kirklandslangasek: i just plucked this crap out of the init script00:00
slangasek(we don't reverse that when stopping the job, so not sure why it should be here)00:00
Keybukkirkland: it wasn't boot speed per se. more than it's easy to trip over upstart bugs00:00
slangasekkirkland: well the init script wasn't very good ;)00:00
slangasekgratuitous use of cat; 4 sed commands where 1 would do; and why are we preallocating /dev/loop nodes here?00:01
kirklandslangasek: will fix the sed00:01
kirklandslangasek: will ditch cat00:01
slangasekah, but those are the minor things, it's the mknods that have me scratching my head :)00:02
kirklandslangasek: the loop devices are necessary for mounting storage for the instances00:02
kirklandslangasek: let me find the bug number with the background00:02
slangasekand "32" just happens to be the right number of loop devices to preallocate? :)00:03
kirklandbug #43084600:03
ubottuError: Could not parse data returned by Launchpad: HTTP Error 500: Internal Server Error (https://launchpad.net/bugs/430846)00:03
* kirkland screams at launchpad00:03
kirklandgoogle cache FTW00:05
wgrantkirkland: edge is working in read only mode, FWIW.00:05
kirklandwgrant: thanks00:06
kirklandslangasek: so there you go .... I asked the kernel team if we could increase the ubuntu default00:06
kirklandslangasek: they told me go make them myself00:06
kirklandslangasek: seems easy enough to do safely at boot00:06
slangasekkirkland: right - I guess I'm just a little wary of the preallocation of a fixed, magic number of loopback devices00:07
slangaseknot a bug now, might be a bug later00:07
kirklandslangasek: and apparently this is something only eucalyptus needs, as the kernel team didn't want to increase it arbitrarily for everyone00:07
Keybukwe need to fix loop00:07
Keybukso that it allocates on demand00:07
Keybukevery one created at boot costs boot time00:07
Keybukit should be more like /dev/pts00:07
Keybukthat's upstreamy though00:08
kirklandKeybuk: i do not disagree :-)00:09
kirklandslangasek: what does this mean:00:10
kirkland<slangasek>         echo -n 1 > /proc/sys/net/ipv4/ip_forward00:10
kirkland<slangasek> not /etc/sysctl.d?00:10
* Keybuk wonders why there's no /sys/module/loop/parameters/max_loop00:10
slangasekkirkland: we already have /etc/init/procps.conf, which will do sysctl tweaks for anything defined in /etc/sysctl.conf and /etc/sysctl.d; should eucalyptus-nc not just drop a conffile there?00:11
Keybukslangasek: I think that kirkland hasn't realised that's a sysctl ;)00:11
slangasekI'm not saying one or the other is right, I'm just pointing out what I thought was the standard mechanism00:11
Keybukkirkland: your echo is equivalent to putting "net.ipv4.ip_forward = 1" in /etc/sysctl.d00:12
kirklandKeybuk: cool, thanks, i can do that00:12
Keybukyeah that's probably right00:12
Keybuksince then all the documentation applies00:12
kirklandslangasek: Keybuk: right well, as i said, this was a straight port of a crappy init script to a crappy upstart script;  we're just now trying to clean it up00:13
kirklandKeybuk: slangasek: what about /proc/sys/net/bridge/bridge-nf-call-iptables ?00:14
kirklandKeybuk: slangasek: is that sysctl-able too?00:14
slangasekanything under /proc/sys00:14
kirklandslangasek: okay, where do i find the docs to look up the var name?00:15
slangasekkirkland: it's straight s,/,./g on the filename00:15
kirklandnet.bridge.bridge-nf-call-iptables = 100:15
kirklandslangasek: ?00:15
* slangasek nods00:16
kirklandthat look about right?00:16
kirklandslangasek: is there a better way to get aoe modprobed?00:18
slangasekapparently not :)00:18
kirklandslangasek: the loops will need to stay for now, until there's a better way to do this00:18
kirklandkirkland@x200:/local/source/eucalyptus/lucid.public/ubuntu$ pastebinit debian/eucalyptus-nc.upstart00:19
kirklandkirkland@x200:/local/source/eucalyptus/lucid.public/ubuntu$ pastebinit debian/30-eucalyptus-nc.conf00:19
slangasekkirkland: looks sane to me00:20
kirklandslangasek: cheers, thanks00:20
kirklandKeybuk: i'll send you an email with a pointer to the bzr branch with all of the eucalyptus upstart scripts, for your leisurely review ;-)00:20
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
=== bdefreese2 is now known as bddebian
=== NCommander is now known as Guest48610
=== spm changed the topic of #ubuntu-devel to: Lucid Alpha 1 released | Archive: open | MoM running (but use bzr!) | Development of Ubuntu (not support, not app development on Ubuntu) | #ubuntu for support and general discussion for dapper-karmic | #ubuntu-motu for getting involved in development | http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://wiki.ubuntu.com/HelpingWithB
=== Guest48610 is now known as NCommander
=== spm changed the topic of #ubuntu-devel to: Lucid Alpha 1 released | Archive: open | MoM running (but use bzr!) | Development of Ubuntu (not support, not app development on Ubuntu) | #ubuntu for support and general discussion for dapper-karmic | #ubuntu-motu for getting involved in development | http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://wiki.ubuntu.com/HelpingWithBugs
=== NCommander is now known as Guest88322
=== nhandler_ is now known as nhandler
=== Whoopie_ is now known as Whoopie
=== Guest88322 is now known as NCommander
=== asac_ is now known as asac
pittiGood morning05:44
StevenKMorning pitti05:44
Hobbseeheya pitti!05:46
pittiWhoopie: just 481448> uload then05:46
* slangasek waves to pitti05:47
pittislangasek: daemon.log> hm, it seems full of debugging spew; but yes, we can certainly also downgrade the debugging level of apps05:48
slangasekdebugging spew - really?05:48
slangasekah, wpa_supplicant05:48
slangasekoh, wait05:49
slangasekpitti: daemon.log, or auth.log?  daemon.log doesn't appear to be synchronous by default05:49
slangasekauth.log is, and is the one I'm concerned about :)05:49
pittislangasek: oh, and I was really talking about daemon.log05:49
pittipowertop regularly complains about stuff writing there05:49
slangasekpowertop doesn't seem to distinguish between synchronous and asynchronous writes05:50
pittiah, then I misunderstood your comment, sorry05:50
slangasekyou probably didn't misunderstand, I was probably just confused :)05:51
pittihey StevenK, morning Hobbsee!05:51
slangaseksince AFAIK the only log we write synchronously by default is one that we /should/, I assumed that was the one your comment in the whiteboard was referring to05:51
slangasekwithout noticing that auth != daemon :)05:52
slangasekpitti: btw, if you're going near SRUs today, hold off for a few hours on the cryptsetup one; I've confirmed the report that 475936 isn't fully fixed in lucid, and am going to track that down today06:02
pittislangasek: cryptsetup> ack06:03
* dtchen wonders if there's a faster way to test-build for armel06:06
dtchen(I'm kinda surprised that upstream PA hasn't encountered these issues)06:06
pittiwith the new LP rollout, do we have 3.0 source format now?07:18
pittiat least sync-source.py still fails07:19
pittibut that could be due to cocoplum having a too old dpkg07:19
pittiwgrant: ^ do you know?07:19
ttxslangasek: looks like we'll need to keep cglib2.1 for the moment.07:28
pittijames_w: thanks for your "distributed dev" summary!07:31
pittijames_w: how often are these branches imported? I just tried lp:debian/cdbs, and it's behind by more than 4 months (there were > 5 uploads afterwards)07:31
pittijames_w: another question, what would happen if I did "bzr push lp:ubuntu/pm-utils-powersave-policy --force-overwrite" so that I can get my native bzr over the auto-import? from what I remember, people can commit individual changes to an auto-import, and they won't get overwritten as long as they are consistent with the .dsc?07:35
dholbachgood morning07:43
persiapitti: Do you know where I might find powerpc ddebs for lucid?07:47
=== ara_ is now known as ara
pittipersia: funny you should ask; we have collected them for years and hardly used them, and I removed them last week because macaroni went out of space07:49
persiaHeh.  And it was last weekend that I went looking :)07:49
pittipersia: if you need them for a recent build, they can still be retrieved from the buildds, though07:49
pittipersia: are you looking for a particular package?07:50
persiaIs that something I can do on a per-build basis?  I rarely need them (my powerpc box mostly just works), but when I do, it's only one or two at a time.07:50
pittithe build log should tell you on which buildd it was done07:50
pittipersia: sure, just install pkg-create-dbgsym locally07:50
persiabinutils 2.20-4ubuntu307:50
pittithen it just happens07:50
persiaI do that, but that's only for packages I build :)07:51
pittipersia: sorry, that was built a month ago already; they are gone, I'm afraid07:51
wgrantpitti: It's waiting on an upgraded dpkg on the various Soyuz machines. Once that's done, a LOSA needs to run some SQL magic to enable 3.0 uploads to Lucid.07:52
persiaAh, so they don't get stored in +archive :(07:52
pittipersia: not yet; however, wgrant was working on that07:52
pittiso, I hope it'll happen "soon"07:52
persiaThat would meet my use case.07:52
wgrantpitti: RT #36873, if you want to watch it.07:52
persiaOn a separate note, if you've removed all the binaries, please also remove the (now invalid) Packages and Release files.07:53
wgrantMost of the ddebs-in-soyuz work is done. I'm nearly there. But then there's the issue of librarian bloat.07:53
persia(it's confusing to have things work and then get 404s for difficult to understand reasons)07:53
pittipersia: well, there's still tons of powerpc ddebs07:54
persiaOh, so I just picked an unlucky package?  Oh well.07:54
pittiI just stopped collecting them07:54
pittibut yeah, I'll remove them for lucid07:54
persiawgrant: How much bloat are we talking about?  Is it something that could be contained by only tracking a subset of packages or something?07:55
* persia is just unsure how to proceed with a segfault in ld07:55
mneptokpitti: "Macaroni Went Out Of Space" sounds like a Hunter S. Thompson book title, or an album by a 1960s psychedelic band.07:57
pittiyou are so right07:57
pittisometimes I don't even read what I wrote any more :)07:57
wgrantpersia: pitti can you probably tell us approximately how much bloat we're talking about.07:58
wgrantBut it's not an insignificant volume :(07:58
pittiright, they are huuuuuuuuuuuuge07:58
pittiI'd estimate some 300 GB on macaroni right now07:58
persiaAh, that is large.07:59
pittiexpect some (elf portion of packages)%507:59
pittierm, *507:59
wgrantPlus I suspect that you have a more aggressive binary removal policy than LP.07:59
persiaI don't know about other people, but I tend to find that most classes of crashes can be replicated on an arbitrary architecture (excepting endian issues, word length issues, and silly assumptions about alignment).07:59
pittiyeah, I only keep the latest07:59
pittiit explodes otherwise07:59
pittiso please let's not keep all of them for ia64 :)08:00
persiaSo I wonder if it might be possible to get away with just the packages that are used as build dependencies for other packages (mostly libraries and toolchain)08:00
wgrantSo it needs an Awful Lot of disk space on both cocoplum and mizuho.08:00
persiaAnd I presume that me running to the shop and shipping disks isn't the solution, because there are chassis limitations, etc.08:02
pittiwe could only keep supported arches, and only for current lts, current stable, and development release08:05
pittiplus ports for dev release perhaps08:05
persiaThat's a set that matches all my use cases :)08:06
wgrantActually, thinking about it, we could safely enough purge ddebs early.08:06
slangasekttx: wah; ok08:06
ttxslangasek: proxool needs asm2, cglib (2.2) needs asm3, cglib2.1 needs asm208:07
ttxslangasek: and asm2 and asm3 don't live well together in the same classpath08:08
ttxso I'd say that debian adding "Provides: libcglib2.1-java" to libcglib-java is somewhat optimistic08:08
slangasekpitti: cdbs import failure info: http://package-import.ubuntu.com/failures/.bzr/failures/cdbs08:09
ttx(and eucalyptus needs proxool and cglib)08:09
pittislangasek: ah, thanks08:10
pittislangasek: btw, do you have sata disks? my laptop doesn't even have sata power link management08:10
=== Adri2000` is now known as Adri2000
pittislangasek: I'm changing the heuristics to be a little more clever, not just looking at total RAM, but at active RAM, too08:10
pittiI'm just curious how it feels like when enabling it08:11
slangasekpitti: yes, I have sata, with the intel chipset needed to use this08:11
slangasekI have 3GB RAM today, which is enough that SLPM doesn't hurt me08:11
slangasek(SLPM? SPLM?)08:11
slangasekwhen I had 1GB of RAM, it was bad :)08:12
pittiI now use "min_power" if inactive/total > 15%, and medium_power if inactive/total > 5%08:12
pittiand if it's even less, I don't touch it08:12
pittisince it seems dangerously close to swapping08:12
pittiI'm not entirely sure if this is a better heuristics, I'll send out a call for testing08:12
slangasekI think it would be fine to just check the amount of RAM; after all, mem usage will change over time, much more rapidly than the pm-utils hook will be triggered08:13
pittiso perhaps just bump it to 2 GB?08:14
* pitti tries to teach himself to use correct units, in light of the month-long units policy TB topic08:14
* slangasek grins08:14
slangasekfine with me; maybe it should be varied by architecture, also?08:14
slangasek1GB might be enough on 32-bit, isn't enough on 64-bit08:15
pittidoes it make such a difference actually?08:15
pittiI had expected most things to just use ints08:15
pittipointers do, of course08:15
slangasekeverything python approximately doubles in size08:16
pittihow much RAM do netbooks have these days? They are prone to have i38608:16
persiaThey vary from 512MB to 3GB08:16
persia(at least that's what I saw last time I looked in a shop about 3 weeks ago)08:17
pittiso, let's say 1 GiB on i386, and 2 on AMD64?08:17
slangaseksounds reasonable to me08:17
pittijames_w: btw, do you ignore language-pack-* for auto-imports? it'd be a waste, and people really shouldn't change the packages directly anyway08:23
slangasektseliot: hmm, did you change anything that should have fixed the input-dropping bug in plymouth?  I don't see any recent upgrades on anything except initramfs-tools, but my passphrase entering is now reliable08:28
tseliotslangasek: no, but that happened to me too once and I don't know what made it more reliable08:29
slangasektseliot: should I close my bug, or leave it open?08:32
tseliotslangasek: please leave it open, I think it's something we should investigate08:32
slangasekpitti: cryptsetup checks out, the problem I'm seeing is mountall+fstab interaction and not a bug in the cryptsetup fix itself08:36
pittinice, so good to go to -proposed then?08:37
slangasekpitti: IMHO yes; I'll update the test case based on my latest findings08:37
slangasekpitti: test case updated08:50
pittiogra_: did you ever happen to test lucid alpha-1 on your touchscreen? was there a major regression due to the udevification?09:11
pitti(of x.org)09:11
ogra_pitti, i'm currently not running lucid on my laptop, will tell you once i do09:27
pittiogra_: thanks09:27
=== tkamppeter_ is now known as tkamppeter
james_wpitti: they are imported continuously, but there was a bug meaning Debian was missed, so some of them are out of date. An LP bug is hampering my effort to get all of Debian up to date.10:09
james_wpitti: you could push --overwrite pm-utils-powersave-policy. I would have some fallout to deal with, but that's not a problem.10:10
james_wand I do not currently ignore language-pack-*10:10
pittijames_w: so is the --overwrite considered a hack or a semiofficial method to move the real-world bzr branches to their preferred naming scheme?10:11
james_wboth :-)10:11
pittijames_w: if it's sanctioned, I'd like to do it for apport, jockey, pm-utils-powersave-policy (since those have full source)10:11
james_wit works fine10:11
* slangasek grins10:11
james_wI can do it by changing the links, which is perhaps a little more elegant10:11
pittiI keep getting bad merge requests for apport, and just pushing the real branch to lp:ubuntu/apport would help a lot10:11
james_wthere's a bug I need to fix before I do that though10:11
slangasekif the package also exists in debian, though, that will lose us some of the history I guess?10:12
pittifor those three that's not the case10:12
pittiI guess for the merge case it might be better to just throw away our old branch, merge with Debian, and push the merge to the auto-import branch?10:12
james_wI have to make it re-import taking in to account what is already there, after adding some tags10:12
james_wI would prefer to do it all properly, but I understand that it must be frustrating waiting for me10:13
pittijames_w: if it's causing you extra work, I'll leave it as it is for nwo10:14
EsatYucei m member one of the Launchpad team which name is https://launchpad.net/~ubuntu-l10n-tr, i want to release translated's package. How?10:14
james_wby doing a push --overwrite you force the issue10:14
pittijames_w: I was just wondering because you said that we can actually commit individual changes to those branches, upload, and the importer would see that it's already there and not clobber it10:14
james_wpitti: you theorised correctly10:15
james_wthere are just a couple of implications for the backend10:15
james_wthe code hasn't quite caught up with reality yet10:15
=== ogra_ is now known as ogra
EsatYuceis this right channel to ask about Launchpad??10:21
slangasekEsatYuce: #launchpad is probably a better channel for questions about Launchpad10:21
=== mac_v is now known as _
=== _ is now known as _7hills
=== _7hills is now known as mac_v
=== jelmer__ is now known as jelmer
=== jelmer is now known as Guest85315
=== Guest85315 is now known as jelmer
=== c0p3rn1c is now known as copernic
dpmEsatYuce, I'd recommend you to ask at #ubuntu-translators, but in general, you might have to get in touch with the developers of the package you have translated11:12
pittitjaalton, tseliot, bryce: FYI, I documented the udev X.org bits on https://wiki.ubuntu.com/X/InputConfiguration11:23
mdzanyone else see this in current lucid?11:23
mdzrm: cannot remove `/tmp/mkinitramfs_cLvNRp/lib/modules/2.6.32-8-generic/modules.*map': No such file or directory11:23
tseliotpitti: it looks good, thanks for working on this11:23
pittitjaalton, tseliot, bryce: (I also linked it from /X)11:25
tseliotpitti: ah, for quick access from X, nice. Thanks again11:25
persiapitti: You report that joysticks have ID_INPUT_KEY : is this for joysticks that actually include a keyboard map, or ones that just provide buttons as well?11:27
tjaaltonpitti: thanks. it's possible that device configuration will not be allowed from the backend, but moved back to the ddx11:27
tjaaltonpitti: even for lucid11:27
tjaaltonpatches upstream11:27
pittipersia: hm, let me check11:27
pittipersia: right, current input_id indeed does't assign _KEYS, fixing11:28
pittitjaalton: would that mean that it stops using udev altogether then?11:28
persiapitti: Well, that was an accident on my part :)  is there another resource I should look at that for deeper grounding (or would that be source)?11:29
tjaaltonpitti: no, it just wouldn't be possible to pass x11_options from it. the xorg.conf format will become far more flexible to handle this11:30
pittipersia: well, I wrote input_id, so if you have a good reason why ID_INPUT_JOYSTICK (and the like) should also have _KEYS=1, I'm happy to add it11:30
tjaaltonpitti: and also support for xorg.conf.d11:30
slangasekmdz: yes, Keybuk said he was going to fix that today11:30
pittitjaalton: ah, I see; well, the doc can always be updated then11:30
tjaaltonpitti: which packages can use to ship conf snippets11:30
mdzslangasek, ok, thanks (and what on earth are you doing awake?)11:30
tjaaltonpitti: of course11:30
slangasekmdz: I work odd hours on Thursday in order to stay in touch with Europe :)11:30
persiapitti: For the general case, they shouldn't.  Oddities like the Saitek Cyborg Command Unit probably should.11:31
persia(where the joystick is just asking to be remapped to send keyboard events)11:31
pittipersia: if the device event mask claims that it can produce EV_KEY, input_id will report that11:33
persiaAh, that makes sense.  OK.11:33
* persia will go read the input_id source this weekend to actually understand11:33
pittiit by and large just tests the bitmasks in /sys/class/input/*/capabilities/{ev,abs,rel,key}11:34
pittiusing the flags in /usr/include/linux/input.h11:34
tjaaltonpitti: there are bugs where evdev grabs joysticks, and that makes the cursor jump to the middle of the screen when the joystick is touched11:34
pittibut this bit mask testing is horrible/impossible in pure udev rules, which is why we need the callout11:34
pittitjaalton: right, I have that when I attach my joystick11:35
persiatjaalton: actually, that depends on the joystick: not all of them are either absolute or autocentering.11:35
tjaaltonpitti: it also happens for hw accelerometers, so tap your laptop.. :)11:35
tjaaltonpersia: ah, ok11:35
pittitjaalton: I was never quite sure whether plugging in a joystick is supposed to behave like a mouse, or whether that was an unintended side effect11:36
tjaaltonpitti: I'd say not. there's a -joystick driver for that11:36
pittiwe can of course easily ignore joysticks as X input devices11:36
persiaThere are definitely two schools of thought by users.11:36
tjaaltonyes, but the last time this happened with hal, there were a bunch of users upset :)11:36
tseliotjames_w: can you join #ubuntu-x, please?11:36
pittitjaalton: ... which we don't install by default?11:37
persiaI think there should be an (optional) way to enable/disable that easily.  Used to be xf86-input-joystick11:37
tjaaltonpitti: right11:37
pittitjaalton: so xserver-xorg-input-joystick should ship an udev rule which sets ENV{ID_INPUT_JOYSTICK}==1, ENV{x11_driver}="joystick"11:37
tjaaltonpitti: that's already in git, just not released yet11:37
pittiah, sweet11:37
persiaA related question: with HAL it was possible to remap stuff, so for instance one could enable a fake mouse somewhere on one of the monster 15-axis 40-button joysticks and still use the joystick in games.11:38
persiaIs there a way to do that without HAL?  Should descriptor files be shipped somewhere?11:39
persia(e.g. in the x52pro package)11:39
persiaOr does that now require custom input drivers?11:40
tjaaltonhal was just the backend11:40
pittipersia: hal itself didn't really add new devices (and wasn't able to)11:41
tjaaltonif it was possible with x11_options, it'll still be possible11:41
pittipersia: you should be able to set configuration in udev just as well as in hal11:41
persiaOK, so that's just constructing the right ENV(x11_options.foo) entries?11:41
persiae.g. ENV(x11_options.event_key_remap)="464=118" and the like?11:44
pittipersia: if that worked in hal, it'll work with udev rules11:45
pittiit's just the messenger11:45
pittipersia: above wiki page has a migration guide11:45
persiaCool.  Thanks.11:45
* persia was reading that, but hadn't looked at input stuff much since intrepid or so, and was catching up slowly11:45
=== ara__ is now known as ara
LucidFoxseb128, regarding xchat-gnome - my patch for the git version is in the upstream bug report.12:58
seb128LucidFox, the current patch still change .glade files there12:59
seb128LucidFox, they named the files .glade rather than .ui?13:00
seb128seems buggy...13:00
LucidFoxYes - the git version uses these .glade files.13:01
seb128ok, I will check that later13:01
seb128that seems wrong13:02
seb128gtkbuilder files are named .ui usually13:02
LucidFoxActually... the git version seems to still use libglade. o_O13:03
seb128weird, http://git.gnome.org/cgit/xchat-gnome/commit/?id=8ba28cb15d6bb365891a495130aed49145a7a2b513:04
LucidFoxOnly the preferences dialog is converted to GtkBuilder.13:04
seb128I think they rather didn't rename their .glade to .ui13:04
LucidFoxThe rest still use libglade13:04
seb128oh ok13:04
LucidFoxyet even for the preferences dialog. they didn't rename .glade to .ui.13:05
=== ikonia_ is now known as ikonia
loolpersia: Latest binary package built wins13:35
loolLast one to build that is13:36
persiaHighest version wins.13:36
persia(for those seeking context, see #launchpad)13:36
qenseIn the upstream conversation about bug 487937 it became clear that Nautilus already has PackageKit integration for unknown file formats, Fedora 11 advertised it in its release notes. Now I'm not sure how Nautilus' PackageKit support in Ubuntu is. It's disabled, isn't it?13:39
ubottuLaunchpad bug 487937 in hundredpapercuts "open up of compressed filetypes should request to install proper extractor" [Undecided,Confirmed] https://launchpad.net/bugs/48793713:39
seb128qense, we don't use packagekit but have a nautilus patch using gnome-app-install for installing things too...13:46
seb128qense, the issue is not a nautilus one though13:47
qenseit is a share-mime-info issue?13:47
seb128it's likely that file-roller claims handling those in its desktop entry13:47
seb128well I would guess that file-roller is used13:47
seb128but it fails13:47
qensethe error message just says no suitable application can be found13:48
seb128there is a bug on file-roller asking for smart install already though13:48
qensereally? I couldn't find any13:48
seb128qense, bug #14808413:48
ubottuLaunchpad bug 148084 in file-roller "totem-like/firefox-like plugin installer for file-roller" [Wishlist,Triaged] https://launchpad.net/bugs/14808413:48
qenseah, I see13:49
qensebut that's for inside file-roller13:49
seb128qense, but if file-roller is not open it might be because the mimetype is not listed13:49
seb128what mimetype are those which give you the error using?13:49
qenseI can confirm it with CAB, which is in packages/freedesktop.xml -- a file of shared-mime-info, the only .xml one -- the reporter talked about lzma and 7zip.13:50
seb128qense, cab = application/vnd.ms-cab-compressed?13:53
seb128qense, no .desktop lists those in their .desktop in /usr/share/applications13:53
seb128nautilus can't guess from nowhere what application could be able to open those13:54
seb128the .desktop need to list the mimetype for that to work13:54
seb128lzma and 7z should work13:54
qensea lot of ahs here ;)13:54
seb128since those are listed in the file-roller .desktop list13:54
qenseI see CAB isn'tin .desktop indeed13:55
seb128I'm not sure file-roller handle those13:55
seb128I'm not sure we have any gui to handle those13:55
seb128it's a somewhat specific format13:55
qensemust have been my memory playing a trick on me, confusing a terminal command with a GUI action13:56
* persia comments that it's possible for plugin packages to add special mime-handler-only .desktop files that end up calling into the executable into which they plug if that helps with non-default formats13:56
seb128it's listing application/x-cabinet though13:56
qensethat's strange13:57
seb128could be a .desktop bug13:57
qenseor an inconsistency from MS13:57
seb128in any case it's not a nautilus issue13:57
seb128it just matches mimetype and .desktop lists13:57
qenseI'll check if application/vnd.ms-cab-compressed can be processed the same as application/x-cabinet and mark the LP bug as a dup of the bug you named. The upstream report could be turned into a bug about the .cab archive handling.13:59
qenseThanks for your help, seb128! (and for your suggestion, persia)13:59
MezHmm, it seems my automount stuff has stopped working in karmic.14:01
MezIs this a known issue?14:01
seb128Mez, no14:01
seb128qense, you're welcome14:01
Mezseb128: any idea of how to "debug"14:02
seb128use ubuntu-bug storage14:02
seb128it will get gvfs, devicekit-disks etc logs14:03
Mez"Package storage does not exist"14:03
seb128install apport-symptoms14:04
* Mez waits for upgrade to finish14:05
Mezhmm... new apport version ?14:05
Mezseb128: #36906714:12
seb128bug #36906714:12
ubottuLaunchpad bug 369067 in pidgin "pidgin crashed with SIGSEGV" [Medium,Invalid] https://launchpad.net/bugs/36906714:12
seb128Mez, doesn't seem right14:12
Mezbug #3690674914:12
ubottuError: Launchpad bug 36906749 could not be found14:12
Mezah, crud14:13
Mezbug #49777114:13
ubottuLaunchpad bug 497771 in devicekit-disks "Discs not automounting in Karmic." [Undecided,New] https://launchpad.net/bugs/49777114:13
=== rmcbride_ is now known as rmcbride
Mezhmm, looks - sane enough to me.14:14
seb128I think it's a dup14:15
seb128the device has no type14:15
Mezseb128: a reboot seems to have fixed this.14:19
seb128is that karmic uptodate?14:20
Mezmaybe some process got killed off at a point i hit high memory or something14:20
seb128there was a fd leak in udev earlier in the cycle14:20
Mezyeah, up to date.14:20
seb128but that got fixed in a stable update14:20
Mezoh, when?14:20
seb128that's fixed for a least one month14:20
Mezoh, it's been updated at the end of last week at least14:20
seb128some similar bugs are due to devicekit processes crashing too14:20
seb128so it might be one of those14:21
seb128could you comment on the bug saying that a reboot fixed it?14:21
Mezalready in the process of doing so14:21
seb128thank you14:21
Mezworth marking as invalid too?14:22
seb128you can yes or you can wait for pitti to triage it14:22
seb128I expect he will need some extra log though14:22
seb128so if you can't get the issue now...14:22
Mezwell, it's working now, so I won't really be able to reproduce14:22
seb128close it and reopen if you get it again14:22
* Mez has no access to14:23
seb128and maybe check your system logs for asserts or crashes14:23
Mezwait, wtf14:23
MezI'm part of bug control14:23
seb128you should be able to close your own bug in any case14:23
Mezyeah, missed the "double login" stuff :)14:23
Mezdamn edge :D14:24
Mezafternoon sabdfl15:17
sivangsabdfl is here ? :)15:17
Mezhis IRC nick is :)15:17
=== kklimonda_ is now known as kklimonda
sabdflhello all15:18
=== Lutin is now known as Guest3058
=== Caesar_ is now known as Caesar
hwildeanybody know what is holding up libjsw for 9.10?15:51
persiahwilde: How do you mean "holding up".15:57
slangasekit's turtles all the way down15:58
=== Guest3058 is now known as Lutin
hwildepersia, everything else seems to work but no joystick support so what are we waiting for15:58
persiahwilde: Ah.  Well, since about 7.10 joystick support was *less good* with libjsw installed.15:59
hwildebut we need jscalibrator15:59
persiaUm no.15:59
persiaThat just broke stuff.15:59
ebroderHey persia - could you add me to u-u-s?15:59
hwildepersia, jscalibrator is the only way the logitech rumble pad works15:59
hwilde(the playstation looking controller)16:00
persiahwilde: Anyway, join me in #ubuntu, and I'll talk to you about joysticks :)16:00
ionIt’s teenage mutant ninja turtles all the way down.16:00
pittiseb128, Mez: I'll have a look16:18
=== dendrobates is now known as dendro-afk
jelmer_beuno, ping16:24
=== dendro-afk is now known as dendrobates
=== dendrobates is now known as dendro-afk
kirklandpitti: hmm, looks like something just regressed in the burndown-chart scripts16:34
kirklandpitti: i had some items marked as DONE, that were picked as DONE by your scripts, but now, they're back in the TODO range16:34
pittikirkland: do you have bugs linked to blueprints?16:35
pittikirkland: today we landed a feature to use bugs as work items16:35
pittiwhich are now counted as todo (open) or done (fix released)16:36
kirklandpitti: ah, yeah, i have a few in "fix committed"16:36
kirklandpitti: okay, thanks, that's it16:36
pittikirkland: would that be a plausible cause?16:36
kirklandpitti: i think so16:36
pittikirkland: so you should probably just remove the WIs from the whiteboard which duplicate the bugs16:36
kirklandpitti: cool, i like the feature16:36
kirklandpitti: i'd like each work item to be a bug, actually ;-)16:37
pittikirkland: https://wiki.ubuntu.com/WorkItemsHowto#Work%20items%20as%20linked%20bugs16:37
pittikirkland: well, for some it's great (e. g. "MIR for foo"), for some not so ("test X", "write documentation")16:37
kirklandpitti: meh :-)16:38
pittikirkland: but the whiteboard stuff won't go away, so it's just another option now16:38
kirklandpitti: i like that bugs have priorities, state, owners, history, etc.16:38
kirklandpitti: cool, thanks16:38
=== dendro-afk is now known as dendrobates
keespitti: so, if we have a workitem that reads  "support backingstore (LP: #470636)"  this is now counted twice?16:44
pittikees: if you linked that bug to the BP, yes16:44
pittithis seems to confuse enough people that this warrants an announcement16:45
* pitti mails platform16:45
keeshow are bugs linked to bps?  :P16:45
keesbecuase our todo chart jumped a lot more than just 1 item.16:45
smoserKeybuk, do you a minute for a mountall question?16:46
Keybuksmoser: sure16:46
smoseron ec2, the root= is going to say 'root=/dev/sda1'16:46
keespitti: or at least those bugs should be explicitly listed in the "Status by work item" section16:46
smoseri'd like to label the filesystem and use "LABEL=" in /etc/fstab16:47
Keybuksmoser: do you have an initramfs?16:47
joaopintodtchen, without apparent reasons my sound got muted, I stated getting flood with the volume muted notificatoin and PA was using 100% of one of the cores, is there a PA specific log I can check ?16:47
pittikees: they should, in the database it's all the same16:47
Keybuk(I would assume not, it doesn't make much sense)16:47
smoseri do not16:47
Keybukright, then you're basically limited ;)16:47
smoserwell, wait16:47
Keybukyou can put whatever you like in fstab16:48
Keybukbut it doesn't make any difference - since the kernel is what mounts the root <g>16:48
keespitti: how are they distinguished between "from a white board" and "from a bug" ?16:48
smoseri do when running in UEC, but not when running in ec216:48
pittikees: how do you mean? the DB doesn't care whether it comes from whiteboard, bug, or a wiki page16:48
smoserKeybuk, thats fine. i'm good with that.16:48
Keybuksmoser: only the options part of the fstab entry for the root are actually used16:48
Keybukie. what mount options to use, whether to fsck, etc.16:48
smoserbut will mountall figure out that my LABEL=myroot is the same as sda1 that was on the command line ?16:48
Keybuksmoser: mountall just ignores the LABEL=myroot16:49
keespitti: I'm trying to understand which items are "bugs", since we never planned to track work items with bug reports.  Our todo list jumped, and I want to eliminate all the non-whiteboard work items.16:49
Keybuksmoser: information present in mountinfo when mountall starts overrides all fstab files16:49
pittikees: just unlink them from the blueprint then16:49
keespitti: right now, everything is listed by bp name, so I can't tell which are bugs and which are bp whiteboard workitems.16:49
Keybukif in fstab you have /dev/sda2 / defaults 0 016:49
Keybukbut you boot with root=/dev/sda116:49
pittikees: oh, I see16:49
keespitti: I don't know which are the bugs to do the unlinking.  :P16:49
pittislangasek: hm, how did you solve this?16:49
beunojelmer, pong16:50
Keybukmountall will see "/dev/sda1" as the device name for /16:50
smoserKeybuk, ok. i think thats actually what i want.16:50
pittislangasek: i. e. do you have an use case for linking a bug which isn't a WI?16:50
Keybuk(but remount -o rw,defaults and not check because that's what fstab says)16:50
pittislangasek, kees: I guess it might help if the WI description includes the bug #16:50
smoseri want to trust cmdline16:50
keespitti: yes, that would help.  a prefix maybe.    LP: #nnnnn: [workitem text]16:51
Keybuksmoser: you'll hit a bug with 2.6.32 of course16:51
Keybukmountall will fail because it can't mount /dev/pts or /dev/shm16:51
Keybukdue to the underlying same problem <g>16:51
Keybuk2.6.32 will *also* mount a devtmpfs filesystem on /dev16:51
smoserthe kernel does that?16:51
Keybuk(tangent: kernel really should mount /proc and /sys automatically too - I think there even patches for it)16:51
Keybukif you don't use an initramfs, the kernel gives you devtmpfs on /dev now16:52
Keybukwhich is good16:52
smoserwow. i didn't realize that.16:52
Keybukbut mountall has a bug where it doesn't mkdir the mountpoint in that situation16:52
smoserso, next question16:52
Keybukso it'll fail to mount /dev/pts and /dev/shm because the empty directories don't exist in the devtmpfs16:52
smoserif i want to modify /etc/fstab during boot (add entries)16:52
jelmerbeuno: hi16:52
smoserwill mountall "just notice" ? does it watch /etc/fstab  ?16:52
jelmerbeuno: I was talking to somebody about gtk imports yesterday. Do you remember who that was?16:52
Keybuk(mountall behaves the same way as root here; kernel put devtmpfs on /dev, so it obeys the kernel - overriding its internal fstab)16:52
jelmerbeuno: FWIW I've fixed the import.16:53
Keybuksmoser: it will not notice16:53
smoserok. thats fine.16:53
Keybukthat would be insane16:53
smoserthats fine16:53
Keybukif you deleted a line of a mount point that mountall had already mounted, what should it do? :p16:53
smoserso i just add an entry and then issue a mount command for it.16:53
Keybukyou can do that16:53
smoseri agree, insane.16:53
Keybuk(and mountall will actually issue a mounted event for that too <g>)16:53
smoserthank you Keybuk16:53
beunojelmer, yes, it was jjardon16:53
Keybukbecause it *does* watch mountinfo16:54
smoseroh, that last tidbit is nice.16:54
Keybukso it'll see your mount command adding to mountinfo16:54
ionkeybuk: Random thoughts: hypothetically, there could be a number of simultaneous input/progress items, a number of simultaneous UIs and a tiny daemon that contains the logic to delegate everything between all of them. Plymouth with a plain virtual console fallback could be the initial UI, then gdm could take over (displaying any active items under the login area) and even the desktop session could have indicators/notifications for them.16:54
Keybukion: we settled on using Plymouth to do the delegation16:55
Keybuk(since it already does)16:55
pittikees: I need to leave in 5 (sorry), so please feel free to commit; should be a simple patch16:55
keespitti: ok, I'll test it16:55
ionkeybuk: How about gdm/desktop interaction? For instance, Upstart warn about jobs that failed to start, there would be a way to readd deferred fscks with the user being able to see the progress at all times etc.16:57
Keybukion: why should Upstart bother a user with such things?16:58
Keybukif the system can't recover from a failed service start by itself, my 8yo niece is also unlikely to be able to16:58
Keybukdeferred fsck -> we're not going to do this16:58
Keybukgdm won't start all the time filesystems are being checked16:59
Keybukany filesystem with passno>0 will hold up gdm16:59
ionYour niece would be immediately able to say “it says cups failed to start” and read the error message it printed just before dying straight from the screen instead of saying “printing doesn’t work” and taking 45 minutes of phone call time just to figure out the information. ;-)17:00
Keybukif there's a problem with printing, the message should go in the print dialog17:00
Keybuka popup that went past while she was concentrating onto facebook to harvest her crops is not going to stick in her mind17:01
Keybukyou could even do some fun adaptaive UI17:02
Keybuklike if the print server is down, the print icon on the toolbar has a line through it17:03
Keybukclicking it explains what went wrong, and gives you a button to try and restart it, etc.17:03
Keybukthen you could do things like check for an update, etc.17:03
ionSpeaking of plymouth, is it planned to implement graphics support for vga16fb, or is there just going to be a text fallback for the input/progress items?17:08
ionIn my VirtualBox VM with ~ubuntu-boot packages, any fscks seem to be cancelled almost immediately, probably since plymouth is unable to show anything.17:09
Keybukion: that is the plan17:14
Keybukit should be mostly just porting code from bogl to deal with the colourspace17:14
Keybuk. o O { if you get bored ... :p }17:14
ionkeybuk: Yeah, i was considering taking a look at it some day.17:15
Keybukion: obviously the plymouth framebuffer renderer is the appropriate one17:23
Keybuk(you may have to move drm.so out of the way, as there might be a bug with the fallback in our package :p)17:23
ftastgraber, do you have a fix for pastebinit to support openid?17:28
mathiazKeybuk: hi - I was trying to modify the hostname.conf upstart job with this: http://paste.ubuntu.com/343531/17:43
mathiazKeybuk: the goal is that I want to run the hostname.conf job when /etc/vm_config/ has been mounted17:44
mathiazKeybuk: it's a local block device17:44
Keybukyou're not uploading that to the real distribution, right? :p17:44
mathiazKeybuk: no :D17:44
Keybukthat won't work today17:45
Keybukbut it will work some point before alpha 217:45
mathiazKeybuk: ah ok.17:45
Keybukyou want "start on mounted" ;)17:45
mathiazKeybuk: that would work today?17:45
Keybukneither works today17:45
Keybukmount happened *before* the given mountpoint is mounted, and blocks it from mounting17:45
Keybukwhich is definitely not what you want :p17:46
Keybukin mountall 2.0, mount is replaced by mounting - and a new mounted gets added which is what you want17:46
mathiazKeybuk: great!17:46
mathiazKeybuk: thanks17:46
blackxoredhello, since lucid is lts and syncing from testing, I should wait my package to migrate to testing to get it into ubuntu, right?17:47
persiablackxored: Unless you're *really* confident and need it as a dependency for something else.17:48
blackxoredpersia, as a matter of fact, both apply for me :P17:48
mathiazKeybuk: would "start on (mount MOUNTPOINT=/ and net-device-up IFACE=eth0)" work today?17:50
mathiazsmoser: ^^ - I was looking into some ec2-config and run into the issue above17:52
blackxoredI suppose I'll wait17:52
blackxoredpersia, ^^^^17:52
ionkeybuk: Now that i think of it, perhaps we should only set iopriority instead of both it and niceness for low-priority fscks. The iopriority should prevent thrashing, and equal CPU priorities could utilize multiple processors better.17:53
smosermathiaz, it doesn't work today17:59
mathiazsmoser: right - I noticed that18:00
smoserneeds update to new mountall, and also needs to say "mounted MOUNTPOINT=/"18:00
smoserheres what i have right now for "early boot"18:00
smoserstart on (local-filesystems and net-device-up IFACE=eth0)18:00
smoseri believe thats generally functional, but i've been playing with a lot of workarounds18:01
kirklandwhat is tileblit and bitblit?18:01
kirklandthese are dragging in the fbcon module on the Lucid ubuntu-server installs18:01
kirklandwhich is definitely not desired (by me anyway)18:01
kirklandblacklisting isn't working either18:01
geserwhere do I file bugs against manpages.ubuntu.com?18:02
kirklandgeser: lp:ubuntu-manage-repository18:03
kirklandgeser: if it's a bug with the generation18:03
kirklandgeser: bugs against manpage content should go against the supplying package18:03
geserkirkland: it's a bug with converting the printf.3 manpage. it stumbles about the \0 inside the manpage18:04
kirklandgeser: okay, sure, file a bug18:04
kirklandgeser: send a patch, that would be great ;-)18:04
kirklandgeser: i don't spend much time working on manpages.ubuntu.com any more18:05
kirklandgeser: usually just a day or two per cycle, to get it working with the new code name18:05
kirklandgeser: and i'll scrub a few bugs while i'm at it18:05
kirklandgeser: fwiw, that has already happened for this cycle, so it might be a while18:05
geserno hurry, I just found this today and want to file it before I forget about it18:06
=== yofel_ is now known as yofel
geserjames_w: I was trying to do a bzr merge of perl (I've already submitted a debdiff for sponsoring) but got stuck as the debian branch is pretty old and perl is also listed at http://package-import.ubuntu.com/failures/.bzr/failures/. Do I need to do something about it?18:39
james_wgeser: that's a bug I need to fix18:40
geserjames_w: so I can only wait on someone looking at the u-m-s queue and sponsoring the debdiff?18:42
james_wgeser: unfortunately so18:42
geserno problem, I justed wanted to be sure with several processes getting updated this cycle18:44
ebroderCan anybody add me to ubuntu-universe-sponsors?19:03
geserpersia: ^^19:07
* persia fiddles with LP a bit19:08
mathiazkirkland: hey19:10
mathiazkirkland: is bridge networking working correctly on NCs?19:11
mathiazkirkland: this is on lucid19:11
persiaebroder: What's your LP ID?  LP is being annoying.19:11
kirklandmathiaz: i can't say yet19:13
kirklandmathiaz: i'm just getting the eucalyptus 1.6.2 packages tested now19:13
ebroderpersia: \o/ Thanks!19:13
ebroderDo new packages in Debian automatically sync to Ubuntu?19:40
ebroderOr do they need to be manually approved the first time?19:40
james_wit's a semi-automatic process19:41
james_wthe autosyncer pulls them in19:41
james_wbut that's not being run very frequently at the moment19:41
ebroderOk, sounds good19:42
kirklandis bzr pull from LP hanging for anyone else?20:05
* kirkland found the #launchpad topic20:09
mathiazkirkland: yop20:09
JontheEchidnaasac: ping20:32
Keybukmathiaz: no20:49
Keybukmathiaz: you'd block any filesystems from mounting until a network device came up20:50
Keybukion: sounds right to me20:50
Keybukion: I'm increasingly convinced *that* behaviour of Upstart is wrong too :p21:06
ionkeybuk: Sorry, what behavior of Upstart? It does priority manipulation?21:13
Keybukion: no, that "on foo and bar" blocks foo until bar happens21:13
Keybukit seemed like a good idea at the time21:13
Keybukit still does have useful effects21:13
Keybukbut it's susprising everyone :p21:14
sebnerrKeybuck: The magic of opensource, Hmm? hoia BTW :)21:14
ionkeybuk: Ah21:16
slangasekpitti: hmm, didn't solve that because I didn't realize it needed solving :)  Had thought about including the bug # in the WI description, but then thought it might be better to extend the database first to include other structure :)21:57
=== dendrobates is now known as dendro-afk
ebroderIf I have a pre-Karmic (site-local) package that's shipping an upstart event.d file, what's the best way to start it in the postinst? Just use initctl?22:14
=== Administrator is now known as Guest50176
nixternalanyone able to use bzr+lp right now?22:27
Guest50176i am using ver 8.04, is there a utility to check the health of the hard drive?22:28
persianixternal: /topic in #launchpad implies not22:28
persiaGuest50176: #ubuntu for support22:30
nixternalthanks persia22:33
superm1is bzr+ssh different than bzr+lp?22:36
superm1i didn't realize there was a bzr+lp22:36
LaserJoc1I thought they were the same thing, just bzr+lp does like the lp:<branch> stuff22:38
=== LaserJoc1 is now known as LaserJock
=== dendro-afk is now known as dendrobates
wgrantsuperm1: 'bzr+lp' doesn't exist. I suspect nixternal just meant bzr and LP.22:55
nixternalwgrant: it is a new feature silly, you didn't know about it? :p22:58
elmojames_w: ping23:01
LaserJockhmm, I swear I've seen bzr+lp URLs23:01
elmojames_w: (urgently)23:01
james_whi elmo23:01
LaserJockmaybe I've finally fallen of the deep end23:01
elmojames_w: can you look at jubany?  I think you're DOS-ing bzr.lp.net23:01
elmoit's OOMKed itself twice in as many hours, and you have a scary number of bzr processes on there; could be random coincidence23:02
james_wyeah, something looks odd23:02
james_wlots of hung ssh processes23:02
james_wI'll stop it and investigate23:02
sebnerrnixternal: another k3b update waiting for you :p23:02
sebnerrjames_w: buh a lot of discussion going on now about bzr :/23:06
james_whi sebnerr23:06
james_wis it talk like a pirate day?23:06
sebnerrjames_w: haha. nah I suppose I can't use my bouncer on my iPod23:08
sebnerrand _ is boring :p23:08
dtchenjoaopinto: user.log for starters, but you'll want to use wiki/PulseAudio/Log23:10
dtchenjoaopinto: that's normally a sign of a badly behaved client not the daemon itself.23:11
joaopintodtchen, ok, i will try to get it next time, i had a new event, but this time it was going to max sound23:11
dtchen(though of course there's a bug in that the daemon shouldn't belly-up)23:11
joaopintoI was wondering if could something realted to the volume changing keys instead23:12
joaopintosince it was providing the visual hint when you press the keys23:13
dtchenjoaopinto: are you using PULSE_NO_SIMD=1 ?23:15
dtchenjoaopinto: also, we should move this to +123:16
=== Administrator is now known as Guest68962
=== jldugger is now known as pwnguin
=== dendrobates is now known as dendro-afk

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