#jogamp @ irc.freenode.net - 20130424 05:05:47 (UTC)

20130424 05:05:47 -CatOut- Previous @ http://jogamp.org/log/irc/jogamp_20130423050634.html
20130424 05:05:47 -CatOut- This channel is logged @ http://jogamp.org/log/irc/jogamp_20130424050547.html
20130424 06:21:54 * DemoscenePassiv (~Lutsche@anon) has joined #jogamp
20130424 06:46:54 * DemoscenePassiv (~Lutsche@anon) Quit (Ping timeout: 256 seconds)
20130424 07:26:41 * [Mike] (~Mike]@anon) Quit ()
20130424 08:12:36 <sgothel> good morning
20130424 08:23:38 <xranby> sgothel: good morning!
20130424 08:53:30 <sgothel> got such a amd-mobile-radeon laptop, will test later after installing windows
20130424 08:55:00 <sgothel> tring to get a whole build through - but jenkins doesn't block jobs properly -> network failure and overlapping builds :(
20130424 08:55:12 <sgothel> however .. I checked the ARM tags etc
20130424 08:55:40 <sgothel> Xerxes: Pls check gluegen, openal-soft, joal, jogamp-scripting commits
20130424 08:56:02 <sgothel> gluegen has a shell script to validate all ARM libaries for example
20130424 08:56:34 <sgothel> we also force align the GLIBC version to be <= 2.4
20130424 08:57:43 <sgothel> I guess we could use a JOAL version NApplet .. using a JOALVersion class (if we don't have it already) w/ extensions etc .. hmm
20130424 08:58:13 <sgothel> besides .. linux-32bit wasn't working, fixed a week ago (I guess)
20130424 09:45:24 <xranby> sgothel: do you want me to populate /usr/local/x-tools/ and run the build.all scripts?
20130424 09:56:53 <xranby> sgothel: hmm [cc] qemu: uncaught target signal 11 (Segmentation fault) - core dumped
20130424 09:57:27 <xranby> ah it is running qemu-x86_64-static on my 32bit system
20130424 10:42:48 <sgothel> you are kidding ? uh :)
20130424 10:47:44 <sgothel> I don't see a qemu source tar ball here .. or compilation artifact .. hmm
20130424 10:48:13 <xranby> my machine automagically started to use qemu in order to run the 64bit arm cross compiler
20130424 10:48:25 <sgothel> oh puhh :)
20130424 10:48:26 <xranby> on a 32bit system
20130424 10:48:58 <sgothel> hmm .. but w/ build scripts our x-tools blobs shall be used
20130424 10:50:02 <sgothel> btw .. I took the user libs: armv4te-soft: debian6-arm (vanilla) and armv6-hard: debian7-arm-hard-raspi
20130424 10:50:30 <sgothel> it is correct that the resulting ARM sections are the common denominator of all linked binaries!
20130424 10:50:46 <xranby> sgothel: the x-tools blobs are ELF 64-bit LSB executable, x86-64
20130424 10:50:51 <sgothel> hence the 'user libs' version shall not exceed the build target
20130424 10:50:55 <sgothel> yes
20130424 10:51:03 <xranby> thus my system will run them using qemu-x86_64-static
20130424 10:51:05 <sgothel> host is x86_64, debian7
20130424 10:51:11 <xranby> and in this case it triggered a crash
20130424 10:51:12 <sgothel> oh
20130424 10:51:29 <sgothel> see - a good reason to avoid qemu :)
20130424 10:51:35 <xranby> since the machine i am on right now is a 32bit Ubuntu 12.04 system
20130424 10:51:41 <sgothel> looks in jogamp-scripting .. I have described the build
20130424 10:51:48 <sgothel> so you can do you own ..
20130424 10:52:06 <xranby> i guess i should start by installing a 64bit debian system someplace
20130424 10:52:47 <sgothel> http://jogamp.org/git/?p=jogamp-scripting.git;a=blob;f=jenkins-server-slave-setup/nodes/crosstools/README.txt;h=f66ba37cc748a602a3a61b7fc03407f550a4cdc3;hb=91f5ab94eb4f2734b7bd554292a71fe2ca7aa5f4
20130424 10:53:03 <sgothel> of commit http://jogamp.org/git/?p=jogamp-scripting.git;a=commit;h=91f5ab94eb4f2734b7bd554292a71fe2ca7aa5f4
20130424 10:53:50 <sgothel> for the sake of name alignment I have used suffix: sf -> armsf and hf -> armhf
20130424 10:55:03 <sgothel> finally build went through .. aggregating ..
20130424 10:56:22 <sgothel> http://jogamp.org/deployment/archive/master/gluegen_666-joal_431-jogl_977-jocl_785/
20130424 10:56:33 <sgothel> the evil gluegen build :)
20130424 10:56:38 <xranby> yes.. sweet
20130424 10:56:46 <xranby> you should do a maven build asap ;)
20130424 10:57:05 <sgothel> right
20130424 11:01:54 <sgothel> maven '2.0.2-rc-20130424' in progress .. deploying ..
20130424 11:02:02 <sgothel> ~1 hours ..
20130424 11:02:24 <xranby> can this be run after all sucessfull aggregate builds?
20130424 11:02:51 <sgothel> w/o signing .. it could
20130424 11:02:59 <sgothel> these are signed though
20130424 11:03:18 <sgothel> however, usually one might want to test it before - but I didn't .. so .. :)
20130424 11:04:18 <sgothel> more awesome would be to use a maven internal repo space (again - forgot the buzz word) .. i.e. populating it while building .. and kick off a maven test
20130424 11:04:27 <sgothel> .. w/ those builds
20130424 11:04:39 <sgothel> later .. I am sure Harvey likes to use it for Java3D .. etc
20130424 11:05:11 <sgothel> so .. if anybody knows how to populate this internal maven repo while building non-maven ..
20130424 11:05:43 <sgothel> I have to focus on the remaining issues .. for 2.0.2 .. I cc'ed you regarding Debian 7
20130424 11:05:55 <sgothel> and added you on the bugreport (just closed)
20130424 11:08:55 <rmk0> .o.
20130424 11:09:24 <rmk0> sgothel: the main problem is that pom files are required to deploy to a repos
20130424 11:10:50 <xranby> rmk0: you may add an activeByDefault mirror inside the pom http://paste.ubuntu.com/5598038/
20130424 11:11:41 <xranby> this allows me to add the jogamp test mirror to individual projects
20130424 11:12:26 * rmk0 eyes it
20130424 11:13:47 <rmk0> not sure i understand what this does
20130424 11:14:12 <rmk0> i mean... isn't this the same as using <activeProfile>jogamp-test</activeProfile>?
20130424 11:14:18 <xranby> a complete pom.xml that is using the test-mirror by default looks like https://github.com/xranby/jake2/blob/master/pom.xml
20130424 11:15:15 <rmk0> ah, right, i see
20130424 11:16:24 <rmk0> .. i may've missed the point
20130424 11:16:35 <rmk0> why would i do this?
20130424 11:18:12 <sgothel> my angle is: a final jogamp job (after gluegen, joal, jogl, jocl ..) would produce some maven repo (FAST), then we could kick-off a maven test, and maybe other maven related builds.
20130424 11:18:28 <sgothel> dunno about that test-maven server thingy ..
20130424 11:19:27 <sgothel> ah .. so a user doesn't need to configure the maven stuff for our test repo .. I see
20130424 11:19:46 <sgothel> I guess it's a matter of taste :)
20130424 11:20:14 <sgothel> so in the future we will have 2.0.2 in official maven repo .. while we continue to have newer test builds ..
20130424 11:20:26 <sgothel> I guess it's up to the dev .. to target either ..
20130424 11:21:04 <sgothel> so maybe, Xerxes you could add this 'flavor' to the maven Wiki .. as an alternative, i.e. builds using our pre-release for testing .. until proper release is done
20130424 11:25:46 <rmk0> if i understand then... what's really wanted is basically quick per-project deployment to the test repos at the end of a build (as opposed to what we do now - rip jar files out of all-platforms.7z and publish them all at once)
20130424 11:26:11 <rmk0> we'd obviously still need that for deploying to central (as they need to be pgp signed)
20130424 11:26:23 <sgothel> ofc ..
20130424 11:26:43 <sgothel> but also a quick 'aggregated' maven repo for maven-only related jenkins builds - maybe
20130424 11:26:59 <sgothel> i.e. we cannot have a build wait for 1-2 hours :)
20130424 11:27:18 <rmk0> yeah, that deployment time is nasty
20130424 11:27:24 <rmk0> i've not found a way to reduce it
20130424 11:27:42 <sgothel> ofc .. it will be faster from jogamp.org itself .. will try to do that next time
20130424 11:27:55 <sgothel> or can I simply run it again ?
20130424 11:28:03 <sgothel> w/ same version numbers ?
20130424 11:28:11 <rmk0> you can
20130424 11:28:18 <sgothel> in the end .. they don't need to be signed ..
20130424 11:28:19 <rmk0> we don't make any guarantees about the contents of that repos
20130424 11:28:42 <rmk0> in fact the wiki page warns of death and destruction \o/
20130424 11:28:46 <sgothel> well .. 'best effort' .. but yes, no indemnification clause :)
20130424 11:28:51 <sgothel> hehe
20130424 11:30:37 <sgothel> since Xerxes kicked off the issue .. what would be a good temporary internet root machine (some call it cloud node :) .. w/o relations to Amazon ?
20130424 11:31:02 <sgothel> so we can kick off a GNU/Linux image .. etc ..
20130424 11:31:24 <sgothel> I assume they use a hypervisor or something ?
20130424 11:31:37 <rmk0> what's this for... to host a mirror of jogamp.org or something?
20130424 11:31:41 <sgothel> can we build such a vanilla machine ourselves ?
20130424 11:31:55 <sgothel> to create a node on the fly .. 1st a test node
20130424 11:32:23 <sgothel> I learned this a bit w/ those Amazon machines .. but I don't like to support them
20130424 11:32:29 <rmk0> oh, for jenkins?
20130424 11:32:34 <sgothel> usually quite cheap ..
20130424 11:32:49 <sgothel> yes .. or for our own manual testing w/ different images .. distris .. etc
20130424 11:32:55 <rmk0> right
20130424 11:32:57 * rmk0 prods jk4
20130424 11:33:08 <sgothel> usually they have to offer images 'on site'
20130424 11:33:13 <rmk0> jk4 has more vps subscriptions than any human being alive
20130424 11:33:24 <sgothel> but I would not mind to have the knowledge ourselves
20130424 11:33:39 <sgothel> but can you script them ?
20130424 11:33:53 <rmk0> no idea... i've only ever used shared hosting
20130424 11:33:53 <sgothel> and we would need real root nodes .. -> GPU
20130424 11:34:10 <rmk0> i've heard someone complain intensely about rackspace
20130424 11:34:17 <rmk0> entirely unfriendly to scripted creation
20130424 11:34:20 <sgothel> e.g. Hetzner (jogamp.org) has a nice root node offer, but not scriptable AFAIK ..
20130424 11:34:45 <rmk0> i think people end up creating base images and then cloning them
20130424 11:34:50 <rmk0> due to a lack of scripting support
20130424 11:34:51 <sgothel> and setup time should be like <3min
20130424 11:34:56 <sgothel> from SCRATCH :)
20130424 11:35:28 <sgothel> must be something like hypervisor I guess w/ bootp access / NFS .. dunno
20130424 11:35:45 <rmk0> yeah, think so
20130424 11:36:37 <sgothel> if we would know how to setup that environment .. we could have more redundancy and independence .. (from individual) .. however, even such setup would require people w/ knowledge to maintain it - so I don't know :) In the end .. you always need the guru who gives the system regular loving :)
20130424 11:37:47 <rmk0> yow
20130424 11:38:03 <rmk0> $899 a month for "gpu servers"
20130424 11:38:14 <sgothel> we could start w/ a scripted setup of well defined distributions .. flexible (scp source) .. - would make sense if somebody new willing to maintain is doing it
20130424 11:38:27 <sgothel> Amazone is like per minute .. pretty cheap :)
20130424 11:38:44 <sgothel> Hetzner is like 600 EUR per year, has an NV GPU
20130424 11:38:57 <sgothel> (but not the scripting thingy .. and not per minute)
20130424 11:39:07 <sgothel> or hour ..
20130424 11:41:14 <sgothel> i.e. if you new baby (CI) can do build/test jogamp .. maybe later this year we can swap ?
20130424 11:41:56 <sgothel> me tired as well of deleting failed builds due to jenkins / networking / whatever problems :)
20130424 11:42:37 <rmk0> there's quite a bit of work still to be done on it
20130424 11:42:45 <rmk0> and it really only supports building maven projects
20130424 11:46:26 <rmk0> eyeing the jogamp-scripting maven scripts
20130424 11:47:31 <rmk0> can obviously run a script from the jogamp ant scripts, but i don't think it's going to make it any quicker to deploy
20130424 11:48:04 <sgothel> yup
20130424 11:48:20 <sgothel> deploy time is >> than transport time
20130424 11:48:34 <rmk0> wondering how much time gpg:sign-and-deploy-file adds compared to deploy:deploy-file
20130424 11:49:00 <sgothel> not much really, machine is quite fast ..
20130424 11:49:18 <sgothel> ok .. so the felt time of 1-2 hours is actually ..
20130424 11:49:35 <rmk0> it takes me about an hour from here, but my connection isn't fast at all
20130424 11:49:49 <sgothel> ~30 min
20130424 11:50:10 <sgothel> maven 2.0.2-rc-20130424 deployed
20130424 11:50:18 <rmk0> so that's 14 projects in 30 minutes
20130424 11:50:30 <rmk0> 2-3 minutes each, then
20130424 11:50:40 <rmk0> hrm...
20130424 11:50:51 <rmk0> wonder what's taking so long...
20130424 11:51:03 <rmk0> can compile and install projects faster than that
20130424 11:51:06 <sgothel> transport is long .. since its a lot of small files hmm
20130424 11:51:39 <rmk0> yeah
20130424 11:51:56 * rmk0 harasses maven people
20130424 11:52:28 <sgothel> :)
20130424 11:52:36 <xranby> sgothel: jake2 keyboard work ok using 2.0.2-rc-20130424, nice
20130424 11:52:45 <sgothel> puhh :)
20130424 11:53:38 <sgothel> Bug 520 and Bug 706 are next .. testing on that machine in a bit .. hope this similar machine exposes the bug .. :)
20130424 11:54:07 <sgothel> Xerxes: on ac100 w/ JOAL ? :)
20130424 11:54:52 <sgothel> who likes to port Jake2 on Android w/ JogAmp (mostly IOUtils hacking for loading artifacts) ?
20130424 11:56:29 <xranby> sgothel: i can check JOAL on the ac100 in a minute
20130424 11:57:08 <sgothel> I tested w/ all basics tests (JOAL + JOGL) on raspi-hf, panda-hf, panda-sf
20130424 11:57:27 <sgothel> panda-hf didn't give me audio output though .. but all could load native libs etc
20130424 11:58:08 <sgothel> and when I compare openal-soft's info app w/ our output .. seems like we don't show all the extensions or something .. dunno
20130424 11:58:42 <sgothel> note: the linux-amd machines (debian7) use our openal-soft, where linux-nv (ubuntu) use the deployed openal ..
20130424 11:58:54 <sgothel> hence the unit test difference .. you can check that ..
20130424 11:58:54 <xranby> did you check volume output using alsamixer
20130424 11:59:00 <sgothel> yes, was working
20130424 11:59:12 <sgothel> at least I saw something :) .. nothing plugged in though
20130424 11:59:41 <sgothel> but I had to stop at one point .. to cont. w/ other bugs .. hmm, will test again after updating panda/raspi machines
20130424 12:00:51 <rmk0> hm! may've found a way to do bulk deployment...
20130424 12:00:54 <rmk0> watch this space...
20130424 12:16:37 <xranby> sgothel: the openal-soft tries to use to oss backend (and fails) on my ac100
20130424 12:17:24 <xranby> and that result in no sound, yet the application runs
20130424 12:17:59 <xranby> sgothel: the cross compilation is a sucess
20130424 12:18:30 <xranby> the cmake options we use to configure openal-soft may need tuning
20130424 12:23:18 <rmk0> sgothel: just pushed a patch to jogamp-scripting that i think will reduce the deployment time significantly
20130424 12:23:47 <rmk0> it takes 40 seconds to deploy everything to a directory in the filesystem here, compared to nearly four minutes prior to the patch
20130424 12:31:25 <xranby> sgothel: there is a alsoft-conf tool in ubuntu/debian
20130424 12:31:41 <xranby> you may use it to generate a ~/.alsoftrc
20130424 12:32:06 <xranby> it allows you to disable certain backends on your panda
20130424 13:22:24 <xranby> sgothel: can you try run joal on a 32bit linux machine: java -cp joal-test.jar:joal.jar:gluegen-rt.jar com.jogamp.openal.test.manual.Sound3DTest
20130424 13:23:06 <xranby> sgothel: i get sound using my own build. while no sound using the latest aggrigate build :/
20130424 13:24:35 <xranby> i have checked out the openal-soft git from the jogamp server git://jogamp.org/srv/scm/openal-soft.git
20130424 13:34:25 <sgothel> back ..
20130424 13:35:17 <sgothel> Mark: Great - awesome (maven deployment speed)
20130424 13:35:31 <sgothel> @Xerxes: So whats the diff ?
20130424 13:35:44 <sgothel> can you check cmake and the crossbuild results ?
20130424 13:35:53 <xranby> cross compiled vs native compile
20130424 13:36:02 <sgothel> shouldn't it use something else if oss fails ?
20130424 13:36:52 <sgothel> I neglect to consider 'native build' :) - lets better find out whats wrong w/ the crossbuild setup and output selection (if one fails)
20130424 13:37:19 <sgothel> IMHO !oss -> pulse ..
20130424 13:37:30 <sgothel> but IMHO it should be pulse, alsa, oss ..
20130424 13:37:41 <sgothel> (order of trying)
20130424 13:38:10 <sgothel> @Mark: You stuff pushed ? Can I use it ? New way to use it ?
20130424 13:38:26 <rmk0> sgothel: interface hasn't changed
20130424 13:38:32 <sgothel> cool
20130424 13:38:35 <rmk0> would try deploying something the usual way and see if it's faster there
20130424 13:39:08 <sgothel> @Xerxes: I guess if it's 'just' OSS which disturbs us, I guess it's safe to remove it from the openal-soft build
20130424 13:39:25 <sgothel> who is using OSS w/o Alsa anyways today ? :)
20130424 13:39:49 <sgothel> but maybe there is something else .. dunno
20130424 13:42:33 <sgothel> 40/4*60 .. awesome
20130424 13:42:47 <sgothel> 40/(4*60) :)
20130424 13:43:12 <rmk0> deployed?
20130424 13:43:42 <sgothel> no not yet .. I can pull ? will do later - swizzling the windows machine right now
20130424 13:43:49 <rmk0> yeah, is pushed
20130424 13:46:33 <sgothel> testing .. :)
20130424 13:46:40 <sgothel> (overriding version)
20130424 13:47:14 <sgothel> error .. overriding .. well
20130424 13:47:23 <rmk0> erk
20130424 13:47:45 <sgothel> [ERROR] Failed to execute goal org.apache.maven.plugins:maven-gpg-plugin:1.4:sign-and-deploy-file (default-cli) on project gluegen-rt: Error deploying attached artifact gluegen-rt-2.0.2-rc-20130424-javadoc.jar: Failed to deploy artifacts: Could not transfer artifact org.jogamp.gluegen:gluegen-rt:jar:javadoc:2.0.2-rc-20130424 from/to jogamp-mirror (scpexe://jogamp.org/srv/www/jogamp.org/deployment/maven/): NullPointerException -> [Help 1]
20130424 13:47:57 <rmk0> ow!
20130424 13:48:09 <rmk0> heh, that's not good
20130424 13:48:19 <rmk0> not seen a null pointer exception before...
20130424 13:49:19 <rmk0> would edit make-deploy-one.sh to run mvn with -e and -X
20130424 13:49:41 <rmk0> to find out where that's happening... might have to file/fix a bug with the gpg plugin
20130424 13:52:30 <sgothel> ay
20130424 13:53:22 <rmk0> if you send me the output, i'll do the rest
20130424 13:53:37 <sgothel> real 7m19.342s
20130424 13:53:37 <sgothel> user 1m31.744s
20130424 13:53:37 <sgothel> sys 0m5.740s
20130424 13:53:38 <rmk0> can't reproduce it here, but it probably won't be hard to track down with a backtrace
20130424 13:53:51 <rmk0> ...
20130424 13:53:53 <rmk0> working?
20130424 13:54:10 <sgothel> no, just the time w/ failure .. still a very good time
20130424 13:54:28 <rmk0> it helps not running mvn once per file...
20130424 13:59:21 <sgothel> sent
20130424 13:59:42 <sgothel> maybe it's just the over-write ?
20130424 14:00:31 * rmk0 eyes source
20130424 14:02:00 <rmk0> org.apache.maven.wagon.providers.ssh.external.ScpExternalWagon.executeScpCommand(ScpExternalWagon.java:275)
20130424 14:03:32 <sgothel> so maybe it's the overwrite .. scpexe doesn't allow it ? hmm .. would bring us back to: how to remove a released version from filesystem and xml files CONVENIENT and RELIABLE :)
20130424 14:04:54 <rmk0> line 275 is cl.setWorkingDirectory( localFile.getParentFile().getAbsolutePath() );
20130424 14:05:28 <rmk0> so... doesn't even get as far as attempting to overwrite
20130424 14:10:17 <rmk0> as for removing artifacts... this is when you start getting into repository manager territory
20130424 14:10:28 <rmk0> something like http://archiva.apache.org/index.cgi
20130424 14:10:49 <rmk0> i've not used it... i might install it here and see how awful it is
20130424 14:33:17 <sgothel> same bug in new file space ..
20130424 14:33:57 <sgothel> hence .. we have a regression :|
20130424 14:34:17 <sgothel> .. but performance is great :)
20130424 14:38:56 <rmk0> hehe
20130424 14:39:04 <rmk0> fast and broken!
20130424 14:40:39 <rmk0> i assume it breaks if you try make-deploy-one on the gluegen-rt project?
20130424 14:44:32 <rmk0> it seems that there's a wagon-ssh and wagon-ssh-external... for some reason we're using the "external"
20130424 14:44:41 <rmk0> i don't know what the difference is... will try with the other one and see what happens
20130424 14:46:37 <rmk0> ah, managed to reproduce the crash
20130424 15:06:18 <sgothel> I 1st tried all as usual - then only one for log files
20130424 15:06:28 <sgothel> cool
20130424 15:27:42 <sgothel> hurray - Windows-ATI_Mobile bugs are reproducible :)
20130424 15:40:35 <rmk0> ok... sent mail with a test case to the maven list
20130424 15:40:44 <rmk0> https://github.com/io7m/maven-wagon-bug-20130424
20130424 15:40:47 <rmk0> will see what comes back
20130424 15:41:24 <rmk0> sgothel: you may want to switch back the old and slow version until this is sorted out...
20130424 15:41:29 <rmk0> >_<
20130424 15:42:06 <sgothel> I am patient .. just waiting for your fix .. np
20130424 15:42:15 <sgothel> I assume the pushed version is fine
20130424 15:42:42 <sgothel> (the previous jogamp maven build .. this is)
20130424 15:43:16 <rmk0> just annoying that you obviously can't push new builds with that script until this is fixed
20130424 16:26:07 * [Mike] (~Mike]@anon) has joined #jogamp
20130424 17:06:22 * [Mike] (~Mike]@anon) Quit ()
20130424 17:28:00 * [Mike] (~Mike]@anon) has joined #jogamp
20130424 18:07:23 <sgothel> I became quite frustrated myself w/ one node offline while I was looking for a good tooling / OS 'match' :)
20130424 18:15:58 * sgothel (~sven@anon) Quit (Ping timeout: 272 seconds)
20130424 18:21:56 * sgothel (~sven@anon) has joined #jogamp
20130424 18:21:56 * ChanServ sets mode +v sgothel
20130424 18:23:53 * DemoscenePassiv (~Lutsche@anon) has joined #jogamp
20130424 19:33:08 * xranby_ac100 (~xranby@anon) has joined #jogamp
20130424 20:04:39 <sgothel> Win7 ATI mobile driver bug: Simply crashes when trying to create a GL 4 ctx, while GL 4 works w/ ARB creation functions :)
20130424 20:07:51 <xranby_ac100> sgothel: nice, and jogl are able to enforce use of ARB creation functions, yes ?
20130424 20:10:15 <sgothel> we walk through all possible GL profiles @ init, to create a mapping of whats available, so the user can choose and decide
20130424 20:10:35 <sgothel> while doing so .. ATI mobile driver of certain versions crash ..
20130424 20:11:13 <sgothel> same WGL-ARB function works w/ e.g. 3.2 ctx .. but crashes w/ 4.2 ctx - diff. is only the argument :)
20130424 20:11:19 <sgothel> good quality :)
20130424 20:13:07 <sgothel> https://jogamp.org/bugzilla/show_bug.cgi?id=706
20130424 20:13:43 <xranby_ac100> btw.. did you file a bug for adding audio to the glffmpeg player?
20130424 20:13:58 <sgothel> nope sorry ..
20130424 20:14:12 <xranby_ac100> FFMPEGMediaPlayer
20130424 20:14:22 <sgothel> you know, me currently 100% doing things ..
20130424 20:15:11 <xranby_ac100> then i will file one.. i have primed myself somewhat by writing small native c libav -> openal test examples to check what to implement
20130424 20:15:48 <sgothel> thx
20130424 20:43:43 <sgothel> oh .. so what is with the no-sound issue ? do you have an explanation ? removing oss-support will help ? isn't this a bug in openal-soft ?
20130424 20:44:23 <sgothel> i.e. order of output devices .. fallback .. etc
20130424 20:44:47 <sgothel> would be very helpful to analyze that
20130424 21:08:10 <xranby_ac100> https://jogamp.org/bugzilla/show_bug.cgi?id=725
20130424 21:09:30 <xranby_ac100> i will look into the no-sound issue from the cross compiled binarys, i have no explenation yet..
20130424 21:11:45 <xranby_ac100> there is no errors reported at runtime :/ everything looks to be working except that there is only silence produced by the speakers
20130424 21:31:59 <sgothel> I remember a 'no output device' .. failure when running joal/make/scripts/tests.sh
20130424 21:32:33 * DemoscenePassiv (~Lutsche@anon) Quit (Ping timeout: 240 seconds)
20130424 21:37:48 <xranby_ac100> i ran a check using the libav openal decode.c from 725 and alcOpenDevice(NULL); returns 0 thus it fail.
20130424 21:38:28 <sgothel> isn't that already a too complex test ?
20130424 21:38:50 <xranby_ac100> that is the first openal line executed
20130424 21:39:05 <sgothel> however, this indicates cross-compiled openal-soft is buggy .. I see (as you mentioned)
20130424 21:39:17 <sgothel> you know strace ?
20130424 21:39:32 <sgothel> you can trace all kernel commands, especially open ..
20130424 21:39:49 <sgothel> and you could see which device is tried to be opened 1st ..
20130424 21:40:05 <sgothel> Q: your native openal-soft was not building w/ OSS right ?
20130424 21:40:26 <sgothel> i.e. cmake caps native vs cross := +OSS ?
20130424 21:41:01 <sgothel> if so .. try disable OSS in crossbuild and see .. would be a great isolation already
20130424 21:41:28 <sgothel> maybe some header files are buggy when crosscompiling ?
20130424 21:41:41 <xranby_ac100> i can see that it parses the system /etc/openal/alsoft.conf before failing
20130424 21:41:54 <sgothel> ah .. alsa
20130424 21:42:12 <sgothel> via strace ?
20130424 21:42:36 <sgothel> oss would use alsa's oss emul ..
20130424 21:43:00 <sgothel> your target has alsa + pulse working ?
20130424 21:43:55 <sgothel> hmm .. we could check whether a version mismatch could lead to such result of (oss, or alsa, or pulse..)
20130424 21:44:11 <sgothel> would be poor design of headers / protocol etc though
20130424 21:45:02 <sgothel> guess I need to find my headphones as well :)
20130424 21:46:16 <xranby_ac100> not sure if alsa can play when pulseaudio is running
20130424 21:47:33 <sgothel> it can ..
20130424 21:47:43 <sgothel> just run aplay ..
20130424 21:49:24 <xranby_ac100> agree. aplay plays
20130424 21:49:50 <rmk0> sgothel: the maven people are telling me that the 'ssh-external' thing we're using is obsolete and that we should be using the plain 'ssh' module. unfortunately, it doesn't seem to support talking to the openssh agent. there's a facility for storing encrypted copies of your ssh key's password in your settings.xml, but... i don't like it
20130424 21:50:12 <xranby_ac100> and i can see using strace that the cross compile openal tries to use OSS
20130424 21:50:31 <rmk0> the alternative, i think, is to not use ssh, but to use a repository manager like they do at sonatype, so that artifacts can be deployed over plain https
20130424 21:50:36 <xranby_ac100> while the native build picks up pulseaudio
20130424 21:50:47 <rmk0> going to look into archiva to see if it's horrible or not
20130424 21:51:40 <sgothel> https .. yikes .. security security
20130424 21:52:06 <rmk0> well, it's hardly worse than ssh
20130424 21:52:18 <sgothel> plus that web-backend .. I really love your scripts dude ! :)
20130424 21:52:27 <rmk0> hehe
20130424 21:52:36 <rmk0> well, we'd still use the same scripts, it'd just be different URIs
20130424 21:52:46 <xranby_ac100> sgothel: have you seen (EE) alcOpenDevice: Option 'format' is deprecated, please use 'channels' and 'sample-type' on your end?
20130424 21:52:55 <sgothel> ssh uses a 4k async key .. negotiating a session key .. IMHO much safer w/ less web-backend
20130424 21:53:14 <sgothel> @Xranby: one sec
20130424 21:53:18 <rmk0> i've not given up on the ssh module, i just don't like how they're storing passwords
20130424 21:53:23 <rmk0> am still looking into it
20130424 21:53:45 <rmk0> i'd rather it talk to my ssh-agent like every other program that uses ssh
20130424 21:54:41 <sgothel> I use ssh-agent too .. well, ~/.ssh is one option of many
20130424 22:00:06 <sgothel> @Xerxes: no .. checking on panda01 .. getting headphones and installing pavucontrol etc
20130424 22:06:18 <sgothel> @Xerxes: which platform are you testing and fails ? OS/ armsf or armhf ?
20130424 22:06:27 <sgothel> ac100 / ubuntu / armhf ?
20130424 22:06:41 <sgothel> how about raspi ?
20130424 22:07:06 <xranby_ac100> i am testing right now on the ac100
20130424 22:07:28 <xranby_ac100> sgothel: while writing a minimal reproducer.. the issue stopped
20130424 22:07:51 <sgothel> ha
20130424 22:08:09 <sgothel> btw .. panda speaker/headphone jack is top or bottom of the 2 ? :)
20130424 22:08:18 <xranby_ac100> http://paste.ubuntu.com/5599535/
20130424 22:08:24 <sgothel> where are the specs ..
20130424 22:09:16 <sgothel> dude .. well, the openal_info would be generated as well .. directly w/ openal-soft (I added build scripts there too)
20130424 22:09:27 <sgothel> maybe sideeffects .. hmm
20130424 22:12:46 <rmk0> ah... some degree of success
20130424 22:12:53 <rmk0> will document
20130424 22:13:04 <rmk0> it's a bit of a pain, but it does work, and uses ssh
20130424 22:14:01 <sgothel> sweet
20130424 22:14:09 <sgothel> I guess we are used to pain :)
20130424 22:16:17 <xranby_ac100> (00:08:10) sgothel: btw .. panda speaker/headphone jack is top or bottom of the 2 ? :) <--- i dont know they look the same :p
20130424 22:16:31 <sgothel> yeah :)
20130424 22:17:30 <xranby_ac100> i think the panda can use audio over hdmi
20130424 22:19:33 <sgothel> /usr/bin/java -DNo com.jogamp.openal.test.manual.OpenALTest
20130424 22:19:33 <sgothel> AL lib: pulseaudio.c:331: PulseAudio returned minreq > tlength/2; expect break up
20130424 22:19:33 <sgothel> output devices:
20130424 22:19:41 <sgothel> no sound - aplay works
20130424 22:19:55 <sgothel> panda: bottom is headphone jack
20130424 22:22:17 <sgothel> jogamp@panda01:~/projects/JOGL/openal-soft/build-armv6$ ./openal-info
20130424 22:22:17 <sgothel> Available playback devices:
20130424 22:22:18 <sgothel> !!! none !!!
20130424 22:22:18 <sgothel> Available capture devices:
20130424 22:22:18 <sgothel> !!! none !!!
20130424 22:22:18 <sgothel> Default playback device:
20130424 22:22:18 <sgothel> Default capture device:
20130424 22:22:18 <sgothel> ALC version: 1.1
20130424 22:22:19 <sgothel> AL lib: (EE) oss_open_playback: Could not open /dev/dsp: No such file or directory
20130424 22:22:19 <sgothel> !!! Failed to open default device !!!
20130424 22:22:25 <sgothel> panda01 == armsf
20130424 22:26:09 * xranby_ac100 (~xranby@anon) Quit (Ping timeout: 256 seconds)
20130424 22:30:43 <sgothel> Building OpenAL with support for the following backends: [exec] -- OSS, WaveFile, Null - on all linux-amd nodes - duh!
20130424 22:32:56 <sgothel> I switched so many distris over weekend - I guess I lost count - duh
20130424 22:33:15 <sgothel> installing libpulse and libasound developer files .. :)
20130424 22:38:55 <rmk0> sgothel: pushed updated instructions
20130424 22:39:09 <sgothel> thx .. a lot
20130424 22:39:19 <sgothel> will update scripts
20130424 22:39:22 <rmk0> if you've got old pom files generated with ./make.sh, you'll need to run ./clean.sh and make again
20130424 22:39:34 <sgothel> ay
20130424 22:39:51 <rmk0> basically.. the ssh provider is specified in each generated pom file
20130424 22:39:56 <rmk0> and we're using a different one now
20130424 22:40:01 <rmk0> so... need to regenerate
20130424 22:40:46 <rmk0> i deployed some files into /tmp on the server, so it seems to work properly
20130424 22:41:09 <sgothel> nice .. will ping you probably tomorrow about success
20130424 22:41:16 * rmk0 holds breath
20130424 22:41:37 <sgothel> ok - can do now if you like .. IRQ :)
20130424 22:41:43 <rmk0> hehe
20130424 22:41:50 <rmk0> i don't mind... not really doing much now
20130424 22:41:52 <sgothel> recursive IRQ :)
20130424 22:41:58 <rmk0> i get to sit here and watch the carnage
20130424 22:44:21 <sgothel> mvn --encrypt-master-password abcdefgh <-- I guess it's really required 'eh ? or 'just' your taste of things to spread the pain ? :)
20130424 22:44:37 <rmk0> hehe.. it's annoying
20130424 22:44:48 <rmk0> they've obviously never heard of people stealing passwords via 'ps'
20130424 22:44:55 <sgothel> now I have to trust maven w/ my ssh passphrase ?
20130424 22:45:10 <rmk0> unfortunately
20130424 22:45:32 <sgothel> jesus
20130424 22:45:33 <rmk0> it will at least let you use an alternate key
20130424 22:45:53 <sgothel> but problem is: the key-pair has access to our server
20130424 22:46:03 <rmk0> exactly
20130424 22:46:23 <rmk0> it's either this or https, unfortunately
20130424 22:46:24 <sgothel> TBH .. I prefer the previous method .. :(
20130424 22:46:30 <sgothel> or .. 30 min
20130424 22:46:30 <rmk0> yep
20130424 22:46:55 <rmk0> if you can stand the deployment times, by all means use the old way
20130424 22:47:12 <sgothel> hmm .. I put this on a branch then .. and go w/ 30 minutes .. yup - however, thank you !
20130424 22:47:42 <sgothel> understand the pain you went through .. sorry
20130424 22:47:55 <rmk0> hehe
20130424 22:48:30 <rmk0> hm
20130424 22:48:44 <rmk0> could create a maven deployment user on the server that only has access to that repository
20130424 22:49:02 <rmk0> no read/write access to anything elsewhere, can only scp
20130424 22:49:08 <sgothel> 'only'
20130424 22:49:19 <sgothel> and you know scp is ssh .. full access
20130424 22:49:35 <rmk0> it's certainly possible to lock it down heavily
20130424 22:49:39 <sgothel> plus - it makes managment much more complicated - and I already have a headache
20130424 22:49:45 <rmk0> hehe
20130424 22:50:17 <sgothel> I earmark this .. and come back to it for sure - after 2.0.2
20130424 22:50:28 <rmk0> right
20130424 22:51:10 <rmk0> really irritating that jsch can't talk to ssh-agent natively
20130424 22:51:35 <sgothel> i.e. that 'provider' is wrong - why can't it work as the other ?
20130424 22:51:52 <rmk0> how d'you mean?
20130424 22:51:57 <sgothel> ssh-agent or just ssh should be enough
20130424 22:52:18 <sgothel> i.e. you would just need to invoke ssh - assuming you already did authorize via agent
20130424 22:52:39 <sgothel> thats how we run slave nodes .. somewhat
20130424 22:52:41 <rmk0> the old provider talked to the scp executable, but the new provider actually implements the ssh protocol... http://www.jcraft.com/jsch/
20130424 22:52:57 <rmk0> no external executable involved
20130424 22:52:59 <sgothel> which IMHO is a waste of time
20130424 22:53:18 <sgothel> i.e. impl. ssh stuff .. is a security science for itself ..
20130424 22:53:30 <sgothel> and now comes the java master :)
20130424 22:53:30 <rmk0> well, the people that wrote it have written their own X servers in java
20130424 22:53:40 <rmk0> so they don't seem averse to ... work
20130424 22:54:18 <sgothel> and it needs you to expose your keys + passphrase, which I spot in the very 1st second - are they blind ?
20130424 22:54:51 <rmk0> i don't think jsch requires that
20130424 22:54:57 <rmk0> it's the maven frontend that's at fault
20130424 22:55:07 <rmk0> it's an apache project... you know how they are
20130424 22:55:15 <rmk0> "oops"
20130424 22:56:13 <sgothel> oh yes .. taking pills from the evil (open office for example) .. times are different now I guess
20130424 23:02:41 <sgothel> http://jogamp.org/git/?p=jogamp-scripting.git;a=shortlog;h=refs/heads/maven_fast
20130424 23:02:49 <sgothel> maven_fast is the branch of your WIP
20130424 23:03:02 <sgothel> I have reset master to 30min slow motion :)
20130424 23:03:07 <rmk0> \o/
20130424 23:17:59 <sgothel> JOAL has pulse & also now .. duh
20130424 23:18:53 <sgothel> and the buggy ATI driver needs an old GL context to be current before using ARB createContext for GL >= 3.0 .. lalala nuts
20130424 23:24:37 <rmk0> ugh
20130424 23:45:38 <sgothel> nice .. our openal-soft works, pre-installed does not :)
20130424 23:45:48 <sgothel> guess we should favor our version then
20130425 00:32:05 <sgothel> joal: panda-armsf OK, raspi-armhf: OK, panda-armhf: OK
20130425 02:32:42 * [Mike] (~Mike]@anon) Quit ()
20130425 03:26:59 * [Mike] (~Mike]@anon) has joined #jogamp
20130425 05:06:01 -CatOut- Continue @ http://jogamp.org/log/irc/jogamp_20130425050601.html