aliendude3500cjwatson, Thanks for the help! It works flawlessly! :)00:06
crimsunslangasek: do I need a UserInterfaceFreeze exception for the pulseaudio task in bug 533877? The effect is an addition of the correct "Connector" in gnome-volume-control's Input tab.00:09
ubottuLaunchpad bug 533877 in alsa-utils "[SigmaTel STAC9228] Recording problem - integrated microphone no longer available on Dell XPS 1330" [Low,Fix released] https://launchpad.net/bugs/53387700:09
crimsun^^ anyone else on the release team00:10
crimsun(err, am I supposed to be asking in -release?)00:10
Riddelltjaalton, bryceh: any X dudes about?00:10
cjwatsonaliendude3500: good, happy to help00:11
brycehRiddell, mm?00:20
Riddellbryceh: has the intel version string changed recently?00:21
RiddellI mean the format00:21
Riddellbryceh: kwin expects something like "20061017" but it seems to be "2009" "Q" "4"00:22
brycehRiddell, it quite possibly could have00:34
brycehRiddell, they have adopted a quarterly release process and it seems quite likely they'd change the nomenclature to suit00:34
brycehRiddell, probably worth making kwin's regex support either style00:35
xnoxthanks to everyone involved in UDD, bzr and lp! I've just did bzr merge-package and it did the right thing without generating a single conflict!00:35
cjwatsonnice, isn't it :)00:36
xnoxcjwatson: It's awesome. Thank you ;-)00:36
cjwatsonnot I, really00:36
xnoxcjwatson: btw - there are no bzr-nightly builds for lucid00:38
cjwatsonI wouldn't know about that00:39
xnoxI can update cookbook branch (or do a merge proposal) but do i need to poke james westby about that?00:39
cjwatsoncheck whoever seems to be uploading the others, or runs the team, or whatever ...00:39
xnoxYeah it's james =)00:39
* xnox is stuck in stone age without bzr nightly builds for lucid ;-)00:40
slangasekcrimsun: I don't think that needs a UI freeze exception00:46
=== jono_ is now known as jono
bdrungcjwatson: what's the status of bug #513197?01:14
ubottuLaunchpad bug 513197 in ntfs-3g "Please merge ntfs-3g 1:2010.1.16-0.1 (main) from Debian testing (main)" [Wishlist,Triaged] https://launchpad.net/bugs/51319701:14
cjwatsonI intend to apply for a feature freeze exception for it, but have not got round to it yet.  no other statuss01:14
cjwatsonArneGoetje: gbrainy.mo is in both base and gnome langpacks, breaking DVD builds.  Where's a good place to report this sort of thing, that isn't specific to a single package?01:29
ArneGoetjecjwatson: those exceptions are handled individually in langpack-o-matic01:36
ArneGoetjecjwatson: so, I take it that it should be only present in the gnome langpacks?01:39
cjwatsonArneGoetje: I don't know; I just know it shouldn't be in both :-)01:42
ArneGoetjecjwatson: ok, I'll check01:42
cjwatsonerr, it's part of what used to be gnome-games isn't it?01:42
cjwatsonand it's in ubuntu-desktop and not kubuntu-desktop01:42
cjwatsonso yes, I think that would make it gnome langpacks01:42
ArneGoetjecjwatson: ok01:43
ArneGoetjecjwatson: I will add an override to langpack-o-matic.01:44
cjwatsongreat, thank you01:45
cjwatsonwill there be another langpack upload before beta-1?01:45
slangasekI think it's important that we have buildable DVDs for beta-101:46
cjwatsonagreed, I was wondering whether new langpacks were planned anyway, or would need to be added01:46
ArneGoetjecjwatson: current situation is that in the -base langpacks (full-export), gbrainy.po is in the base package. In one of the delta updates it seems to have switched over to the gnome langpacks. So, what I can do now is to manually delete it from the gnome langpack updates and re-upload them.01:47
ArneGoetjecjwatson: next full-export is scheduled for Saturday night (UTC), so new langpacks will be available on MOnday01:48
ArneGoetjecjwatson: and with the next langpacks gbrainy.po should be in the gnome langpacks only.01:50
cjwatsonthat sounds workable01:50
cjwatsonas long as we get the delta fixes before Saturday, since we'll need to iterate some more on DVD builds and need as much lead time as we can get01:50
ArneGoetjecjwatson: yes, I will do it right away01:50
cjwatsonthey've been failing since January on and off, so we can't expect them to be particularly sound :(01:50
=== yofel_ is now known as yofel
cjwatson(not your fault, this is the first time I've seen this particular error)01:51
cjwatsongreat, thank you01:51
ArneGoetjecjwatson: :)01:51
maco2Any archive admins around? https://edge.launchpad.net/ubuntu/+source/spim/8.0-0ubuntu1 should be in universe, not multiverse01:56
maco2(license changed this release; it's now 3-clause BSD)01:56
ArneGoetjecjwatson: ok, so gbrainy will be without translation updates for this short period, but will get them back on Monday.02:04
maco2cjwatson: can i borrow you a moment?02:07
slangasekmaco2: is there a bug report open for this?  There's no interface that lets us comment on component changes, so a bug report is our best bet for figuring out why it moved if there are questions later02:11
maco2slangasek: i can go file one02:12
slangasekappreciated :)02:12
maco2slangasek: sure, now there's bug 53706302:14
ubottuLaunchpad bug 537063 in spim "spim 8.0 should be moved from multiverse to universe" [Undecided,New] https://launchpad.net/bugs/53706302:14
ArneGoetjecjwatson: gaa, doesn't work this way... need to run the complete delta update again, otherwise it will mess up the packages... should not take that long...02:15
slangasekmaco2: and promoted02:17
maco2slangasek: thanks02:18
=== cjohnston_ is now known as cjohnston
micahgArneGoetje: I wanted to ask you about bug 53441702:32
ubottuLaunchpad bug 534417 in thunderbird "Thunderbird: localization not installed automatically" [Undecided,Confirmed] https://launchpad.net/bugs/53441702:32
ArneGoetjemicahg: right, need to move that bug to software-center. That function is already implemented in language-selector and software-center just needs to call a function to get this. I will add a comment and move it over later, when I'm done with the language-packs02:35
micahgArneGoetje: great...I also wanted to ask you about the template for the firefox langpacks02:36
micahgdo  you want to switch from firefox-3.6 to firefox after beta 1?02:36
ArneGoetjemicahg: template has been fixed02:36
micahgArneGoetje: it's firefox now?02:36
ArneGoetjemicahg: oh, that... no, we can only switch when 3.0 and 3.5 are out of life and firefox has been upgraded to 3.6 for all releases.02:37
micahgArneGoetje: oh, I thought it was going to be sooner...ok, I'll talk to asac about the translation menu link in ubufox then02:38
ArneGoetjemicahg: ok, thanks02:38
leagrisHello, anyone interested in kernel BUG at /build/buildd/linux-2.6.31/fs/btrfs/inode.c:735 ?03:00
persialeagris: I suspect a number of folk are.  I'd recommend running `ubuntu-bug linux` to gather up a lot of the information about the bug and your system, and then report with as much additional detail as you can to the bug tracker.03:12
leagrispersia, here is a extract from dmesg: http://pastebin.ca/183294903:14
persialeagris: Thanks, but I'm not someone who can understand the problem from that.  Please do file a bug.  IRC is a poor place for bug reports, because lots of folk sleep.03:16
persia(well, other reasons too, but that's a big one)03:16
leagristhanks persia03:18
ccheneyfor some reason i can no longer ssh to my systems foo.local address, anyone know what could cause that?04:04
ccheneyhmm it works after resetting the ethernet connection04:05
emgent`aloha jono04:35
jonohey emgent`04:37
nonix4How can I use the *-dbg* packages with gdb?05:53
persianonix4: Install the -dbg package.  Start gdb.  If you have to do something special, file a bug.05:54
* nonix4 ponders (a bit offtopicly) which alternate browser to use since the buggy app in this case is firefox in infinite loop and gdb says "no debugging symbols found"05:59
micahgnonix4: what's wrong with firefox?06:05
nonix4It got to infinite loop switching mouse cursor between pointer and hand. Thoughts on how to analyze situation better? gdb backtrace doesn't tell much to untrained eyes with missing symbols06:06
micahgnonix4: what were you hovering over?06:08
nonix4Not sure anymore... one of {a login input, submit button, help href, another tab}. But regarding the -dbg packages I was apparently just missing some of them (guess there're no dependancies between them)06:12
nonix4boh great, X froze on attach to firefox...06:16
* nonix4b makes a mental note of debugging X apps requiring remote connection for the control06:17
micahgnonix4b: well, if you get enough info feel free to file a bug and I'll try to get to it at some point06:18
micahgpitti: could you please check on the lucid retracer (bug 533357 has been waiting 4 days) -- thanks06:22
ubottuError: Could not parse data returned by Launchpad: list index out of range (https://launchpad.net/bugs/533357)06:22
* nonix4b considers filing a bug on debug symbols of gdb itself missing06:43
pittiGood morning06:45
RAOFpitti: Good morning.06:45
nonix4'k "kill -CONT $ff; kill -9 $gdb" got X back alive, so may actually be able to get some information out of it :)06:56
pittiev: oh, nice! usb-creator has completed 140% :)07:17
emgentRiddell, ping.07:38
emgentmorning pitti07:38
emgentapachelogger, ping too07:38
dholbachgood morning07:47
amitkmorning dholbach07:49
emgenthey dholbach ! how goes?07:50
emgentlong time to saw you07:50
dholbachhey amitk, hi emgent07:50
dholbachhow are you guys doing?07:50
emgentreally busy with my job :-\07:51
emgentand anyway i have a not good news for us..07:51
emgentlifeless, ola, are you in involved in canonical webapps ?07:52
emgenti have some security hole to notify.07:53
lifelessso do so07:53
lifelesswe use launchpad, it has support for that07:53
emgentare not ubuntu stuff bug.. are canonical webapps issue07:54
emgentbut if you prefer.. i will notify in LP.07:54
emgentdholbach, when back query me07:55
dholbachI'm all here07:55
spaetzwould opening a kernel bug with patch to support the Marvel 8431 (which is in newer netbooks) be appropriate?07:56
persiaspaetz: You could, but I think they kernel team likes git-formatted patches to their mailing list (although I may be mistaken).07:58
persiaspaetz: You ought get guidance in a bug if you file it though.07:59
dholbachspaetz: https://wiki.ubuntu.com/KernelTeam/KernelPatches07:59
dholbachspaetz: having the patch in a bug is good, but the kernel team prefers discussion on their mailing list08:00
alkisgI'm trying to use pam_ck_connector to make local LTSP sessions CK-aware, but I don't understand how I'm supposed to export the CKCON_X11_DISPLAY_DEVICE environment variable (i.e. export doesn't work). It does work if I put it in /etc/environment, but I want it to have different values per session, so I don't want it in a global file. Would someone care to give me some hint?08:00
dholbachhttps://wiki.ubuntu.com/MeetingLogs/devweek1001/KernelPatches has some good discussion about it too08:00
spaetzpersia, dholbach thanks. I can point to the corresponding git commit for the opensuse kernel, hope that is fine. I will send a mail to the mailing list.08:01
lifelessemgent: we have products in lp for many/most of our webapps08:09
lifelessemgent: which ones are affected? you can privmsg if you want08:09
spaetzdholbach: filed bug 537168. Thanks for your pointers.08:15
ubottuLaunchpad bug 537168 in linux "[lucid] add support for Marvel 0x8431 (88E8059)" [Undecided,New] https://launchpad.net/bugs/53716808:15
dholbachno worries :)08:15
dugger5688Anyone got some tips on where to get started in bug fixing? I want to get into developing for ubuntu, but for obvious reasons don't want to dice into the deep end trying to start my own project.08:35
slangasekbigon: gupnp-igd FTBFS in lucid because of a problem with the test suite, which I assume was enabled as an MIR requirement; there's a fix upstream for getting the test suite to *build*, but enabling 'make check' in debian/rules still causes the package to FTBFS even with version 0.1.6-1 - do you have any ideas on this? (bug #534311)08:37
ubottuLaunchpad bug 534311 in gupnp-igd "gupnp-igd ftbfs on all archs in lucid" [Critical,Triaged] https://launchpad.net/bugs/53431108:37
dholbachdugger5688: try https://wiki.ubuntu.com/MOTU/GettingStarted08:38
slangasekbigon: it looks like it might need dbus running for the testsuite08:38
dugger5688Thanks, found my way there already :-) Will go through it in the morning.08:39
bigonslangasek: well there is the same bug in debian08:46
slangasekbigon: in Debian the testsuite isn't enabled - do you mean this is why it's not enabled?08:47
bigonwait I check08:47
sabdflKeybuk: purple consoles, +1 ftw08:48
bigonslangasek: it could be related to nm support (and the fact that dbus-glib is not working properly when initialised inside a thread, dixit gupnp-igd upstream08:49
bigonslangasek: I'm planning to make a gupnp-igd without nm support in debian to fix a related FTBFS08:50
directhexseb128, moonlight 2.2 uploaded. good luck to whomever needs to clear that through NEW.08:51
seb128directhex, thanks!08:51
directhex1.0.1-3ubuntu0.xul191build2 to 2.2-0ubuntu1 (36.1 MiB)08:52
slangasekbigon: ok - if you decide to fix it in Ubuntu as well, please note that I've merged up to 0.1.3-4 (testing) already on the lp:ubuntu/gupnp-igd branch, so if we continue with 0.1.3+fixes, would be good to preserve that08:52
cjwatsonmaco2: did slangasek cover what you wanted, or was there something different you wanted to ask me about?08:53
bigonslangasek: do you have logs of gupnp-igd 0.1.6 FBTFS?08:55
slangasekbigon: posted to the bug08:56
bigonslangasek: oh it's not a crash (I thought so) well then It could be http://bugzilla.openedhand.com/show_bug.cgi?id=1979 then09:01
ubottubugzilla.openedhand.com bug 1979 in gupnp "g_warning make gupnp-igd test fails" [Normal,New]09:01
slangasekbigon: ah, looks right, yes09:02
alkisgDoes /etc/security/pam_env.conf have access to the normal environment variables, or it can only access pam-environment? E.g. I'm trying to put MYVAR DEFAULT=${LANG} there, but it doesn't work for me...09:06
spaetzcjwatson: thanks for fixing ubiquity so fast. Glad I could help with my logs09:08
pittiapw: do you think there's a chance to fix bug 397734 for lucid? it still causes a lot of grief, and OEM team also has trouble09:09
ubottuLaunchpad bug 397734 in devicekit-disks "can't eject cdrom with hardware button" [Medium,Fix released] https://launchpad.net/bugs/39773409:09
pittiapw: (it's basically flipping the default value of /proc/sys/dev/cdrom/lock to 0)09:09
pittiapw: we could set it in /etc/sysctl.d somewhere, but first that's racy (doesn't get active if the cdrom module isn't loaded yet), and also means extra parsing during boot, etc.09:10
pittiapw: hm, I guess it could be a modprobe.d file, but it'd still feel cleaner to set the default right at the source09:11
apwpitti, i'll have a look09:12
pittiapw: thanks09:12
seb128siretart`, hello09:15
siretart`hi seb128! how are you?09:15
seb128siretart`, good, you?09:15
seb128siretart`, I'm starting that gst-ffmpeg against ffmpeg discussion with pitti again09:15
pittihey siretart`, wie gehts?09:15
siretart`thanks, fine.09:15
siretart`danke! :-)09:15
seb128since he was not around the other day09:15
seb128siretart`, can you summarize the issue between both?09:15
seb128does gst-ffmpeg need a newer or older ffmpeg to work correctly?09:16
siretart`ah, another round in the gst-ffmpeg vs ffmpeg deathmatch :-) - okay, let the fun begin09:16
seb128is downgrading or upgrading ffmpeg to that version an option?09:16
pittisiretart`: background is, ffmpeg is pretty prone to security flaws (and bugs, too, I suppose), so if we can avoid a copy, we should09:17
siretart`okay, short recap: ffmpeg 0.5 is now pretty exactly 12 months old, and has since then gained quite a lot of additional codecs09:17
siretart`gst-ffmpeg upstream has decided that these additional codecs are important for them and switched from tracking the 0.5 release branch (which we use as system ffmpeg) to ffmpeg trunk09:18
pittiseb128: 0.5+svn was in karmic as well, so I guess that had the same problem?09:18
siretart`naturally, some details have changed, and gst-ffmpeg has strongly advised me against compiling it against 0.509:18
seb128pitti, ok, seems I got my version wrong, better to read what siretart` is writting09:18
siretart`slomo however has identified which parts don't work against 0.5, and patched them out09:18
pittisiretart`: ah, so the issue is that gst-ffmpeg has a _newer_ copy than our system ffmpeg?09:18
siretart`pitti: exactly. gst-ffmpeg will track the 0.6 branch as soon as it opens (maybe we'll create that branch this weekend, maybe next weekend, we'll see)09:19
siretart`pitti: regarding security issues, does the security team really care about the security patches in the ffmpeg package? I did write a mail to security, but they were rejected because "no testcase"09:20
siretart`in the mean time the patches I did for the debian package got integrated in 0.5.109:20
pittihm, I had assumed they did; kees, jdstrand, mdeslaur ^ ?09:20
siretart`and are now in lucid. but for karmic the patches are still missing09:20
seb128siretart`, should we understand from that using the system 0.5 ffmpeg will mean cutting or features and codecs which would work using the copy then?09:20
pittisiretart`: ah, so is there actually a problem in lucid then?09:20
keesI really care about security patches.  :P09:21
keesI'm not sure I understand the question, though.09:21
siretart`pitti: 0.5.1-1ubuntu is in lucid and has these patches. all sec related patches I'm aware of are included there09:21
seb128kees, the question is "do you care if gst-ffmpeg uses a ffmpeg copy" ;-)09:21
pittisiretart`: I meant, you said that slomo patched out the bits of gst-ffmpeg which don't work with our 0.5.1 ffmpeg lib09:22
siretart`seb128: yes, trunk always has (almost per definition) more codecs then a release branch09:22
pittisiretart`: so it seems that in lucid it's actually fine?09:22
kees`I did write a mail to security, but they were rejected because "no testcase"` ?  surely not security@ubuntu.com ?09:22
siretart`pitti: lucid should be fine09:22
keesseb128: yes, I care -- it absolutely should not use an embedded copy.09:22
pittisiretart`: *phew* great to hear09:22
siretart`kees: I'll fetch the message id09:22
seb128siretart`, if lucid should be fine I'm getting confused about your ping from the other day09:22
siretart`seb128: security wise. gst-ffmpeg in lucid is compiled against 0.5, which upstream strongly opposes09:23
seb128why, if it works fine?09:23
seb128I had the impression you told me it was open door for crasher issues09:23
seb128but now you say it's fine09:23
siretart`pitti: kees: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=550442#35 this is the last message I got from the ubuntu security team, I think09:24
ubottuDebian bug 550442 in ffmpeg "ffmpeg: deluge of crashes due to missing input sanitization" [Serious,Fixed]09:24
siretart`I understood the last sentence as "we need an proof-of-concept exploit" for each crash09:24
siretart`which I really don't have the time for09:25
pittithat would be weird09:25
pittimost security issues we fix don't have one, AFAIK?09:25
keessiretart`: he meant "someone needs to analyze these crashes to see if they're really security issues."09:25
siretart`seb128: yes, and that is right. we are having two seperate discussions here09:25
persiaI'd hope that we weren't writing test cases for every security hole.  That code would be dangerous.09:26
keespersia: we do try -- how else to better validate a vulnerability has been fixed?09:26
siretart`kees: I've discussed each single patch with other ffmpeg developers, and I think only one or two were classified as 'not security relevant'09:26
persiaOh my!09:26
lifelesspersia: kees has bee doing this for years :)09:27
keespersia: have you not see lp:~ubuntu-bugcontrol/qa-regression-testing/master  ?09:27
persiakees: I've not looked in detail, no.09:27
* persia wonders how the publication of that branch intersects with UEC images09:28
siretart`seb128: I'm saying that *ffmpeg* itself should be fine security wise. I didn't say that about gst-ffmpeg!09:28
seb128siretart`, I've to admit I'm confused about the discussion now09:28
seb128siretart`, I though the discussion was about having something which is stable against something not tested and crashy09:29
keessiretart`: if there are open security issues in ffmpeg in releases of ubuntu, we need to track them.  mdeslaur seemed to be identifying where various things were fixed already?  this list needs to be clarified, etc.09:29
seb128siretart`, and now it turns to be a discussion about ffmpeg security which I've no real clue about or interest in09:29
siretart`kees: pitti: anyway, in the mean time a DSA has been released about these issues, so the patches could be taken straigt from there. however, looking at the latest commits svn://ffmpeg.org/branches/0.5 might be helpful as well09:29
seb128siretart`, so I will let you guys sort that one09:29
siretart`seb128: okay, let's discuss open security issues first and then return to gst-ffmpeg09:30
seb128siretart`, thanks09:30
siretart`kees: I haven't heared anything back from mdeslaur since then, no idea if he continued to track that issue09:31
keessiretart`: okay, I'd like to get it clarified.  specific what patches are missing for which ubuntu releases of ffmpeg.09:32
siretart`kees: for intrepid, I think taking the patches from the debian DSA would probably be easiest, although I didn't check if I haven't missed to backport some patched back then. I think there might be even some newer patches to check09:34
keessiretart`: can you perhaps open a new LP bug (tracking this in a debian bug probably isn't the clearest way to handle this)?09:35
siretart`kees: for jaunty and karmic (they are effectivly both 0.5), replaying the patches from the 0.5 release branch should be easiest09:35
siretart`okay, I can do that09:35
siretart`shall we return to the issue of gst-ffmpeg then?09:36
keessiretart`: cool thanks.  when mdeslaur wakes up, he can poke at it more.  I'm overdue for sleep.  :) g'night!09:37
evpitti: :)09:37
pittisiretart`: so now I'm confused; I thought it would work well in lucid now, with slomo's patches?09:37
siretart`kees: good night! :-)09:37
pittikees: sleep well!09:37
siretart`pitti: well, it is the 'well' that upstream objects. They highly doubt that it works 'well'09:37
siretart`it builds, but nobody really tested that combination09:38
siretart`and I have to admit that I'm not going to look to deeply into crashers that come via that way09:38
pittisiretart`: if you feel too unsure about it, perhaps we could downgrade to an earlier gst-ffmpeg from a time when 0.5.1 was current?09:38
siretart`pitti: yes, sticking to the version of gst-ffmpeg from karmic seems to me a safe solution09:39
pittisiretart`: so we should go back from 0.10.10 to 0.10.9?09:39
seb128or use the gst-ffmpeg copy?09:40
seb128I don't fancy using year old code09:40
seb128especially for multimedia, it often means lack of support for codecs, etc09:40
pittiwell, we should either upgrade ffmpeg to trunk or downgrade gst-ffmpeg09:41
siretart`pitti: that would make us stick to a codebase that we know, but it also would make seb128 and many users sad because we're missing lots of codecs09:41
pittihaving two copies of ffmpeg really doesn't make sense, either from security or maintenance POV09:41
pittiif we keep a newer ffmpeg in gst-ffmpeg, then we'd still have those codec/bugs problems in the system library09:41
seb128pitti, it would make sense in the sense it would provide a decent multimedia experience to our user09:42
pittiseb128: how wouldn't that be provided by updating the system library?09:42
siretart`pitti: as for long term plans, I plan to work on the 0.6 release soon, and bilboed (gst-ffmpeg upstream) has indicated that they will start tracking that branch as soon as it opens09:42
apwpitti, this CDROM lock thing ... i just played with mine and i am unsure as to what the real issue here is... as far as i can tell the button works when you are allowed to eject the CD and not when you are not (its in use) ... are we saying we want to allow eject even on mounted FSs?09:42
seb128pitti, you are trading distro work against user experience, ie arguing that we care about having less work and not about providing a good user experience there09:42
pittiseb128: but if it's not ready for release yet, then that's just bad luck and timing09:42
siretart`pitti: I didn't push 0.6 for lucid because I cannot commit to transition all package in ubuntu from 0.5 to 0.6. I have really no idea how much breakage that will cause09:43
siretart`for me, the plan was to go with 0.5 for both lucid and squeeze09:43
seb128pitti, by experience upgrading system ffmpeg is not trivial09:43
pittiwe are saying that ffmpeg 0.5 is stable, and 0.6 is not there yet09:43
apwpitti, also having just tried it, when you rip out the disk the fs remains mounted from there on ... and nautilus (i think) triggers 5 errors to syslog every few seconds until its reinserted or unmounted09:43
seb128pitti, right09:43
siretart`bilboed was fully aware of that plan but decided to go for trunk anyways09:43
pittiapw: that's a separate bug indeed09:43
siretart`and I can udnerstand him in some ways. there are really very interesting new features in 0.609:44
pittiseb128: so either 0.6 is ready for release, then we can update to it, or it's not, then I don't see a major reason why we shouldn't stay with 0.5?09:44
apwpitti, i also can't see what would stop eject working in the middle of a disk burn ...09:44
siretart`but since we are way past feature freeze, I did considered upgrading as an option09:44
pittiseb128: do you really think 0.5 was so bad? video codecs have worked great in linux for at least a decade, after all09:44
siretart`upgrading gst-ffmpeg to use ffmpeg 0.6 would require a FFe anyway AFAIUI..09:45
pittiseb128: do we have lots of bug reports which would be fixed with 0.6?09:45
siretart`pitti: well, we could also see it that way:09:45
seb128pitti, I don't know enough about ffmpeg to say09:45
siretart`pitti: for lucid+1, I will very soon upload a ffmpeg 0.6 package as system ffmpeg09:46
seb128pitti, I just know that a year is lot in multimedia things09:46
pittiseb128: but you just argued that we need 0.6 for good user experience?09:46
seb128pitti, and sirestart said slomo had to turn codecs and things off since they didn't work with ffmpeg 0.509:46
siretart`pitti: so as soon as ffmpeg 0.6 is releasd, the security team would have to care for 0.6 as well anyways..09:46
seb128pitti, I'm just saying that based on siretart`s comment09:46
seb128"users sad because we're missing lots of codecs"09:47
pittisiretart`: that's the natural flow of things, yes; but they'd have to care about it much longer if it's in lucid, too;09:47
pittisiretart`: and apart from the seucrity side, if 0.6 isn't stable yet, it sounds like we would introduce more problems than we solve09:47
pittiI mean seriously, how many video formats cause problems on recent linux? My feeling is "hardly any"09:48
pittibut we do have a lot of _bugs_ with playing back videos indeed09:48
siretart`pitti: seb128: http://paste.ubuntu.com/393150/ is the change log in ffmpeg trunk so far since the 0.5 release09:48
pitti(thinks like searching in totem, etc.)09:48
keespitti: AVC HD is miserable currently.  /me really goes to bed now09:48
siretart`I even had been presented a backport of wmapro for 0.5...09:49
pittikees: in totem?09:49
keespitti: right.  and not even mplayer can play it right.  the only thing that seems to get it right is mythtv's internal player.09:49
siretart`pitti: AVC HD playback is pretty bad shape in all ubuntu packages right now09:49
pittiseb128: so, I guess we can argue either way (for 0.5 or 0.6), but I really object to having _both_ as a rational answer09:49
keesthere will always be new codecs.09:50
* kees votes for 0.6 just because it'll be easier to get fixes for in 3 years. :)09:50
* kees really really goes to bed09:50
=== yofel_ is now known as yofel
pittiwell, that's fair enough09:50
seb128pitti, I've not spent enough time looking at bugs and I don't know enough about gst-ffmpeg and ffmpeg to have an argumented answer there09:50
pittibut that wouldn't help you at all if gst used 0.6 and everything else 0.509:51
siretart`pitti: there are no 'stable' releases. ffmpeg development is way to fast to even think about that. It took years to convince upstream to do at least snapshot releases. ATM, its mainly me who tries to identify security related changes in trunk and backports them to release branches09:51
pittisiretart`: so, what's your recommendation right now?09:51
seb128pitti, I'm mainly trying to avoid us realizing in one month that we are flooded by totem gst-ffmpeg crash bugs because we use a non support ffmpeg and gst combinaison09:51
ttxasac: about bug 526480 -- you think you'll have time for it ?09:51
ubottuLaunchpad bug 526480 in liblog-log4perl-perl "[MIR] liblog-log4perl-perl" [High,New] https://launchpad.net/bugs/52648009:51
ttxotherwise may be best to assign it to someone else in MIR team09:51
seb128pitti, and being stucked in a sucking position then for the lts because it's too late for changes09:51
pittiseb128: right, and I'm trying to avoid that flood by suddenly switching to a new upstream ffmpeg in gst which nobody tested yet..09:52
seb128well it's an internal copy09:52
seb128it has been tested by upstream09:52
seb128and switch the configure flag back is easy09:52
siretart`pitti: with following three options: A) compiling new gst-ffmpeg against 0.5 B) reverting to old gst-ffmpeg and C) new gst-ffmpeg with internal copy09:52
siretart`pitti: I'd probably vote C>B>A09:53
seb128same here09:53
pittisiretart`: I'm interested why you favor C09:53
seb128since the first one is unsupported and crashing09:53
pittiit seems like the least sensible option to me09:53
siretart`pitti: it is the only option blessed by upstream09:53
seb128it's the one providing the best user experience09:53
pittisiretart`: like, "get twice the amount of bugs for double the price"?09:53
seb128at a cost for us to fix issues in that copy09:53
seb128pitti, well, we have to choice is we care about having less work or about having a better user experience09:54
siretart`let me ask in a different way09:54
seb128since the less work solution will bring trouble and crashes09:54
siretart`who cares for gst-ffmpeg bugs these days?09:54
pittiso what about D) update system ffmpeg?09:54
siretart`in terms of crashers?09:54
seb128care in which sense?09:54
seb128nobody work on those09:54
seb128but I would prefer not having crashers there09:55
seb128I do care about what we ship to our users09:55
pittiseb128: you are sure that updating ffmpeg won't bring new problems?09:55
seb128pitti, we would not update, we would use the gst-ffmpeg ffmpeg copy with is what upstream work on and test09:55
seb128and which is the version which probably match best what gst-ffmpeg handle09:55
pittiseb128: sure we would09:55
siretart`pitti: D) would be an option as soon as we have an 0.6 release branch, or even better, the 0.6 release. I don't think that will happen in time for lucid09:56
pittiwhether it's linked or internal, it's still a new version09:56
persiaseb128: But why not take the ffmpeg gst-ffmpeg recommends and use that for system ffmpeg?09:56
seb128D) would also mean a transition09:56
seb128how much other things would that break?09:56
siretart`pitti: and I cannot commit for doing that transition in a couple of weeks. the last ffmpeg upgrade was pretty painful09:56
seb128persia, because we are not sure how it would impact on everything else using ffmpe09:56
seb128persia, ie xine, vlc, ...09:56
seb128persia, what sirestart says09:57
persiaAnd given schedule, urk.09:57
pittisiretart`: yes, I wasn't going to ask you to actually; (I favor neither C nor D, FWIW, but it seemed prudent to at least complete the option table)09:57
siretart`I can say for sure that our mplayer won't remotely compile against 0.609:57
siretart`but mplayer is special09:57
seb128pitti, well new version right, but with limited scope, it just needs to be tested on gst-ffmpeg, we don't risk stability issue for ie xine09:57
seb128or mplayer or vlc or etc09:58
pittiso, I don't think I have any new arguments at this point; seems we just need to agree to disagree then :)09:58
siretart`I think the safest option would be to revert to lucid's gst-ffmpeg, and promote a PPA with a newer gst-ffmpeg09:59
pittisiretart`: s/lucid/karmic/ ?09:59
siretart`err, of course09:59
siretart`I think the safest option would be to revert to karmic's gst-ffmpeg, and promote a PPA with a newer gst-ffmpeg09:59
siretart`and upload a newer gst-ffmpeg to lucid-backports that uses the internal copy of ffmpeg10:00
seb128how much drawback is that compared in format supports, bug fixed, etc?10:00
pitti(we should probably go to 0.10.9-3, not -1, but details..)10:00
seb128siretart`, do you know what slomo plans to do for debian?10:00
siretart`seb128: no idea, I havn't seen him on irc for quite some time now10:01
siretart`I think he is going with newer gst-ffmpeg compiled against 0.510:01
siretart`read: option A10:01
seb128so it seems pitti and security team stand strongly against C10:01
seb128so we have choice between downgrade or stay with that unsupposed binaison10:02
seb128how can we get feedback on well or not gst-ffmpeg + ffmpeg 0.5 is working or crashing or...?10:02
seb128on *how* well10:02
pittimy personal gut feeling is A>B>D>C, but since it seems that D) is not practical, it reduces to A>B>C10:03
siretart`I'm told that gst-ffmpeg comes with an extensive testsuite10:03
pittiseb128: hm, we could look at gstreamer/totem/gst-ffmpeg bugs since lucid?10:03
RAOFI guess we could run totem over a test-suite of video files; I know they exist.10:03
siretart`perhaps running that would give us more insight?10:03
pittisiretart`: sounds like a good idea10:03
pittieveryone watch pr0n for a week and report the results!10:04
seb128ok, so let's stay on A and try to do some testing with it?10:04
seb128maybe talk to the qa team about organizing some testing on that?10:04
seb128just to know where we stand10:04
siretart`seb128: perhaps you could try talking to upstream about that issue?10:05
siretart`I mean with your ubuntu gnome guy hat on?10:05
pittilooks like http://live.gnome.org/GStreamerMediaTest10:05
seb128what issue? how to test?10:05
siretart`about option A) anyway despite their explicit disapproval10:05
siretart`this choice is not going to please them10:05
seb128siretart`, who did you talk to upstream about that before?10:06
seb128I will get in touch with them10:06
siretart`seb128: to bilboed10:06
seb128ok thanks10:06
siretart`seb128: he approached me in #ffmpeg-devel10:06
siretart`not sure if there is an gst-ffmpeg specific dev channel10:07
seb128siretart`, #gstreamer will do10:07
siretart`I'll lurk10:07
Keybukwell, I can replicate the press-ENTER-to-kill-X bug10:07
pittiKeybuk: ah, when you see the mouse cursor on a VT?10:08
siretart`anyway, need to return to work now, will file the sec bug later10:08
pittisiretart`: thanks a lot for the heads-up!10:08
Keybukpitti: no, fixed those bugs - now you see full X with gnome panels10:08
ografunny, that bug vanished for me exactly today10:08
pittiogra: oh, how? nothing changed in that regard, except perhaps the moon phase to influence the chances?10:09
sladenI got the mouse-on text vt... it even changes cursor when you hover over the GDM login box10:09
sladen(not that you can see it)10:09
ograpitti, well, then its the moon phase :) for the first time i could hit enter in the keyring dialog without ending up at gdm10:09
pittiogra: I suppose it's just because your CPU now starts slightly warmer with the winter end, so the booting is 0.01 s faster10:10
ograhmm, its actually slower10:10
ograbut i think thats due to ureadahead reprofiling, i didnt reboot twice10:11
apwKeybuk, as some additioanl data i think that the die on return is occurs every time if you use nomodeset ... seemed to on my mini1010:12
apwpitti, ok this new behaviour is already being complained about ... that you can eject and it stays mounted ... which it would10:12
apwpitti, presumably due to the devkit-disks cahnge which makes it work for some types of mount10:13
apwbug #47665410:13
ubottuLaunchpad bug 476654 in udisks "CD eject but not unmount when using drive button" [Medium,Triaged] https://launchpad.net/bugs/47665410:13
pittiapw: right, it's that bug10:13
pittiapw: it's not clear yet whether that's a kernel bug or feature10:13
pittiapw: the kernel already cleans up stale mounts for e. g. USB sticks, if you yank them out10:14
pittiapw: if it's a feature that this isn't done for CD-ROMs, then we need to add userspace code to listen to remove events and call umount -l10:14
apwpitti, its not at all clear to me letting you take it out is good at all10:14
pittiapw: I talked to cooloney-Beijing, and he was going to look into it10:14
ograpitti, but as you explained to me yesterday the kernel also powers down the device10:14
ogra(in the usb case)10:14
apwpitti, seems we are chaniging good behaviour just because people don't know what their eject button means10:15
pittiapw: there's no other way for e. g. DVDs10:15
pittiapw: udisks unlocks the door after mounting10:15
pittiapw: but if totem open()s it again, it again triggers the lock, and there's nothing we can do to unlock it again10:15
apwpitti, yep and i personally think thats a mistake10:15
apwpitti, its locked cause its in use10:15
pittiwell, it's one interpretation; but it basically renders the eject button useless10:16
pittiand it seems people are used to it10:16
pittiand taking out a readonly medium like a DVD is pretty harmless, at least compared to yanking out USB sticks10:16
apwwell in reality people don't think generally thats why we have to handle them just pulling out USB disks when we know its something you should never do10:16
* apw grumbles ... yes but what about non-read only media10:17
apwlike in the middle of a burn10:17
ograwouldnt that lock ?10:17
pittiapw: just passing the message here.. OEM team currently adds a sysctl.d file to set the flag..10:17
apwogra, not with the requested change to the kernel to turn off the lock10:17
pittiogra: we can probably make it to lock (not sure if it does right now)10:17
ograwe should10:18
cooloneyapw: we found all the eject button do not work10:18
pittinot locking by default doesn't mean that we can't lock manually, surely?10:18
ograpulling out in the middle of a burn seems bad10:18
cooloneyso we have to add sysctl.d trick to make it works10:18
apwcooloney, they don't work till you unmount the media in software right?10:18
pittiogra: well, YAFIYGI.. :)10:18
cooloneyapw: exactly10:18
pittiapw: (in which case you'd just use the eject button in nautilus to get it ejected)10:18
cooloneyif the data disc is mounted, eject button does not work10:19
apwand now you tell the kernel to allow you do something stupid, and complain when it does something stupid?10:19
pittiogra: but yes, we should lock during burn10:19
ograpitti, i didnt ask :)10:19
apwcooloney, that is as designed though10:19
pittiapw: why is taking out a mounted CD-ROM stupid?10:19
apwits stupid if you didn't unmount it cause you didn't tell me it wasn't needed (me == kernel)10:19
cooloneyapw: actually, 99% normal users will press that button10:19
apwyep, and does it work in windows?10:20
cooloneyinstead of umounting it in the GUI10:20
pittiapw: yes, it does10:20
cooloneyapw: right10:20
pittiwindows doesn't lock the tray door10:20
apwbah ...10:20
cooloneywindows does not do that10:20
ograapw, but but .... there is a button on the device and i pressed it ... must be broken if the Cd doesnt come out10:20
* apw cuts off one of ogra's fingers ... and now?10:20
* ogra takes a toe 10:20
cooloneyogra: exactly, user will think that is a hardware problem10:20
cooloneybutton does not work10:21
pittiapw: so, I agree that both bugs (locking and unmounting of stale mounts) should be fixed together; if the latter requires an userspace monitor, I'm happy to look into that10:21
apwhrm you won't convince me that ill informed users are right10:21
apwbut i seem to be in the minority here10:21
ogradoesnt have to do with being informed10:21
ograjust with natural expectations10:22
persiaI'm confused.  Will pressing the button just eject, or tell the kernel to force-umount and then eject?10:22
apwchanging the default in the kernel is 'easy' but there will be two instant issues: 1) burn will be vunerable, 2) the errors will puke till it gets unmounted10:22
pittipersia: it will just eject, and kernel/userspace have to tidy up after it10:22
smbWell sort of yes. I would expect the button not to work when the cdrom is mounted because its in use10:22
apwpersia, the button doesn't go to the kernel10:22
apwthe kernel doesn't find out till the tray has opened and the media is gone10:22
ograsmb, whats mounting ? can you explain to my mother ?10:22
cooloneyapw: actually burning issue is the same issue as you yanking a usb stick when you are writting something into the usb stick10:23
ograshe popped that CD in and played some music ... now she pressed the button ...10:23
apwogra, no i can't but then i can explain 'click here' ... and thats what i do10:23
smbogra, Thats about ill informed users. Linux != Windows != Mac10:23
Keybukreading documentation on Linux terminal devices, while working for this company, is confusing10:23
ograsmb, i would agree if there were only slot-in CD roms10:23
ograas soon as you have a button the button should just work10:24
apwbut ogra the button does just work, the error is that you think its an eject button10:24
ograKeybuk, stop reading ! we dont want you to resign to understand the prob !10:24
apwwhen its actually an eject the media if its not being used, button10:24
apwalways was10:24
ograapw, there is a eject logo on it10:24
apwyou think thast what the logo is, thats not my fault10:25
ogra(bar with a little triangle above)10:25
cooloneybut the problem is the data disc is auto mounted defautly10:25
apwyep, but thats a design issue, i can't help they have used the wrong logo10:25
cooloneyif you insert the data disc, the eject button won't work10:25
ogracooloney, right and the button should properly unmount it10:25
apwand by design the button _cannot_ do that10:25
ograand then eject the CD10:25
apwas it does not have a channel to tell the kernel it has been pressed10:26
ograapw, why ? userspace surely gets a button event10:26
apwwhy do you assume that, cause it has an eject logo?10:26
ograyes :)10:26
apwthe button is a physical eject, it tell the drive to open if its not locked10:26
cooloneyapw: yeah, since I use my cd player, we think that button means open the tray10:27
ograok, thats a hardware flaw10:27
apwthe drive if unlocked will open and send a 'i just opened and your media went away' event10:27
ograstill ... user expectations are differently and we should apply10:27
apwits far too late by then to cope10:27
* apw goes back into his troll cave, and erects his anti-user force field10:27
pittiwhy? you unmount, and userspace copes just fine10:28
ograwhy is it to late ? you can still unmount then10:28
smbIn theory maybe you could. But question would be is it done10:28
pittiudisks and friends get the removal events, nautilus switches away from it, etc.10:28
apwyou can't unmount a r/w device then ... if yo uhave any data to write10:28
ograits not like you write to a readonly media10:28
pittithe fun thing is that the kernel unmounts the r/w usb stick just fine10:28
smbogra, And my cd burner?10:28
pittibut doesn't for a read-only CD-ROM mount10:28
ograsmb, your CD burner software should acre for locking10:28
ograand remove the lock when its done10:29
apwpitti, 'just fine' is a little unfair10:29
smbogra, Agreed, do we know it does :)10:29
apwogra, it doesn't need to ... cause its locked on use ... currently10:29
ograsmb, if it doesnt its a bug :)10:29
apwanyhow ... i'll go change the default, you can handle the fallout :)10:29
ograits not like there are millions of linux CD write apps10:29
apwpitti, i think ogra just volunteered to fix all apps which _write_ to cd media10:30
ograits a handfull of patches in max if burning SW really misses here10:30
pittisweet ;)10:30
ograwhat ?10:30
* ogra glares at apw 10:30
apwogra, note that that includes things which mount r/w disks for instance10:30
* apw heard you10:30
pittiogra: I can look at brasero10:30
sorenI could have sworn that it used to work that way.10:30
ograpitti, well, arent there only two low level tools atm ? we should fix the backend10:31
pittisoren: in the hal era, hal polled the CD-ROM for button presses and did a lazy unmount/eject10:31
sorenpitti: Back in the hal days, didn't we somehow detect that the user had pressed the eject button=?10:31
sorenRight, exactly.10:31
sorenThere's and SG_IO ioctl thingamabob to do that.10:31
cooloneythat is funny, because the button works very well in jaunty10:31
pittiogra: yes, probably10:31
cooloneybut we failed since karmic10:32
pitticooloney: that was still the hal polling10:32
cooloneypitti: i understand10:32
KeybukI fixed the enter-kills-X bug10:32
pittibut it causes an awful amount of wakeups, etc.10:32
* pitti hugs Keybuk10:32
* seb128 hugs Keybuk10:32
sorenKeybuk: Oh, no!10:32
apwKeybuk, sounds good ... what the heck was it10:32
ogragrouphug !10:32
Keybukapw: plymouth was setting the VT back into Canonical Mode10:32
sorenKeybuk: I rely on that bug to get to gdm occasionally. :)10:32
pittidoes that prove that Canonical is bad then?10:32
apwoh no time for a name change10:33
pittiapw: is that "oh, no time" or "oh no, time"?  :)10:33
Keybuksince Raw Mode is better, let's call ourselves RAW Ltd10:33
apwoh no,10:33
cooloneyapw: actually, i wanna talk about the second cdrom issue10:33
sorenEvery now an again when I boot, I will have a console login prompt and a mouse cursor(!). Typing yields nonsense on the console. Pressing return kills X, it restarts, and yay, I have gdm.10:34
cooloneyapw: https://bugs.launchpad.net/ubuntu/+source/udisks/+bug/47665410:34
ubottuLaunchpad bug 476654 in udisks "CD eject but not unmount when using drive button" [Medium,Triaged]10:34
Keybuksoren: I fixed that one last week ;)10:34
apwi am assuming someone else will look after that one cooloney10:34
pittisoren: that's bug 523788 which Keybuk supposedly fixed in his local branch10:34
ubottuLaunchpad bug 523788 in plymouth "Only see X mouse cursor on VT during boot" [High,Confirmed] https://launchpad.net/bugs/52378810:34
Keybuk"supposedly" ... what, you don't trust me? :p10:34
sorenKeybuk: I'm /almost/ sure I've seen it this week. I'll keep an eye out to make sure.10:34
pittisoren: yes, it's in today's lucid10:34
pittiKeybuk: upload it! upload it!10:35
Keybuksoren: I haven't uploaded fixed plymouth yet10:35
sorenKeybuk: Fixed and uploaded last week?10:35
sorenAh, ok.10:35
sorenthat explains :)10:35
Keybukon the basis that if I don't fix all of the known bugs, I'll get lynched10:35
seb128Keybuk doesn't like to share fixes apparently ;-)10:35
smbKeybuk, Just doit :)10:35
cooloneyapw: i am working on this. heh10:35
apwcooloney, ok what you thinking10:35
Keybukplus am testing to make sure I don't introduce new ones10:35
Keybuklike right now, the text renderer is broken again10:35
pittiKeybuk: does it render Klingon now?10:36
cooloneyapw: i failed to find where does kernel umount or cleanup the removed usb stick10:36
Keybukpitti: you could write a klingon plugin if you like10:36
cooloneywhen you yanking a usb stick from a machine10:36
apwcooloney, are we sure the kernel even does that?10:36
cooloneyit should be similar to press button eject t a file system10:36
Keybukcooloney: the kernel doesn't10:36
apwKeybuk, who _does_ do that10:37
cooloneyapw: i am not sure10:37
Keybukapw: userspace10:37
cooloneyi guess is user app10:37
smbKeybuk, udev?10:37
cooloneybut pitti told me that kernel does that10:37
Keybukkernel gives us a "device removed" uevent10:37
cooloneyright Keybuk10:37
Keybukudev tells udisks tells session, tells udisks to unmount10:37
pittihm, didn't it use to?10:37
apwKeybuk, do we know who in userspace, and why cdroms don't ?10:37
Keybukpitti: don't think so - it'd be pretty strange for the kernel to just vanish half the filesystem without userspace doing it10:37
* pitti tries10:37
cooloneyright, why cdrom fails10:37
pittiKeybuk: ok, then we need to teach udisks about that10:38
apwcooloney, cause the 'drive' isn't gone like is in usb i assume10:38
cooloneyi monitors the udisks in lucid10:38
Keybukbut I don't know for ausre10:38
Keybukapw: right, that's what I suspect10:38
cooloneyand devicekit-disk in karmic10:38
Keybukapw: which would fit what I believe to be true10:38
cooloneyit will get cd remove event as usb stick10:38
Keybukusb key removed => remove event for device => userspace cleans up10:38
Keybukcdrom ejected => device is still there => userspace unaware10:39
KeybukI thought we locked the CD tray ? :)10:39
smbapw, So the "media gone away" probably not produce a uevent?10:39
davmor2Keybuk: I have the enter to get gdm issue on une this morning and last night10:39
apwKeybuk, i wish :)10:39
Keybukin fact, I'm pretty sure this is *why* we locked the CD tray :p10:39
Keybukdavmor2: I had it 10 minutes ago, before I fixed it :p10:39
cooloneyok, remove usb key10:39
pittiKeybuk, apw: so, if I yank out a mounted usb key, it gets unmounted10:39
davmor2Keybuk: so the fix should be in for tomorrow then?10:40
pittibut I guess it's not the kernel then10:40
ogradavmor2, you need the secret sources.list entry to pull directly from Keybuks laptop and then dist-upgrade10:40
cooloneygot 2 uevent, one is sdb, the other is sdb110:40
Keybukdavmor2: should be10:40
Keybukogra: sources.list?!10:40
ogradavmor2, or wait :)10:40
pittiKeybuk, apw: ah, if I kill udisks-daemon, it doesn't, so I guess we need to fix udisks to lazy-unmount CD-ROMs then10:40
cooloneyremove cd disc, a one remove event10:40
Keybukyou need to cd into co/plymouth, then run "make install", then move bits around a bit to match :p10:40
apwpitti, how would it know to do it10:40
ogradetails :)10:40
davmor2ogra: no I just need to travel over to keybuks hand him all my lucid boxes and say fix these ;)10:40
cooloneypitti: agree10:40
Keybukpitti: as apw said, I suspect the problem here is that we don't know the CD-ROM has been ejected :p10:40
apwpitti, without polling ... when you might as well do what hal did10:41
cooloneyso is there any event from cdrom.c driver for ejecting?10:41
pittiwe do get an uevent with CD-rom eject10:41
Keybukcooloney: no10:41
Keybukpitti: no!10:41
cooloneyKeybuk: too bad10:41
pittiUDEV  [1268304083.700451] change   /devices/pci0000:00/0000:00:1d.7/usb1/1-8/1-8.4/1-8.4:1.0/host2/target2:0:0/2:0:0:0 (scsi)10:41
Keybukpitti: the drive is *still there* :)10:41
Keybukright, there might be a change event10:41
pittiKeybuk: not a remove, a change10:41
Keybukthat would be worth investigating10:41
pittiwhich is just fine10:41
pittiI just tried it here10:41
apwpitti, so you could use that ... excellent10:42
Keybukchange => check for media => media gone => unmount10:42
KeybukI was building up to that10:42
apwok so that sounds like a plan ...10:42
cooloneyhow about add a new uevent in cdrom.c driver to do that10:42
pittiapw, Keybuk, cooloney: ok, I'll look into fixing the cdrom unmount in udisks10:42
apwpitti, thanks10:42
pitticooloney: uevents are all fine10:42
apwcooloney, sounds like we have what we need ...10:42
mattiHi apw10:42
* cooloney hugs pitti 10:42
cooloneyapw: thanks,10:42
cooloneyKeybuk: thanks10:42
pittibug 476654 targetted to lucid adn assigned to me10:43
ubottuLaunchpad bug 476654 in udisks "CD eject but not unmount when using drive button" [Medium,Triaged] https://launchpad.net/bugs/47665410:43
cooloneyi will let those guys know this issue10:43
Keybukslangasek: I guess you're not up?10:44
apwpitti, do we need to track the 'fix burners' side of this issue?10:44
apwget a nice bug assigned to ogra10:45
pittiapw: I'll add a task to bug 397734 for checking this10:45
ubottuLaunchpad bug 397734 in devicekit-disks "can't eject cdrom with hardware button" [Medium,Fix released] https://launchpad.net/bugs/39773410:45
cooloneyapw: you scared ogra away, heh10:47
ograah, back to normal ... http://people.canonical.com/~ogra/osiris-lucid-20100311-3.png10:50
apwpitti, where will i fine the usdisks changes you made which unlock the drive?  i need to disable them to test this change10:57
pittiapw: sudo python, open("/dev/sr0") should lock it again10:58
pittiapw: but the thing you asked for is http://cgit.freedesktop.org/udisks/commit/?id=1064be64a7398b9b5dec4ece39b738ff15073d5610:59
directhexso does anyone know what would cause an ntfs entry in /etc/fstab to lock up lucid during boot?11:07
asaccjwatson: thanks for klibc fix11:09
apwdirecthex, what does it say?  waiting for it?11:10
directhexapw, yeah, i think that was it11:10
apwsmb, you had something about mount lines needing a new option (that they should always have had) to make boot work11:11
smbnobootwait or so...11:11
smbdirecthex, Right, try "nobootwait" in fstab as one of the options11:13
directhexi'll give it a shot when i'm feeling brave enough to go potentially unbootable again11:14
directhexhow about the switch to nouveau? totally killed my x, with no obvious way to go back to nvidia-glx (ended up dropping to a terminal & installing things manually)11:15
smbdirecthex, It worked well for me as I had entries in there with "user" for usb sticks which was not enough anymore11:15
directhexsmb, if the lack of nobootwait, which i'd never heard of until today, is enough to make the system unbootable, even in recovery mode, perhaps something needs to behave better?11:16
* matti seconds directhex 11:16
apwdirecthex, see if thats the fix, i think that someone was meant to be making the prompt easier to understand11:17
smbdirecthex, Actually it gives you some "imo cryptic" hint. Though I am not sure its carried over to plymouth. Was it displaying some text waiting for xxx [SM]?11:17
apwdirecthex, it did prompt you to tell you there was an issue right?11:17
directhexsmb, yes11:17
smbAt that point you can press 's' to skip11:18
apwpitti, was it you who said that those messages were on the list to fix?11:18
directhexsmb, really? THAT wasn't made clear11:18
* smb is not even sure what 'm' would do. But 's' was going on11:19
* ogra always thought SM was something that involves whips11:20
directhexwait, [sm] is a prompt?11:20
smbWaiting indefinitely for a mount really counts as "sm" to me11:21
ograit tells you the keys you can press11:21
smbdirecthex, yep11:21
directhexson of a...11:21
ograon fsck there is also C11:21
ogra(for cancel)11:21
ograif the theme doesnt handle that now though11:21
pittiapw: "those messages"?11:22
apwpitti, sorry, the plymouth mesages when it wants you to skip a mount and says /boot [SM]11:23
pittiapw: oh, I'm afraid I have no idea about that one11:23
apwsomeone was saying they were under review, and wanted to make sure that i was correct11:23
apwmay have been slangasek11:23
cjwatsonasac: upstream backports are the easy ones ;-)11:37
Keybukthat's interesting11:43
Keybuka bug is fixed by a random line of code that slangasek added, without any reference to a bug number, and that he didn't send upstream11:43
Keybukand I can't find any reason for him adding that other than "he noticed it was probably needed"11:44
siretart`pitti: mdeslaur: ffmpeg bug as discussed earlier today filed as bug #53729712:09
ubottuError: Could not parse data returned by Launchpad: list index out of range (https://launchpad.net/bugs/537297)12:09
=== MacSlow is now known as MacSlow|lunch
=== tkamppeter_ is now known as tkamppeter
sabdflis anyone here particularly focused on chromium?13:06
persiafta seems to spend the most time on it (dailies, etc.)13:07
persia(assuming you mean chromium-browser rather than chromium-bsu)13:07
ograor chromium OS :)13:08
persiaThat's "Chrome OS"13:08
ograi thought there was a chromium branch of it as well ?13:08
cjwatsonChromium OS is the open-source bit of Chrome OS13:09
* persia isn't 100% sure13:09
cjwatsonanalogously to Chromium/Chrome13:09
sabdflto see if we can add http://www.ubuntugeek.com/ambiance-theme-for-google-chromechromium.html to the package13:11
sabdflor maybe it needs to be in the light-themes package?13:12
sabdflseems odd to have bits there for a Universe app13:12
sabdflbut it's a *popular* universe app :-)13:12
persiaCould do a light-chromium-theme package.13:12
ograasac, ^^^13:12
persia(or chromium-theme-light)13:12
cjwatsonlamont: is it feasible to update util-linux to 2.17.1?  see bug 53007113:16
ubottuLaunchpad bug 530071 in ubiquity "Lucid Default live-cd install fails with 4K sector / Advanced Format drives" [High,Triaged] https://launchpad.net/bugs/53007113:16
asacfta: ^^13:23
asacfta: (see a few lines above for chromium theming)13:23
* ogra thinks we should just set them as defaults in the package13:24
ogra(both, the scrollbar and the theme)13:25
persiaHow does that play with xubuntu/kubuntu?  (I'm not sure folks using mythbuntu, ubuntustudio, or ubuntu-server will run chromium-browser so much)13:26
superm1ogra, the scrollbar is an extension, so i doubt it can be set up by default like that13:26
superm1and it doesn't work on internal pages (chrome://)13:27
cjwatsonpitti: so, you removed libjpeg7 from the archive saying "not needed in lucid", but libjpeg-progs still depends on it13:45
cjwatsonpitti: what's up with that?/13:45
pittiuh? I ran checkrdepends on it13:45
cjwatsonit has a bunch of rdepends, and we don't have libjpeg8 in lucid13:45
cjwatson(and I'm a bit sceptical about upgrading to it at this point)13:46
cjwatsonyou apparently left the libjpeg-progs binary in there, even though it's in the same source package, and libjpeg7 wasn't NBS13:46
cjwatsonso I'm confused about what was going on :)13:46
pittiI spotted libjpeg7 while reviewing duplicated libraries13:47
pittiso I ran checkrdepends on those, and it seemed to have none13:47
pittiapparently I screwed up on that one then13:47
cjwatsonpitti: shall I just reupload libjpeg7 to put it back?13:48
cjwatsonpitti: what was the other duplicate, though?13:48
pitticjwatson: I'd rather build it from 6b again13:48
cjwatsonI wouldn't13:48
cjwatson6b's build system sucked13:48
cjwatsonIIRC it wasn't cross-buildable13:48
pittihm, having a second implementation just because of that? it was built from 6b in karmic as well13:49
cjwatsonplus libjpeg-progs was already bumped to 7, so going backward would be hard13:49
cjwatsoncan we just transition to 7?13:49
pitti$ apt-cache rdepends libjpeg62|wc -l13:49
pittithat's what I wanted to avoid13:50
yofelhey, will there be 64bit installation media for ubuntu-netbook? as there are 64bit atom cpu's out there now13:50
cjwatsonDebian's libjpeg-progs is built from 813:50
cjwatsonso we could do 7+really6b-1 as the version for just the libjpeg-progs binary13:50
pittiright, I just thought the same13:51
pittiand in manic we can just sync it again13:51
pitticjwatson: ok, I'll get that fixed ASAP, thanks for pointing out13:51
cjwatsonpitti: oh, ok, thanks, I can do it if you're busy13:51
pittiI am, but since I broke it..13:52
cjwatsonwho isn't busy at the moment I guess13:52
pitticjwatson: I created bug 537370 as a reminder13:55
ubottuLaunchpad bug 537370 in libjpeg6b "build libjpeg-progs again" [High,In progress] https://launchpad.net/bugs/53737013:55
cjwatsonpitti: thanks13:55
mr_pouitthere was already bug 535629 opened13:55
ubottuLaunchpad bug 535629 in libjpeg6b "package libjpeg-progs is not built from any source package but several packages in lucid depend on it (dup-of: 537370)" [High,New] https://launchpad.net/bugs/53562913:55
mr_pouiterf, ok13:55
pittimr_pouit: sorry, noticed too late; duped13:56
\shpitti: does that mean, for lucid+1 total transition to 7 or 8 ? hopefully mostly done by debian at some point?13:56
pitti\sh: to 8 presumably13:57
pittiI guess we'll get a lot through autosyncs13:57
=== MacSlow|lunch is now known as MacSlow
\shwell...I get some new hardware for an additional buildserver in the next 2 months...it should be ready for lucid+1 ;)14:01
lamontcjwatson: Keybuk is the better one to ask about 2.17.1 though I think he might poke me about it, istr14:07
zulasac: ping14:15
=== Chex_ is now known as Chex
asaczul: hi14:15
asacon MIR?14:15
zulasac: you read my mind ;)14:15
asaci approved the log thing (with a few comments)14:15
zulasac: cool thanks14:15
asacanything else i havent checked that is on your list?14:15
zulasac: not that I know of14:16
asacok great14:16
=== dmart_ is now known as dmart
tumbleweedgarr, excues that, lag--14:40
pittitumbleweed: better change your password now :)14:44
tumbleweedpitti: that wasn't a password, thankfully14:44
=== jamie is now known as Guest19337
maco2cjwatson: slangasek got it15:25
=== mcasadevall is now known as NCommander
=== plars` is now known as plars
=== deryck is now known as deryck[lunch]
=== yofel_ is now known as yfoel
=== yfoel is now known as yofel
=== binarylooks_ is now known as binarylooks
ftaasac, ogra, persia, cjwatson: i don't think chromium has a system wide location for extensions like firefox, but i can sure check16:53
persiafta: Look a little higher up in the scrollback :)  But good luck.16:54
asacfta: thanks16:54
ftapersia, how far?16:56
ografta, up to the person who requested it i guess :)16:59
ftasabdfl, ^^ + i've been working on chromium since day one. I'll have a look at what's possible for that theme but i think we need a scheme for chromium extensions in general17:00
=== MacSlow is now known as MacSlow|capoeira
stefanlsdAnyone else have an issue running /usr on a seperate partition on lucid? Trying to debug why. 1/2 times mountall doesnt mount /usr and need to hard reboot. Having an issue now where noveua drivers wont blacklist either17:06
\shstefanlsd: last night I booted lucid with /usr on a separate partition successfully17:10
stefanlsd\sh: any issues since? my first boot always fails, then reboot works...17:12
\shstefanlsd: nope...even my problem with LVMs disappeared yesterday17:13
sabdflfta: yes. i don't know if they allow both system-level and per-user extensions17:13
sabdfltook ff years to get extensions working well17:13
stefanlsd\sh: kk. thanks. maybe i just need to re-install and start fresh :)17:14
ftasabdfl, i still have to push the ffmpeg codecs to lucid (currently only in the 3 chromium PPAs). the package is ready but i'm unsure about the redistribution license of some of those codecs (h264, aac..). I wouldn't mind some help here17:17
fta(this is needed for the HTML5 audio/video tags)17:18
sabdflif they are open source, go ahead17:18
ftait's basically ffmpeg, which we already have, but configured/packaged differently17:19
\shfta: talk to siretart` ;)17:30
=== deryck[lunch] is now known as deryck
siretart`I got hilighted?17:34
persiasiretart`: You were highlighted as a contact for fta to discuss ffmpeg packaging for the chromium-browser ffmpeg stuff17:35
siretart`fta: what do you mean with 'push ffmpeg codecs to lucid'?17:35
siretart`AFAIUI, chromium tracks astrange's ffmpeg-mt branch, is that right?17:36
ftasiretart`, http://bazaar.launchpad.net/~chromium-team/chromium-browser/chromium-codecs-ffmpeg.head/files/head:/debian/17:55
* Keybuk teaches popey about =17:58
ftasiretart`, yes, it's the -mt branch, with a bunch of patches and a build system from the chromium devs, and different flavors of codecs depending on the branding (chrome vs chromium)17:59
Keybukshells used by Real Men have a neat =thingy shortcut ;)17:59
Keybukit's basically `which thingy`17:59
Keybukso I just do dpkg -S =skype17:59
mathiazCould an archive admin promote vlan-modules-2.6.32-16-generic-di and vlan-udeb to main?17:59
popeyahhh neato!17:59
popeythanks :)17:59
mathiazThese are binary packages only17:59
Keybukdoesn't work in bash though17:59
popeydo I have to use keybuksh18:00
davmor2popey: emacs ;)18:00
ograKeybuk, now whats a real mans shell ?18:00
KeybukI resisted the urge to make a \sh! joke18:00
zygahas anyone seen mvo recently?18:05
Keybukhe's on holiday I believe18:08
blueyedzyga: just wanted to ping him also, about bug 50264118:08
ubottuLaunchpad bug 502641 in apt "[Lucid] apt-get source always selects highest available version instead of the specified one" [Unknown,Fix released] https://launchpad.net/bugs/50264118:08
blueyedit should be fixed, but does not appear so..18:08
blueyedthe proposed fix (http://bazaar.launchpad.net/~mvo/apt/mvo/revision/1693) appears to be in Ubuntu.18:10
blueyedNobody affected by this? Even the "--only-source" workaround mentioned in the Debian bug does not work for me.. have to resort to using dget..18:11
kamusinpitti, thanks for patching tzdata at time, you are our hero18:11
zygaKeybuk: I see18:13
ftasiretart`, persia: i guess i should just send chromium-codecs-ffmpeg to lucid and wait for the NEW to be reviewed, unless you already have comments i should address18:15
* persia is just passing messages18:16
mathiazlamont: re bug 532587 - is it important/mandatory that system users/groups are removed when the package is removed/purged?18:17
ubottuLaunchpad bug 532587 in puppet "removal of package does not stop daemon" [Medium,Fix released] https://launchpad.net/bugs/53258718:17
cjwatsonzyga,blueyed: I think he's off ill18:18
\shKeybuk: please do...I need to laugh today ;)18:19
lamontmathiaz: uh...  see policy?18:21
lamontbut stopping the daemon is kinda critical18:21
mathiazlamont: right - that is fixed in lucid now18:21
mathiazlamont: hm - where in the policy is the user/group remove on package removal documented?18:24
mathiazlamont: http://www.debian.org/doc/debian-policy/ch-opersys.html#s9.2 deals with system users/groups18:24
mathiazlamont: but doesn't refer to what should be done on package removal18:25
persiamathiaz: I'll suggest not removing the users/groups on package removal (which I've seen in a number of packages): this avoids the problem of potentially chowning files that are to be preserved, etc.18:26
lamontyeah - and I don't see it anywhere either18:26
lamontnot removing is a good answer18:26
persia(plus it means end-users get the *same* uid/gid if they reinstall later)18:27
lamontpostfix package removes the users and groups it created, many other packages don't18:27
mathiazpersia: right - not removing the user/group seems to safest option to me as well18:27
mathiazlamont: do you need an SRU for karmic on the puppet bug?18:28
persialamont: Do you happen to know why postfix does it that way?18:28
mathiazlamont: or lucid is fine?18:28
lamontmathiaz: romantic notions of cleaning up after myself?18:28
persiaOK.  As long as there's a good reason :)18:29
lamontbind9 likewise - most likely I copied the bind behavior into postfix18:29
lamontmathiaz: ultimately, I'm certain it was "the standard reason": it seemed like a good idea, at the time18:29
persiaBeing tidy seems more defensible.18:30
persialamont: On a completely different note: did my last comment to bug #67544 contain enough information, or is more useful?18:31
ubottuLaunchpad bug 67544 in fpc "PPC build of fpc fails" [Unknown,Fix released] https://launchpad.net/bugs/6754418:31
persia(not really Fix Released)18:32
Keybuk<grin> at bug #53724818:37
ubottuLaunchpad bug 537248 in plymouth "does not format usb flash" [Undecided,New] https://launchpad.net/bugs/53724818:37
Keybukmost people think that Plymouth not formatting usb flash devices is a feature18:38
Keybukfilesystem corruption is about the only bad thing it *doesn't* do :p18:38
zygadoes anyone know if it's possible to link PPA with a launchpad project somehow?18:59
jpdszyga: Try asking #launchpad.19:00
slangasekKeybuk: the line in src/main.c?19:20
slangasekKeybuk: that was: after fixing the proximate cause of bug #506717, plymouth would helpfully segfault19:20
ubottuLaunchpad bug 506717 in plymouth "[Lucid] plymouth does not display when using nvidia drivers" [High,Fix released] https://launchpad.net/bugs/50671719:21
Keybukwas the fix for 506717 the declining to show the script plugin thing?19:21
alexmoldovanHello guys, I have a q: I am running apport and it asks me "Has this issue been confirmed to exist with the upstream kernel?". I don't know what this means, so I don't know what to answer. I would like to help, so I need to understand this.19:27
bdrunga package FTBFS in the archive, because a network test failed. i tried to reproduce it locally by disabling the network, but the package still builds. how can i reconstruct the buildd environment?19:30
bdrungi am talking about bug #52120419:37
ubottuLaunchpad bug 521204 in lucene2 "lucene2 2.9.1+ds1-5 fails to build from source" [Undecided,New] https://launchpad.net/bugs/52120419:37
kkszysiuwell anyone working to integrate somerhing like http://www.omgubuntu.co.uk/2010/03/easy-gui-window-button-switcher-for.html into gnome-appearance-properties ?19:39
johnsgruberIs there anything afoot to get plymouth to start gdm on a tty >=7 ?19:40
slangasekKeybuk: 506717> well, what I was fixing was that plymouth wouldn't fall back to the text plugin when script was selected19:49
Keybukthat's ok, it fits what I thought19:51
=== MacSlow|capoeira is now known as MacSlow
viyyerhi kirkland there?19:56
YokoZarWhat component do I file this against?  I've noticed that when I change from noveau to proprietary nvidia drivers my laptop stops working well with my docking station / monitor (have to manually tell it to switch displays and then manually change resolution)19:58
YokoZarSomehow it seems to me that's not entirely an nvidia mode setting thing, but I don't know too much about it19:58
KeybukYokoZar: file it with your NVIDIA support contact?19:59
YokoZarso is it all a proprietary drivers thing?19:59
YokoZar*s/package/component, been filing too many bugs upstream lately20:00
Keybukwell, if it works with nouveau, and doesn't with nvidia-glx - it's a binary driver problem, no?20:00
persiaCould file a bug against the nvidia-glx package, but it won't get fixed there unless it's fixed upstream.20:01
YokoZarKeybuk: Not necessarily, no.20:01
YokoZarIn my experience usually X is involved in all unwanted resolution changes20:01
viyyerI wanted to try out testdrive . While installing apt-get gives an error. viyyer@Atri:~/Desktop/panel$ sudo apt-get install testdrive20:02
viyyer[sudo] password for viyyer:20:02
viyyerReading package lists... Done20:02
viyyerBuilding dependency tree20:02
viyyerReading state information... Done20:02
viyyerSome packages could not be installed. This may mean that you have20:02
viyyerrequested an impossible situation or if you are using the unstable20:02
viyyerdistribution that some required packages have not yet been created20:02
viyyeror been moved out of Incoming.20:02
viyyerThe following information may help to resolve the situation:20:02
viyyerThe following packages have unmet dependencies:20:02
viyyer  testdrive: Depends: cpu-checker but it is not installable20:02
viyyerE: Broken packages20:02
persia!paste | viyyer20:03
ubottuviyyer: For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic.20:03
tesukiIf I make a proprietary program could I get it into the standard repos? (a program that needs a license file and accepting of a license)20:03
KeybukYokoZar: X *is* whatever driver you're using20:03
Keybukthere isn't much more to it than that, really20:03
persiatesuki: Depends on the license.  If it can be distributed freely, but needs a license to use, it can be in multiverse, but you may have a hard time interesting developers in uploading it.20:04
YokoZarKeybuk: But X has multiple components, and not all are the drivers, hence the questions.  I remember specifically having a resolution problem in the past that was an XRandR issue, for instance.20:05
tesukiIf I make a deb (that I can install with dpkg -i name.deb) would it still be hard? Could money help?20:06
YokoZarIf it's packaged right20:06
persiatesuki: making a .deb makes it impossible.  It needs to be a source package.  The sun-java6 package in jaunty is a good example of how that works.20:06
YokoZar^^ what persia said20:07
persiatesuki: I believe Cannical offers a program where you can pay to have proprietary appls inthe partner archive.  No idea on costs.20:07
viyyerkirkland,  http://pastebin.ca/1834309  what package is cpu-checker ?20:07
KeybukYokoZar: I don't think nvidia-glx *does* xrandr20:07
KeybukYokoZar: it does mad TwinView crap20:07
YokoZarKeybuk: ahh ok, hence the problem.  You've sated me.20:08
YokoZarviyyer: packages.ubuntu.com can answer that for you20:08
viyyerYokoZar, there is none20:09
YokoZarviyyer: change "karmic" to "lucid", it seems it's a new package20:10
tesukipersia: that program is namned?20:10
YokoZarregardless ubuntu-bug packagename will work even if it's a binary package name and not a source package name20:11
\shtesuki: http://www.canonical.com/services/packaging <- canonical packaging service for partner apps20:11
tesuki\sh, persia thank you.20:12
viyyerYokoZar, the purpose of installing testdrive is to test out lucid. I don't want to install lucid right now20:13
viyyerYokoZar, testdrive allows one to test lucid on a VM20:13
=== Caesar_ is now known as Caesar
debfxAdri2000: hi, I just noticed you filed a FFe to get blobby into lucid20:30
debfxI'm the debian maintainer and wanted to make it build-depend on the (just accepted) tinyxml package (instead of building its own copy) for the next upload20:31
Adri2000debfx: hi20:34
Adri2000debfx: hmm then in this case maybe we're going to sync the current blobby and not wait for your next upload...20:36
Adri2000otherwise it would require another FFe for tinyxml20:36
debfxAdri2000: yes, I guess that's the best way as these architectures aren't supported by ubuntu anyway and chances that someone wants to play blobby on them are pretty low :)20:41
=== blueyed_ is now known as blueyed
=== viyyer_ is now known as viyyer
Adri2000debfx: I've just added a comment about this in the FFe bug. thanks for notifying me of this. do you think you could wait a bit before uploading the new version in debian, until we've decided for sure what version we sync? (so if we choose -1, it will still be possible to "sync")20:55
ccheneywhat util does line ending conversion from dos to unix?20:58
* ccheney can't find it with apropos20:58
ccheneycjwatson: thanks20:59
geniicjwatson: Isn't also dostou ?21:07
kirklandviyyer: hi, sorry about that ... I'm uploading cpu-checker to the ppa now21:13
kirklandviyyer: that was a recent change for lucid, forgot to backport until you reminded me :-)21:14
viyyerkirkland, thanks .. I think test drive is great idea.21:14
viyyerkirkland, when would you be able to upload the same ?21:15
kirklandviyyer: what do you mean?21:16
kirklandviyyer: https://edge.launchpad.net/~testdrive/+archive/ppa/+packages21:16
kirklandviyyer: the builds are queued, should be done in ~45 minutes21:16
viyyerkirkland, thanks :)21:16
kirklandviyyer: i'm also looking at the code, trying to remove (soften) the cpu-checker dependency21:17
\shwhy does sun-java6-bin need libnss-mdns and therefore avahi-daemon?21:26
kirklandviyyer: okay, cool ... I just modified the code to use cpu-checker if there, but not depend on it; this should solve the issue much better :-)21:35
kirklandviyyer: i'm uploading to the PPA now21:36
kirklandviyyer: again, will build in a bit21:36
mathiazkirkland: what is responsible for creating /dev/kvm?21:36
kirklandmathiaz: in lucid?  /etc/init/qemu-kvm.conf21:36
kirklandmathiaz: <lucid, the init script21:37
mathiazkirkland: on lucid, on an NC I get this: http://paste.ubuntu.com/393565/21:37
mathiazkirkland: that was *after* installing eucalyptus-nc as a package21:37
kirklandmathiaz: does /dev/kvm exist?21:37
kirklandmathiaz: lsmod | grep kvm21:37
mathiazkirkland: hm - now yest21:38
kirklandmathiaz: ?21:38
mathiazkirkland: because I manually modprobed kvm_intl21:38
kirklandmathiaz: pastebin your /etc/init/qemu-kvm.conf21:38
kirklandmathiaz: status qemu-kvm21:38
slangasekKeybuk: any idea what's causing these "plymouth not stopping at shutdown" bugs suddenly?21:38
mathiazkirkland: http://paste.ubuntu.com/393569/21:39
kirklandmathiaz: modprobe -r those two kvm modules21:40
kirklandmathiaz: and sudo start qemu-kvm21:40
kirklandmathiaz: make sure that that get's the module loaded correctly21:40
mathiazkirkland: is qemu-kvm started by the postinst?21:40
kirklandmathiaz: hmm21:40
mathiazkirkland: I haven't rebooted the NC after the eucalyptus-nc package has been installed21:41
kirklandmathiaz: i'm not sure, actually21:41
kirklandmathiaz: let me check21:41
slangasekKeybuk: ah, an apport change21:41
kirklandmathiaz: oh, one more thing ... dpkg -S /etc/init/qemu-kvm.conf21:41
slangasekpitti: this new "unkillable_shutdown" hook is broken21:42
mathiazkirkland: according to /var/lib/dpkg/info/qemu-kvm.postinst there is nothing that start qemu-kvm in there21:42
kirklandmathiaz: yeah, i agree21:42
mathiazkirkland: qemu-kvm is the package shipping /etc/init/qemu-kvm.conf21:42
kirklandmathiaz: cool ... so just a "start qemu-kvm | true" toward the end of that?21:43
mathiazkirkland: how do you install the qemu-kvm upstart job?21:43
mathiazkirkland: with dh_installinit?21:43
kirkland        dh_installinit -s --no-restart-on-upgrade --error-handler=true --noscripts21:43
kirklandmathiaz: oh, i think the --noscripts is breaking me21:44
mathiazkirkland: yeah - probably21:45
mathiazkirkland: should I report a bug?21:45
slangasekpitti: specifically, it's reporting plymouth as unkillable (bug #537248 et al.)21:46
ubottuLaunchpad bug 537248 in plymouth "does not format usb flash" [Undecided,New] https://launchpad.net/bugs/53724821:46
kirklandmathiaz: yes, please21:46
kirklandmathiaz: i'm fixing it now21:46
kirklandmathiaz: i'd like a bug to reference21:46
mathiazkirkland: bug 53768221:47
ubottuLaunchpad bug 537682 in qemu-kvm "qemu/kvm not started after package install" [Undecided,New] https://launchpad.net/bugs/53768221:47
kirklandmathiaz: thanks21:47
cjwatsongenii: it's the sort of thing that is sufficiently easy to write that many people have written it.  I don't know that enumerating all the versions of it is useful :-)21:55
mdzsay, why does my system slow to a crawl with disk I/O whenever apparmor is upgraded and reloads its profiles?21:55
cjwatsonhey, I noticed that too, had been meaning to report it21:55
jdstrandit regenerating its cached files21:56
jdstrandcached profiles21:56
jdstrandwe now have binary blobs that load super fast on boot rather than having to generate them on the fly21:57
jdstrandthat helps with boot times, but slows down install a bit21:57
cjwatsonit absolutely hammers my laptop on upgrade21:57
jdstrandthe system shouldn't slow to a crawl though...21:58
cjwatsonis it trying to massively parallelise the caching or something?21:58
jdstrandI'm not sure-- jjohansen is the one to ask21:58
jdstrand(or possibly kees, if he did something in packaging that would affec it)21:58
jjohansencjwatson: no21:59
* sebner waves at jdstrand and reminds him of moin 21:59
jjohansencjwatson: its single threaded, compiling the policy can hit the cpu pretty hard though21:59
cjwatsonfrom what I very vaguely recall, the load seemed to be in a kernel thread, but I couldn't look closely because I couldn't do very much :-)21:59
cjwatsonso that might be bogus data21:59
jdstrandsebner: don't need to remind-- I've pretty much only been working on it this week :)22:00
cjwatsonjjohansen: if it's just that, then maybe nice(1) would be in order22:00
jdstrandsebner: I need to get the stable releases out, then I'll review your debdiff22:00
jjohansencjwatson: sure that could be done22:00
sebnerjdstrand: aye aye, just thought you might have forgotten it :)22:00
keescjwatson: yes, it uses all available CPUs when generating the caches22:02
keescjwatson: using "nice" would defeat the purpose of loading it as fast as possible, though?22:03
jjohansenhrmmm, thats right kees parallelized it launch multiple compiles22:03
jjohansenI had forgotten that22:03
* kees nods22:03
jjohansenthe compile is itself single threaded22:04
micahgdidrocks: could I get you to test your firefox LP bug in the upstream build as well?22:05
jdstrandmaybe 1 - N-1 would be the way to go22:05
jdstrandkees: ^22:05
keeswell, the actual issue was mdz claiming heavy disk I/O, which is unrelated to the parallelization, I would imagine.22:05
jjohansenit does very little disk I/O22:06
mdzjdstrand, it really hammers my system, to the point where the desktop is unresponsive22:07
cjwatsonkees: ^- indeed, I'm seeing the same thing as mdz.  "as fast as possible" becomes counterproductive at that point22:08
cjwatsonit wasn't directly clear to me whether it was disk I/O or something else - unresponsiveness was the main symptom22:08
mdz        xargs -n1 -P$(getconf _NPROCESSORS_ONLN) "$PARSER" "$@" --22:08
mdzkees, is that what you're referring to?22:09
cjwatsonI think I may have been getting some kind of disk thrashing as a second-order effect22:09
directhexokay, am i the only one experiencing weirdness when gdb appears? if i hit enter to select myself from the user list, gdm seems to die & restart22:09
directhexthe second time, it's fine. just buggered post-boot22:09
RAOFdirecthex: No, you're far from the only one.22:10
mdz /etc/init.d/apparmor reload with a hot cache is reasonably sane22:11
RAOFdirecthex: That's plymouth being evil.22:11
keesmdz: yup, that's the parallelization22:11
sbeattiemdz|cjwatson: are you guys near memory limits where non-trivial amounts of memory consumption might cause swapping to occur?22:12
cjwatsonsbeattie: nowhere close22:12
keesmdz: seb128 has been complaining about how long some of the reloads were taking when doing package installs, so we pulled out the stops to speed it up.22:12
mdzsbeattie, nope22:12
keesI guess we could nice it.22:13
cjwatsoneven with two kvm instances running (which I didn't at the time), I have over 1GB free22:13
keesseems weird to me.22:13
mdz4.72user 0.15system 0:03.63elapsed 133%CPU (0avgtext+0avgdata 141552maxresident)k22:13
mdz0inputs+1528outputs (0major+40010minor)pagefaults 0swaps22:13
cjwatsonkees: or use only up to n-1 processors if n > 122:13
mdzkees, what does apparmor_parser do?22:14
keesI wonder if xargs correctly handles -P0 ...22:14
cjwatson-P0 means "run as many as possible" per the man page22:14
cjwatsonequivalent of make -j22:14
keescjwatson: eek22:14
ionmdz, cjwatson: I take it it’s not bug #458299 you’re encountering, though?22:14
ubottuLaunchpad bug 458299 in linux "apparmor_parser: page allocation failure. order:5" [Medium,Confirmed] https://launchpad.net/bugs/45829922:14
keesmdz: it builds the kernel rules for apparmor from the text file rules22:14
mdzion, I don't see that message, no22:14
mdzkees, right, but what sort of system calls does that involve?22:15
keesmdz: i.e. parses them into the binary DFA that the kernel uses for following AA rules22:15
mdzwhy should it be so I/O intensive?22:15
cjwatsonion: hmm, I do have one of those in my logs22:15
cjwatsonso perhaps mdz and I have different problems22:15
keesmdz: it just uses open read write.  the files are relatively small.22:15
mdzkees, something else is going on here, then22:15
keesmdz: what does du -sh /etc/apparmor.d/cache say ?22:15
mdzkees, there are only 1.1MB of files in /etc/apparmor.d in total22:16
mdzthat doesn't even come close to explaining the behavior I' see22:16
keesmdz: yup, sounds about right22:16
ioncjwatson: The symptoms of that issue seem be something in kernel space constantly allocating all free memory in an infinite loop, making the system unusably slow.22:16
mdzkees, what happens to running processes when those are reloaded?22:16
cjwatsonion: that sounds plausible22:16
mdzkees, is it possible that it's paging in a lot of stuff?22:17
keesmdz: do you mean "how are the AA protections for the processes handled?" they are atomically swapped for the new profiles once they're loaded.22:17
keesjjohansen: ^^ is it possible the kernel is hauling in lots of memory to attach/swap the profiles?22:18
mdzkees, does that involve touching more than a trivial amount of process memory?22:18
keesmdz: if that were the case, I would assume a standard hot-cache "sudo service apparmor reload" would show the same problem.22:18
mdzkees, except that it's all paged in22:18
mdzI have plenty of RAM22:19
jjohansenwell it pulls in some memory dependent on what the policy is but trivial on todays machines 2-4 MB22:19
keesmdz: the blobs in /etc/apparmor.d/cache are what is being loaded in the kernel, so that's the extent of the kernel memory.  the userspace parser uses more memory than that when parsing, but it's not huge.22:19
ionCache in /etc? Eww. :-)22:19
jjohansenuserspace compilation does get large22:19
keesion: :)22:20
mdzjjohansen, so you think it's apparmor_parser itself which is the likely culprit?22:20
jjohansenmdz: we can experiment with running this without doing the actual kernel load22:21
jjohansenvs. just loading from prebuilt cache22:22
=== jono_ is now known as jono
slangasekKeybuk: so I'm getting casper-md5check to talk to plymouth; is there any sort of distinction in plymouth between displaying errors vs. high-prio text vs. normal text?  I don't see anything along those lines in the API22:38
Keybukwe can add things like that22:40
Keybukthere are two types of message our theme supports22:40
Keybukgeneric blah blah types like "Checking your disks"22:40
Keybukand beneath that, a message about what keys you can press22:40
Keybuk"Press C to cancel" etc.22:40
slangasekwell, this certainly doesn't seem like something that should be added into a theme if there's no protocol-level distinction22:41
KeybukI think you misunderstand me22:42
Keybukthis kind of thing, in plymouth, is deliberately left up to the theme22:42
Keybukplymouth provides basic ways to communicate with the theme22:42
Keybukand you build from there22:42
Keybukthat's how our fsck progress works, for example22:42
slangasekI don't think I misunderstand you, I think I disagree with you that this is a reasonable design :)22:43
Keybukreally, why?22:44
Keybukplymouth is just a plugin architecture for dealing with displays really22:44
Keybukwhere the most common use is to make splash screens22:44
persiaKeybuk: So if I get text overlaying an entry box during boot, that's a plymouth bug, or a theme bug?22:44
Keybukpersia: theme bug22:44
Keybukrelevant aside: there's new gdm, mountall + plymouth packages in my PPA22:45
Keybukwould appreciate testing22:45
slangasekKeybuk: that the client and the theme have a private agreement for the meaning of the text strings?  Yes, I think that's a bad design22:45
slangasekit means anyone who wants to use a different theme gets weird handling of the annotated strings the client is sending22:47
Keybukthough that being said, I never really got how some text can be "urgent" and some can't be22:47
Keybukslangasek: but in plymouth, switching a plugin can be switching from a splash screen to a login screen to an OS switcher, etc.22:48
slangasekplugin, or theme?22:48
slangaseknone of the themes shipped here seem to be login screens or OS switchers22:49
Keybukplugins and themes are the same thing22:49
Keybukslangasek: half the themes in plymouth don't even support message ;-)22:49
slangasekwell - that's better than having them support it and wind up with "keys:" prepended? :)22:50
Keybukthe alternative is that you put a massive burden on theme authors to support dozens of commands that they might not care about22:50
slangasekI half think that if a theme isn't going to behave sensibly with mountall, we shouldn't ship that theme by default22:51
slangasekdump it in plymouth-bad-for-ubuntu or something :)22:51
KeybukI don't disagree :p22:51
Keybuksplitting the themes out is on the todo22:51
Keybukthat being said, I did make the text theme work with mountall ;)22:51
=== jono_ is now known as jono
ion(add-apt-repository ppa:scott)22:53
=== robbiew is now known as robbiew_
=== sconklin is now known as sconklin-gone
ionkeybuk: Worksforme™23:09

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