaving.)
20130206 16:28:36 <sgothel> people are so 'fan boy', " I love it! I'll promise to match it with a small red hat ;)" .. better doing a bald bold software fork :)
20130206 17:11:03 <sgothel> for fun: glibc fixes value of TWO .. http://sourceware.org/ml/glibc-cvs/2013-q1/msg00115.html :)
20130206 17:18:29 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130206 17:38:54 <hharrison> Lookig at bug 681...gluegen can _do_ that? I really need to find more time to play with it
20130206 17:39:21 <sgothel> Of course .. otherwise we could not access return structures at all :)
20130206 17:39:30 <hharrison> Excuse the crap link above, I meant https://jogamp.org/bugzilla/show_bug.cgi?id=681
20130206 17:39:35 <sgothel> we even handle all the alignment per platform and validate that ..
20130206 17:40:01 <hharrison> Ahh alignment, I have a special place in my heart for that after doing some kernel work
20130206 17:40:02 <sgothel> 681 .. using StructAccessor .. yes
20130206 17:40:35 <sgothel> yeah .. and we do some fancy validation and dropping all the magic numbers in the generated struct-class
20130206 17:41:02 <sgothel> we have a static assumption at compile time which must match the runtime validation ..
20130206 17:41:29 <sgothel> it's pretty much elaborated in the gluegen package html page ..
20130206 17:41:55 <hharrison> OK, gluegen is officially moving up on my cool shit list
20130206 17:42:06 <sgothel> thank you :)
20130206 18:48:24 <rmk0> there's been quite a bit of talk here lately of getting something running so that we can still write java and target platforms without a jvm
20130206 18:48:35 <rmk0> so... anyone looked at: http://vmkit.llvm.org
20130206 18:48:35 <rmk0> ?
20130206 18:50:58 <sgothel> interesting .. look @ Xerxes blog .. RoboVM http://labb.zafena.se/ http://www.robovm.org/ - GCJ doesn't catch up (still java5) .. then Avian-VM bundle ..
20130206 18:51:14 <sgothel> we should have a matrix on wiki .. for all combinations
20130206 18:52:25 <sgothel> another good question is .. how is this related to our 'free 'java' talk .. Oracle's license (GPLv2 enough? TCK license impossible?) ..
20130206 18:52:55 <rmk0> i'm still thinking about that
20130206 18:53:07 <rmk0> i've not come to enough of a conclusion to dump in an email...
20130206 18:53:25 <rmk0> removing the stuff from the javax.* namespace would be a good first step, i think
20130206 18:53:28 <sgothel> yeah - almost looks like a 'religious' thing
20130206 18:53:58 <sgothel> meaning to obey the view of Oracle regarding their Java/TCK license :)
20130206 18:53:59 <rmk0> maybe offer a compatiblity package that allows old programs to run with those old names
20130206 18:54:14 <sgothel> no - either or :)
20130206 18:54:34 * rmk0 bursts
20130206 18:54:46 <sgothel> I don't care here .. since I don't like to obey Oracle's view on their license :)
20130206 18:54:48 <rmk0> back in a few hours!
20130206 18:55:03 <sgothel> ay - however - if you get all our folks together to move package name - I am fine w/ it !
20130206 18:55:20 <sgothel> this would be a major communication undertaking though!
20130206 19:10:14 * ebrayet (~Adium@anon) has joined #jogamp
20130206 19:10:32 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=682 - for discussion
20130206 19:10:58 * DemoscenePassiv (~Lutsche@anon) has joined #jogamp
20130206 19:11:45 <sgothel> Hi Dominik .. when did we plan to meet ?
20130206 19:16:08 <sgothel> adding 'voting' in bugzilla again :)
20130206 19:41:28 <DemoscenePassiv> hiho :) feb 16. was the plan :)
20130206 19:41:53 <sgothel> ah - good, thx
20130206 19:43:13 <DemoscenePassiv> is the date still fine with you ?
20130206 19:43:18 <sgothel> YUP
20130206 19:43:25 <DemoscenePassiv> roger that :)
20130206 19:43:58 <DemoscenePassiv> btw: awesome fosdem video, just watching it
20130206 19:44:32 <sgothel> thx
20130206 19:44:49 <sgothel> see my kdenlive notes above .. in the log
20130206 19:48:55 <hharrison> Truthfully,I'm comfortable with anything OpenJDK derived
20130206 19:49:39 <hharrison> Oracle only had even half a leg to stand on in the Google case as Dalvik was a new implementation
20130206 19:50:28 <sgothel> yup - no patent case, no real copyright case (1 USD)
20130206 19:51:11 <hharrison> Well, the copyright case was even more of a joke re: API copyrightability
20130206 19:53:13 <hharrison> The implicit patent grant in GPLv2 is good enough for me
20130206 19:53:31 <hharrison> re: no further restrictions on distribution/use
20130206 19:54:03 <sgothel> but is that true ?
20130206 19:54:03 <hharrison> The TCK/JCK stuff is way more about Trademark IMHO
20130206 19:54:32 <sgothel> i.e. is there such a thing like patent grant via GPLv2 - GPLv3/APLv2 differs
20130206 19:54:38 <hharrison> Hasn't been tested any where I've seen
20130206 19:54:50 <sgothel> the latter have it in explicitly ..
20130206 19:55:01 <sgothel> yes, I don't care the name of the VM either :)
20130206 19:55:14 <hharrison> Yes, in the GPLv2 there is only the implicit grant
20130206 19:55:39 <sgothel> and they didn't raised it in the google case .. hmm
20130206 20:06:19 * jk4 (~jk4@anon) has joined #jogamp
20130206 20:23:59 <sgothel> Updated Bugzilla to 4.2.4
20130206 20:34:44 <hharrison> I'm curious what people like Azul systems do, are they licensing the TCK and getting certified, or just doing teir own thing
20130206 20:36:02 <hharrison> Nevermind, they have signed the TCK agreements
20130206 20:53:00 * rmk0 reappears
20130206 21:39:40 * xranby (~familjen@anon) has joined #jogamp
20130206 21:44:39 <xranby> http://blog.krisk.org/2013/02/packets-of-death.html whoopsie! a reply packet that kan kill your intel network NIC!
20130206 21:46:54 <xranby> "This customer location, for some reason or another, could predictably bring down the ethernet controller with voice traffic on their network."
20130206 21:48:01 <xranby> "Nothing but a power cycle would bring it back."
20130206 21:48:44 <sgothel> ooopsie .. hope I won't get those
20130206 21:48:58 <sgothel> no - I don't tell you my ethX :)
20130206 21:49:56 <xranby> ddos attackers will have a fun year
20130206 21:50:03 <xranby> or decade
20130206 21:50:35 <xranby> apparently if you installa level 2/3 firewall you might be safe
20130206 21:51:02 <sgothel> whatever that is - me running iptables .. but external eth faces the world ofc
20130206 21:51:14 <sgothel> well - at least the firewall machine
20130206 21:52:43 <sgothel> hmm .. another low-power gbit fw-route w/ linux/iptables ..
20130206 21:52:50 <hharrison> BTW, we've switched to the atomic jars in our jogl use, was rather painless, well done
20130206 21:53:16 <sgothel> nice - yes, I validate content w/ a script every now and then :)
20130206 21:53:52 <sgothel> jogl/make/scripts/lstjars.sh
20130206 21:54:00 <xranby> hharrison: nice to know, i think libgdx might be a good target for the atomic stuff as well
20130206 21:54:14 <xranby> and jme3
20130206 21:54:20 <sgothel> however - IMHO it's more slow to download many little JARs - compared w/ one big JAR
20130206 21:54:38 <sgothel> @Xranby: Pls don't - the lag would explode, all well tested
20130206 21:54:58 <sgothel> enough that we have gluegen/jogl ..
20130206 21:55:16 <sgothel> consider that for each JAR file - even if cached - a connection is established
20130206 21:55:29 <sgothel> that will kill all internet applet/jnlp usage
20130206 21:55:33 <hharrison> Internally I just wrap it in exe4j as we deploy a all-in-one exe so there's very little penalty, and a great size savings (~2.5M)
20130206 21:55:46 <sgothel> ok .. thats fine then
20130206 21:56:01 <sgothel> for internet .. one might also use jar-in-jar ..
20130206 21:56:01 <hharrison> If we ever have to deploy externally, the calculus would likely change
20130206 21:56:35 <sgothel> but what are 2.5M today ? .. some nice ubuntu fonts for example :)
20130206 21:56:46 <hharrison> Speaking of hitting the internet per file
20130206 21:56:51 <sgothel> (I know .. graph will be dislocated ..)
20130206 21:57:08 <sgothel> hehe .. speaking of launching java < 0.3s
20130206 21:57:10 <hharrison> You know that calculating the hashcode of a URL object in Java does DNS resolution, right?
20130206 21:57:32 <hharrison> So those nice hashtables you have in gluegen/jogl with URLs as the keys....
20130206 21:58:02 <sgothel> really?
20130206 21:58:05 <hharrison> Yep
20130206 21:58:10 <hharrison> Saw that last night
20130206 21:58:17 <hharrison> It's blocking too
20130206 21:58:17 <sgothel> opsa - and now you tell me :)
20130206 21:58:26 <sgothel> thats a major bug
20130206 21:58:38 <sgothel> when does it happen ?
20130206 21:58:48 <hharrison> Enjoy:
20130206 21:58:50 <hharrison> http://findbugs.sourceforge.net/bugDescriptions.html#DMI_BLOCKING_METHODS_ON_URL
20130206 21:58:59 <hharrison> I was triaging some findbugs stuff
20130206 21:59:09 <sgothel> I start to love your tools ..
20130206 21:59:15 <sgothel> how to enable w/ jenkins ?
20130206 21:59:16 <hharrison> Fixed this last year in our code, sometimes latency can be a real bitch
20130206 21:59:31 <rmk0> findbugs is pretty good
20130206 21:59:45 <hharrison> findbugs has a standlone jar your can tie in as a jenkins task....lemme see if I can find it
20130206 21:59:54 <sgothel> I remember .. I had a colleague in Russia while working on plugin stuff - he always had a different opinion new to us :)
20130206 22:00:01 <hharrison> Forget pretty good, it is _amazing_
20130206 22:00:24 <sgothel> who files this bug ?
20130206 22:00:55 <hharrison> I can if you wish, I'm away from my home machine currently, but I will tonight with the file/line numbers that pop out
20130206 22:01:10 <sgothel> ok .. then I do it - but thx
20130206 22:01:20 <hharrison> https://wiki.jenkins-ci.org/display/JENKINS/FindBugs+Plugin
20130206 22:01:40 <hharrison> Although I have no experience with the above, I do the srcgen step and then build in eclipse
20130206 22:01:42 <sgothel> I saw it once .. hmm .. I guess we have it .. but maybe not configured or so ..
20130206 22:02:09 <sgothel> so FindBugs in Eclipse you use ..
20130206 22:02:13 <hharrison> Yeah
20130206 22:02:32 <hharrison> It uses a tonne of ram for some analysis, so I only do it periodically
20130206 22:02:43 <sgothel> ah .. so better not on jenkins :)
20130206 22:02:53 <xranby> sgothel: can you use it for the aggregate builds?
20130206 22:02:57 <sgothel> post build / srcgen
20130206 22:03:00 <hharrison> yep
20130206 22:03:08 <xranby> scan the src zip
20130206 22:03:13 <sgothel> any special setup or DB you use ?
20130206 22:03:22 <sgothel> @XRanby: 1st manual - then we see ..
20130206 22:03:34 <hharrison> Nope, just sort by findbugs 'severity' and start reading
20130206 22:03:37 <xranby> thumbs up on that
20130206 22:03:41 <sgothel> no rush hush here or even too much hype/excitement please :)
20130206 22:03:51 <xranby> it would be sad panda to trash an aggregate build :)
20130206 22:03:53 <sgothel> ay .. will do later
20130206 22:04:18 <hharrison> I basically use the ant build to get me to the gensrc, and have my eclipse set up with a bunch of linksource targets
20130206 22:04:29 <hharrison> loosely following Wade Walkers stuff in the wiki
20130206 22:05:15 <hharrison> It found some really neat stuff in the j3d repo a year or more back...my personal favorite:
20130206 22:05:23 <sgothel> so - "Don't use java.net.URL. That's the simple answer to your question. Use java.net.URI instead, which won't do hostname resolution."
20130206 22:05:35 <hharrison> http://findbugs.sourceforge.net/bugDescriptions.html#ML_SYNC_ON_FIELD_TO_GUARD_CHANGING_THAT_FIELD
20130206 22:05:44 <hharrison> yep
20130206 22:06:09 <hharrison> REad the javadoc for URL::hashcode, it's criminally insane
20130206 22:06:37 <hharrison> But they couldn't change the decision later...which is most of the reason URI exists really
20130206 22:06:38 <sgothel> haha
20130206 22:07:02 <hharrison> Oh yeah, URL::equals is also blocking for the same reason
20130206 22:07:02 <sgothel> just read your last one .. sweet, yeah .. learned that in 2006 or so
20130206 22:07:27 <hharrison> I never saw it in practice until Java3d
20130206 22:07:34 <sgothel> very good, will do that plus FindBugs .. i.e. you already found a bunch ..
20130206 22:08:29 <hharrison> Trust me, jogl is in good shape, the URL thing is a papercut, most of the stuff it finds is 'bad/confusing practice' stuff
20130206 22:08:46 <hharrison> Not 'this code is going to blow up' stuff
20130206 22:08:48 <sgothel> nice then - less scared
20130206 22:09:11 <hharrison> I already sent patches for things that were even remotely worth it
20130206 22:09:15 <xranby> sgothel: is any of the jogamp demos using openmax nowdays?
20130206 22:09:19 <sgothel> yeah .. I tend to confuse w/ final and inner separating blocks :)
20130206 22:09:40 <xranby> i am chatting with noxo on twitter trying to find any working OpenMAX java audio library
20130206 22:09:47 <sgothel> @XRanby: well .. yes, but I had to strip down OpenMAX impl. due to license issues
20130206 22:09:53 <hharrison> There's one class of bug you may want to check for, it makes some debug statements way tless useful
20130206 22:10:17 <sgothel> @hh: shoot
20130206 22:10:24 <hharrison> http://findbugs.sourceforge.net/bugDescriptions.html#DMI_INVOKING_TOSTRING_ON_ARRAY
20130206 22:10:26 <xranby> sgothel: really? so the khronos royality free etc spec is all vapour?
20130206 22:12:06 <sgothel> @XRanby: 46f17013c7cd59d551371edb2c1a4a57f8cbd84f
20130206 22:12:49 <sgothel> and 1be7b5271ec1d08b6929a88dd811754d33d149de
20130206 22:13:30 <sgothel> it was using an NV extension for parsing multiplexed streams ..
20130206 22:13:36 <sgothel> this is not included in OMX :(
20130206 22:13:51 <sgothel> w/o .. it's sort of useless .. so we would need that back
20130206 22:14:07 <sgothel> it's like the missing WAV 'parser' in JOAL :)
20130206 22:14:26 <sgothel> .. which is ofc very easy to do .. on should be on your list ..
20130206 22:14:44 <xranby> hmm.. i have to take a look how the raspberry pi openmax audio implementation work and is licensed
20130206 22:15:09 <xranby> at least there is an openmax audio and video example in /opt/vc/src on the Raspbian distribution
20130206 22:22:43 <xranby> @sgothel yes an platform agnostic parser world be good
20130206 22:22:55 <xranby> how did you parse the wav on android now again then?
20130206 22:23:08 <sgothel> I didn't - see unit test :)
20130206 22:23:09 <xranby> without using the java standard lib
20130206 22:23:25 <sgothel> TestResource .. or something .. using raw data
20130206 22:23:28 <xranby> i will look it up
20130206 22:23:48 <sgothel> the WavData class is platform agnostic ..
20130206 22:24:03 <sgothel> WavUtil .. has to be fixed
20130206 22:25:16 <sgothel> when working on JOAL audio output .. for our GLMediaPlayer libav backend .. we may see some overlap of use case, i.e. decoding lossy audio codecs ..
20130206 22:36:43 <sgothel> @Xerxes: https://jogamp.org/bugzilla/show_bug.cgi?id=684 + you are default assignee for all JOAL modules (as maintainer)
20130206 22:38:33 <xranby> thumbs up
20130206 22:40:02 <xranby> The world is filled with engineering puzzles, engineer game on
20130206 22:40:33 <sgothel> hehe
20130206 22:42:57 <hharrison> We've got a trivial implementation of a VP8 encoder in pure java...useful for doing in-app screencasting
20130206 22:43:22 <hharrison> If we're on the lookup for random format utils :-)
20130206 22:43:49 <sgothel> I was thinking of adding WebM (vp8?) decoder via libav to our default deployment .. could add the encoder as well ?
20130206 22:43:55 <sgothel> which license ?
20130206 22:44:09 <sgothel> vp8 <= webm .. I guess ?
20130206 22:44:09 <hharrison> gpl3, but could be relicensed very likely
20130206 22:44:19 <sgothel> I thought encoding is very costly
20130206 22:44:33 <hharrison> Not if you don't ry to compress very hard
20130206 22:44:54 <sgothel> how about the libav/webm though ?
20130206 22:44:56 <hharrison> We use it to get a quick intermediate file we then can pass easily to a real video editing prog
20130206 22:44:57 <sgothel> *thought
20130206 22:45:23 <hharrison> Prob not a bad idea
20130206 22:45:33 <sgothel> sounds very nice, ofc a dead-end for mobile
20130206 22:46:17 <sgothel> so you think I am safe deploying it ? patents .. bla ? probably should add another bug entry for that .. since my fingers are still warm :)
20130206 22:46:48 <hharrison> For vp8? I'm willing to take google at its word that it is not patent encumbered
20130206 22:47:10 <sgothel> you know - the MPEGLA is still trying hard ..
20130206 22:47:22 <hharrison> But at some point, just making it easy for people to add it themselves instead of doing explicit integration becomes more scalable
20130206 22:47:22 <sgothel> (well - webm is a shitty format .. however ..)
20130206 22:47:48 <hharrison> Curious, what did you find lacking in WebM?
20130206 22:47:50 <sgothel> like system installed libav you mean ?
20130206 22:47:57 <hharrison> yeah
20130206 22:48:11 <sgothel> thats the way we do it now - ok for linux, bad for Windows/OSX
20130206 22:48:27 <sgothel> - or we need to write backends for their native crap
20130206 22:49:31 <sgothel> webm .. quality .. I guess I posted it here .. hmm, was regarding webm'ish audio I guess .. one sec
20130206 22:49:53 <hharrison> Or do you explicitly mean the WebM container format, the VP8 bitstream is odd, but understandable
20130206 22:50:22 <hharrison> As for audio, that's still a wee bit of a clusterfuck until libOpus becomes widespread (fingers crossed)
20130206 22:51:20 <sgothel> the mozilla folks bitch about the Google owned specs etc .. which are not even there yet..
20130206 22:51:37 <hharrison> Oh, the specs are shit
20130206 22:51:50 <hharrison> But the code is a way better spec anyways ;-)
20130206 22:51:54 <sgothel> however, the x264 dev .. dude, compared video formats .. let me find it - the images were clearly in favor of some other open source format :)
20130206 22:52:04 <hharrison> I actually found libVPX pleasent to explore
20130206 22:52:50 <sgothel> the guys said .. webm is SNR optimized, but gives a shitty picture due to the lack of those psycho profiles
20130206 22:52:54 <hharrison> dark_shikari?
20130206 22:53:30 <hharrison> I remember being thoroughly unimpressed with that blog post
20130206 22:54:06 <hharrison> It was a great critique of one _implementation_, but sorely lacking analysis of what the format allowed
20130206 22:54:37 <hharrison> And to be fair, the early libVPX implementation was pretty slow and awful
20130206 22:56:50 <hharrison> But (if I remember correctly) VP8 compares very favorably to h264 base profile
20130206 22:57:21 <sgothel> trying to find the picture comparison .. where it shows the difference, h264 >> webm
20130206 22:57:22 <hharrison> Which is all the vast majority of devices have on the hardware accel side
20130206 22:57:30 <sgothel> even theora was better
20130206 22:57:53 <hharrison> Hmm, maybe I'm thinking of a different post then
20130206 22:59:30 <sgothel> (trying my memory hard) .. right .. I picked it up about WebP https://bugzilla.mozilla.org/show_bug.cgi?id=600919
20130206 23:00:22 <sgothel> so it's the guy you mentioned .. http://x264dev.multimedia.cx/archives/541
20130206 23:00:40 <sgothel> but look at the pictures ..
20130206 23:00:55 <sgothel> maybe google will add those psycho-profiles ..
20130206 23:01:57 <hharrison> I was thinking of this:
20130206 23:01:58 <hharrison> http://x264dev.multimedia.cx/archives/377
20130206 23:03:43 <sgothel> http://x264dev.multimedia.cx/archives/541#comment-7294 .. too-long .. but in essence w/ post: Better to have a proper open source patent-free framework codec to apply and 'learn' those hard-to-get psycho-profiles
20130206 23:04:27 <sgothel> ofc .. google is also pretty arrogant .. no-specs, half-ass ready stuff, expecting the world to follow suite b/c, hey, it's Google :)
20130206 23:04:48 <sgothel> (the Mozilla .. point)
20130206 23:05:51 <sgothel> back to the real world - yeah, so webm-libav .. maybe packed .. at least for OSX/Windows
20130206 23:06:03 <sgothel> .. fallback solution at least
20130206 23:06:59 <sgothel> it really sucks that the provided libav in chrome and mozilla (?) cannot be used - I get an error 'network connection refused, or alike'
20130206 23:15:12 <sgothel> enough making myself busy w/ bug entries :)
20130206 23:32:34 * ebrayet (~Adium@anon) Quit (Quit: Leaving.)
20130206 23:55:44 <xranby> good, fosdem now links to the jogamp slides and video page as well! https://fosdem.org/2013/schedule/event/jogamp_fast_media_processing/
20130206 23:56:25 <sgothel> don't hold your breath waiting for a positive message from the overlords :)
20130206 23:56:33 <sgothel> i.e. on openjdk or java.net :)
20130206 23:56:59 <xranby> we get possitive feedback from the raspberry pi community
20130206 23:57:00 <sgothel> usually they think 'it's not invented here' .. even though, in this case, it is :)
20130206 23:57:47 <xranby> http://www.raspberrypi.org/phpBB3/viewtopic.php?f=26&t=32556&p=280634&hilit=java#p280634
20130206 23:58:15 <xranby> note. teh raspberry pi forum pages are locked after edit
20130206 23:58:24 * DemoscenePassiv (~Lutsche@anon) Quit (Ping timeout: 264 seconds)
20130206 23:59:13 <xranby> so in a nutshell the users like what we do and share the love
20130206 23:59:21 <sgothel> too bad Dominik is gone already, maybe you can convince him of a RaspPi demo entry
20130206 23:59:47 <sgothel> nice nice
20130207 00:00:17 <sgothel> I am [also] satisfied by our work and communication
20130207 00:00:33 <sgothel> good stuff, fun and self directed
20130207 00:02:22 <sgothel> OMX: noox likes to jump on this task ?
20130207 00:03:21 <sgothel> sure, if we fix & test the OMX deplex/streamer we have omx_util .. we can use it for a/v or audio only as well
20130207 00:03:45 <xranby> he is currently working on a c64 java emulation demo and is having some issues with the SID emulation sounding like a modulated /dev/random instead of the funky bazz you expected
20130207 00:03:58 <sgothel> before my NV ext. removal - it worked well on Tegra 2500 thingy back in 2006 or so
20130207 00:04:14 <sgothel> fascinating :)
20130207 00:04:20 <xranby> so in a nutshell he want an OpenMAX IL implementation
20130207 00:04:28 <xranby> for audio processing
20130207 00:05:04 <sgothel> our omx_util hooks to OMX IL, yes - just not a complete binding, since we only used a minor subset
20130207 00:05:43 <sgothel> but: stream -> [buffer|device] .. why not ?
20130207 00:06:12 <sgothel> afaik that is already somwhat implemented
20130207 00:07:27 <xranby> needs testing for sure, sounds promising
20130207 00:10:05 <sgothel> ok .. where to find an executable/lib of the JVM in use to test ElfHeader ?
20130207 00:11:01 <xranby> sgothel: try the libjvm.so
20130207 00:11:26 <sgothel> in one of these:
20130207 00:11:29 <sgothel> contains(System.getProperty("sun.boot.library.path"), gnueabihf) ||
20130207 00:11:29 <sgothel> contains(System.getProperty("java.library.path"), gnueabihf) ||
20130207 00:11:29 <sgothel> contains(System.getProperty("java.home"), gnueabihf) )
20130207 00:11:29 <sgothel> ?
20130207 00:11:55 <sgothel> I have to load them via file-io
20130207 00:12:01 <xranby> relative to java.home perhaps
20130207 00:12:40 <xranby> jre/lib/*/*/libjvm.so
20130207 00:13:04 <xranby> it is usually something like jre/lib/arm/server/libjvm.so
20130207 00:13:19 <xranby> note that the os.arch name arm is here
20130207 00:13:31 <sgothel> yup .. 'just' need to locate it properly .. hmm
20130207 00:13:56 <sgothel> jre/lib/<os.and.arch>/*/libjvm.so .. :) or something
20130207 00:14:57 <xranby> on linux there is a shortcut by reading /proc/self/exec
20130207 00:15:20 <xranby> sorry /proc/self/exe
20130207 00:15:48 <xranby> it is a symbolic link to the runtime
20130207 00:16:31 <xranby> you will then red the elf header of the java / javac etc launcher
20130207 00:17:20 <sgothel> nice
20130207 00:18:08 <xranby> sweet and short
20130207 00:18:26 <sgothel> for linux, yes .. well .. we will see
20130207 00:18:51 <xranby> at least same for android!?
20130207 00:19:06 <sgothel> should be -> linux, let's see ..
20130207 00:19:40 <xranby> http://stackoverflow.com/questions/1023306/finding-current-executables-path-without-proc-self-exe
20130207 00:20:07 <xranby> mark4o: Mac OS X: _NSGetExecutablePath() (man 3 dyld) Linux: readlink /proc/self/exe Solaris: getexecname() FreeBSD: sysctl CTL_KERN KERN_PROC KERN_PROC_PATHNAME -1 BSD with procfs: readlink /proc/curproc/file Windows: GetModuleFileName() with hModule = NULL
20130207 00:20:23 <sgothel> you are funny :)
20130207 00:20:43 <sgothel> we cannot use native code until we have analyzed something .. :)
20130207 00:20:59 <xranby> yes that raises the bar a little
20130207 00:21:04 <sgothel> hehe
20130207 00:21:15 <sgothel> but env vars might be helpful ..
20130207 00:21:43 <sgothel> ok .. I just start testing w/ self/exe .. then we see
20130207 00:26:31 <xranby> the bruteforce way is to try load all native libs and rely on the runtime exception when loading fails
20130207 00:26:37 <xranby> should be fast if you have all libs on disk
20130207 00:26:58 <xranby> recent linux distributions ld is fixed to not crash
20130207 00:27:05 <xranby> when loading the wrong lib
20130207 00:27:15 <xranby> and report back nicely
20130207 00:27:41 <xranby> this is a platform neutral way to do it
20130207 00:27:43 <sgothel> again: since we do internet connections .. this is not a solution
20130207 00:28:24 <sgothel> this method is also utilized when requesting the native JAR file from an URL ..
20130207 00:28:26 <xranby> well... for jnlp we have to rely on os.arch yes? or else load all libs like we do on arm already
20130207 00:28:43 <xranby> when os.arch is not specific enough
20130207 00:29:17 <sgothel> IMHO the ElfHeader way is good for at least GNU/Linux, hence most of the mobile market for which we have no proper os.and.arch
20130207 00:30:29 <sgothel> and it would be great to have this in libGDX etc as well .. brute force is something not reliable
20130207 00:43:43 <sgothel> [0..4]: 0x7f, 0x45, 0x4c, 0x46, valid: true .. identity worked .. now the flags :)
20130207 00:44:52 <sgothel> do you remember the hard_float soft_float ABI/ELF stuff ?
20130207 00:54:04 <xranby> readelf -A /proc/self/exe | grep Tag_ABI_VFP_args
20130207 00:54:42 <sgothel> yup thx - will fiddle through the bits and readelf sources / specs
20130207 00:55:46 <xranby> for reference last years discussion: http://icedtea.classpath.org/wiki/New_os.arch_namespace_Architecture
20130207 00:57:57 <sgothel> hehe .. thx
20130207 00:59:40 <xranby> i dont know about the music or artists, but the dude on to the right >> looks ok https://twitter.com/Pancho_D6/status/298968822369624064/photo/1
20130207 01:01:12 <xranby> the cover is apart from that rubbish and provoking vomits
20130207 01:01:56 <sgothel> ah ja .. well, for some reason my eyeballs were focusing the lower center :)
20130207 01:02:26 <sgothel> who pays you to distract me from the elf-reader task ? :)
20130207 01:02:53 <xranby> really sorry.. but there is some trend printing free java on tshits taking off on twitter
20130207 01:03:46 <sgothel> hehe
20130207 01:04:01 <sgothel> ask them where we can print our jogamp t-shirt ..
20130207 01:04:28 <sgothel> good quality - not that flocked stuff .. plastic .. i.e. real color imprint
20130207 02:01:41 * hharrison (~chatzilla@anon) Quit (Quit: ChatZilla 0.9.89 [Firefox 19.0/20130123083802])
20130207 03:47:09 * xranby (~familjen@anon) has left #jogamp
20130207 03:55:08 * hharrison (~chatzilla@anon) has joined #jogamp
20130207 05:05:29 -CatOut- Continue @ http://jogamp.org/log/irc/jogamp_20130207050529.html