[00:21] bah https://bugzilla.gnome.org/show_bug.cgi?id=607302 [00:21] Gnome bug 607302 in Plugins (other) "cover art search fails with 500 Client Refused" [Normal,Resolved: notabug] [00:22] "Discogs staff have confirmed that rhythmbox users were simply placing too much [00:22] load on their systems, so they had to block us. " [00:24] jcastro, do you know what banshee uses for coverart? [00:25] seb128: rhapsody->amazon->musicbrainz->somethingelse last I checked [00:25] I though amazon broke clients too? [00:25] it tries a bunch and snags the highest resolution one [00:25] I can find out [00:25] oh hey, gabaug is in here. [00:27] jcastro: the order is listed here: http://banshee-project.org/support/faq/ [00:27] thanks!! seb128 ^ [00:30] gabaug, jcastro: thanks [00:35] seb128 - this menu issue goes away if i delete the desktop file cache [00:35] chrisccoulson, weird [00:35] the cache is not by user [00:35] jcastro, gabaug: https://bugzilla.gnome.org/show_bug.cgi?id=593163 [00:35] Gnome bug 593163 in Plugins (other) "Rhythmbox failed to fetch covers" [Normal,Resolved: notabug] [00:36] yeah, that is pretty weird [00:36] jcastro, gabaug: this bug states [00:36] "http://developer.amazonwebservices.com/connect/ann.jspa?annID=483 [00:36] We can't really implement request signing since we'd have to distribute the [00:36] secret key to all users, so we can't use amazon's web services any more." [00:36] and the issue doesn't come back again after i recreate the cache [00:36] do you know what banshee is doing to use amazon? [00:36] seb128: for cover art? [00:36] gabaug, yes [00:37] seb128: we construct the URL based on the ASIN (which we get from MusicBrainz) [01:04] ArneGoetje, hi [01:15] rickspencer3: hi [07:12] asac: ah, thanks for merging; it was applied upstream as well already [07:21] didrocks, kenvandine: netbook CD exploded by 40 MB (!); I'll upload a new rb-ubuntu-music-store which drops the erroneous python-distutils-extra dependency; I'm using the p:ubuntu/rhythmbox-ubuntuone-music-store, is that right? [07:21] pitti, ewww [07:21] yes [07:21] ok, thanks [07:21] and good morning pitti [07:21] hey kenvandine; still awake? [07:22] yup [07:22] :) [07:25] uploaded and pushed [07:35] thx [07:36] good night all! [08:05] good morning [08:05] didrocks: Good morning [08:06] hey al-maisan [08:09] bonjour didrocks [08:10] guten morgen pitti [08:10] installed android 2.1 on my samsung spica (== lite) yesterday. Really good :) [08:12] didrocks: oh, got a new phone now? [08:12] pitti: fortunately, the repair service lost my previous one. So, I had this one in exchange :) [08:12] heh [08:13] but it was a terrible android 1.5 branded with my phone provider operator [08:13] really buggy (you have at each boot to uninstal one app to install it again and others) [08:13] at each boot [08:13] a lot of defaults are lost, etc. [08:14] so, I decided to install vanilla android + some drivers for the phone on it :) [08:32] seb128: bonjour [08:33] seb128: slept well? [08:33] hey pitti [08:33] not enough but well yes! [08:33] pitti, you? [08:34] bit short, but fine [08:34] they are having the olympic games on the wrong coast [08:34] salut seb128 [08:34] lol [08:34] I watched hockey again [08:34] but that starts at 1am [08:34] lut didrocks [08:35] seb128: so, notify-osd is gone from today's chart :) [08:35] waouh [08:35] and I just found out the reason for the udisks CPU increase [08:37] oh? [08:37] what is it? [08:37] sorry about the rhythmbox store depends overlook [08:37] it came with the new libatasmart [08:37] I should have spotted when I sponsored [08:37] the previous version had a bug which caused ata smart status to be broken [08:37] so it wasn't done :) [08:37] now smart reporting works again [08:37] oh ok [08:37] but apparently that's quite expensive [08:37] I'll look into that [08:38] I need to look into smart and how it's working [08:38] my parents computer open a failing disk warning at every boot [08:39] seb128: bug 438136 ? [08:39] Launchpad bug 438136 in libatasmart "palimpsest bad sectors false positive" [Unknown,Confirmed] https://launchpad.net/bugs/438136 [08:39] but nothing reported wrong on winxp with the tools my father tried [08:39] pitti, looking [08:39] I though that one was fixed in a sru [08:40] ok, it's not the same I found by then [08:40] pitti, thanks [08:41] pitti, oh it's waiting for sponsoring? [08:41] is that on your list? [08:41] no [08:41] there is no solution so far [08:42] should sponsors be unsubscribed then? [08:42] yeah, they should [08:42] seb128: btw, does the screensaver work for you? [08:42] pitti, define work? [08:42] seb128: when I try to unlock, it spins forever, and I have to ctrl+alt+f1 and killall gnome-screensaver [08:43] could that be related to the new keyring? [08:43] (it's working for me) [08:44] ok, perhaps it's something local then [08:44] pitti, https://bugzilla.gnome.org/show_bug.cgi?id=610678 [08:44] Gnome bug 610678 in keyring files "starts eating cpu when trying to unlock screen" [Normal,Unconfirmed] [08:44] pitti, I get it on coming back from guest session [08:44] but not if I lock and unlock the screen using ctrl-L [08:44] I get it during the timeout [08:44] oh timeout… [08:48] lol [08:48] "I want to see love" [08:48] nice bug report topic ;-) [08:48] indeed :-) [08:49] (it's about having a <3 smiley in empathy) [08:49] pitti, speaking of which I fixed the icq aim issue [08:50] backported a 1 liner change to libpurple [08:50] that fixes it in pidgin and empathy [08:50] I saw on -changes, merci! [08:51] de rien [09:14] good morning everyone [09:14] hey chrisccoulson, how are you? [09:15] hey didrocks, yeah, good thanks [09:15] how are you? [09:15] hey chrisccoulson [09:15] hey pitti, how are you today? [09:15] I'm great, thanks! [09:15] chrisccoulson, hey [09:15] you had quite a late night last night didn't you? [09:15] hey seb128 [09:16] chrisccoulson: yeah, wanted to get that notify-osd thing done; and it's gone from today's charts :) [09:16] chrisccoulson: good thanks :) [09:16] chrisccoulson: how are you? [09:16] fantastic! [09:16] yeah, i'm good thanks [09:16] looking forward to the end of the week:) [09:17] chrisccoulson, end of this week or start of next week? ;-) [09:18] seb128 - both ;) [09:18] :-) [09:19] morning [09:19] lut huats [09:19] hello mister seb128 o/ [09:21] hello huats [09:21] * seb128 just installed meld for patch review [09:21] that's quite a nice software to use [09:22] i've never tried that before [09:22] that looks quite nice [09:22] seb128: right, it's nice and shiny :) [09:22] chrisccoulson, it show you before and after side to side [09:22] upstream accepts patch quite easy and is reactive [09:22] with changed part colored [09:22] meld's been around for ages. I used it a bunch in my previous job, so at least 5 years ago. [09:22] didrocks, what upstream? [09:23] seb128: meld's upstream [09:23] soren, yes, I know about it but I don't do patch review so much usually [09:23] or I do using vcs-es [09:23] but for review those appindicator package changes that handy === cypher_ is now known as czajkowski [09:24] I could also put the source in bzr and bzr diff it [09:24] seb128: It's just interesting that several people have mentioned it over the last couple of days, given that it's an old project. [09:24] I forget who mentioned it yesterday. [09:24] oh other people did? [09:24] * soren greps through logs [09:24] Yup, over in #kernel. [09:24] Yesterday. [09:25] Funny, that. [09:30] * mvo stil has a open bugreport to integrate it in synaptic when availalbe [09:31] mvo: Oh, for doing conffile merges? [09:31] yes [09:31] Neat. [09:32] it would be so nice, but I never got around to it, should be pretty trivial to do too [09:32] * mvo will try to find a contributor for doing it :) [09:35] * seb128 grrrrr at rsync deleting his cd iso again on ctrl-C [09:42] seb128: do a hardlink before [09:42] seb128: http://people.canonical.com/~pitti/scripts/rsync-desktop [09:45] meh! [09:45] p-distutils-extra is still on http://cdimage.ubuntu.com/ubuntu-netbook/daily-live/20100224.1/lucid-netbook-i386.manifest [09:45] and it's still oversized [09:48] what changed it? the seed change done yesterday by stevenK was on [armel] only [09:49] (apart from the bluez and rhythmbox plugin things) [09:49] didrocks: I fixed rb-u1-music-store (that depended on it) [09:49] but now that new version is on that CD, and it's still there [09:50] apt-cache rdepends python-distutils-extra only has two pacakges [09:50] pitti, thank you! [09:50] * pitti scratches head [09:51] * pitti rsyncs desktop CD to try [09:51] hum, right, and no rdepends are installed (promissed, I didn't add Quickly to the netbook ;)) [09:51] erm, netbook [09:51] didrocks: I already checked that [09:51] germinate doesn't say anything either [09:51] http://people.canonical.com/~ubuntu-archive/germinate-output/netbook.lucid/rdepends/python-distutils-extra/python-distutils-extra === tkamppeter_ is now known as tkamppeter [10:34] \o/ [10:34] pitti: new size is ok after rebuild? :) [10:34] udisks is back to two thin light blue CPU bars, hardly noticeable [10:34] didrocks: rebuild is still ongoing [10:35] now I need to get this patch past David [10:35] congrats for udisks :-) [10:35] ♪ another one bites the dust ♫ [10:35] * didrocks hugs pitti [10:38] pitti, is there any iso respin now or scheduled? [10:38] pitti - what did you do to udisks? [10:38] seb128: netbook is ongoing, server is scheduled [10:38] pitti, would a gnome-bluetooth upload be disruptive? [10:38] chrisccoulson: I deferred the initial ata smart check by half a minute; no need to do this during boot, when everything needs both CPU and IO [10:39] ah, that makes sense [10:39] seb128: depends on how much regression potential it has [10:39] pitti, well it's only bluetooth ui, how much that can break? ;-) [10:39] ie I wouldn't consider gnome-bluetooth broken being an alpha blocker anyway [10:39] ooh, i've got access to canonicaladmin.com now [10:40] pitti, it's an appindicator port but it's a fairly trivial one [10:40] seb128: no, but if it causes crashes on startups it's not a nice experience on the alpha, and will get us lots of bugs, etc. [10:40] seb128: TBH I'd wait with that until after a3 [10:40] ok :-( [10:40] * seb128 hates alpha weeks [10:52] didrocks: http://cdimage.ubuntu.com/ubuntu-netbook/daily-live/20100224.2/lucid-netbook-i386.manifest [10:52] didrocks: it's gone now, but still oversized [10:53] and more than a little oversized :/ [10:54] well, language packs seem to be the same than yesterday [10:55] * didrocks makes a diff against manifest [10:55] didrocks: 23.1 against 24.2? [10:55] (23.1 was the last good one) [10:56] yes === czajkowski is now known as cz === cz is now known as cztab [11:00] pitti: nothing really noticeable: the bluez-utils + bluetooth stuff, rhythymbox-ubuntuone-plugins and its depends (python-ubuntuone, libubuntuone). The rest is only new package version [11:01] (maybe we should have a manifest with the size of the packages as a third column) === cztab is now known as czajkowski [11:07] chrisccoulson, pitti: bug #526747 [11:07] Launchpad bug 526747 in indicator-applet "battery applet popup is empty" [Undecided,Triaged] https://launchpad.net/bugs/526747 [11:07] there is a 2 liners fix there [11:07] if you want to do an update or queue that next time you do gpm changes [11:07] it's a bug in the gpm appindicator patch [11:11] didrocks: som, *puzzled* [11:12] s/som/so/ [11:12] pitti: I've just sync the new iso, I'm looking at the new packages first [11:12] if not, there is one updated package which is *big* [11:12] right [11:12] didrocks: woudl you mind investigating this? I need to do some server ISO juggling [11:12] pitti: sure :-) [11:13] * pitti hugs didrocks, merci [11:13] * didrocks hugs pitti back [11:42] seb128 - thanks, i'll take a look at the gpm patch [11:42] it looks like i'm going to be spending more time on gpm later anyway ;) [11:43] ;-) [11:43] chrisccoulson, thanks [11:48] didrocks: any luck yet? If not, we'll just declare the one as alpha-3 and document it in the release notes that you need to use an USB stick [11:50] pitti: sorry, I was just ensuring that the ubiquity installer create the wallpaper on live mode first. Now that it's ok, let me reboot and make some tries [11:52] didrocks: oh, so the caching works? [11:53] pitti: if installing in live mode, right. Apparently, not in autoinstall mode [11:54] didrocks: maybe talk to Keybuk and ask him to check the caching after an auto-install? (you know much more how this works now, sorry for delegating) [11:55] didrocks: nevermind :) [11:56] :) [11:57] asac: I just installed network-manager 0.8-0ubuntu4~nmt4 from the PPA and my wireless connection broke. the nm-applet icon is spinning all the time. starting it with --no-daemon I get a usage info from dhclient [11:59] ouch [12:04] yeah gnome-user-share release [12:04] cool [12:05] IT IS FOR ME !!!IT IS FOR ME !!!IT IS FOR ME !!!IT IS FOR ME !!! [12:05] :) [12:05] chrisccoulson, I was joking, if you want to do it, you can go [12:05] asac: in git log it says Merge remote branch 'origin/dhcp6', maybe this is where the problem comes from - just a wild guess [12:06] baptistemm, it's for nobody we are frozen for alpha today [12:06] ;-) [12:06] pitti - if you have 5 minutes, would you be able to verify that gnome-user-share is not in the ubuntu-desktop packageset (I would do that myself, but i can't run edit_acl.py from work). if it's not in that packageset, i will ask cjwatson about it this afternoon [12:06] as this is the 1st release since 2.28.x, there should a bunch of new changes, like a new share bar in ~/Public [12:06] it has a nautilus extension now? [12:07] sucks [12:07] we might not want to update for lucid [12:07] stop adding things which are init-ed at login... [12:07] we'd need to review the impact on login time [12:07] yeah [12:07] chrisccoulson, does it requires to be an extensions ? [12:07] baptistemm: yeah. those bars are usually from nautilus extensions [12:08] there is no way to register tha through dbus [12:08] so, i suspect it installs its own extension now [12:08] i just checked the git log [12:08] it's definately got a nautilus extension now [12:08] TeTeT: i asked cypthermox to check that [12:09] will upgrade later today to daily again ... then i will see too ;) [12:09] chrisccoulson, perhaps you can just patch to disable the extension [12:09] asac: good luck [12:14] TeTeT: can you paste the error you got in --no-daemon? [12:18] asac: http://pastebin.ubuntu.com/382930/ [12:22] asac, we don't use a dhcp with d-bus capabilities? (I remember to see such) [12:23] baptistemm: that was dropped long ago (intrepid?) [12:23] ah okay [12:23] * baptistemm is totally lagging in certain part [12:23] thx [12:23] dhcpdbd [12:23] i think was the package name [12:23] thats gone now [12:24] TeTeT: so yes, the dhpc6 commit seems to be the best guess [12:24] i remember dan talking about our dhcp being sooo outdated [12:25] looking at code it seems to add the -S or -N option [12:25] for ip6 [12:25] - return dhclient_start (client, "-6"); [12:25] + return dhclient_start (client, "-6", info_only ? "-S" : "-N"); [12:25] but not for ip4 [12:25] in nm-dhcp-dhclient.c [12:26] right [12:26] asac: our dh_client doesn't seem to understand either [12:30] asac: I compile one without the -S and -N and test it === MacSlow is now known as MacSlow|lunch [12:43] asac: seems I can'\t build it anymore, gtk-doc is missing [12:43] TeTeT: how are you building? [12:44] asac: in the upstream source, copied the debian directory from ubuntu-head and then 'debuild -b -uc -us' [12:47] asac: building now from our apt-get source network-manager base instead of upstream [12:51] asac: doesn't work for me, still the same error, hope I did build it right [12:53] strange [12:54] thats the same version that biult on the builders? [12:54] TeTeT: ? [12:54] asac: yes, I've built it, but the problem with dhclient does not go away [12:55] ok .thats what you mean [12:58] TeTeT: from configure: [12:58] + AC_MSG_CHECKING(for dhclient) [12:58] + # NM only works with ISC dhclient - other derivatives don't have [12:58] + # the same userland. NM also requires dhclient 4.x since older [12:58] + # versions do not have IPv6 support. [12:58] so ... we need newer dhclent [12:58] or ... try dhcpcd [12:58] --with-dhcpcd=yes [12:58] in rulesa [12:59] asac: ok, I give it a try [13:12] chrisccoulson: sorry, was off for lunch; looking [13:13] chrisccoulson: g-u-s has no uploaders at all; this needs to be fixed by cjwatson [13:14] pitti - thanks :) [13:14] i'll ping cjwatson about that then. it would be useful if i could upload it again [13:14] chrisccoulson: just done [13:14] (the ping) [13:14] pitti - thanks:) [13:29] asac: doesn't seem to work with dhcpcd and without dhclient - I have to give a class in 30 minutes, so I will continue testing tomorrow [13:30] TeTeT: guess you also have to say: --with-dhclient=no ... if you already did that then thats bad luck ;) [13:30] have to think about it [13:30] probably have to update dhcp in the daily ppa for now [13:30] but i hope for dhcpcd atm ;) [13:42] seb128, kenvandine: seems we currently get flooded with indicator patches; ok for you if I fan them out between the three of us, for review and applying? [13:43] pitti, yes, that was the idea when I asked them to assign the team rather than to kenvandine [13:43] ok, fine [13:43] bah [13:43] gwibber crashes on start on current a3 image for me [13:43] as do desktopcouch-service [13:44] hey guys [13:44] seb128, i have a fix for that [13:45] kenvandine: good morning! [13:45] kenvandine, seems to bug #519557 [13:45] Launchpad bug 519557 in gwibber "gwibber-service crashed with DBusException in call_blocking()" [Medium,Confirmed] https://launchpad.net/bugs/519557 [13:45] rather a work around for desktopcouch being slow to start at boot [13:45] and bug #525542 [13:45] Launchpad bug 525542 in desktopcouch "desktopcouch-service crashed with BadArgumentsError in __init__()" [Undecided,New] https://launchpad.net/bugs/525542 [13:45] * kenvandine looks [13:45] hey kenvandine [13:45] pitti, that is what i was working on last night :) [13:46] bug #519558 [13:46] Bug 519558 on http://launchpad.net/bugs/519558 is private [13:46] have seen bug #525542 yet though [13:46] Launchpad bug 525542 in desktopcouch "desktopcouch-service crashed with BadArgumentsError in __init__()" [Undecided,New] https://launchpad.net/bugs/525542 [13:46] I get it on today's livecd booted on the mini with an usb key [13:47] when I tried to open "broadcast account" I think [13:47] yeah [13:47] the bad admin account thing? [13:47] it does it every time [13:47] dunno what you are talking about ;-) [13:47] hehe [13:47] 519558 [13:47] just open the me menu and pick broadcast account [13:47] bug # [13:48] bah [13:48] ah [13:48] I get all 3 crashes together [13:48] ok [13:48] right same here (gwibber crashes) [13:48] every time I try to open gwibber [13:49] seb128, from the live image right? [13:49] yes [13:49] ok, i'll boot the live image this morning [13:49] i have a fix to handle gracefully waiting for desktopcouch to start [13:49] which seems pretty reliable [13:49] but it feels like a real hack [13:50] kenvandine: increase the d-bus timeout? [13:50] pitti: so, I've made a script to get the size of each package as well and runned it in today and yesterday's image: http://people.canonical.com/~didrocks/ubuntu-netbook/ [13:50] can you do that for service activation? [13:50] kenvandine: it's not related to activation as such [13:50] pitti: I can't find anything noticeable looking at the default. At most, we should have get something like 2MB [13:50] pitti, i think it actually is [13:50] didrocks: weird [13:50] kenvandine: well, of course activation causes the call to take much longer [13:51] but from an API POV activiation is invisible [13:51] i'll look at that [13:51] kenvandine, well 525542 indicates desktopcouch crashes [13:51] my hack was to catch the exception and try again [13:51] pitti: are there any additional files uninstalled on the live, out of any package? [13:51] kenvandine, so it's not a waiting to start issue [13:51] yeah, that isn't related [13:52] kenvandine, I doubt it's a start race... [13:52] i agree [13:52] kenvandine, I've waiting at least 5 minutes between tries [13:52] not sure what that is [13:52] seb128: upstream rhythmbox helped me out, they said --debug is too much to be asked for every bug report [13:52] seb128, i'll make sure chad looks at that [13:52] nigelb, see ;-) [13:52] kenvandine, thanks [13:53] hehe, so I'm going with just nonstandard gconf data === MacSlow|lunch is now known as MacSlow [14:00] seb128: how do you want me to do this? debdiff or request a bzr merge? [14:00] as you prefer [14:00] just put the bzr url in the bug you opened [14:00] or update the same bzr location [14:00] easier [14:00] I'll update the same bzr location [14:01] hi, which netbook launcher should be used in today's version of UNE, the classical netbook-launcher or netbook-launcher-efl? [14:02] thekorn: netbook-launcher if you have a 3D driver and hw, then n-l-efl should be launched as a fallback if you don't have it [14:03] didrocks, ah, ok, this explains why I have different launcher in kvm and real hardware, thanks [14:03] thekorn: it just show it works ;) [14:05] ok, one bug I don't need to file, let's find some more ... :) [14:06] thekorn: let's say for each bug you file and I fix, you implement one of the missing feature in LP API I need? ;) [14:06] seb128: hm, the gsd/libgnomekbd ones are pretty intrusive :( [14:06] didrocks, deal! [14:06] :) [14:06] thekorn: oh, does the efl launcher actually work for you? [14:07] it just crashes for me in kvm [14:07] pitti, yes :-( [14:07] pitti, yes, it works on real hardware (eeepc) but crashes when shutting it down, [14:08] but on kvm it indeed crashes [14:08] it crashes on startup for me; I don't have any launcher at all [14:08] ah, ok [14:08] * pitti fixes the lucid apport retracer chroots [14:08] right, no launcher in kvm [14:09] what do you prefere? should I reopen old bugs (bug 277133) or create a new one? [14:09] Launchpad bug 277133 in payson "text under icons wraps badly in netbook-launcher" [Wishlist,Confirmed] https://launchpad.net/bugs/277133 [14:11] personally I prefer reopening old bugs, unless it's not 100% clear that they are identical, or the old bug has two million duplicates and 200 comments, so that working with it is impractical [14:11] but since it's didrocks's bug, it's his call :) [14:13] thekorn: pitti: it's not the same based code, right for launcher < karmic? [14:13] base* [14:13] njpatel: ^ [14:13] when in doubt, open a new one rather [14:13] if it's not the same, I prefer a new bug :) [14:13] duplication is much easier than cloning [14:14] thekorn: just ensure it's not one related to bug #459367 [14:14] Launchpad bug 459367 in netbook-remix-launcher "Menu anomalies for categories with >= 8 rows" [Medium,Triaged] https://launchpad.net/bugs/459367 [14:14] thekorn: if you want to have a look at my high priority bug, you can find them at https://wiki.ubuntu.com/UNE/lucid-bugs [14:15] pitti: found the cause of the CD oversize [14:15] great page, looks like alot of work ;) [14:15] thekorn: just hope to be able to empty it before lucid's release :) [14:16] pitti: it's not on the squashfs system. It's in pool/main/l (there are now the linux/ linux-firmware/ and lupin/ directories) [14:16] ooh [14:16] I'm sure you will, UNE is going to rock! [14:16] and as the linux package is something like 30 MB :) [14:17] pitti: so, my next logical question is "where is definied the packages that goes to pool/ ? It's normally additional package not installed by default, right? [14:17] correct, ship-live [14:18] didrocks: none of those are in earlier CDs? [14:18] /mnt/iso-23/pool/main/l$ du -sh * [14:18] 197Klinux-wlan-ng [14:18] 12Klupin [14:19] /mnt/iso-24/pool/main/l$ du -sh * [14:19] 29Mlinux [14:19] 7,0Mlinux-firmware [14:19] 10Klinux-meta [14:19] ok, lupin-support is [14:19] 197Klinux-wlan-ng [14:19] 12Klupin [14:20] didrocks: right, diffed the two .list files [14:20] +/pool/main/l/linux-firmware/linux-firmware_1.31_all.deb [14:20] +/pool/main/l/linux-meta/linux-image-generic_2.6.32.14.15_i386.deb [14:20] +/pool/main/l/linux-meta/linux-image_2.6.32.14.15_i386.deb [14:20] +/pool/main/l/linux/linux-image-2.6.32-14-generic_2.6.32-14.20_i386.deb [14:20] I should have started by that instead of writing the script to get the package manifest size [14:21] pitti: I don't see any change to the seed related to that [14:22] no, there isn't; I asked slangasek/cjwatson in #u-release [14:22] (well, not sure what the supported file is really about, but well ;)) [14:22] thanks [14:27] didrocks: as always, cj "hero" watson fixed it [14:27] didrocks: bug in the seeds [14:27] didrocks: r1447 in netbook seed [14:28] pitti: seeing. I don't understand why we hadn't the issue before [14:30] Processing triggers for libglib2.0-0 ... [14:30] gio-querymodules: symbol lookup error: /tmp/tmpatgqLq/usr/lib/libgobject-2.0.so.0: undefined symbol: g_variant_type_copy [14:30] oh-uh [14:31] needs pre-depends? [14:32] it was just in the apport chrooots, so I can't claim that it's a real bug [14:33] apt-get -f install cured it [14:33] so, lucid retracers should be happy again [14:33] fuse-utils postinst needed more permissions than fakechroot can handle [14:34] shouldn't this statement work for an apport hook? "attach_gconf('GconfData', 'rhythmbox')"? [14:34] (I'm not seeing the information attached) [14:35] nigelb: do you import the complete apport.hookutils namespace? [14:35] nigelb: also, the first argument needs to be the report object, not a staring [14:35] s/staring/string/ [14:35] pitti: oh, ah [14:36] nigelb, looks to other installed examples [14:36] I did, then dont have this thingie [14:36] they just use gconftool -R [14:36] source_gnome-power-manager.py: attach_gconf(report, 'gnome-power-manager') [14:36] there [14:36] ah === al-maisan_ is now known as al-maisan [14:39] seb128: that seems to contain some usernames, passwords, shall I mask it out? [14:39] yes please [14:39] :) [14:45] I dont seem to be able to manipulate the data to mask it when I use "attach_gconf(report, 'rhythmbox')" [14:45] should I go back to apport.hookutils.command_output? [14:47] nigelb: this function creates report['GConfNonDefault'] [14:47] nigelb: so you could do the filtering on this key's value after attach_gconf() [14:48] attach_gconf() is handy because it already filters out the default values, etc. [14:48] but I havent understood how to filter the attach_gconf() [14:49] nigelb: attach_gconf(report, 'rhythmbox') [14:49] report['GConfNonDefault'] = my_filter_function(report['GConfNonDefault']) [14:50] ah. hacking this is not easy :) [14:50] nigelb: filtering that is not any harder than calling gconftool certainly? [14:51] pitti: no [14:51] nigelb: but of course you are welcome to call it directly, if it's any easier [14:51] I meant hacking through apport for the first time is not easy ;) [15:07] kenvandine, hiya [15:07] hey rickspencer3 [15:08] hi pitti [15:08] hey rickspencer3 [15:08] hi seb128 [15:08] soooo ... [15:08] sabdfl asked us to turn the crank on more time on gwibber [15:08] kenvandine has the details [15:09] ? [15:09] seb128, pitti how hard would it be start gwibber up say 30 seconds after boot? [15:09] trivial [15:09] but do we want that? [15:09] rickspencer3, i filled pitti in a little last night [15:09] never bad to hear that [15:09] rickspencer3: easy (but only if you actually have configured accounts) [15:09] we probably have 80% of our users who don't need it [15:09] rickspencer3, and sabdfl and i traded a few more emails after the one i forwarded you [15:09] I mean, we should only start it if you have configured accounts [15:09] pitti, right [15:09] +1 [15:09] rickspencer3, sabdfl did agree to that as well [15:09] kenvandine, ok, so what's the status? [15:10] we want to enable it in the session if you have configured accounts [15:10] can we start by making gwibber start and not crash on today's image? ;-) [15:10] (I would think not starting an unconfigured service would go without saying, but ... heh) [15:10] it would be nice to delay start a little [15:10] how do we turn it off? [15:10] kenvandine, can you share estimates and necessary work with what you actually agreed to with sabdfl? [15:10] seb128, it would be off by default [15:10] I've a configured account but I do not want gwibber to start [15:10] uncheck it in the session :) [15:11] ok, easy enough [15:11] or have an "autostart" preference in gwibber [15:11] seb128, how would we handle delayed starts? [15:11] so it's a 2 lines chang [15:11] in fact, it shouldn't really start until after you have a network connection, I would think [15:11] using autostart-delay key in the desktop entry? [15:11] it would be nice to have a general way of handling that [15:11] + an autostart condition [15:11] seb128, ok, so that exists? [15:11] cool [15:11] ie checking a gconf key to know if there is a configure account [15:11] kenvandine, yeah, chrisccoulson did that this cycle [15:11] excellent [15:12] we use it to delay other things already [15:12] there are two problems [15:12] hello [15:12] chrisccoulson, who is that guy? why do I keep hearing his name? [15:12] one, you have to check desktopcouch to see if there are accounts [15:12] chrisccoulson, hey ;-) [15:12] my ears are burning ;) [15:12] lol [15:12] we don't want to do that :) [15:12] the hack would be to write a gconf key saying it was configured [15:12] just define a key false by default [15:12] and make gwibber enable it [15:12] kenvandine, can you capture these points in the blueprint please? [15:12] when you first define an account [15:13] (after discussing here) [15:13] we need a place to document it, and we need work items [15:13] seb128, right, i was thinking we would just ship the autostart file as disabled (like we do now) [15:13] then enable it when you configure accounts [15:13] and having that delay is great :) [15:13] that is hackish now [15:13] it's either to tweak gconf keys in a programmatic way [15:13] the problem i see with that is syncing accounts [15:14] we shoudl start firefox by default, too [15:14] either -> easier [15:14] and gtkblog [15:14] pitti, and empathy! [15:14] seb128, ok [15:14] hehe [15:14] seb128: right [15:14] pitti - and evolution [15:14] "ready to use" [15:14] ;) [15:14] let's autostart eveything installed [15:14] then it will take 10 minutes to log in.... [15:14] ;) [15:14] if we installed things that's to use those ;-) [15:14] rickspencer3, can i just add this to the existing sfts blueprint? [15:14] and bughugger! [15:14] haha [15:15] don't forget photobomb :) [15:15] kenvandine: please do (sfts) [15:15] anyway... [15:15] kenvandine, that's what I meant, yeah [15:15] we would need a way to do the same thing to enable it to start if it has synced accounts from u1 [15:15] which i just don't know if there is a sane way to do that [15:16] i think we need to just punt on that, and first time you run gwibber and it sees there are configured accounts and the key isn't set... enable it [15:17] pitti, about the dbus timeout... got an example how to set the timeout? [15:17] i can't find any docs on that... [15:18] kenvandine: you want to use a timeout other than the default 25s for dbus calls? [15:18] is it 25s? [15:18] it is failing faster than that [15:18] for service activation, it needs to start desktopcouch [15:19] and if you start it in your session, it fails frequently [15:19] it's 25 seconds by default, but we've had issues with this before [15:19] how long does it take to fail? [15:19] * kenvandine glances at the log [15:19] we seem to be having lots of issues with activation right now.... [15:19] about 2s [15:20] hmmmmm [15:20] ah, you guys rock, you know that? :-) [15:20] hey sabdfl :) [15:20] kenvandine: is the service name claimed before registering the object you're trying to communicate with? [15:20] chrisccoulson, if i let it wait about 10s it works fine [15:20] i don't think so [15:20] sabdfl, hey, thanks ;-) [15:21] obj = dbus.SessionBus().get_object("org.desktopcouch.CouchDB", "/") [15:21] cdb = dbus.Interface(obj, "org.desktopcouch.CouchDB") [15:21] cdb.getPort() [15:21] fails in about 2s [15:22] kenvandine - what happens on the desktop couch side? (which I assume is what you're trying to communicate with) [15:22] yeah [15:22] * kenvandine hasn't looked at that code [15:22] one sec [15:22] does it claim the name on the session bus first [15:22] and then register the object "/" [15:23] i've got a feeling it's racy if you do it in that order [15:23] yes [15:23] ah! [15:23] could you try swapping that? [15:23] bus_name = dbus.service.BusName("org.desktopcouch.CouchDB", [15:23] bus=dbus.SessionBus()) [15:23] self.death = death [15:23] dbus.service.Object.__init__(self, object_path="/", bus_name=bus_name) [15:23] wait... how do you swap that? [15:23] yeah, i'm just wondering how to do that [15:24] i'm sure it's possible to claim a temporary name, register an object and then change your name later on [15:24] seems you have to have the bus_name first [15:24] i haven't tried it though [15:24] you can change the name [15:24] not sure how... but can find out [15:24] that might be the way to do it [15:25] i bet this would fix lots of the problems people are having with desktopcouch [15:25] aquarius, you have any experience with that? [15:25] not sure how much of a dbus guru you are :) [15:25] i don't know if that's your issue there though, but i think dbus will consider the service activated once the bus name appears [15:25] yeah, it kind of makes sense [15:25] so, if that happens before you've registered the object you're interested in, i'd expect it to be an issue [15:26] kenvandine, I am in no way a D-Bus guru. Every time I need to use it I have to go back to the tutorial... [15:26] and the tutorial is very bare [15:26] there is definately a problem with desktopcouch starting with dbus activation [15:26] you ain't kiddin'. [15:26] * kenvandine is digging in [15:27] kenvandine: we have a similar issue with udisks too, so i'd be interested to know how to fix it ;) [15:27] haha [15:27] :) === gnomefreak76 is now known as gnomefreak [15:27] so activation notices that you've claimed your name on the bus and then calls you before you've registered an object? [15:27] that's dense. [15:27] aquarius, we don't know for sure [15:27] that is chrisccoulson's speculation [15:27] aquarius: i'm guessing that's how it worls, but i'm not sure [15:28] that suggests that absolutely everyone who uses activation ought to do this temporary-name shuffle, or they'd surely get the same problem? [15:28] aquarius: possibly, but i'm only speculating at the moment [15:28] we keep getting activation issues though [15:29] *nod* I certainly agree there seems to be an activation problem [15:29] * kenvandine looks at some C code for examples [15:29] I'm just puzzled by how everyone in the world isn't bitten by this, if you really have to do the hot-shoe-shuffle with temporary names [15:30] chrisccoulson: there's been a number of activation-related fixes in dbus recently [15:30] but yes, those races also need squashing [15:30] mclasen: thanks. have you been noticing issues too? [15:30] what's the best way of trying to debug an activation issue? [15:30] there were issues with reloading configuration dropping pending activations [15:31] and issues with activating things that daemonize [15:31] kenvandine: oh, does the couchdb thing daemonize? it shouldn't [15:31] I had all sorts of trouble with that when I tried that years ago [15:31] aquarius, does it? [15:32] * kenvandine thinks CardinalFang should hang out in #ubuntu-desktop so we could leave aquarius alone :) [15:32] mclasen: yeah, we've had issues with things which daemonize in the past (system-tools-backends being one of them) [15:33] speak of the devil :) [15:33] CardinalFang, does desktopcouch daemonize? [15:33] i don't think it does [15:33] mclasen - our main issue right now is bug 521481, which seems to be related to udisks activation [15:33] Launchpad bug 521481 in gnome-disk-utility "gvfs-gdu-volume-monitor crashed with SIGSEGV in gdu_pool_get_presentables()" [Medium,Triaged] https://launchpad.net/bugs/521481 [15:34] kenvandine, er, when one calls the getPort() dbus call, programs are started by something that is daemonized, the DBus daemon. [15:34] yeah, but desktopcouch-service doesn't daemonize [15:34] right? [15:34] chrisccoulson: yeah, we thought we had that nailed with the reload fixes, but it is still around [15:35] It does not detach from dbus daemon, no. [15:35] ok [15:35] so that isn't the issue [15:35] kenvandine, the couchdb process we start gets started as daemonised. desktopcouch-service *itself* doesn't daemonise, though [15:35] I don't think it's allowed to, but I'm not certain about that. [15:35] mclasen - yeah. that one gets lots of duplicates [15:37] CardinalFang, when using a getPort() call to start desktopcouch, we are see frequent failures if done at session start [15:37] and looking at the gwibber logs, it is failing in about 2s [15:37] the timeout should be 25s [15:38] Huh. I found a problem lately where I had set couchdb log file to a place I couldn't write to, and startup failed about the same way as that. [15:38] didrocks: has the idea of replacing FF with epiphany or something lighter (chromium) for UNE been considered that you know of? [15:38] CardinalFang, chrisccoulson speculates it might be a problem with claiming the name on the bus before registering the object [15:40] but it looks silly, since you need a bus_name to register the object [15:40] Yeah, I'm trying to parse that. [15:40] so if that was the case, we would need to claim a temp name, then register the object [15:40] then... change the name [15:40] LaserJock: we discussed it but no decision has been taken. Let's see at lucid UDS [15:40] lucid+1 UDS :) [15:41] CardinalFang, which seems like a real hack [15:41] didrocks: I noticed that chromium is much faster start up, but uses a lot more memory so I'm not sure that makes them even :-) [15:42] didrocks: but epiphany seems to be really making headway, and is more tightly integrated [15:43] LaserJock: it's too late and too controversial for lucid anyway, let's see in +1 when the time comes :) [15:44] didrocks: sure sure, I was not thinking for Lucid at all, I just wondered if it had been discussed before [15:44] kenvandine, Okay, I'm trying to see how something might be out of order. aquarius, PortAdvertser class, in __init__(), we call dbus.service.Objext.__init__(). But the methods we're exporting are decorated with dbus.service.method . Isn't the decorator run at compile time, and __init__ only at instance-creation time? [15:46] aquarius, I do not know if that is a problem. Just fishing. [15:46] hrm. [15:46] I...am not sure. [15:47] Me either. Probably a blind alley. [15:48] kenvandine, are there no useful log files of that happening? [15:48] didrocks: I was able to change icon sizes last night, but I don't know how to change the size of the container for the categories [15:50] LaserJock: I don't have the time to look at that before tomorrow, I saw yesterday it was in the same file [15:51] didrocks: ok, I'll keep digging around then [15:52] LaserJock: if you need more help tomorrow, do not hesitate to ask [15:52] didrocks: is ubuntu-netbook-default-settings in bzr somewhere? [15:56] CardinalFang, hang on let me reproduce and get a log [15:57] aquarius, confirmed, decorators run when the class is parsed. [15:59] LaserJock: lp:ubuntu/ubuntu-netbook-default-settings [16:00] CardinalFang, but d-bus can't start exporting those methods until __init__ is run, because we're not on the bus at that point [16:01] what's the bug number? [16:01] I fixed an activation issue a few months ago, so might be able to help [16:02] james_w - we're still seeing the same issue with devicekit-disks / udisks [16:02] aq: The Python DBus tut has exactly our code, too. [16:02] it never really went away, even after the issue you fixed a few months back [16:02] chrisccoulson: is my fix in the dbus package? [16:02] james_w - yeah, it's in there [16:03] didrocks: I'm assuming bug #524333 is just adding vlc to the maximus exclude_class? [16:03] Launchpad bug 524333 in ubuntu-netbook-remix-default-settings "VLC Full Page" [Low,Triaged] https://launchpad.net/bugs/524333 [16:03] james_w - as long as we're talking about the same fix though ;) [16:03] (the timeout issue) [16:03] yeah [16:03] yeah, we've got that fix [16:04] that rules out one thing then [16:04] LaserJock: right :) [16:04] LaserJock: hence the "good and easy opportunity" :) [16:05] didrocks: ok, I'll do that one and onboard [16:05] LaserJock: sweet, thanks! [16:05] chrisccoulson: can you reproduce? [16:05] didrocks: yeah, I thought I'd start with some low-hanging fruit [16:05] chrisccoulson: there's no xsession-errors on that bug [16:06] LaserJock: that's the good way, indeed [16:07] kenvandine: what's the actual error you get when you make the call? [16:07] straight timeout error? [16:07] james_w - is there a bug number? [16:07] james_w, i don't think so [16:07] heh, bug 527112 is useful [16:07] reproducing in a VM [16:07] chrisccoulson: the udisks one from mario [16:07] Launchpad bug 527112 in ubuntuone-client "xx" [Undecided,New] https://launchpad.net/bugs/527112 [16:08] bug 521481 [16:08] Launchpad bug 521481 in gnome-disk-utility "gvfs-gdu-volume-monitor crashed with SIGSEGV in gdu_pool_get_presentables()" [Medium,Triaged] https://launchpad.net/bugs/521481 [16:08] didrocks: I guess we don't want to get crazy with it, but what about Tomboy? [16:08] james_w - i've never recreated that bug, but it still gets people reporting it [16:09] the only way for gdu_pool_new to return NULL is if the dbus call fails somehow [16:10] yeah [16:10] I want to know what error dbus is sending back [16:10] it will give some clue as to how dbus is seeing the failure to activate [16:10] LaserJock: hem, let me think and have a try [16:10] i think in the desktopcouch case it is desktopcouch tracebacks, like the find_port failure [16:11] LaserJock: yeah, it seems logical, right [16:11] but if you give it plenty of time to start up, it never fails [16:11] plenty meaning 8-10s [16:11] so it is like it responds to the name, but it isn't ready to respond [16:12] CardinalFang, i guess the problem probably is more related to getPort failing [16:12] right [16:12] so perhaps it doesn't wait long enough for couch to start [16:12] so don't create a dbus.service.BusName [16:12] until you are ready to handle everything [16:13] that's what dbus watches for [16:13] CardinalFang, ^^ [16:13] kenvandine: so, that sort of confirms my earlier theory [16:13] yeah [16:13] so that creates the name first thing on init [16:13] either that or have the method call block until it is ready if that is possible [16:15] oh... interesting [16:15] find_port is the call that tries starting couch [16:15] james_w, kenvandine, okay. Fortuitously, I have a branch that should already do this right. [16:15] aquarius, ^ I'm not magical, I promise. [16:15] :) [16:15] CardinalFang, can i test your branch? [16:16] CardinalFang, ! [16:16] kenvandine, Give me two minutes to upload. [16:16] mclasen - you might be interested in the comments from james_w in the scrollback there [16:16] CardinalFang, you've already done the work for this even though you didn't know it was needed? that's a useful talent :) [16:16] thx [16:16] that might be what causes the udisks activation problem too [16:17] chrisccoulson: that's why I want to see the xsession-errors, if it's a timeout error you get then it's likely a different issue [16:17] the name gets claimed on __init__ but the underlying service it needs to start doesn't start until we call the method [16:18] aquarius, part of the get_port through DBus cleanup that's occupied my week. [16:18] james_w - i'll try and find that from a later report. [16:18] chrisccoulson: thanks [16:18] chrisccoulson: I'm pretty sure that udisks does things in the right order [16:19] mclasen - in devicekit-disks (i assume udisks does the same), it claims the service name and then registers the object you're trying to communicate with [16:19] CardinalFang, aha, yeah [16:19] that's the wrong way round isn't it? [16:19] chrisccoulson: talk to david, I'd say [16:20] mclasen: will do. thanks [16:20] kenvandine, this is *not* reviewed yet. Caveat Bzrker. [16:20] lp:~cmiller/desktopcouch/get_port_through_dbus [16:22] CardinalFang: care to propose that for merging so that we could easily look at the diff [16:22] ? [16:22] you can use the "Extra options" to set it as Work In Progress if it's not ready for review yet [16:23] james_w, Proposed. [16:23] thanks [16:29] CardinalFang: that should help indeed [16:29] CardinalFang: it's still taking the bus name before creating the object, but that's less likely to be an issue with it starting couch up front [16:29] seb128: do we have a page somewhere that outlines what kind of hardware support rhythmbox has? [16:30] no [16:33] james_w, Do I not need the bus name already before I can init the object? Maybe there's a Object.set_bus() or something .... /me looks [16:33] CardinalFang: did you read the docstring for Object.__init__() ? [16:33] `bus_name` : dbus.service.BusName or None [16:33] Represents a well-known name claimed by this process. A [16:33] reference to the BusName object will be held by this [16:33] Object, preventing the name from being released during this [16:33] Object's lifetime (unless it's released manually). [16:34] so you can just do self.bus_name in your class, rather than having Object do it for you [16:38] james_w, I don't see that the member name is the same as the arg name, yet. Also, it seems I need a dbus.connection.Connection instance instead. [16:41] CardinalFang: you can still use dbus.SystemBus() to get the connection [16:41] self.conn = dbus.SystemBus() [16:41] self.death = death [16:42] dbus.service.Object.__init__(self, object_path="/", conn=self.conn) [16:42] Okay. I meant as param to Object.__init__. """Either conn or bus_name is required""" [16:42] slef.bus_name = dbus.service.BusName("org.desktopcouch.CouchDB", bus=self.conn) [16:43] Cool. [16:55] seb128, kenvandine how is csd looking? still buggy? [16:55] CardinalFang, better... sort of [16:55] now i actually get a timeout error from dbus starting the service :) [16:55] rickspencer3, good from there [16:55] rickspencer3, i think much better [16:55] seb128 knows more [16:56] CardinalFang, so i definately think it is an improvement... but still something not quite right [16:56] great to hear [16:57] thanks seb128 kenvandine [16:58] rickspencer3, there is one small bug where decoration are displayed where they should not [16:58] hmmm [16:58] ok [16:58] rickspencer3, otherwise things look good [16:59] kenvandine: is the timeout in 25s, or much shorter? [17:00] kenvandine, Hah. That's not better. [17:00] haha... [17:00] ok, added some logging [17:00] it is getting the timeout error in 2s [17:00] org.freedesktop.DBus.Error.ServiceUnknown: The name :1.284 was not provided by any .service files [17:00] not timeout [17:00] service unknown [17:00] hummm [17:01] wtf! [17:03] CardinalFang, james_w: that is really strange [17:03] kenvandine: what's the call that is triggering it? [17:03] obj = dbus.SessionBus().get_object("org.desktopcouch.CouchDB", "/") [17:03] cdb = dbus.Interface(obj, "org.desktopcouch.CouchDB") [17:03] port = cdb.getPort() [17:04] i shouldn't need to restart dbus? [17:04] nope [17:04] it's making a call on a unique name [17:04] the service file is installed [17:04] and looks correct [17:04] or, we're getting confused and this is not actually triggered by the getPort call, but by something the daemon is doing in response to that [17:05] no it is [17:05] i just put mine in a try except and printed the exception [17:06] http://pastebin.ubuntu.com/383124/ [17:06] log output is [17:06] 2010-02-24 12:02:28,784 - Gwibber GNOME Client - INFO - KEN: Before calling DC [17:06] 2010-02-24 12:02:36,380 - Gwibber GNOME Client - ERROR - Failed to connect to desktopcouch, org.freedesktop.DBus.Error.ServiceUnknown: The name :1.291 was not provided by any .service files [17:06] so in under 2s it fails with ServiceUnknown [17:06] yes [17:07] and desktopcouch is actually not started [17:07] but if it is actually trying to call a method on :1.291 then that is expected [17:07] before it did eventually start [17:07] true [17:12] kenvandine: so, it may be you have something on the bus at org.desktopcouch.CouchDB when you call get_object and it then leaves the bus before you call the method on it [17:13] kenvandine: do a dbus-monitor --system while running the code again please [17:13] james_w, CardinalFang: a reboot did wonders [17:14] cool [17:14] I don't want to hear that. [17:16] the follow_name_owner_changes parameter to dbus.bus.BusConnection.get_object is what led me to that idea [17:16] I think it may want to be false for interactions with desktopcouch [17:16] so now at boot, it starts in 9s [17:17] and gwibber waits patiently for it [17:17] no idea why dbus needed a restart to find the service [17:17] james_w, name changes aren't a problem for us [17:18] desktopcouch restarting doesn't break gwibber [17:18] afaict anyway :) [17:18] if you only call getPort once then you are ok [17:18] apart from the race I highlighted above [17:18] james_w, i might have missed that race you mentioned [17:19] can you repo? [17:19] repeat? [17:19] my laptop overheated again... [17:19] kenvandine: so, it may be you have something on the bus at org.desktopcouch.CouchDB when you call get_object and it then leaves the bus before you call the method on it [17:20] I haven't confirmed it's actually an issue, and it could be seen as a bug in the bindings [17:20] oh, like desktopcouch crashes? [17:22] ok, in my VM desktopcouch start time (time to respond to getPort) has varied between 6 and 13s when starting in the session [17:23] in my sampling of 10 logins of my VM [17:23] and killing gwibber, couch and friends and starting gwibber [17:23] dropping caches? [17:23] getPort responds in about a second [17:23] it was with reboots [17:24] seems to be working well, still slower than i would like [17:24] but deferring it a bit should make it start fast [17:25] at least no apport crash dialogs :) [17:25] CardinalFang, ^^^ [17:25] CardinalFang, so in my testing no very little code review... big thumbs up from me :) [17:25] s/no/and [17:26] Hrm. Thanks, kenvandine. [17:27] CardinalFang, i would buy you at least 3 beers if you can make desktopcouch start faster :) [17:27] kenvandine: did you fix the dbus issue yet? [17:27] chrisccoulson, this branch does [17:27] ( i was afk and haven't checked the scrollback yet) [17:27] lp:~cmiller/desktopcouch/get_port_through_dbus [17:28] kenvandine: excellent :) [17:28] right, home time for me now [17:28] bbl [17:28] chrisccoulson, later [17:28] * kenvandine goes to lunch... bbiab [17:29] kenvandine, desktopcouch crashes the same way on the installed system [17:29] same argument error [17:30] when trying to join admin_username admin_password [17:30] could it be a missing depends or init? [17:33] pitti: do I had the bg wallpaper extra work as a WI somewhere? (the one with "ubiquity doesn't use gnome-destkop in install mode, so no cache and we are screwed" and the second "netbook-launcher doesn't use gnome-desktop to draw the bg") [17:33] not sure about 1/ apart again from the hakish way [17:34] kenvandine, ok, gotcha [17:34] kenvandine, it's an issue when there is no gnome-keyring available [17:38] Hi. I'm having what I think is a PolicyKit problem. I've exhausted all other resources and could really use some help. [17:42] ah polkit [17:42] Yeah, fun stuff... [17:42] DanEngholm, I can't help you, but I know that robert_ancell, who won't be online till much later, has some experience based on his work with GDM [17:43] OK. I can check back later. Can you be more specific on when? [18:17] kenvandine, still around? [18:17] seb128, i am at lunch :) [18:17] what's up? [18:17] kenvandine, ok, the desktopcouchdb is a gnome-keyring issue at least partially, I will fix it [18:17] oh? [18:17] kenvandine, you should get the "do you want to create a keyring" [18:18] oh! [18:18] it happens when there is none yet [18:18] and couchdesktop doesn't handle nicely this case [18:18] why isn't that working on the live image? [18:18] kenvandine, no keyring there or on new installs? [18:18] i get that in a guest session... or at least i did in the past [18:19] the keyring is usually created when you first run empathy or evo [18:19] it asks then your keyring password [18:19] you know the dialog with 2 entries to confirm the password [18:19] right [18:19] so that isn't working? [18:19] anyway I've tested that the libgnomekeyring http://git.gnome.org/browse/libgnome-keyring/commit/?id=5e37e8cc09712fd8cab60e42636f260f23bacd7e change fixes it [18:20] great [18:20] thx [18:21] np [18:21] that probably breaks empathy too [18:21] yes [18:21] I tried empathy to create the keyring [18:22] which failed too :p [18:22] kenvandine, anyway enjoy your lunch [18:22] thx [18:22] * kenvandine goes to finish eating :) [18:22] seb128, did you see that there is a desktopcouch branch that should fix up the other startup issues? [18:23] kenvandine, I didn't follow details [18:23] seems to work well :) [18:24] cool [18:36] good evening everyone [18:36] good evening chrisccoulson ;) [18:36] hey didrocks [18:37] have you had a good day? [18:37] chrisccoulson: well, yeah, good but discovered that my *love* for wallpaper cache isn't finished :) [18:37] * didrocks hugs netbook-launcher and ubiquity install mode :) [18:38] heh ;) [18:38] yours? ;-) [18:38] yeah, it was ok thanks [18:38] i'm in wind-down mode now :) [18:38] heh [18:38] time for dinner, bbl [18:38] good night everyone! [18:39] have a good night pitti [18:52] reviewing app-install-data updates is always fun, I just came accross "happydigger" - Program for cataloging archaeological finds" what a well choosen name :) [18:53] lol [19:10] colorname … woah [19:12] the description sounds clever :) [19:12] its pretty cool [19:16] pitti, should we think about removing nouveau-kernel-source from the archive since we're loading it from l-b-m? === nick_ is now known as Guest35451 [19:22] bryceh: so, when I use Alt + 7, I got a square around the current windows like when you press alt + tab. I have the nvidia-current package installed, any idea how to debug that? (it just happened when switching to nouveau and then to nvidia). I can't change the WM too to compiz [19:24] didrocks, just debug it in the usual ways [19:24] bbiab (lunch) [19:24] ok === mpt_ is now known as mpt [20:01] didrocks - have you figured out what causes your issue yet? (with alt+7) [20:01] chrisccoulson: no, I hadn't the time to debug it. But I can't run compiz as well anymore. So, switching to the ubuntu nvidia driver package had caused those regression to me [20:02] I'm waiting for another user to confirm that as well [20:02] that's a strange one [20:02] perhaps i should try installing lucid on my desktop, which has nvidia hardware [20:02] although, i don't think my gf would like that too much [20:05] chrisccoulson: heh, don't make her angry :) [20:05] I try to catching up with things I have to do first ;) [20:44] does anyone here use any virtualization other than virtualbox? [20:44] chrisccoulson, i use kvm [20:45] kenvandine, is that quite easy to set up? [20:45] no :) [20:45] i use virtualbox on my desktop, primarily because it's quite old and other virtualization technologies suck on it [20:45] not compared to virtualbox [20:45] but not bad [20:45] use virt-manager [20:45] but i'm wondering if i should try something else on my laptop [20:45] i'll have a look at virt-manager [20:45] thanks [20:45] virt-manager isn't bad [20:45] UI around kvm [20:46] * seb128 uses kvm [20:46] seb128 - do you find it works quite well too? [20:46] yes [20:46] for what I do at least [20:47] i might try setting that up on my laptop [20:47] which is mainly testing on fresh iso boots [20:47] and doing iso testing for milestone too [20:47] i tried it on my desktop, but it sucks on there because my CPU doesn't have the necessary extensions [20:47] but my laptop is obviously much newer:) [20:48] yeah [20:48] just activate the option in the bios [20:48] and run kvm on an iso [20:48] kvm -cdrom .iso [20:49] you might want to allow some extra memory too I think the default is low [20:49] cool, thanks [20:49] i'll give that a try this evening :) [20:57] chrisccoulson: I use "testdrive" to test ubuntu things in a VM, it's a convenience wrapper around kvm [20:57] chrisccoulson: launchpad.net/testdrive [20:57] chrisccoulson: it'll update your isos, etc. [20:58] jcastro, thanks. i'll take a look at that too [21:21] seb128 - my menu bug is back again now [21:22] and i've just noticed something [21:22] oh? [21:22] i've got 2 cache files in /usr/share/applications, with slightly different names [21:22] one of them is updated with all the latest applications, and the other one isn't [21:22] what names? [21:23] i have a "desktop.en_GB.utf8.cache" and a "desktop.en_GB.UTF8.cache" [21:23] what and the UTF is wrong? [21:24] the UTF8 one has an application in it which i just installed, and the utf8 one doesn't [21:24] and the application is missing from my menu [21:24] what LC_MESSAGES do you use? [21:25] seb128 - "en_GB.utf8" [21:25] and the user who doesn't have the bug? [21:26] they're both using the same LC_MESSAGES, so that still doesn't explain why the user accounts were behaving differently [21:27] seb128 - actually, i'm wrong [21:27] you can do MENU_VERBOSE=1 gmenu-simple-editor [21:27] the other account has "en_GB.UTF8" [21:27] so there's the issue [21:27] chrisccoulson, ah! [21:28] the question is "why" now [21:28] yeah. technically, the latter locale is not supported on my system [21:28] "locale -a" doesn't list it [21:31] python-gmenu.postinst is weird [21:31] eval `locale|grep LC_MESSAGES` [21:31] cache="/usr/share/applications/desktop.$LC_MESSAGES.cache" [21:32] so it depends of the locale when the update runs [21:32] and it will cache only one locale [21:32] need to talk to pitti about that [21:32] yeah, it seems like it should do it for all supported locales [21:33] gdm sets the environment doesn't it? [21:33] I was going to say that [21:33] i'm trying to trace it back, and figure out where this bogus value comes from [21:33] $ locale -a | grep en_ | wc -l [21:33] 16 [21:33] *shrug* [21:34] we should maybe create a cache dir [21:34] chrisccoulson, I think gdm does yes [21:34] seb128 - yeah, i just grep'd for "LC_MESSAGES" in the gdm source, and it seems the greeter sets it in several places [21:36] what Language do you have in your dmrc? [21:36] in /var/log/gdm/ [21:37] ah, i was just about to ask where my dmrc is now, as it's not at ~/.dmrc [21:37] it has been moved which doesn't need a password to be unlocked ;-) [21:38] +in a dir [21:38] seb128 - my dmrc has the correct locale "en_GB.utf8" [21:38] and the broken account has no dmrc [21:40] time to attach gdb to the greeter now so i can work out where that value comes from [21:40] chrisccoulson, not gdm I guess [21:41] yeah, possibly not, seeing as dpkg has seen the bogus locale at some point too [21:41] trying to start a session without gdm it has UTF-8 [21:43] yeah, i confirm that too [21:43] just logging in to a console [21:44] seb128 - it's the same on karmic though [21:44] so perhaps the value in my dmrc is the wrong one :-/ [21:44] i'm confused now [21:44] it's utf8 there too [21:45] seems it's what gdm do [21:45] so, gdm is messing up the case somewhere then [21:45] the cache should probably handle any utf(-)8 without case [21:45] well I've an extra "-" too on the non gdm session [21:46] and listing the locales using locale use "utf8" too [21:47] yeah, i get the same too [21:47] I think the easier would be to make the cache work with any case and "-" [21:47] rather than trying to normalize the values [21:48] yeah, that would make sense [21:48] iz pitti bog [21:50] yeah, we can bug him in the morning ;) [21:50] well, i'm glad we understand that one now [21:51] yeah [21:51] hey robert_ancell [21:51] seb128, hello [22:04] * kenvandine heads out for some family time [22:04] bbl [22:05] yay, kvm is working :) [22:12] kvm is easy to setup [22:12] I used it this week end to validate LTS upgrade [22:14] wowo I have my notification area wich is moving to the left as comes and disappear the g-p-m battery icon [22:31] hum [22:31] g_printerr ("Error opening %s: %m\n", device_file); [22:31] how does that work? [22:32] seb128, yuck, what is that from? [22:32] robert_ancell, udisks [22:32] hey robert_ancell, that's a long time o/ [22:32] not a good sign... [22:32] didrocks, hey [22:32] robert_ancell, there is plenty of those, but no compiler warning in the build log [22:32] I'm not sure why [22:32] I feel there is something I don't get [22:33] " g_printerr ("Error seeking to position %" G_GSSIZE_FORMAT " for %s: %m\n", [22:33] pos, [22:33] device_file);" [22:33] too for example [22:34] does %m automagically get replaced with strerror(errno) [22:34] ? [22:34] seb128: %m: (Glibc extension.) Print output of strerror(errno). No argument is required. (from man 3 printf) [22:34] that's handy [22:35] geser, robert_ancell: thanks [22:35] learning every day ;-) [22:40] heh. i didn't know that either ;) [22:41] seb128 - you're looking at udisks crashes? [22:42] chrisccoulson, to one yes [22:42] bug #527202 [22:42] Bug 527202 on http://launchpad.net/bugs/527202 is private [22:42] I though it would be something easy maybe [22:43] often those "crash when printing an error" are a GError not set to NULL [22:43] or similar [22:44] seb128 - the string passed to printf not being NULL terminated is another common cause [22:53] kenvandine: My gwibber just crashed but apport didn't fire off. Looking in /var/crash I see a bunch of gwibber related .crash files [22:53] kenvandine: should I file a bug by hand or ...? [22:55] seb128: hey? [22:56] desrt, hello [22:56] seb128: have you been shipping unstable glibs? [22:56] seb128: hi :) [22:56] desrt, yes [22:56] sweet [22:56] desrt, we have 2.23.4 [22:56] jcastro, double click on those in nautilus? [22:56] so, implicitly, you'll have .24 in lucid [22:56] jcastro, double click on those in nautilus? [22:56] ups [22:57] desrt, yes [22:57] ok. i'll try not to abuse you too much, then :) [22:57] seb128: I had no idea I could do that, that's awesome [22:58] desrt, good work on gvariant btw! [22:58] seb128: not done just yet :) [22:58] two small parts remain [22:59] i've already gotten email about "wtf?!? you removed varargs support?!?" [22:59] "no. just not landed yet." [22:59] seb128: i guess we're really far past new package freeze? [23:01] desrt, not far === robbiew is now known as robbiew_ [23:01] eh. probably better off with a PPA anyway [23:01] desrt, I would say that new packages in universe are non issue [23:01] not a potential to break a lot [23:01] i made a test coverage tool [23:02] using a ppa works good too :-) [23:02] ya. i'll do that. it's easy enough. [23:02] it'll be approximately 4 lines worth of dh rules :p [23:03] it's a vala (read: .c in the tarball) autotools package that depends on gtk and installs a single binary file [23:04] even i can handle that one :) [23:19] robert_ancell: I hear you're the guy with whom I need to chat about a polkit problem I'm having. This is bug number 485586. [23:19] seb128 - i'll put the latest gnome-user-share in to the ubuntu-desktop PPA if you want to do some benchmarks with the new nautilus extension (if i get the chance to before i go to bed) [23:19] did you get your mini fixed btw? [23:19] DanEngholm, looking... [23:20] robert_ancell: Thanks. [23:20] bug #485586 [23:20] Launchpad bug 485586 in policykit "PolicyKit: "Not Authorized to make changes"" [Undecided,New] https://launchpad.net/bugs/485586 [23:21] chrisccoulson, ok thanks [23:21] chrisccoulson, yes [23:21] chrisccoulson, the mini, yes [23:21] chrisccoulson, not reinstalled yet though I'm fighting with ubiquity [23:21] DanEngholm, perhaps you could post the output of "ck-list-sessions" to the bug report (from the session that has the issues) [23:21] I spent my afternoon filling ubiquity crash bugs [23:21] and waiting for new isos [23:22] seb128 - it's good that you got it fixed now :) [23:22] yeah ;-) [23:22] DanEngholm, so what changed when the problem started? [23:22] bratsche, bug #527431 [23:22] Launchpad bug 527431 in gtk+2.0 "gtk_window_set_decorated(window, FALSE) does not work when called before showing the window" [Undecided,New] https://launchpad.net/bugs/527431 [23:22] bratsche, could you look at that bug? [23:24] robert_ancell: I'm not sure. I did put /etc under Subversion control but I did it in place. That is, all the files remained, just the .svn directories got added. Other than that, I have no idea. [23:24] DanEngholm - thats a known issue [23:24] Oh? [23:24] the svn files in the seats.d folder screws up consolekit [23:24] Ah hah! [23:24] and all local sessions get added to dynamic seats [23:25] I can easily cull them out. [23:25] DanEngholm, just to confirm though, please run "ck-list-sessions" [23:25] Will do... Working... [23:25] i suspect that you are not on Seat1... [23:25] and that is the issue [23:25] chrisccoulson, nice [23:26] robert_ancell, yeah, it seems CK is quite fragile there [23:26] chrisccoulson, is that fixed for Lucid? [23:26] robert_ancell, not yet AFAIK [23:26] chrisccoulson, is there a CK bug on that? [23:26] i'll check in a bit though. i know it's fixed in it [23:26] s/it/git [23:27] robert_ancell, yeah, there's a bug [23:27] Posted and you're right. Seat = 'Seat4'. [23:27] i'll just check my bug mail [23:27] DanEngholm, cool. thats the issue [23:27] chrisccoulson, sweet, please update bug #485586. and we should take that change into Lucid if it is safe [23:27] Launchpad bug 485586 in policykit "PolicyKit: "Not Authorized to make changes"" [Undecided,New] https://launchpad.net/bugs/485586 [23:27] OK. Is the only fix now to get rid of .svn files from some director[y/ies]? [23:28] DanEngholm, yeah. i'll tell you which folder in a second [23:28] OK. [23:29] robert_ancell, bug 478274 [23:29] Launchpad bug 478274 in consolekit "cannot mount usb volumes: "Not authorized", after upgrade to Karmic" [Unknown,Fix released] https://launchpad.net/bugs/478274 [23:30] DanEngholm, /etc/ConsoleKit/seats.d [23:31] seb128: Sure. [23:31] bratsche, thanks [23:32] DanEngholm, you will need to restart afterwards (you could try just restarting consolekit, but I wouldn't really recommend that) [23:32] chrisccoulson, nice work [23:32] seb128: Already fixed in my next patch. [23:33] bratsche, good I was wondering if that was the same issue you had yesterday [23:33] bratsche, thanks [23:38] seb128 - OOI, did those gsd and gpm warnings in your xsession-errors go away with the latest versions? [23:39] chrisccoulson, yes [23:39] cool,thanks [23:39] "GLib-CRITICAL **: g_propagate_error: assertion `src != NULL' failed" [23:39] still get that gsd one though [23:39] seb128 - yeah, i'm going to look at that one too [23:39] i thought i'd fixed that [23:39] chrisccoulson, thanks to you for the fix ;-) [23:39] but it seems not ;) [23:40] what gpm change do you work on? [23:40] just asking because I'm wondering if we should upload the gpm icon fix [23:40] it's easy enough... [23:40] the only other gpm change i'm looking at is gnome bug 609720 [23:40] Gnome bug 609720 in gnome-power-manager "Can sometimes miss idle reset alarm, causing display to blank when it shouldn't do" [Normal,Assigned] http://bugzilla.gnome.org/show_bug.cgi?id=609720 [23:40] but that's not ready yet, so we should upload the icon fix [23:41] btw, the nautilus icon has the same issue as the gpm one [23:43] oh? [23:43] that's weird [23:50] bratsche, bug #527431 have you read the new comments about vmware? [23:50] Launchpad bug 527431 in gtk+2.0 "gtk_window_set_decorated(window, FALSE) does not work when called before showing the window" [Low,In progress] https://launchpad.net/bugs/527431 [23:50] bratsche, we might want to make sure to test those before beta [23:50] chrisccoulson: Thanks. I'm restarting now. Will BRB to let y'all know how it worked. [23:51] ArneGoetje, bug #407300, does that break something? [23:51] Launchpad bug 407300 in ubuntu-translations "/etc/gdm/Xsession breaks LANGUAGE" [Low,Confirmed] https://launchpad.net/bugs/407300 [23:51] not sure why you assigned it to the team now [23:52] it has been sent upstream and with open questions [23:56] Ahhhhhh..... All better. Thanks! [23:59] good night everybody