/srv/irclogs.ubuntu.com/2014/04/24/#ubuntu-devel.txt

=== vibhav is now known as Guest42515
pittiGood morning02:55
=== jamesh_ is now known as jamesh
pittiapw: hm, I don't remember this at all02:56
pittiogra_, apw: yes, systemd pid 1 mounts /proc and /sys, similarly to mountall in the upstart world02:56
infinitypitti: Hey.  You have some tasks with your name on them on NewReleaseCycleProcess02:56
infinitypitti: retracer, ddebs, autopkgtest...02:57
pittiogra_, apw: but not sure what these have to do with mtab -- /etc/mtab is a zombie which never quite dies sufficiently to get rid of it :/ but either way, at the time when we mount /proc and /sys it's definitively too early to even consider (or read) /etc/mtab02:57
pittiinfinity: hey02:57
pittiinfinity: yeah, finally :)02:57
pittiinfinity: did we already get any uploads to utopic? i. e. anything to catch up with on ddebs?03:04
infinitypitti: A few, yeah.03:04
infinitypitti: https://lists.canonical.com/archives/utopic-changes/2014-April/thread.html03:04
pittiinfinity: ddebs set up; cron job starts in 30 mins, I'll check ddebs when I start my day for real03:11
infinitypitti: Lovely.  I assume autopkgtest will be more effort.03:12
pittiinfinity: I'll deal with that with jibel this morning03:13
pittiinfinity: utopic retracers set up03:18
infinitypitti: Dude, I just realized what time it is.  What sort of crazy person wakes up at 5am?03:41
pittiinfinity: heh, you of all people complain about weird working hours? :-)03:42
pittiI woke up at 4, couldn't sleep any more, so I thought I could just as well do something useful03:43
pittiand maybe get tired over it :)03:43
infinitypitti: Ahh, kay.  *That* happens to me all the time.  I was afraid this might have been an intentional wakeup, and was going to send nice men in pretty white coats to take you away.03:43
pittiheh; I'd give a lot for being able to sleep through a whole night; I've had some trouble with that for a year or so03:44
infinitypitti: I feel your pain.03:45
infinitypitti: I make it through maybe one night a week, if I'm lucky.03:45
pittiurgh03:45
infinity(And it usually seems to be when I actually needed to be up early...)03:46
infinityBasically, I can't win. :P03:46
dholbachgood morning06:35
MacSlowhey dholbach06:35
dholbachhey MacSlow06:35
ari-tczewhello06:44
dholbachhi ari-tczew06:55
alebHow can I become a maintainer of the Launchpad package I'm developing (upstream)?07:29
mlankhorstwho's the current maintainer?07:38
aleb"registry" https://launchpad.net/pitivi07:41
alebI guess nobody07:41
jameshaleb: ask on #launchpad07:42
jameshaleb: wgrant or cprov should be able to help07:42
ogra_pitti, mtab gets linked to /proc/mounts very very early in the initrd (pretty much the first thing the /init script does there)08:24
pittiogra_: oh, do we do this now? niiic!08:24
pitti/etc/mtab -> /proc/mounts08:25
ogra_i wonder if we couldnt just leave it that way instead of having something create a file08:25
pittiI have it on my desktop, but I have a suspicion that's systemd, not ubuntu standard08:25
ogra_iirc there were reasons to still have it08:25
pittiI guess to keep track of non-kernel mount options?08:25
ogra_like tools parsing it and expecting a certain fromat that /proc/mounts doesnt give us08:25
pittiit's mostly the same format, but /proc/mounts doesn't retain userspace options08:25
* ogra_ would love if we could just keep it a link now08:26
pittime to08:26
pittitoo08:26
pitti/etc/mtab has been a horrible horrible misconception08:26
ogra_right, therewas info missing, but we might need to fix some external tools to not regress08:26
ogra_(if they still expect a certain format)08:27
pittiogra_: what's your /etc/mtab on your desktop?08:28
ogra_its a file ... i'm on precise08:29
pittioh, wow08:29
ogra_i'm pretty sure its a file on trusty too08:29
ogra_(not on the phone weher mountall is largely a no-op (all mounting happens in initrd there)08:29
pittiok, so I guess systemd sets that up08:30
ogra_is it a link under systemd?08:30
pittiyes08:30
ogra_cool08:30
* pitti tries something08:31
pittiok, that works08:31
pittiuhelper=udisks208:31
ogra_i think we can rely on tools to be fixed then08:31
pittithat's an userspace mount option that udisks uses so that you can call "umount" as user on auto-mounted drives08:31
pittiand that works08:31
ogra_great08:32
pittiand that's the bit which is *not* in /proc/mounts08:32
pittibut "mount" does show it08:32
pittiso it keeps track of those someplace else now08:32
ogra_hmm08:32
pittihah! /run/mount/utab08:32
pittiSRC=/dev/sdb1 TARGET=/media/martin/PittiUSB ROOT=/ OPTS=uhelper=udisks208:33
pittiso util-linux had this fixed ages ago08:33
pittinot quite surprising, given how long Fedora etc. have run with /etc/mtab -> /proc/mounts08:33
ogra_heh08:33
pittiso, user mount options: check08:33
pittithat was the only reason AFAIK why we still needed that08:33
ogra_:)08:33
=== mbiebl_ is now known as mbiebl
jameshmardy: could I pick your brain about online accounts again?09:35
xnoxogra_: on the phone most mounting is done by mountall =) like the gazillions of bind mounts.09:40
ogra_xnox, nope09:40
ogra_its all done in the initrd09:41
ogra_oh, wait, you are right, the binds are done by mountall09:41
ogra_the actual mounts arent09:41
xnoxpitti: i think we do need newer util-linux, we've been staggering behind for a while now....09:42
xnoxlamont: infinity: what's the status about getting newer util-linux?09:42
pittixnox: we don't need it for this particular issue, but having a non-ancient one would be good nevertheless09:42
mardyjamesh: hi, sure09:42
* xnox needs to check the partman/d-i bugs about it.09:42
jameshmardy: I put together .provider and .service files for SoundCloud via the desktop.  I put together this short test program to try and retrieve the access token, but it seems to log me out again: http://bazaar.launchpad.net/~jamesh/unity-scope-soundcloud/go-port/view/head:/accounts/accounts.c09:44
pittixnox: there are some attempts to moving to a newer version by some community members, but as it doesn't have broken-out patches it's quite a mess I'm afraid09:44
xnoxpitti: ok. imho everything should partse mountinfo, not mtab / mounts. but i guess loads of things will never learn.09:44
jameshI get the error "GDBus.Error:com.google.code.AccountsSSO.SingleSignOn.Error.UserInteraction: userActionFinished error: 10" and a notification bubble saying I've been logged out09:44
pittixnox: we certainly should keep the /etc/mtab symlink around for some time, yes09:44
xnoxpitti: yeap, that should be fine.09:45
jameshI used the signon-ui logging environment variables you gave me before to see that an access token was retrieved correctly, so I would have thought it would just be provided to me directly09:45
xnoxpitti: how is lightdm/plymouth integration with systemd? do i need to steal/port units over?09:45
pittixnox: I have lightdm with a systemd unit in my PPA (and soon in utopic); there is no plymouth integration ATM09:46
pittii. e. boot is just with a black screen until lightdm starts09:46
mardyjamesh: but then, if you go to the Online Accounts panel, you should see an exclamation mark next to the soundcloud account09:46
pittinot that I'd see it for a long time (it's just a second or so)09:46
mardyjamesh: which means it needs to be authenticated09:46
mardyjamesh: the reason is that we need to popup a UI in order to authenticate09:46
pittixnox: plymouth integration certainly sounds like one of the more/most complicated bits in that transition09:47
jameshmardy: yes.  I log in again in the control panel, rerun my program and get the same result09:47
mardyjamesh: OK, this is weird then. Can you enable the logging in /etc/signond.conf, try again (both from the control center and from your app) and send me the syslog?09:48
mardyjamesh: you might have to clear it from sensitive informations, maybe09:48
jameshmardy: okay.  I'll give that a shot09:48
xnoxpitti: also re:utopic-plans-of-running-system-upstart-as-pid2 do we want that at all? i was still unsure how to achieve that.09:49
pittixnox: I don't think we want it for anything packaged; just ensuring that for every upstart-only job that we have (i. e. no init.d script) we add an accompanying unit seems easier to me09:49
pittixnox: but I think during the UDS session we said we'd want it for custom upstart jobs written by the admin?09:50
xnoxpitti: right, have you seen https://code.launchpad.net/~upstart-devel/upstart/upstart-jobs ?09:51
xnoxpitti: that branch has all upstart jobs, all initd scripts and all unit files present in ubuntu at the moment.09:51
pittixnox: heh, no; is that somehow magically updated from all packages from the archive?09:52
pittihttp://bazaar.launchpad.net/~upstart-devel/upstart/upstart-jobs/view/head:/fetch-contents.mk09:52
pittinice09:52
pittixnox: I added a link to that to https://blueprints.launchpad.net/ubuntu/+spec/core-1403-systemd-transition09:53
pittihah! there, bluetooth fixed; it's quite nice, now it only starts up if you actually have a bluetooth adapter09:56
ogra_pitti, not so sure about that ... how does it know ?09:57
ogra_on the phone you might end up that you actually dont see all of the BT device09:58
pittiogra_: it doesn't start by default; but there is this rule in /lib/udev/rules.d/99-systemd.rules:09:58
pittiSUBSYSTEM=="bluetooth", TAG+="systemd", ENV{SYSTEMD_WANTS}+="bluetooth.target"09:58
ogra_ugh09:58
pittithen it starts bluetooth.target09:58
* ogra_ hopes that will still work with the hardware being handled inside the container 09:58
pittiogra_: how can bluez work on the phone if there are no devices?09:58
ogra_we have custom per-device BT upstart jobs atm09:59
pittiogra_: in the worst case we put it back on "always start" on the phone09:59
ogra_(thats about to change and the full logic will move into the container, on the ubuntu side only bluetootd should start)09:59
pittiogra_: custom jobs? on the phone you mean? because on teh desktop we just always start it09:59
ogra_yes, we run hciattach with the per device options atm ... but that will move into the container10:00
ogra_note that we suppress starting of udev too10:01
ogra_until after the container is up ...10:01
pittiogra_: in theory this should be fine; bluetooth needs to see a device to actually work; and once you see it, udev ought to see it as well10:01
pittiand as I said, if due to our weird hacks this is broken, we just enable it to always start10:01
ogra_so the events get used by udeventd (android) instead of udev10:01
pittiogra_: we don't run an udev coldplug?10:02
ogra_you cant "just always start" BT on a phone10:02
ogra_that might eat your battery ...10:02
pittiogra_: well, or transition whichever custom upstart jobs we have (I didn't look at them yet)10:02
pittibut binding this to "I have BT hardware" sounds like the right thing in general?10:02
ogra_pitti, udev starts in initrd ... then gets shot down, once the container starts ueventd grabs the events ... once the container is up we start udev normally10:03
ogra_pitti, lxc-android-config has all of the custom jobs in one place10:03
ogra_(well, everything container related at least)10:03
ogra_(sorry, i'm a little over-cautious ... )10:04
pittiogra_: hm, I see a few udev rules which change permissions of BT devices, but no custom jobs?10:05
pittiogra_: sure :) well, it's not like we'd dump that on the phone and don't check :)10:05
ogra_oh, for BT they are in bluetooth-touch10:05
ogra_i thought you wanted to see the udev side :)10:05
pittibluetooth-touch-mako.conf just has a few setprop calls10:06
ogra_yeah, looks at the others :)10:08
ogra_*look10:08
pittiogra_: I did, but they all just seem to fiddle with the killswitches or poke some bits into the android side, or load firmware10:09
ogra_really depends what chipset you have all devices with broadcom need to load some blob dynamically using brcm_patchram10:09
pittiogra_: I don't actually see any change to bluez startup?10:09
ogra_no, but to the device startup10:09
pittiyeah, that's fine10:09
ogra_if udev doesnt run at that time ueventd might steal the event10:10
pittiogra_: we need to do an udev coldplug run after starting udev (just like on the desktop/server if you have an initramfs)10:10
pittiotherwise a lot more stuff will be broken anyway, so I suspect we already do that10:11
ogra_hmm, i hope that doesnt slow us down ...10:11
pittiwell, we have a lot of udev rules there, including the ones in lxc-android-config10:11
pittithey won't actually run if we don't coldplug udev :)10:11
ogra_ok :)10:11
pittiso it would be very weird if we don't do that already; also for device permissions, creating persistent storage/usb/etc. links10:12
ogra_ah, right, that we do10:13
ogra_(device permissions)10:13
pittiogra_: yeah, they will also run during coldplug, as on the phone all the devices are already there when you start udev (i. e. no hotplugging)10:14
pittiogra_: so in theeeory, if we write unit counterparts for the bluetooth-touch upstart jobs, it should just work10:14
pittiin practice, it'll all break horribly initially, of course :)10:14
=== MacSlow is now known as MacSlow|lunch
=== alkisg is now known as work_alkisg
mptbdmurray, where are the “failed” colors used?12:03
mptI don’t know if I’ve ever seen them before.12:03
xnoxkenvandine: mardy: hey! i'm poking ubuntu-system-settings, online-accounts, on the unity7 desktop. When i open-up "add u1 account" i just see a blank screen =(12:22
kenvandinexnox, i don't think you should even see that on the desktop12:24
mardyxnox: you are probably missing the U1 plugin. Let me see how it's called...12:24
LaneyIt's greyed out for me12:25
xnoxkenvandine: i want to, though =)12:25
kenvandineoh... in uss, not unity-control-center :)12:25
pittidoko_, xnox: please ignore the utopic-adt failures for now; we are currently setting up autopkgtest for utopic, and will fix/retry these12:25
xnoxkenvandine: yeap, i want to see if qml/qt5 u1 works and try to somehow re-use it in the ucc with black magic.12:26
xnoxkenvandine: cause it's a pain to maintain python2-qt4-webkit on the dekstop images12:26
kenvandineindeed12:26
mardyxnox: it should just be account-plugin-ubuntuone12:26
xnoxmardy: not installed. installing. Strange that "u1" was listed in the "add-account view" even though i don't have it =/12:27
xnoxmardy: now it "just works" even without restarting uss =)12:27
mardyxnox: yep, for some reason /usr/share/accounts/providers/ubuntuone.provider is provided by ubuntuone-credentials-common instead12:27
xnoxmardy: excellent this works now, thanks.12:29
xnoxNow on to the black magic of getting it to work with ucc12:29
mardyxnox: yep, I'm afraid you are entering a minefield12:29
mardyxnox: ucc is Gtk-based, so you'd have to use XEMBED12:30
kenvandinehaha12:30
xnoxmardy: that was my plan, indeed.12:30
mardyxnox: which we are also using BTW in other plugins :-)12:30
xnoxmardy: oh really?! like which?12:30
seb128xnox, mardy: bug #1287640 is the "provider is installed in common which leads to buggy entries"12:30
ubottubug 1287640 in ubuntuone-credentials (Ubuntu) "UbuntuOne account plugin does not work" [Undecided,New] https://launchpad.net/bugs/128764012:30
xnoxmardy: or is it the other way around to get gtk plugins to show up in the qml app?12:30
=== MacSlow|lunch is now known as MacSlow
mardyxnox: most of them, except the empathy ones. Basically all those which are authenticated via OAuth (facebook, google, flickr, twitter)12:31
xnoxmardy: cool.... and u1 is not using OAuth because *reasons*12:31
mardyxnox: because U1 is using OAuth 1.0 (not 1.0a or 2.0), or a variant of it12:32
mardyxnox: indeed, if it used OAuth, making it work would be just a matter of setting the right data in the provider XML file12:33
mardyxnox: I had this discussion several time with the U1 folks, but the idea was that migrating to a standard OAuth would be too much work12:33
mardyxnox: your starting point should probably be http://bazaar.launchpad.net/~online-accounts/gnome-control-center-signon/trunk/files/head:/libaccount-plugin/12:34
mardyxnox: the oauth-plugin.c file implements a subclass of ApPlugin, which authenticates via OAuth by embedding the window which will be created by signon-ui12:35
xnoxmardy: but not implementing 1.0a means that both U1 and Launchpad are vulnerable to the session-fixaction attack.12:36
* mardy googles, never heard of that thingie12:36
xnoxmardy: and i have managed to complete oauth authentication with 1.0a client against launchpad, simply specify any string as verifier.12:37
xnoxmardy: http://oauth.net/advisories/2009-1/12:37
xnoxmardy: essentially to exchange request-token for an access-token, one should supply "oauth_verifier" which is returned to the user upon authorizing request (and/or in the callback url)12:39
xnoxmardy: this mirrors the challenge - to verify that the same client that authorized access, exchanges the request-token.12:41
mardyxnox: please get in touch with the U1 guys (dobey?), they'll certainly know more12:41
mjtHello.  Is it okay to add a package to Suggests: if it is only available in universe?  I remember it is not good to Recommend a package from universe, but how about Suggests?12:41
xnoxmjt: suggests is fair game.12:42
xnoxmjt: we drop things from recommends to suggests all the time, when we want to keep certain stacks in universe, and not get pulled into main.12:42
xnoxmjt: main is closed set of (build-depends, depends, recommends)12:42
mjtso basically, one can use Suggests for everything, even for something not in archive? :)12:43
xnoxmjt: it's against debian policy to suggest something that is not in the archive. Or suggest non-free things in main.12:43
mjthm. Is it #ubuntu-devel, or #debian-devel? :)12:44
xnoxmjt: so if you do that, you'll get an RC bug in debian. But in ubuntu we don't have a concept of RC bugs, so although we have reports we don't currently have people actively investing time to enforce that part of debian policy.12:44
mjtspeaking of non-free things, suggests: is the only way to have, say, firmware12:45
xnoxmjt: surprisingly, debian-policy is still a requirement in the ubuntu, sans were we explicitly diverged / applied changes.12:45
mjtsure, hence the smile at the end of my statement12:45
xnoxmjt: firmware in ubuntu, can go into multiverse and/or restricted, since it's hardware-dependent blobs. And in ubuntu one can suggest those.12:46
=== _salem is now known as salem_
xnoxmardy: looking at account-plugins -> all of them are oauth2.0. is there support for 1.0 at all?12:53
mardyxnox: flickr and twitter use 1.0a12:53
cjwatsonpitti: Are you able to set up autopkgtest for utopic?  Would be nice to get results for the runs that were triggered by binutils13:00
pitticjwatson: jibel and I are working on that furiously :)13:00
pitticjwatson: http://d-jenkins.ubuntu-ci:8080/view/Utopic/view/AutoPkgTest/13:00
pitticjwatson: still some fallout, I'm on that13:00
pittilike missing ddebs indexes for utopic-updates/security (just fixed), and some autopkgtest bugs13:00
pitticjwatson: jibel sent an RT to publish that to the public jenkins, too13:01
cjwatsonpitti: ah, right, thanks13:03
pittiman, everything related to the DC feels like a tarpit today; must be release time :)13:05
xnoxmardy: right, so it looks like canonical-sso and launchpad-oauth are two different beasts. launchpad-oauth uses standard 1.0 way to request token, whereas canonical-sso does not.13:12
=== bfiller_afk is now known as bfiller
dobeyxnox: launchpad doesn't use standard oauth 1.0 so much13:16
dobeyxnox: and the u1 plug-in isn't using the oauth plug-in, it's a custom plug-in that talks to the sso v2 api13:16
dobeyxnox: the gtk+ online-accounts UI requires providing a plug-in written in gtk+ for UI, and we only have a qml UI, so it doesn't work under unity-control-center; you have to use system-settings to add/manage a u1 account in uoa13:18
bdmurraympt: here - https://errors.ubuntu.com/retracers-results/13:18
xnoxdobey: under https://launchpad.net/~xnox/+oauth-tokens one totally uses bog standard oauth 1.0 with https://launchpad.net/+request-token , https://launchpad.net/+authorize-token, https://launchpad.net/+access-token OAuth 1.0 endpoints.13:19
xnoxdobey: e.g. launchpadlib and all scripts that drive launchpad use it (e.g. syncpackage, requestsync, import-bug-from-debian, etc.)13:19
xnoxdobey: where is sso v1 api docs?13:19
xnoxdobey: and how come sso v2 doesn't provide a normal oauth 1.0 api?13:20
xnoxdobey: or oauth 2.0 api...13:20
dobeyxnox: yes, but launchpad doesn't quite follow the oauth spec exactly. the oauth plug-in in UOA doesn't work with getting a token on launchpad.net (i made a plug-in to try it :)13:20
dobeyxnox: oauth api is separate from sso api13:20
dobeyxnox: sso just doesn't provide an oauth api at all13:20
xnoxdobey: is oauth api available on sso.... oh *damn*13:20
xnoxdobey: well, i may have a merge proposal against launchpad and python-requests-oauthlib to make it standard compliant ;-)13:21
dobeyxnox: i'm not sure where sso v1 api docs are; it's discouraged to use it13:21
xnoxdobey: at the moment +access-token page must use "body" signature type, instead of header. And i have merge proposal to fix that.13:22
dobeyxnox: the problem with launchpad mostly, is that it doesn't use a consumer key, iirc13:22
xnoxdobey: nah, that works just fine. One specifies any, with no secret. and launchpad just accepts any.13:22
xnoxdobey: do you have that UOA plugin around? i wonna try it against my launchpad dev instance.13:23
dobeyi don't know, i'll have to look13:23
xnoxideally canonical-sso would support oauth api for requesting a token.13:26
dobeyxnox: no need to convince me. i would ♥ oauth2 support13:27
xnoxdobey: no, no, no =) oauth1 api for requesting tokens ;-)13:27
xnoxdobey: oauth2 is *evil* =)13:27
dobeyno13:28
dobeyoauth1 is evil13:28
xnoxdobey: "When compared with OAuth 1.0, the 2.0 specification is more complex, less interoperable, less useful, more incomplete, and most importantly, less secure." http://hueniverse.com/2012/07/26/oauth-2-0-and-the-road-to-hell/13:28
xnoxdobey: a whole bunch of authors withdrew their names and association from oauth 2.0....13:29
xnoxdobey: oauth 2.0 has one benefit - easy client code.13:29
dobeyoauth 2 has lots of benefits13:30
dobeynot just easy client code13:30
dobeywell, oauth any version is evil; but oauth 2 is slight less evil for what we're doing13:31
=== arun is now known as arunpyasi
sergiusensxnox: I'm an oauth newbie, but are you looking for something like this? curl -H "Content-Type: application/json" -X POST -d '{"mail": "email", "password": "password", "token_name": "mytokenname", "otp": "2fa"}' https://login.ubuntu.com/api/v2/tokens/oauth13:41
xnoxsergiusens: yeah, i have reference to that. that's canonical-sso specific api to get u1 tokens.13:42
xnoxsergiusens: although it returns an oauth 1.0 token, that's not how oauth 1.0 spec mandates to create tokens.13:43
sergiusensxnox: yeah; sso isn't fully oauth compliant; you can talk to nessita or pindonga for details I guess13:44
dobeyxnox: it returns a token that is compatible with oauth 1.0; it is not an oauth token :)13:47
dobeyit's an sso token13:47
dobeyand signed requests are oauth compatible13:49
xnoxdobey: the devil is in the details =) fun13:49
xnoxdobey: duck-typing for the win =)))13:49
dobeyoauth 2 would be great though, so i can throw away 90% of the code in ubuntuone-credentials, and we can avoid having a consumer_secret on the client, and etc etc13:50
dobeyxnox: though, what are you trying to do? replace ubuntu-sso-client in 14.10 with the UOA stuff?13:51
xnoxdobey: yeah.13:51
xnoxdobey: at least such that login works, possibly without signup. I could refactor ubiquity's ubuntuone plugin into accounts plugin.13:52
dobeyxnox: maybe fix the UOA gtk+ UI to allow embedding the qml plug-ins, and not require gtk+ ones13:52
xnoxdobey: exploring that as well with xembed type of things.13:53
xnoxdobey: but that is kind of against mir strategy.... or does xembed work under mir? =/13:53
dobeyxnox: well, afaik the webkit view for the oauth plug-ins is just xembedded13:53
xnoxdobey: correct.13:53
dobeyxnox: are we going to keep having two separate system settings apps when we swtich to mir?13:54
dobeyxnox: or will we just use the ubuntu-system-settings then?13:54
xnoxdobey: no idea... i just want to kill python2-qt4-webkit on the desktop, with least amount of effort possible.13:54
dobeyxnox: port ubuntu-sso-client to qt5? :)13:55
dobeyxnox: fwiw, i don't think it's using anything that is no longer in qt5, but i don't really know where to begin with porting a python app from qt4 to qt513:55
dobeyxnox: alternatively, we could possibly bring back the gtk+ UI instead, and then kill the qt UI13:56
* dobey thinks the gtk+ UI was much nicer anyway13:57
xnoxdobey: there was gtk+ UI?! where?13:57
dobeyxnox: there used to be one yes. many releases ago13:57
xnoxdobey: ubuntu-sso-client porting to qt5 is not going to happen, because that needs python3 porting as well, which falls apart in twisted not twisting well under python3.13:58
xnoxdobey: do you at all recall which release? or like a package name or some such?13:58
dobeyxnox: i was wondering about the python3 thing. i saw mark's blog mentioned "finally dropping python2" but i don't see that happening unless we pull software-center out of the default install14:00
xnoxdobey: well.... lenses replace software-center, no?! =)14:01
xnoxdobey: but once ubuntuone stuff at the moment is the biggest python2 chunk on the desktop, which we have to sort out.14:01
dobeyxnox: scopes? sure, when we switch to unity8 and only support click installs :)14:01
dobeyxnox: right, software-center and ubuntu-sso-client is the sticking python2 bit, afaik (not sure if there's anything else using it still)14:02
xnoxdobey: oh right.14:02
xnoxdobey: and we have mvo back, and he can port software-center quickly =) or something-rather.14:02
xnoxdobey: there is myon qt store in kubuntu, we can use that.14:03
xnoxdobey: there is also kylin software-center in qt, which would fit on ubuntu desktop right in.14:03
dobeymuon you mean?14:03
dobeyhmm14:03
dobeydo those both use ubuntu-sso-client? or do they do their own login thing?14:04
cjwatsonpitti: autopkgtest> I see there's a testbed error of some kind on binutils/i38614:05
mvoxnox: heh :) wasn't the plan to use unity exclusively instead of s-c ?14:07
xnoxmvo: "the plan" is one thing, unicorns is a different thing =)14:08
xnoxdobey: muon uses ubuntu-sso-client, kylin doesn't use anything i believe.14:08
dobeykylin doesn't support ratings/reviews?14:09
xnoxdobey: hey support alternative archive with packages not from archive.ubuntu.com...14:10
dobeymvo: come back in two years, we de aren't installing debs any more :)14:10
dobeywtf i can't type14:10
dobeys/we de/when we/g14:10
mvodobey, xnox: heh :)14:11
xnoxdobey: http://goo.gl/vRCzkk14:12
dobeyxnox: u-s-c supports alternative archives too; that's what PPAs are14:12
xnoxdobey: you haven't seen kylin store yet, have you?! =)14:12
xnoxdobey: i am not quite sure why a new one was written, let me paste you a screenshot.14:13
dobeyno, but i'm also not in china :)14:13
xnoxdobey: http://people.canonical.com/~xnox/kylin-store.png14:14
mvoxnox: woah14:18
sergiusensthat looks... interesting14:19
dobeyomg14:19
dobeylol @ dota though14:20
* highvoltage hopes that's even in the archives but is afraid to ask14:20
xnoxhighvoltage: furthermore it installs packages not from the archive =)14:20
mvoit does *what*?14:23
highvoltagescary.14:24
highvoltageand weird considering that it's an official flavour?  are they still?14:24
dobeyxnox: where is it installing them from?14:25
mdeslaurthey have their own repo that was approved by the tech board14:25
dobeyoh14:25
dobeydoesn't ubuntu gnome do that as well, with a ppa?14:25
mptbdmurray, I suggest using exactly the same colors for “(failed)” as for “(by 12.04 standards)”14:26
pitticjwatson: yep, on my radar14:26
* highvoltage frowns at ubuntu kylin and ubuntu gnome14:26
pitticjwatson: it timed out, pretty much everything is dog slow today :/14:27
pittiI'll see how to avoid that14:27
mvomakes me wonder why their http is listening on port http://archive.ubuntukylin.com:10006 but well :)14:28
ogra_well, at least it has a roof ... so you wont get wet when it rains14:28
mvopfff :P14:28
dobeyanyway, that is too weird.14:30
xnoxhighvoltage: see tech-board resolutions, it's installing from a ppa mirror, thus packages are build in a ppa singed by a canonical key, which is stored in launchpad.14:32
xnoxmvo: yeah, 10006 was inquired about....14:32
highvoltagexnox: yeah but it just doesn't feel right, it's like it goes against the Ubuntu way (if that's even a thing)14:56
highvoltagexnox: for example, they have non-free software in their main repository: http://archive.ubuntukylin.com:10006/ubuntukylin/pool/main/w/wps-office/14:56
LaneyIt's supposed to be analogous to the Canonical partner archive14:57
NikThpitti: Hello14:57
highvoltageLaney: does the system make it even vaguely easy for the users to identify that they're installing non-free software when installing from that archive?14:58
NikThpitti: I think you should consider to update the systemd PPA for Utopic Unicorn. What's your thoughts ?14:59
Laneyhighvoltage: I can't read the language to tell what that software centre is saying, I'm afraid14:59
mdeslaurhighvoltage: does software-center?15:00
highvoltagemdeslaur: that's why I very carefully chose the word "vaguely", at least in software-center you have to enable the partner archive manually and you can read up on what it's about15:01
mdeslaurhighvoltage: well, the extras.ubuntu.com repo is enabled by default15:02
mdeslaurand there's non-free stuff in there15:02
LaneyThere sure shouldn't be15:04
=== doko_ is now known as doko
highvoltagemdeslaur: when did exrtas.ubuntu.com start accepting non-free packages?15:05
dobeyi thought extras went away15:05
dobeyoh i guess not15:05
mdeslaurwell, maybe I'm wrong, one sec15:06
dobeymdeslaur: i'd think if non-free stuff was in extras, the ARB wold be failing at its job, no?15:07
pittiNikTh: no, I won't -- I'll upload the PPA contents to utopic :)15:07
pittiNikTh: I tested it quite a bit today with just upstart, and I didn't see any breakage15:08
pittiNikTh: I'm now also running my workstation with systemd as pid 1, also working quite well; this exposes problems much better (I fixed bluetooth, lxc is broken, etc.)15:08
NikThpitti: Yes, you have fixed it I guess. I added the PPA again today and I did not notice the problem. I upgraded to Utopic yesterday and I added your PPA manually ;)15:10
pittiNikTh: yep, just keep the PPA source as "trusty", it'll work fine on utopic (the only real change is base-files so far..)15:10
mdeslaurhighvoltage, dobey: ah, sorry, the non-free software isn't installed from extras15:10
NikThpitti: so , is there a plan for systemd as default system management daemon in 14.10 ?15:11
NikThpitti: and how you made that change "as pid 1" ?15:12
pittiNikTh: yes, https://blueprints.launchpad.net/ubuntu/+spec/core-1403-systemd-transition15:12
pittiNikTh: as I wrote in the blog, set init=/lib/systemd/systemd in grub15:12
NikThpitti: Ok, I have done this already :)15:12
NikThpitti:  Now I want to test nvidia drivers , I had a problem on install. Installation should be done in upstart mode. I guess the packaging should be change, because it requires upstart.15:14
dobeymdeslaur: right. just partner and multiverse, no?15:14
pittiNikTh: please report such things and tag it with "systemd-boot", so that we have a nice list15:14
pittiNikTh: i. e. that it appears on https://bugs.launchpad.net/ubuntu/+bugs?field.tag=systemd-boot15:14
pittiNikTh: but please label the title with [systemd] to make it clear it doesn't affect default ubuntu15:15
NikThpitti: OK.15:15
directhex$ dpkg -c buildroot/pool/main/c/curl/curl-udeb_7.35.0-1ubuntu2_amd64.udeb15:15
directhexdrwxr-xr-x root/root         0 2014-04-01 18:44 ./15:15
directhexwhoops15:15
mdeslaurdobey: no, software-center installs specials PPAs for the non-free software that's in it15:16
dobeymdeslaur: oh, you mean for the "for purchase" apps?15:16
mdeslaurdobey: yeah, or the for purchase $0 apps15:16
dobeyright, yeah, those are all magical private PPAs15:16
dobeydoesn't matter if the app is "free software" or not15:17
mdeslaurright15:17
slangasekseb128: what makes you think bug #1311488 is a freetype issue, rather than a driver issue?15:38
ubottubug 1311488 in freetype (Ubuntu) "Missing letters " [Undecided,New] https://launchpad.net/bugs/131148815:38
infinity" The problem is intermittent - I just rebooted and there are15:39
infinityno missing letters."15:39
infinity^-- I'd be inclined to blame that on heat.15:39
infinityOverheating video cards do awesomely confusing things.15:40
infinityWell, heat or bad driver that corrupts video memory the same fun ways that heat does (the Intel driver has been infamous for this in the past)15:42
pitticjwatson: binutils is PASS now, so http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html is now only complaining on the freeze and some unblock request version mismatch15:43
pittijibel: hm, http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html has "lintian PASS", but it actually failed (on an actual error)15:44
pittiit's fine to let it propagate in that case, it's certainly not a binutils regression; but formally it should hold it back15:45
infinitypitti: Yeah, I'd like to sort out why we're getting passes for fails before we just let things go.15:45
infinitypitti: Pretty please. :P15:46
pittihm, and now on reload, lintian is gone altogether15:46
pittiinfinity: yeah, sounds like the same old sorting bug that jibel was working on15:46
infinitypitti: passes fall off the display after a bit, IIRC.15:46
pittiinfinity: they are not meant to, though15:46
infinity(Though, not sure why)15:46
infinitypitti: I've certainly seen that with eglibc in the past, where my list starts out with hundreds and ends up with only a few.15:47
pittiinfinity: right, but it's still not meant to fall off, AFAIK15:47
NikThpitti: https://bugs.launchpad.net/ubuntu/+source/nvidia-prime/+bug/1312255 (done) :)15:47
ubottuLaunchpad bug 1312255 in nvidia-prime (Ubuntu) "[systemd] nvidia-prime package failed to install due to missing upstart service" [Undecided,New]15:47
pittiNikTh: thanks15:48
infinitypitti: Kay, so.  Blocking opening on this is probably silly, if none of these bugs are regressions, but we really need to make fixing the autopkgtest<->britney interface a top priority.15:48
pittiinfinity: fully agreed15:48
infinitypitti: Since the whole point of that integration is lost if it doesn't work.15:48
infinity(see the two doko-induced breakages last cycle that should have been caught)15:49
infinityNot to pick on doko, he just happened to be the lucky one who tricked adt-britney twice. :P15:49
pittiit has happened way more often, just with less catastrophic results15:50
infinitypitti: Indeed, I don't doubt that for a second.15:50
infinitykirkland: Your vigpg.1 manpage appears to be for wifi-status...16:03
dokobarry, could you point mvo to the way how to install into the system dist-packages using a virtual env?16:14
stokachustgraber: ping16:17
stgraberstokachu: pong16:17
stokachustgraber: is it even remotely possible to access /dev/kvm for something like nova-compute running inside an lxc container?16:17
stokachunova-compute provides an lxc virt-type but i haven't researched that yet16:18
stgraberstokachu: have you been talking with zul? :)16:18
stokachustgraber: haha yea16:18
stokachustgraber: he's helping me with some openstack specifics for a project16:18
stgrabergetting the same question twice in 30s seemed like too much of a coincidence :)16:18
stokachuhaha16:18
stokachuzul: :)16:18
stgraberso as I was telling him, so long as the kvm module is loaded on the host, /dev/kvm should be accessible from the container by default16:19
stgraberI suspect you may have more problems with libvirt than kvm itself16:19
stokachuso /dev/kvm exists on the host16:19
stgraberI believe hallyn did get that all to work in the past, unfortunately he's not around this week16:19
stokachushould i install the virt specfici stuff16:19
stgraberstokachu: ah, if it doesn't exist in the container, just mknod it16:19
stokachuok lemme try that16:19
stokachustgraber: ok it added it and i got a little farther16:23
stokachuback to the charm stuff, thanks16:23
stgrabernp16:23
=== bfiller is now known as bfiller_afk
seb128slangasek, lack of knowledge of the font rendering stack?16:47
seb128slangasek, I though that those issues were more likely to be freetype/fontconfig problems16:47
slangasekseb128: ok, well when glyphs randomly go missing, I've only ever seen that be a video driver problem16:47
seb128but that was poor guess from my part maybe, so feel free to reassign where you see fit16:47
seb128k, good to know16:47
infinityslangasek: driver, or random memory corruption from heat (as I mentioned above).17:10
infinityslangasek: The user probably just needs to remove the gum from his fan. :P17:11
slangasekinfinity: I'll let tseliot sort that out17:11
=== roadmr is now known as roadmr_afk
=== roadmr_afk is now known as roadmr
=== sz0 is now known as sz0`
=== sz0` is now known as sz0
barryinfinity: are you SRU'ing a backport of debootstrap into trusty (for the utopic symlink)?17:55
infinitybarry: I hadn't prepped one but, yes, I should do so for all supported releases.18:00
barryinfinity: cool, thanks18:01
infinitybarry: (Though, doing it to lucid is a bit of an unfortunate lie, as you can't debootstrap trusty on lucid right now, so maybe I'll leave that one out)18:01
barryinfinity: i mostly only care about trusty anyway :)18:01
infinitys/right now/ever/ since debootstrap doesn't look at post-release pockets, so fixing the packages that aren't gzip in trusty base is something we can't do now. :/18:01
infinityOh, wait.  Yeah.  lucid is probably already a lie if someone added the trusty link.  My brain wasn't even thinking utopic.18:02
infinitybarry: I'll do utopic symlinks for precise->trusty now.18:02
barryinfinity: rock on18:02
* infinity curses himself for having let the gzip base thing fall off his TODO before release, since it's literally unfixable now.18:03
slangasekthe what?18:04
infinityslangasek: When dpkg started defaulting to xz, there was a rough concensus that packages in the debootstrap set should be compressed with gzip, which requires manual intervention.  Some were fixed, some were not.18:04
slangasekhmm18:05
infinityslangasek: The argument was so that debootstrap on other OSes would be less painful to use, but the other side effect is that lucid can't debootstrap trusty.18:05
slangasekah18:05
infinityslangasek: Anyhow, can't fix it now, without overhauling deboostrap to understand post-release pockets (perhaps a good thing to do anyway, but not something worth backporting to lucid), so just a bit of an "oh darn", I guess.18:09
slangasekyeah, not seeing the cost/benefit there coming out in our favor18:10
dokobarry, you told me that you were able to install with pip into the system location18:43
barrydoko: that was with system pip in a virtualenv18:45
barryi.e. if ensurepip's pip wasn't used18:45
dokobarry, so using system pip in a virtualenv lets you install packages into the system location, not into the virtualenv?18:47
barrydoko: right.  that's why we can't use system pip in an venv18:47
barrywe have to use ensurepip's pip18:47
barrydoko: also, https://bugs.launchpad.net/ubuntu/+source/python3.4/+bug/1290847/comments/1518:48
ubottuLaunchpad bug 1290847 in python3.4 (Ubuntu) "pyvenv fails due to mising ensurepip module" [Undecided,Confirmed]18:48
dokobut people *can* do that now, so we should disable that18:48
barrydoko: i think the most reasonable thing is to re-enable ensure pip, so that bundled pip gets installed in the venv.  then people will use that instead of trying to use system pip18:49
barryif they activate the vm18:50
barryif they don't, then they're not in a venv ;)18:50
dokobarry, sure, we have to build the wheels for that18:50
barrydoko: read dstufft's comment.  that's not going to work18:50
dokobarry, read the debian policy, including the wheels is not going to work18:51
barrydoko: we had a long discussion about that over in #d-p yesterday.  i'm not so sure it's a clear violation of policy, especially if dstufft gives us a script to rebuild the bundled whls, which he promised to do18:52
barrydoko: specifically, it probably doesn't violate $4.13 of policy18:53
barrythe whls are just zips of the source, and with dstufft's promised script, we can always recreate them if we need to18:54
barrydstufft makes a compelling argument for why rewheeling + unvendorizing is never going to work18:54
dokobarry, wheels are not data18:56
dokoand we have to remove the windows binaries, so source code included18:56
barrydoko: can we take this over to #d-p and pull dstufft in?18:57
=== bfiller_afk is now known as bfiller
=== roadmr_ is now known as roadmr
=== elopio_ is now known as elopio
infinitypitti: Did we get anywhere on adt-britney sadness?20:18
infinitypitti: Everything now seems to be stuck in RUNNING, which is just as confusing...20:18
=== sz0 is now known as sz0`
zygaxnox: are debian syncs now open again?20:43
infinityzyga: autosyncing will happen soonish.20:44
zygainfinity: thanks20:45
hggdhbdmurray: hi, you might want to have a look at https://bugs.launchpad.net/bugs/131189521:12
ubottuLaunchpad bug 1311895 in gettext (Ubuntu) "0.18.3.1 can fail when tracing due to missing files" [Medium,Triaged]21:12
hggdhbdmurray: it will only affect people trying to build coreutils from git source on 14.04. Not very usual, but might be bothersome when it happens21:13
=== salem_ is now known as _salem
infinityhggdh: Looks like https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=73032121:58
ubottuDebian bug 730321 in autopoint "autopoint: bootstrapping coreutils: autopoint from gettext 0.18.3.1 fails" [Important,Fixed]21:58
hggdhinfinity: indeed, thank you. I wil add the debian bug to the report22:00
infinityhggdh: http://paste.ubuntu.com/7325581/ <-- This was the Debian fix.22:01
infinityhggdh: The upstream fix ignores one more, so maybe that's better to pull in.22:01
hggdhinfinity: thanks. I will try to work on it this weekend22:03
infinityhggdh: I can do it right now, if you want to validate it.22:04
hggdhinfinity: perfect, for that I do have the time :-)22:04
infinityhggdh: Fix uploaded.22:11
infinitybdmurray: If you're around, want to review that gettext in the trusty queue so hggdh can go verification-happy?22:11
infinityhggdh: And uploaded to utopic as well (by way of a merge).22:28
hggdhinfinity: right now installing two trusty VMs, one trusty itself, and one that will go utopic22:32
infinityarges: *poke*22:33

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