/srv/irclogs.linaro.org/2012/01/04/#linaro.txt

plarszyga, zyga-nc10: still there?00:42
rsalvetijcrigby: https://ci.linaro.org/jenkins/view/Ubuntu%20CI%20PKG/01:32
rsalvetithere's also a view for the packaging related jobs now01:32
Dr_Whoany linaro-infrastructure types around?01:50
mwhudsonDr_Who: ish02:04
Dr_Whohi mwhudson,  trying to kick off a build on offspring,  it's telling me I don't have authorization tho I'm logged in02:06
mwhudsonah02:06
Dr_Whonano - oneiric and LinaroTV - xbmc oneiric to be exact02:06
Dr_WhoI dunno .. maybe rsalveti gave me the boot :-)02:07
mwhudsonDr_Who: can you usually do this?02:07
mwhudsonyou need an offspring admin i guess, that's not me02:07
Dr_Whook … thanks02:07
rsalvetiDr_Who: well, you should be able to schedule a build02:07
rsalvetialways 02:07
rsalvetiDr_Who: the owner was tgall, I removed it02:08
rsalvetidon't think we're currently using that field 02:09
rsalvetischeduled a build, if it fails, try scheduling another build with your user02:09
rsalvetifailed already02:09
Dr_Whorsalveti: no worries,  I'm not able to even kick off a general build like oneiric nano so I'm thinking maybe something is up between offspring and wherever it might get my credentials from02:09
rsalvetiDr_Who: could be, are you sure you're logged in?02:10
Dr_WhoI am …  logged out and in again …02:10
Dr_Whomaybe time to restart chrome02:10
rsalvetijust checked, and everyone that's part of the linaro group should be able to schedule a build02:11
Dr_Whohmm  well given you can kick off a build, thinking it must be something on my end02:13
Dr_Whorsalveti: still around ?03:02
rsalvetiDr_Who: yup03:03
rsalvetigreat, seems video playback is working with latest kernel03:03
rsalvetieven with sound03:03
Dr_Whoawesome!03:03
Dr_WhoI've got the new alsa-lib pushed up into overlay already03:03
rsalvetibut noticed that we need a way to bump the kernel abi, otherwise the pvr module will not build for the newer kernel03:04
Dr_Whothe pulse audio update is nearly ready to go but there's one bit here I'd like your opinion on03:04
rsalvetiDr_Who: today?03:04
rsalvetiDr_Who: sure03:04
Dr_Whothere's a omap specific udev rule03:05
Dr_Whorsalveti: I don't like the idea of packing it in the pulse audio package as I'm not sure of a way to conditionally install it correctly03:06
Dr_Whothis really seems like linaro-media-create time type install item03:06
rsalvetiDr_Who: let me check it03:08
Dr_Whowant me to forward your way ?03:08
rsalvetiDr_Who: got it here already03:09
rsalvetiDr_Who: guess you can easily change the pulseaudio package and include this file03:09
rsalvetiit's not going to hurt any other board03:09
rsalvetithere's a very easy way to do it, let me check here03:09
rsalvetiDr_Who: all you need to do is to have the file as <package_name>.udev at your debian folder03:10
rsalvetiif it's already using debhelper03:10
rsalvetisame as we have for other packages, like ti-uim and such03:11
rsalvetiguess you're modifying pulseuadio anyway already03:11
Dr_Whoyup…  have a few patches added already03:11
Dr_Whothe package does already install a 90-pulseaudio.rules03:12
Dr_Whowondering if this 90-pulseaudio-omap.rules couldn't be combined with the existing rules file03:12
rsalvetiat the pulseaudio package you already have /lib/udev/rules.d/90-pulseaudio.rules03:12
Dr_Whoyup03:12
rsalvetiDr_Who: yup, guess you can just have the changes as a patch03:13
rsalvetiDr_Who: and append them at this file03:13
Dr_Whoyup so it would seem03:14
rsalvetiit's hardware specific already03:14
Dr_Whogood deal03:14
Dr_Whothanks rsalveti 03:16
rsalvetinp03:16
robclarkrsalveti, btw, is there a porting guide for armhf?  I'm attempting to rebuild the constellation of xorg trees on a 12.04 armhf filesystem and I get this error:  http://pastie.org/312238703:17
rsalvetirobclark: guess you can find more info about armhf at http://wiki.debian.org/ArmHardFloatPort, but not sure if exactly what you need03:19
rsalvetirobclark: what are you trying to do?03:19
rsalvetijust rebuilding a package or building another tree (hash)?03:20
rsalveticould be that this was fixed with the current package but the fix is just a package patch03:20
Dr_Whorobclark: that looks like between the two libs some different compiler parms were used03:20
robclarkrsalveti, I'm rebuilding stuff from git trees for devel (ie. not using deb pkg build system)03:20
robclarkI'm just trying to track down why parts of the tree end up build w/ armhf and other parts not03:21
rsalvetiwould be weird to have conflicting libs, as they are all probably using the same compiler options 03:21
rsalvetithat's weird 03:21
Dr_Whowell begs the question of what are the correct compiler parms03:21
robclark(possibly something didn't get cleaned when I migrated.. I'm just looking for hints about what might go wrong)03:21
rsalvetioh, could be 03:21
robclarkI'm giving the -c arg to util/modular/build.sh so in *theory* it is doing a 'make clean' first..03:23
robclarkbut I'm not sure if I should expect to need to patch some Makefiles (which I haven't done)03:24
robclarkor if it should all be automagic thanks to compiler defaults03:24
* robclark is anyways having to rebuild a bunch of xorg stuff to rebase dri2video on latest 1.12 pre-release.. but maybe it was a mistake to try and make armhf switch at same time03:27
robclark(doing two things at once, and all)03:28
rsalvetirobclark: but is this a clean checkout or still with stuff from armel around?03:30
rsalvetijcrigby: bug 91137703:30
ubot2`Launchpad bug 911377 in u-boot-linaro "u-boot-linaro fails to build for efikamx with 11.12 release" [Medium,Confirmed] https://launchpad.net/bugs/91137703:30
robclarkrsalveti, not a clean checkout.. but in theory build script did a 'make clean' and re-ran autoconf/automake/configure stuff03:30
robclarkI manually did a 'make distclean' and trying it again to be sure03:31
robclarkbut even just cloning all the trees from scratch takes a lot of time, so I was trying to shortcut03:31
rsalvetigot it03:32
robclark(possibly haste makes waste.. but OTOH !haste takes too d*** long :-P)03:32
rsalvetiI'd assume that automake/autotools would handle armhf just fine by using the compiler defaults 03:32
rsalvetiunless it's hardcoded somewhere 03:32
robclarkok.. just wanted to check if there where known gotcha's to look out for03:32
rsalvetirobclark: not that I know of, any suggestion that comes to my mind is kind of a brutal force one03:33
robclarkie. re-clone all the trees?03:33
rsalvetior checking with libs are not armhf and try recompiling them 03:34
rsalvetito understand the logs and such 03:35
robclarkok03:35
rsalvetijcrigby: it seems that the ci sync is happening everyday for the ti-lt job 03:37
rsalvetiprobably because of the shared tree/branch 03:37
rsalvetiat least from the changelog it was the only thing that changed at the last 2 days 03:38
rsalvetiwhich is also a ci job it seems03:38
rsalvetiguess it's probably because of the ubuntu tree/merge when creating the shared package03:40
=== doko_ is now known as doko
dokomichaelh1, about bug 910791, can you reproduce it with armhf defaults?07:19
ubot2`Launchpad bug 910791 in gcc-snapshot "[armhf] trunk 20111223, ICE in output_move_double, at config/arm/arm.c" [High,Confirmed] https://launchpad.net/bugs/91079107:19
michaelh1doko: hmm, perhaps - a different float ABI07:24
doko-mfloat-abi=hard07:25
michaelh1doko: no luck on /tools/toolchains/arch/armv7l/gcc-4.7~svn182780-armv7l-natty-cbuild220-ursa2-cortexa9r1/bin/g++ -O2 -c -g -mfpu=neon -mfloat-abi=hard lp910791.ii07:26
michaelh1(or with -mfpu=vfpv3-d16)07:26
dokomichaelh1, strange, as the snapshot builds don't have any patches applied. could you recheck this in a precise chroot? it failed for ~10 packages in the precise rebuild test with 4.707:29
michaelh1doko: what revno is the snapshot?  I'll find something closer as well.07:30
doko182617, 20111222, I'll build a new one as well07:32
michaelh1Hmm.  That's pretty close to mine.07:33
michaelh1I'll put that revno through the builders.07:33
benjiGHi all, who can help me with git.linaro.org ? since 2 days I can't see my changes on git trees ...08:35
=== michaelh1 is now known as michaelh1|away
benjiGibiris,asac: do you know who can help me on my git.linaro.org issue ?09:17
fabobenjiG: IS09:19
faboI poked them now09:20
benjiGfabo: do you also have issue with git.linaro.org ? 09:21
fabobenjiG: I'm trying now09:21
fabobenjiG: it works for me -> http://git.linaro.org/gitweb?p=people/fboudra/benjig-test.git;a=summary09:25
fabothough it's a fresh created repo09:25
fabobenjiG: which repo is it?09:32
benjiGfabo: this one http://git.linaro.org/git/people/bgaignard/cmatest.git09:34
fabobenjiG: 2nd test, older repo/fresh clone, works for me09:36
fabobenjiG: have you tried that?09:36
benjiGfabo: I have try to create a new repo (cmatest2) but it isn't display in gitweb 09:37
fabook09:37
benjiGfabo: I can clone my repo but the latest commit isn't fetched 09:38
alf_fabo: The "Using pre-built image" instructions for 2011.12 Android Origen and Snowball don't work. Trying to download the file with either curl or wget, gets the html license agreement page, not the image.10:15
faboalf_: obviously, you can't grab the files with curl/wget. fixing..10:17
danilosgesha, deepti, hi :)10:22
geshadanilos, hi and hny :)10:23
daniloshny too! I hope you had nice holidays :)10:24
geshayep, thanks... you too?10:24
pm215http://patches.linaro.org/project/qemu-devel/ # patches.linaro graphs look kinda silly this month :-)10:26
danilosgesha, yeah, I did, thanks :)10:27
ramana_awayI think that's true of all patches.linaro.org graphs. 10:28
ramana_awaytoday. 10:28
=== ramana_away is now known as ramana
fabodanilos: you're back? :)10:30
danilosfabo, I am, hi :)10:30
fabodanilos: hi and happy new year :)10:30
danilosfabo, thanks, same for you!10:31
fabodanilos: see backlog, benjiG has an issue with git.linaro.org10:31
fabohmm no log for you :) 10:33
fabodanilos: < benjiG> Hi all, who can help me with git.linaro.org ? since 2 days I can't see my changes on git trees ...10:33
danilosfabo, yep, I saw it on irclogs.l.o aboev10:34
faboah cool10:35
benjiGdanilos, fabo: to summarize: I can push my commit on cmatest ($ cat refs/heads/master give the correct commit ID) but git clone (or gitweb) seems to be blocked on the last commit I did on 12th december10:39
danilosbenjiG, fwiw, doing "git log" on git.linaro.org directly also points at the December 2nd commit as the latest10:43
danilosbenjiG, also, it seems to me that the refs/heads/master contains that commit ID as well (from dec 2nd)10:44
danilosbenjiG, so, it all seems consistent to me at least10:45
benjiGdanilos: on cmatest directoty $ cat refs/heads/master give me 34ae9cc88054fb09c4f1c82596563c2c2c9c12e1 10:46
danilosbenjiG, you seem to be using the old git.linaro.org machine10:47
benjiGdanilos: that is my feeling too ...10:47
danilosbenjiG, try ssh-ing directly into git.linaro.org10:47
danilosbenjiG, and check your .ssh/config or wherever you might have a config for the domain set up10:48
deeptidanilos Hello10:49
danilosdeepti, hi10:49
deeptidanilos Happy New Year10:49
danilosdeepti, thanks, I wish a happy 2012 for you as well :)10:50
deeptidanilos Thanks10:50
benjiGdanilos: thanks, it was a problem in .ssh/config 10:51
zygafabo, hi10:52
fabohi zyga 10:55
danilosbenjiG, yw10:56
zygafabo, quick update from yesterday: I pushed my deployment tool branch to launchpad10:56
zygafabo, it has a lot of good patches already10:56
fabozyga: could I poke you later? syncing with asac10:56
zygafabo, I'm currently testing celery support and that will be the final patch10:56
zygafabo, ok10:56
danilosdeepti, any news on the ci.l.o snapshots pass-through?11:03
deeptidanilos, ah! I am working on it .. but another bug came up (911205) yesterday .. so working on that now11:04
deeptidanilos, should be able to fix it by this week 11:04
danilosdeepti, ok, thanks11:05
danilosdeepti, I wonder if 911205 is something that's new or just something that you have discovered recently?11:05
deeptidanilos, its new in a way, I should have started using the hwpacks present under oneiric sometime back itself, but did not realise it till yesterday :)11:09
danilosdeepti, ack, thanks11:15
ramana`mru: ping11:42
markos_doko, hi, both gcc-snapshot and gcc-4.7 built for armhf, both fix #641849 (webkitgtk+ and now qtwebkit ICE), #646163 (neon code) and #630752 (shibboleth and ace ICE). Any chance of having those fixes backported to the gcc-4.6 package on Debian armhf?11:58
mruramana`: holiday-pong11:59
dokomarkos_, is this fixed in gcc-linaro?12:00
markos_didn't check, but it's fixed in gcc-snapshot :)12:01
ramana`mru: ah - won't bother you too much then - the question really was asking if there was an easy way to test ffmpeg. I couldn't remember if it had a testsuite or not. 12:01
markos_doko, would testing ubuntu armhf be sufficient?12:01
loolisn't the Linaro patch enabled on Debian armhf too?12:02
markos_it was for 4.4/4.5 not sure about 4.612:02
lool4.6.1-15 says "Update the Linaro support to the 4.6-2011.10-1 release"12:02
markos_ok then12:02
loollast update was from 8th december, but not sure what that corresponds to12:03
mruramana`: firstly, you should use libav, not ffmpeg12:04
dokomarkos_, you explicitly told me to disable linaro for debian armhf12:04
loolmarkos_: Hmm actually it seems it's not enabled12:04
mruand there is indeex a test suite12:05
loolsee debian/rules.defs around line 28212:05
mru*indeed12:05
markos_doko, yes you're right, to be honest I'm unsure what is the best solution here, are these fixes in gcc linaro already?12:06
dokomarkos_, I didn't look12:06
loolmarkos_: the best way to find out might be to try Ubuntu's gcc-4.612:06
markos_lool, yes, I'll do that asap12:06
markos_doko, lool in general would you recommend enabling linaro for debian armhf?12:07
markos_I'd definitely be in favour of that if it meant having fixes for such bugs earlier than FSF12:08
mruramana`: see http://libav.org/fate.html12:08
ramana`mru: I wanted to test my patch for the fixed to floating point conversions on something more than simple tests. 12:09
ramana`mru: ok - if libav will have cases where there would be uses of that I'd be interested. 12:10
dokomarkos_, who will then care about armhf only issues in Debian?12:10
ramana`markos_: I would rather have the fixes in FSF 4.6 first rather than Linaro or Debian or Ubuntu. Then it doesn't matter. 12:11
ramana`fixes = bug fixes where appropriate. 12:12
mruramana`: such conversions happen in the dts (aka dca) decoder12:13
loolmarkos_: I would think the armhf porters should decide at large; there might be regressions introduced by the patch, but we get them in Ubuntu too and they usually get addressed decently fast12:13
ramana`mru: so running fate would be good enough ?12:14
loolmarkos_: It's a question of whether we have man power to backport just for Debian the specific changes you care about, or whether Debian can live with the compiler issues you mentioned until the move to 4.712:14
loolthere's also the question of whether this is just a temporary one-off situation, or whether this will last for a while -- will gcc-4.7 also have a lot of ARM bugs that only get tackled on 4.8 and need backporting that gcc-linaro 4.7 brings?12:14
mruramana`: that and making sure the instructions are actually used12:15
markos_ramana, doko, that's fine, my question is purely hypothetical, the main problem is as lool said, how long to wait until such ICEs get fixed, I understand that as gcc matures this will get better over time12:15
markos_lool, I would expect that things will gradually improve and as with other arches, ICEs will be a thing of the past soon :)12:15
loolI don't have evidence fo these things; if you ask for my personal opinion, there's good coordination between people working on ARM in Debian and in Ubuntu, so if we could have close toolchains to work from the same base, that would be ideal12:15
mrulibavcodec/dca.c is where I came across the code12:16
markos_-eg I haven't seen an ICE on amd64 or powerpc for years and I do compile a lot of stuff on both :)12:16
ramana`mru: ok. that's fair enough.12:16
mrumight be other places too12:16
loolmarkos_: it's really a question of when rather than if; I agree it should go away, but is it a 3 months thing or a 2 years thing?12:16
ramana`I've seen it get triggered in ffmpeg code in quite a few places when I built it with --disable-asm. 12:16
ramana`markos_: I am happier with ICE's than silent wrong-code.12:18
ramana`mru: thanks for the pointers - I'll give it a go later today 12:18
danilosgesha, btw, you may find http://httpd.apache.org/docs/2.1/programs/httpd.html interesting (note that the process on ubuntu is /usr/sbin/apache2 though)12:18
markos_lool, that's why I asked doko about his opinion. In practice, I'd be fine with things as they are and only backports for specific fixes that produce the ICEs. But if that proves to be too much work, then perhaps enabling the linaro patchset for Debian as well, might be a good alternative12:18
markos_ramana, I agree, it's just that an ICE on a package which gets built for 10+ arches can be difficult to fix in the source package, esp when it builds everywhere else12:19
ramana`markos_: agreed in that case I'd rather negotiate for the fix in FSF 4.6 if the release managers agree and then take it on a case-by-case basis. 12:20
geshadanilos: thanks :)12:20
ramana`markos_: each patch has to be judged upstream with risk vs reward. 12:20
deeptidanilos, can u review the changes for the bug 911205.. I just submitted the changes12:32
ubot2`Launchpad bug 911205 in linaro-ci "Use the latest hwpacks available under http://snapshots.linaro.org/oneiric/" [High,In progress] https://launchpad.net/bugs/91120512:32
=== davidrusling_ is now known as davidrusling
markos_doko, so just tried with ubuntu gcc-4.6 4.6.2-9ubuntu1, #641849 (webkit), #646143 (neon) were fixed, but not #630752 (ace+shibboleth)12:38
loolmarkos_: Could you check whether there's an open bug against gcc-linaro for the last one?12:42
markos_lool, LP:#73666112:45
dokobug 73666112:46
ubot2`Launchpad bug 736661 in gcc "armel FTBFS due to compiler ICE" [Medium,In progress] https://launchpad.net/bugs/73666112:46
ramana`doko, markos_ : http://gcc.gnu.org/ml/gcc-patches/2011-12/msg01550.html 12:48
dokoramana, yes, waiting on feedback12:49
markos_oh it's present on x86_64?12:49
markos_as well12:49
rsalvetitgall_foo: seems using the lightdm login for xbmc works just fine after installing the xbmc package13:01
rsalvetiso guess it should work just fine if you just make the xbmc as the default lightdm session for the set-top box image13:01
akgranerplars, do you have some time today for a call? 13:19
zygaplars, there?13:21
hrwFinally finished writing my 2011 timeline: http://marcin.juszkiewicz.com.pl/2012/01/04/2011-timeline/13:25
=== bero|2 is now known as bero
dzinasac_: mumble?14:02
asac_dzin: yep.14:06
_ThomasIs there any support for sound over hdmi on the origen ubuntu-image?14:11
=== davidrusling_ is now known as davidrusling
=== nytowl_away is now known as nytowl
nytowlHi _Thomas once HDMI gets fixed for the 3.2 kernel sound over HDMI should just work14:15
_Thomasnytowl: Gets fixed?14:15
_Thomasnytowl: How is the sound part done, in that case?14:15
_Thomasnytowl: We're using parts of your hdmi-code to do fb->hdmi14:16
nytowl_Thomas, the  ALC5625 support for Origen somehow works it in14:18
nytowl_Thomas, I don't know the details but you could talk to giridhar and he can fill you in14:19
_Thomasnytowl: ok, but that's an external chip?14:19
nytowl_Thomas, yes14:20
nytowl_Thomas, it is now supported from the staging drivers14:20
_Thomasah, so it's probably hooked up to the hdmi on the outside as well?14:20
_Thomasor there's a feedback-loop?14:20
nytowl_Thomas, that has always been my assumption , I can check the schematics for you14:21
_ThomasAre the schematics for the origen board available somewhere I can download them from?14:21
nytowl_Thomas, there are partial schematics available on the origenboard.org website14:21
nytowl_Thomas, the full schematics aren't available14:22
_Thomasnytowl: Didn't find much on what's available in public, at least14:26
nytowl_Thomas, I didn't think there would be anything there, I'm just looking at the private one right now. , trying to find some interconnect14:27
_Thomasnytowl: I'm working on xbmc right now, btw14:30
_Thomasnytowl: I expect to have an early version of xbmc using the mfc ready some time this evening / night14:30
_Thomas(mfc + fimc, that is)14:31
nytowl_Thomas, nice that should make a good demo14:32
_ThomasWhich is why I'm curious about the sound part14:32
_ThomasSince all we have is exynos -> hdmi14:33
_ThomasSo I'm wondering if I need to do dummy alsa -> hdmi or something14:33
Sagehi, I took ICS image from http://snapshots.linaro.org/android/~vishalbhoj/tracking-4.0.3/5/target/product/pandaboard/ but that only draws the android logo and "desktop" is not coming up. Is that normal and is there some other image that should work?14:42
Sageah, forgot to load the proprietary blobs.14:48
daniloscjohnston, hi, hny :) I wonder if you know off the top of your head if we can easily change the pad to use for summit.linaro.org to pad.linaro.org instead of pad.ubuntu.com?14:50
amitkmturquette: call?15:00
plarsakgraner: hi, sure, when are you free?15:02
faboSage: :)15:04
* Sage is tired15:05
amitkamitdk: call? :)15:05
amitdkamitk: some problem in the line15:06
amitkamitdk: all 3 numbers?15:06
akgranerplars, on a call now - can I ping you when I an finished?15:07
amitdki tried bangalore, india international..15:07
amitdki will try few more15:07
plarsakgraner: sure, take your time, maybe early afternoon or late morning would be best for me15:07
amitkamitdk: try mumbai or worst case skype to the us toll free15:07
markos_doko, also, what's with gdc-4.6 and libphobos? the BR says it only needs porting for s390/s390x but it also fails on other arches15:46
dokomarkos_, please send a patch15:48
DaveDavenporthmm getting: 15:50
DaveDavenportPVR:(Error): PVRSRVBridgeCall: Failed to access device.  Function ID:3223086860 (Invalid argument). [250, /pvr_bridge_u.c]15:50
markos_doko, requesting what? disabling libphobos for arm? or implementing arm support for libphobos?15:50
JesseBarkeralf_, ping?15:51
alf_JesseBarker: pong15:51
JesseBarkeralf_, mumble?15:51
alf_JesseBarker: I am in15:51
zygaplars, hi15:56
plarszyga: hey, you're back!15:56
zygayeah, network woes :/15:56
zygaplars, I have some news 15:56
plarszyga: I saw your development branch yesterday, but didn't see the changes we discussed15:57
zygasome good some bad15:57
plarszyga: ok, shoot15:57
zygathe good news is that pip bundle is a perfect cure for deployment 15:57
plarszyga: or would you prefer I call?15:57
plarsok15:57
zygaI've adapted most of the script to only call pip once15:57
zygaand only to make a bundle 15:57
plarsand if it fails?15:57
zygaonce that is done you can deploy the bundle which looks like "l-d-t install" did before15:57
zygaif it fails you just try again, all it's doing is writing one file15:57
plarsok15:58
dokomarkos_, the latter, and maybe speak to the normal uploader (Ian)15:58
zygainstall on the other hand is 100% offline15:58
zygaso I like that15:58
zygait's good15:58
zygaand all15:58
zygathe bad stuff15:58
zygais that pip bug 416 is still open and currently prevents us from installing anything from source without a patched pip15:58
ubot2`Launchpad bug 416 in gpa "Can't use keyring-editor" [Medium,Fix released] https://launchpad.net/bugs/41615:58
zygaand I'm not sure how to solve this15:58
zygacurrently I think that the best course of action is:15:59
zyga1) move pip installation to be part of 'setup' where a patched pip is kept for lava15:59
zyga2) put more resources into fixing pip upstream15:59
markos_doko, gdc-4.4 did work on armhf, did it not use libphobos as well?15:59
zygaplars, bah, I need to go for a moment, please think about pip and tell me what you'd like see done16:00
dokomarkos_, it's Dv1, -4.6 is Dv216:00
markos_right16:00
markos_ok, i'll try to take a look16:01
plarszyga: well, looking at the bug, it looks like the pull request was closed16:01
=== salgado is now known as salgado-lunch
zygahmm16:04
* zyga checks16:04
zygawrong number 16:05
zygahttps://github.com/pypa/pip/pull/33716:05
zygaplars, anyway, issue stands, I need to run again for dinner16:06
plarszyga: ok, ping me when you return please16:06
DaveDavenporthmm fixed the pvr module issue.. not stil unable to connect16:16
zygaplars, back now16:25
plarszyga: ok, so the bug... it's not even our patch, but the guy that submitted it seems to have withdrawn it16:26
zygaplars, wrong patch, please look at my merge request for pip16:26
plarszyga: oh, I see... I was going by the previous number16:27
zygahttps://github.com/pypa/pip/pull/33716:27
plars:)16:27
zygaright16:27
* plars looks again... sorry, window flipping a lot right now and didn't see the update16:27
plarszyga: so what would it take to add a test?16:27
zygaplars, I did not look at that yet16:28
zygaplars, presumably I can add the test with some insight into pip (I already read quite a lot of the code, just not the tests)16:28
plarszyga: sure... not a high priority though.  We can pull our own for now, but I don't see it as useful for us to invest a lot of time and make it a priority16:29
zygaplars, it's not a big priority but we should fix it soon to ensure that subsequent (precise+1, etc) oses that bundle pip (also virtualenv) are working out of the box16:30
zygacurrently a lot of the workflow is broken because of those few lines16:30
plarszyga: fair enough, I just don't want fixing pip to turn into a time suck16:31
zygaplars, understood16:31
zygaplars, so shall we pull pip into the setup step for now?16:31
plarszyga: yes, thought it was already there in the script16:31
zygaplars, no, currently it is done in a special way per-instance16:32
plarszyga: we've been having to pull the fixed version already right?16:32
zygayes16:32
zygaok16:32
plarszyga: ahh, so you mean installing it on the whole system and not per-instance?16:32
zygain /srv/lava/common maybe16:32
zygabut I need to re-think that16:32
zygawe can still have it in an instance as before16:32
zygajust that 'install' will require network access again16:33
plarszyga: what's wrong with having it in the instance as we do now? it's not as if it's huge16:33
zygaand that may be flaky16:33
plarszyga: so it's not 100% offline for now... that's ok.  The fully offline operation is a plus, but wasn't really a critical thing for us16:33
plarszyga: as this would only affect a new instance, failure isn't that hard to recover from if the one thing (pip) fails to download16:34
zygaok16:34
zygaI'll see if we can simplify that too16:34
plarszyga: my bigger fear is that if we're trying to upgrade an instance, and it blows up, I don't want it to leave it in an inconsistent state16:34
zygaI guess we can bundle pip16:34
zygaand I never though about that :)16:34
plars:)16:35
zygaso pip would install itself16:35
zyga(bundle as in put in the bundle)16:35
plarssure, we are already bundling other things, so why not just add it16:35
plarsright16:35
faboplars: thanks for the blueprints16:37
plarsfabo: need to draft them still, but they are pretty straightforward I think16:38
fabook16:38
faborsalveti: 2 build-deps are missing in Ubuntu to push xbmc anyway. I started to look into the build failure16:39
joeyakgraner: I can go now if you want 16:40
akgranergreat - I'll call ya now16:42
zygaplars, I think that it is a little better than I thought16:42
zygaplars, you only need fixed pip to create a bundle16:43
zygaplars, later on you can use regular pip to unpack16:43
plarszyga: but what about updates?16:43
zygaplars, and since regular pip works with tarballs most people won't see any problems16:43
plarszyga: you would need it again for that I assume16:43
zygaplars, you update with another bundle16:43
zygaplars, anyway, it needs some more experiments but otherwise I think we are good16:44
joeyakgraner: was that you? if so, call again16:44
joeyakgraner: it had callerid blocking on it and I don't answer those... too much spam16:45
akgranerjoey yep16:45
akgranerjoey yeah pete changed it16:45
rsalvetifabo: yup, did you check at debian? don't remember if xbmc was even available at debian16:48
rsalvetiafaik it was almost there16:48
faborsalveti: it's available in debian-multimedia16:48
rsalvetiok, great16:49
rsalvetijcrigby: hey, did you see my email about the abi problem with lt-omap?16:55
rsalvetijcrigby: I remember you were fighting with this in the past, but not yet sure if your auto-bump abi script was done 16:55
asac_danilos: any clue how often status.l.o cronjob runs?16:59
asac_e.g. at whta times. i added loads of blueprints to one card around lunchtime, but dont see the lane/card view pick them up (yet)17:00
jcrigbyrsalveti, yes I responded, will fix that asap17:00
rsalvetijcrigby: oh, just saw your email17:03
rsalvetigreat, thanks17:03
rsalvetijcrigby: let me know once you fix or have at least a workaround for the lt-omap kernel17:03
rsalvetibecause the newer one fixes the video decoding for panda 17:03
jcrigbyok17:03
rsalvetiand we'd like to have it available at the overlay asap :-)17:03
rsalvetithanks17:03
danilosasac_, it runs every hour17:03
danilosasac_, there's the problem of some blueprints not being picked up17:10
asac_danilos: hmm17:10
asac_danilos: its about http://status.linaro.org/card/ANDROID2011-ICS-LEB.html17:10
asac_maybe i got the syntax wrong17:10
=== salgado-lunch is now known as salgado
danilosasac_, right, there were already blueprints not being picked up for it17:10
asac_danilos: how do you see that?17:11
danilosasac_, Zach told me when he filed them17:11
asac_danilos: example https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-snowball-mali-ics-support17:11
asac_danilos: i didnt spot any blueprint that had the roadmap id17:11
asac_danilos: i added like 15 today17:11
danilosasac_, first, I think it needs to be in the "Meta:" section17:12
asac_meta?17:12
asac_ok. where is the doc again? :)17:12
asac_danilos: can headline and acceptance be multi line fields like in there?17:13
asac_are there empty lines allowed within meta section?17:13
danilosasac_, I am trying to find the docs myself to confirm that (or I'll have to dig into code)17:13
asac_hehe17:14
zygaplars, looks good17:14
zygaplars, pip bundle even reuses my download cache!17:14
asac_so i would anticipate that there must be no empty lines and headline/acceptance also has to be in a single line17:14
asac_so here: Meta:17:14
asac_Roadmap id: ANDROID2011-ICS-LEB17:14
asac_Headline: Linaro Android for Snowball now has support for Mali hardware17:14
asac_Acceptance: It is possible to boot Android on Snowball with HW accelerated graphics17:14
asac_let me try that :)17:14
danilosasac_, yeah, that's what I think it should be like17:15
asac_done. lets see if it picks that up. if you have docs that would be cool17:15
asac_otherwise lets trial and error :) ... just not sure how long to wait17:15
plarszyga: does this get us a bit closer to the codelines stuff like michael was discussing?17:15
zygaplars, huh?17:16
zygacodelines?17:16
plarszyga: it was in an email before Christmas17:16
zygaah17:18
zygaI remember17:18
zygamaybe17:18
zygadoing what he described is harder to get right17:18
zygathis gets the basic use case working reliably I guess17:18
danilosasac_, I think production run is at half an hour, and staging run is on the hour, you can use both to see if it has picked up your BP17:19
danilosasac_, http://status.linaro.org/staging/lane/17:19
plarszyga: so it's ready to land? I can fire up a cloud instance and test if that would help17:20
asac_danilos: cool17:20
asac_thx17:20
zygaplars, I'm changing how cmd_install works now, stay tuned17:20
asac_danilos: and average time it takes is like 15 minutes or what?17:20
=== asac_ is now known as asac
danilosasac, yeah, I suppose the HTML generation time helps answer that as well (i.e. 7-8 minutes actually)17:22
asaccool. /me takes a break17:23
rsalvetirobclark: http://paste.ubuntu.com/792764/17:25
rsalvetithe log from xbmc 17:25
rsalvetiLinuxRendererGLES.cpp:1440: fail: GL ERROR: 128217:25
rsalvetiI also sent by email17:26
rsalvetimaybe this is the cause of the black screen17:26
zygaplars, I just did offline installation here :)17:31
zygaplars, do you have some time to burn?17:31
plarszyga: what do you want me to try?17:32
=== michaelh1|away is now known as michaelh1
zygaget a vm ready17:35
zygaplars, oneiric 17:35
zygaplars, branch my development version of l-d-t17:35
plarszyga: will do17:37
zygaplars, get http://people.linaro.org/~zyga/lava.pybundle (still uploading17:40
zygaplars, readme is out of date17:41
zygaplars, but install / upgrade / remove should work 17:42
zygaplars, you can also try getting a new bundle yourself with the bundle command17:42
zygamichaelh1, hi17:45
zygamichaelh1, a friend of mine working on non-ARMv7 (mips and older arm) is asking if by using linaro GCC he's getting something apart from bugfixes that are not yet released by FSF17:46
zygamichaelh1, err, sorry, he wants to use our toolchain on ARMv7-m17:47
michaelh1zyga: yeah, that should be fine.  We don't support 7M but it will work fine and we'll look into any bugs17:49
zygamichaelh1, I understand that, I think my friend wanted to know if there are any obvious benefits?17:49
JesseBarkerfred23`, https://code.launchpad.net/~linaro-graphics-wg/nux/nux-gles2-11.1217:53
jaytaokohttps://code.launchpad.net/~linaro-graphics-wg/nux/nux-gles2-11.1217:53
zygaplars, bundle uploaded17:53
zygaplars, how is that VM of yours?17:53
plarszyga: not running, not sure what's up with it, about to just run on ec2 instead17:54
zygaplars, ok17:54
zygaplars, I'd like to have an instance for testing clean install & upgrade from previous release eventually17:54
zygaplars, since now it is entirely self contained it could be easy to say 'release 2012.01 works, upgrades failed on xyz'17:55
* zyga fires up virtualbox17:59
hrwhave a nice day18:00
zygaplars, I'm going away now18:09
zygaplars, check if you can install an instance from the bundle I posted on people.linaro.org18:09
zygaplars, also check if you can recreate a bundle yourself18:09
zygaplars, lastly, look at my merge proposal18:09
zygaplars, and comment there, thanks!18:10
JesseBarkerjaytaoko, https://code.launchpad.net/~linaro-graphics-wg/unity/unity-gles2-11.1218:10
michaelh1zyga: embedded work normally has different requirements - more correctness and long term support than speed.  We're as correct as upstream though...18:13
DaveDavenportreliability, analysability and durability.. the 3 ility's18:16
robclarkrsalveti, any chance you aren't using latest sgx stuff from ppa?  (re: log you sent)18:35
rsalvetirobclark: nops, using the latest one18:36
rsalvetirobclark: 1.7.10.0.1.4-118:36
robclarkhmm, I'm not sure why PVRSRVWrapExtMemory() would fail..18:37
robclarkI don't suppose you have a setup to try w/ ubuntu kernel?  I'm not sure if there is some tiler difference between newer linaro kernel and what I was using18:37
robclark(my xbmc fs is just plain vanilla 11.10 kernel)18:37
plarszyga: this is not working at all for me18:37
rsalvetirobclark: sure, let me try 18:38
markos_rsalveti, did I read correctly that you're preparing settop box images with xbmc?18:39
rsalvetimarkos_: yup, almost done 18:39
markos_for pandaboard?18:39
rsalvetiwe have xbmc at our ppa already, and trying to finish the remaining issues now18:39
rsalvetimarkos_: yup18:40
markos_cool, would definitely like to try that one out :)18:40
robclarkrsalveti, but yeah, PVRSRVWrapExtMemory() failing means creating the eglImage would fail, which would -> black screen18:40
rsalvetirobclark: guess this function is internal to the pvr driver18:40
rsalvetiwonder why it failed 18:40
rsalvetimarkos_: we will announce once we get it running :-)18:41
rsalvetiprobably in the next few days18:41
robclarkyeah, but only reason I could think of it failing is running out of gpu virtual address space..  but if you are only running xbmc this shouldn't happen18:41
markos_rsalveti, cool :)18:41
DBOhttps://code.launchpad.net/compiz-core18:43
robclarkrsalveti, yeah it is OoM18:44
rsalvetirobclark: hm, weird 18:45
robclarkrsalveti, ok, one more thing to try:  XBMC_DEBUG=decoder,renderer /usr/bin/xbmc18:45
=== salgado_ is now known as salgado
salgadocarif, around?18:46
rsalvetirobclark: will try, let me just reboot with the newer kernel now18:47
rsalvetinewer == 3.0 from ubuntu18:47
robclarkok thx18:47
rsalvetirobclark: same error, even with ubuntu's kernel19:05
robclarkhmm19:05
robclarkI wonder what else could be different about your environment..19:06
rsalvetirobclark: boot args?19:06
robclarkwhat are you using?  I was using stock 11.10 setup..19:06
robclarkcould you try w/ XBMC_DEBUG=decoder,renderer19:06
robclarkit seems strange that it would be out of memory on the first buffer, but this would help tell..19:07
rsalvetirobclark: with xbmc debug: http://paste.ubuntu.com/792996/19:07
rsalveticonsole=tty0 console=ttyO2,115200n8  root=UUID=f75e3511-4d48-4f1a-b109-98a6804715e7 rootwait ro earlyprintk fixrtc nocompcache vram=48M omapfb.vram=0:24M mem=456M@0x80000000 mem=512M@0xA000000019:07
robclarkrsalveti, ok, one difference is I have vram=32M (omapfb.vram doesn't matter)..   although I also did try w/ vram=40M19:08
robclark(but the flipchain stuff is not very reliable so I went back to 30M)19:08
robclarkrsalveti, yeah, it is on very first buffer.. which too strange a coincidence..19:10
rsalvetirobclark: weird, fails all the time, doesn't matter with which arg I use 19:16
robclarkrsalveti, can you rebuild pvr kernel module if I give you a small patch for debugging?19:17
rsalvetirobclark: sure19:17
rsalvetihm, the ppa provides 1.7.10.0.1.3 instead of 1.7.10.0.1.419:17
robclarkrsalveti, http://pastie.org/312663519:20
robclarkmight have to tweak path slightly, but I guess you can see what to do..19:20
robclarkI guess if I know where it fails on kernel side I might have some ideas... but I have a hard time believing that it is really out of memory19:20
rsalvetirobclark: yup, will try this patch19:21
robclarkthx19:21
rsalvetirobclark: same issue with the exact same sgx version from the PPA, time to patch the kernel module19:23
robclarkrsalveti, fwiw, tooks like there is a 1.7.10.0.1.4-1 in omap-trunk ppa19:24
robclarknot sure what the difference between .3-119:24
rsalvetirobclark: yup, that's the one we're using by default at linaro19:24
robclark(not sure if it is too late in the day to catch Xav?)19:24
rsalvetibut tried with the .3-1 now, same issue19:24
rsalvetiprobably19:24
robclarkfwiw, I didn't make any changes on kernel side pvr module on that filesystem19:24
=== ramana is now known as ramana_away
mwhudsongood morning19:46
=== cody-somerville_ is now known as cody-somerville
rsalvetiholy, sd is horrible slow19:49
rsalveti*horribly19:50
rsalvetirobclark: http://paste.ubuntu.com/793048/19:54
rsalvetifrom the point I started xbmc19:54
robclarkthx19:54
rsalvetiin theory the only thing that's different now is the xbmc and boot loader 19:58
robclarkrsalveti, I think I know the issue.. 19:59
robclarkand I think you maybe aren't using latest gst-ducati (I switched it to not use tiled buffers by default)20:00
rsalvetirobclark: yeah, not using the one you sent the link yet20:00
robclarkahh, ok20:00
rsalvetiguess I can easily recompile the latest on20:00
rsalvetistill using the one from the PPA20:01
robclarkyeah, memwrap doesn't work well w/ 2d buffers20:01
robclarksince tiled region is pure pfn (no 'struct page *'s)20:01
rsalvetirobclark: should I just get trunk? or any other specific commit hash?20:01
rsalvetihm, ok20:01
robclarkrsalveti, gst-ducati latest master from git would be fine20:01
rsalvetirobclark: ok, great, let me try building it20:02
rsalvetiget back to you in a few20:02
robclarkrsalveti, from gitorious tree (fwiw).. my github tree may not always be up to date 20:02
CcSsNETftw, xbmc is a waste of resources. although no one here probably cares of my opinion20:02
rsalvetirobclark: ok, good to know 20:03
robclarkCcSsNET, as in uses too much memory, etc?20:04
CcSsNETyes all of that, and video card mem and gpu as well20:05
robclarkI think the code could be structured a lot better, but from user standpoint, it isn't too bad to use20:05
CcSsNEToverall waste of resources20:05
robclarkI've not seen one better, anyways..20:05
CcSsNETfor example. i can stream video to a dedicated pentium 2 machine with minimal lag if i set it up with minimal xorg configs. but try doing that with xbmc it will be laggy even with a p4 and higher end vid card20:07
robclarkwell, in theory xbmc could work without even xorg, although not sure if mesa or any of the desktop gpu's support non-xorg setup..20:08
CcSsNETwell i havent used it nor researched it enough to care how its useable. once i see an app is slow, i mark it useless in my head :)20:09
robclarkanyways, if I was going to write RMC (Robs Media Center), I might design it differently..  but then you run into the problem that it generally sucks because I'd be the only one using it, and the scrapers for downloading movie/tv info wouldn't work, etc, etc20:10
CcSsNETlol20:10
CcSsNETthe only advice i could give the xbmc guys is stop using opengl for the interface and do it without gpu acceleration in general.20:11
CcSsNETthe xmbc interface that is. sure accelerate the videos!20:12
robclarkpart of the issue is that xv doesn't really cut it if you want eye-candy..20:13
robclarkie. translucency and shadows over video,  on-screen menus type of stuff20:13
CcSsNETmeh, i personally use fluxbox for my desktop. i dislike eye candy anyway20:13
robclarkand so far GL has been the only API around to do that..20:13
CcSsNETfairly sure amarok does on screen menu's without touching GL20:14
CcSsNETalthough, i like amarok 1.4, and none of the newer versions20:15
robclarkI guess you could do it w/ xrender20:15
robclarkalthough you end up still using the gpu20:15
CcSsNETeither way xbmc is horrificly wasting resources20:16
robclarkonly difference if you use xrender, is you move the memory allocation into the xorg context..  so sort of a false economy20:17
robclarkI guess you can also do distructive overlays, ie. write to the YUV buffer..20:17
robclarkbut that is absolutely horrible for performance20:17
CcSsNETwell amarok uses an overlay thats 100% faster or more, they should research that approuch20:18
CcSsNETnice. someone is buying my socket 604 xeon cpu's on ebay20:19
CcSsNETtook only a year of listing...20:19
robclarkxbmc does have some support for overlay20:21
DaveDavenportrobclark: on what platform did you get your xmbc running fine.20:28
DaveDavenportI rebuild xbmc, and compiled the ducati stuff from git20:29
CcSsNETmy guess is core2 or newer ;)20:29
DaveDavenportyet still xbmc is to slow20:29
robclarkDaveDavenport, I was using ubuntu 11.10 + updated gst plugins20:29
DaveDavenportfor panda.. 20:29
DaveDavenporthmm same for me20:29
robclarkyou are using my gstreamer-eglimg branch on xbmc?20:29
robclarkor just mainline?20:29
robclarkand if former, check config.log to confirm it detects gst?20:30
robclark(probably some -dev pkgs I had to install for that)20:30
DaveDavenporthmmm that could be  it.20:30
DaveDavenportugh.. I don't have time for this... 20:33
robclarkit is a pain in the a** to build, and takes about a year and a half too20:34
CcSsNETand it doesnt pay  you overtime ^20:35
* robclark was building on a panda w/ hardrive instead of from sdcard.. which helps a *lot*20:36
michaelh1robclark: I find /tmp on tmpfs helps too20:37
DaveDavenportyeah it sucks20:37
DaveDavenportxbmc is to big for tmpfs20:38
DaveDavenportmostly it is time20:38
michaelh1mounirb: ping?  Did you want to talk about training?20:38
DaveDavenportwhile working I am fiddling with fpga boards, students and other stuff.20:38
DaveDavenporthome I have little time to play with panda20:38
robclarkmichaelh1, yeah, if you have sdcard then moving /tmp to tmpfs is imperitive..  but xbmc is waay to big to put src tree in tmpfs20:38
robclarkyeah, .deb pkgs for xbmc would be a big help20:39
DaveDavenportI did the compile on a nice external disk20:39
mwhudsonwhere are these calxeda machines with lots of ram and fast io damnit20:39
DaveDavenportI also have gmpc/gpxviewer/gtodo/stuffkeeper/etc to maintain20:40
robclarkseriously..  I wouldn't mind one to use as a workstation20:40
DaveDavenportaah time time time20:40
DaveDavenportI was planning to use a pandaboard as always on desktop at home20:41
DaveDavenportbut now I am wondering of making it a xbmc box for me parents20:41
robclarkone of my pandas is in the process of becoming the always-on/xbmc pc..  still go a little way to go, but it is getting there20:42
robclarkalthough the case is a bit of a diy project (=> more time ;-))20:43
DaveDavenportwould be nice if the stuff could be in a ppa20:44
mounirbmichaelh1,  yes - currently I am with Andy And Mike talking about training sessions at next connect, can I ping you later?20:49
michaelh1mounirb: hmm, the event was about training...20:49
mounirbmichaelh1:  do you have a calendar event schedule for that?20:55
michaelh1mounirb: yeah, strange eh?20:55
michaelh1mounirb: ah!  The title is "Mike and Mounir sync on training" but it's Mike Levine.  Wonder why it shows on my calendar...21:02
mounirbmichaelh1, I had added it to the linaro events, I should have it on my calendar only21:03
=== salgado is now known as salgado-afk
=== michaelh1 is now known as michaelh1|away
Marexcjb: hey, around ?22:13
cjbMarex: yes22:15
Marexcjb: you should definitelly totally fix the "bigger" keys on the XO on the rubber keyboard22:16
Marexit's really hard to type on that22:16
Marexeg. "backspace" doesn't react well22:17
mwhudsonhm22:56
mwhudsonplars: around?22:56
plarsmwhudson: yep22:57
mwhudsonplars: i got an im from zyga a few hours ago asking me to talk to him around now22:57
mwhudsonplars: do you know anything about that?22:57
plarsmwhudson: I think he was going to ping you about the codelines stuff, I was asking him about it later.22:57
mwhudsonok22:58
plarsmwhudson: I've been poking at the latest stuff in developer branch of ldt22:58
mwhudsoni'll review his branch i guess22:58
plarsmwhudson: basically, he added a bundle command - which needs to be run before trying to install or update22:58
plarsmwhudson: install and update now operate on a bundle, rather than a requirements file22:58
plarsmwhudson: and bundle operates on the requirements file, pulls all the stuff, then install or update can be performed on another machine, or even offline22:58
mwhudsonah right22:59
plarsmwhudson: README/help needs updating of course, and it doesn't work with bzr sources at the moment22:59
plarsmwhudson: otherwise, seemed to work ok for me with a fresh install22:59
plarsmwhudson: incidentally, the documentation on bzr hooks is pretty hilarious23:01
mwhudsonheh23:01
* mwhudson looks at zyga's branch23:04
mwhudsonthis is going to require having lunch first i think23:04
plarsmwhudson: after looking at this for a few minutes, I think I'm starting to feel more confused... is there an easy way of just telling it that after committing to a branch, I need it to tickle a url so that the docs rebuild if there's anything new?23:19
mwhudsonplars: i don't think there's anything built in for that23:26
mwhudsonplars: you can write shell hooks now i think?  so just stick an appropriate curl invocation in there?23:26
plarsmwhudson: in where? I tried adding a post_commit = touch /tmp/foo under [hooks] in .bzr/branch/branch.conf to see if that would have any effect, but no23:27
mwhudsoni don't know the details23:28
* mwhudson takes a look at the docs23:28
mwhudsonplars: i think writing a plugin is going to be the easiest23:30
mwhudsonplars: http://doc.bazaar.canonical.com/bzr.dev/en/user-guide/hooks.html23:30
mwhudsonyou can't make a hook 'stick' to a branch, for good security reasons23:31
plarsmwhudson: yeah, but for that matter then, I may as well just have a script locally that I run that hits them all, and do it every time we release23:32
mwhudsonyeah23:32
=== nytowl is now known as nytowl_away
* mwhudson afk for a bit, will review zyga's branch properly when i get back23:37

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!