/srv/irclogs.linaro.org/2011/08/24/#linaro-meeting.txt

=== asac_ is now known as asac
marcoilJesseBarker: The conf call details are the ones on https://wiki.linaro.org/WorkingGroups/Middleware/Graphics/WeeklyCallDetails , right?11:58
JesseBarkermarcoil, yes sir.11:59
marcoilthanks :)11:59
marcoilthe toll free number works horribly, but the paying one doesn't, I'm starting to get suspicious :P12:00
JesseBarkerAmaranth, call?12:03
AmaranthJesseBarker: yeah, trying to get in now12:03
rsalvetiJesseBarker: also trying to join12:13
JesseBarkerrsalveti, don't know if you saw the mail, but the retrospective has been postponed a bit....12:14
rsalvetiJesseBarker: ok, fine by me12:14
rsalveti:-)12:14
rsalvetiJesseBarker: only thing good to note is that compiz-plugins-main breaks unity and compiz12:17
rsalvetiJesseBarker: as I posted to Amaranth at the irc yesterday12:17
Amaranthoh, eep12:17
rsalvetiseems the changes at the workaround plugin is causing this issue12:17
AmaranthI've been running without ccp and I guess I never loaded workarounds12:18
rsalvetireverted them and it seems to be fine12:18
AmaranthDo you have the stacktrace?12:18
rsalvetiAmaranth: I can try to get it for you, but this is what I got: *** glibc detected *** compiz: free(): invalid pointer: 0x0005b88c ***12:19
Amaranthodd, it's crashing on old code but only after something new was added12:23
rsalvetiAmaranth: were you able to reproduce the crash?12:24
Amaranthyeah12:24
rsalvetiok, great12:24
Amaranthit's crashing on XFree(classHint.res_name);12:24
Amaranthand the documentation says to free those...weird12:25
rsalveti:-)12:26
Amaranthand after I add something that shouldn't have changed that function at all it crashes somewhere else unrelated12:29
* Amaranth fires up valgrind12:29
Amaranthdunno what ended up happening to my panda, it finally stopped booting entirely after trying to switch kernels to something the drivers would work with so I've wiped it with a 2011.07 image12:32
Amaranthso I should at least have that to test a natty 2011.09 if we do one12:32
Amaranthoh lovely, when I run compiz with valgrind it doesn't even crash12:32
AmaranthI guess I'll just back out the change as well for now12:36
marcoilI got disconnected and the dumb number recognition is not working, grrr..12:49
JesseBarkermarcoil,  I think we're about to wrap up...12:52
marcoilJesseBarker: I'm back there :)12:52
Amaranthrsalveti: I don't understand this at all, I reverted the memory leak fix commit and it's still crashing at a random place12:57
Amaranthand now it's just falling over in a way the crashhandler plugin can't deal with12:59
Amaranthrsalveti: So you said you got it fixed on your end?12:59
AmaranthI just reverted the typo fix and the memory leak fix and it's still crashing which makes no sense at all13:00
mansson_#startmeeting13:00
linarobotMeeting started Wed Aug 24 13:00:34 2011 UTC. The chair is mansson_.13:00
linarobotUseful Commands: #action #agreed #help #info #idea #link #topic.13:00
mansson_Hello Android Team! Who's in here?13:00
* bero 13:00
mpoiriero/13:00
* patrikryd is here13:01
botao_sunI'm here.13:01
botao_sunmpoirier: Hi Mathieu, I will discuss something with you after the meeting.13:01
mpoirierok13:01
mansson_no fgiff, cyang?13:02
* cyang here13:02
* fgiff here13:02
* dzin is here13:03
mansson_Great. how about bhoj?13:03
bhojim here :)13:03
fabo/o/13:03
patrikrydbhoj, Welcome!13:04
bhojpatrikryd:thanks13:04
botao_sun+113:04
mansson_Cool. bhoj are you in busines yet?13:04
bhojno.I will starting next week13:04
bhojmonday13:04
mansson_Ah. Cherish your freedom :-)13:05
mansson_#Topic rc2 status.13:05
mansson_botao: Do you care to explain what is happening now, and how it looks?13:06
* tgall_foo takes a seat13:06
botao_sunLong story.13:06
botao_sunwell, actually, the origen currently RC is based on the 2.3.413:06
botao_sunand I'm trying to create a build which will based on the 2.3.513:07
botao_sunand I got a series error during the build generating13:07
botao_sunand I'm trying to solve them with guys here.13:07
botao_sunI'm not sure it can be done successfully at the 16:00 UTC tomorrow.13:08
botao_sunIf not, just use the current RC as our release.13:08
botao_sunBecause this is the first Origen Android release, I think a bootable version is acceptable.13:08
mansson_Did this error materialize when the "correct" kernel came in in a hurry recently?13:09
botao_sunAlthough Angus and his landing team have made some improvement, but I got this news yesterday, time is short.13:09
botao_sunyes, one of Angus' update is the kernel. I have integrated it into the 2.3.5 branch, but not in the current RC13:10
patrikrydIt is caused by building MLO from source :) Origen was configure to build MLO.13:10
patrikrydSame problem we had for Snowball13:10
botao_sunFor that error, bero has submitted a patch and it has been merged into the git repository. The new build is running.13:10
botao_sunThat's the brief for what is happening now.13:11
botao_sunsorry, a update, the new build failed again...13:12
mansson_OK. I can see frantic activity, so I'll even try to hold this meeting short13:12
mansson_but plan B for Origen is not a total disaster I think, It is still progress.13:12
mansson_I won't go through every BP, takes too long..13:13
botao_sunyes, because this is the first release for Origen.13:13
mansson_I have sent out personal mails to each one of you and got mostly reassuring replies.13:13
botao_sunat that time, I picked up the first bootable pinned manifest - that's 2.3.413:13
mansson_Please remember that people are looking at the BPs for information, so please try to keep them updated.13:14
mansson_Is there anything else that is giving last minute headaches?13:15
mansson_Silence is good.13:16
cyanghi botao_sun, it is better to get rid of my manifest if possible since my manifest does not go to any boards13:16
beroJust the "not fully knowing if ffmpeg can be in in time" bit, but since it's not that important to have in the 11.08 build, it's not a huge headache13:16
fabobero: the RC toolchains will be the 11.08 release ones?13:17
berofabo: In all likelihood, yes -- so far, nobody complained about problems with them (at least not to me)13:18
fabook, thanks13:18
mansson_bero: It's really good if we can keep it intact, and we owe you for putting the toolchain straight well in advance of the release date.13:19
mansson_#topic post-mortem topics13:19
mansson_Was it just me or did we experience a little confusion regarding Android version vs. toolchain version?13:20
beroA bit -- probably because the 11.08 toolchain wasn't fully working for a while, so we had a lot of 11.08 builds with toolchain 11.0713:21
mansson_Having both the tools and the target in the same time window is a classic problem, and I know that several of us misunderstood the goal so topic #1 is to improve clarity of the planning the next time.13:22
mansson_Second topic (for next meeting) is Gerrit reviews - How to and how much.13:23
beroI think ideally the toolchain WG would make its releases with a 2wk offset from the OS groups - giving the OS groups more time to integrate the new toolchain and actually report bugs and get them fixed13:23
mansson_Any other topics ?13:23
beroIt would be great to have our releases around the 1st and toolchain WG releases around the 15th or vice versa13:23
mansson_Yes, thats a good suggestion but it also has repercussions for the team work. Iḿ just collecting topics right now.13:24
mansson_OK.13:24
mansson_I think the best way to spend time right now is to work, so the word is free until 15:30 for AOB, then we quit. OK?13:25
botao_sunOK13:26
fabomansson_: libjpeg-turbo 1.1.x port/integration maybe?13:26
cyanghi mansson_, with regard to libjpeg-turbo13:26
fabofor post-mortem topics13:26
mansson_How would you phraze the problem?13:26
cyangTom has sent email to us about it again this morning13:28
tgall_foofabo, yes I suggested that topic to mansson_ as well for this meeting13:28
mansson_It was a long mail, can we catch the problem in one sentence?13:29
tgall_foogetting to the point, are we all aware of the plan for libjpeg-turbo + android ?   .. any questions / comments / feedback ?13:29
patrikrydI'm not aware of the plan...13:29
mansson_Plans should be in BPs.13:29
tgall_foomansson_, and they are13:29
tgall_foohttps://blueprints.launchpad.net/libjpeg-turbo/+spec/engr-multimedia-codec-jpeg-android-1.213:30
fabotgall_foo: nobody from Android is subscribed13:30
tgall_foofabo, if there's a way to subscribe the wg all in one go, that'd be great13:31
patrikrydtgall_foo, How do you benchmark?13:31
cyangI think information should also be updated in https://blueprints.launchpad.net/linaro-android/+spec/linaro-android-integrate-panda-leb-libjpeg-turbo13:31
patrikrydtgall_foo, skia_bench?13:31
tgall_foointernal to cjpeg and djpeg we have some timers at stand and end of operations13:31
beroWhat changes were needed to get libjpeg-turbo to work on Android initially? Just the build system, or anything else?13:32
fabotgall_foo: done13:32
tgall_foobero, android extended the libjpeg API13:32
tgall_foobero, and that extension is what needs t obe ported and submitted upstream to the libjpeg-turbo community13:32
berotgall_foo: Ah, that will make it a bit of a pain13:32
* patrikryd is planning to write an BP...13:32
tgall_fooit was ported to our 1.1.1 version but it was done very quickly and in supportof the demo13:32
tgall_foobero, quite13:32
patrikrydCopied from todays Engineering Report "Write a new BP to benchmark and compare libjpegturbo with the optimised libjpeg that will soon show up on igloocommunity.org."13:33
tgall_foopatrikryd, you could... or you could just use whatis already there13:34
tgall_foopatrikryd, and benchmarking is in the existing bp that covers the 1.2 focused port13:34
patrikrydtgall_foo, There is a benchmark program in Android called skia bench.13:35
tgall_fooso as to this plan, there's been several notes on it trying to get comment.  in the future, it seems like attending this meeting might be a better way to get feedback ?13:35
tgall_foopatrikryd, ah ...  that measures mor than libjpeg performance tho doesn't it ?13:36
mansson_"the problem in one sentence" seems to be BP management. The Android team has a list of BPs for each Milestone. Thats all we care about during the sprint, external BPs gets synced before the sprint. So let's be more clear about that next time.13:36
patrikrydtgall_foo, true (you have remove the other tests or wait for them to finish)13:36
tgall_foopatrikryd, still sounds useful13:37
cyanghi tgall_foo, so the benchmark will be done again libjpeg-turbo 1.1.1 without the fix of https://bugs.launchpad.net/libjpeg-turbo/+bug/826642?13:38
ubot2Ubuntu bug 826642 in libjpeg-turbo "jpeg image is corrupt with libjpeg-turbo on Android" [Medium,Triaged]13:38
tgall_fooalso is the android team maintaining a wiki page targeted for the average bear for getting started, and doing common operations within the code?     That's really been my main frustration with providing android support for libjpeg-turbo13:38
mansson_tgall_foo: You would appreciate a "Getting started with xxx"? Where xxx is=?13:40
berotgall_foo: What exactly would you need? Docs on accessing the code? Or an overview of how the parts of the system are supposed to work together?13:40
tgall_foobero, the later13:41
patrikrydThe are a number of how tos on https://wiki.linaro.org/Platform/Android13:41
tgall_foopatrikryd, ok I'll have a look, thanks13:41
tgall_foothat's all I have unless there's other q's or c's13:42
mansson_OK. But the idea is noted. We have many external contributors to Android milestones every month. They may need some support.13:42
mansson_AOB?13:43
mansson_Not long a go we discovered that this sprint had 8 official working days...13:43
mpoiriermansson_: what sprint ?13:44
mansson_Firstly, I have a feeling that that turned into a very large 8. I think all of us has done 8++13:45
mansson_mpoirier: I mean 11.0813:45
mansson_Secondly, looking at the result it looks like 8 days well spent.13:46
mansson_So we end there. Thanks everybody.13:46
mansson_#endmeeting13:47
linarobotMeeting ended Wed Aug 24 13:47:11 2011 UTC.13:47
linarobotMinutes:        http://irclogs.linaro.org/meeting-logs/linaro-meeting/2011/linaro-meeting.2011-08-24-13.00.html13:47
linarobotMinutes (text): http://irclogs.linaro.org/meeting-logs/linaro-meeting/2011/linaro-meeting.2011-08-24-13.00.txt13:47
linarobotLog:            http://irclogs.linaro.org/meeting-logs/linaro-meeting/2011/linaro-meeting.2011-08-24-13.00.log.html13:47
linarobotWiki:           http://irclogs.linaro.org/meeting-logs/linaro-meeting/2011/linaro-meeting.2011-08-24-13.00.wiki.txt13:47

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