[00:05] <Keybuk> heh, gotta love the forums
[00:05] <Keybuk> someone whinging that plymouth isn't working, and it should be removed
[00:05] <Keybuk> reply asking if they might record a video of what they see
[00:06] <Keybuk> they reply "I'm so angry! I don't deserve this abuse" type rant
[00:06] <ion> You visited the forums and survived?
[00:11] <Caesar> egad
[00:11] <YokoZar> It looks like Transmission 1.92 fixes a few serious issues with 1.91, including a few security-related things.  1.91 is also apparently banned from a few trackers...
[00:12] <Caesar> Keybuk: I love how Plymouth no longer looks like utter crap on systems with Nvidia graphics
[00:12] <Caesar> It's very shiny now
[00:13] <Keybuk> yeah nouveau ftw
[00:15] <Caesar> Oh that's nouveau doing that?
[00:15] <Caesar> Nice
[00:18] <kklimonda> YokoZar: we know, it's going to be uploaded soon
[00:18] <kklimonda> well, probably after the weekend as chrisccoulson is probably sound asleep already :)
[00:18] <chrisccoulson> sleep? me?
[00:18] <chrisccoulson> lol
[00:18] <kklimonda> :D
[00:19] <chrisccoulson> i might look at it before i go to bed ;)
[00:23] <Keybuk> oh
[00:23] <Keybuk> wow
[00:23] <Keybuk> if I span the VGA rows *this way* it goes much faster
[00:23] <Keybuk> you can almost not see it draw now ;-)
[00:44] <directhex> nouveau doesn't work for me as far as i can tell. i could re-try i suppose.
[00:46] <Keybuk> directhex: which card?  what does it do?
[00:46] <Keybuk> it seems that many problems can be solved by sending upstream an mmiotrace of nvidia-glx starting X on your card
[00:46] <directhex> gtx260, and x just wouldn't start.
[00:46] <Keybuk> they send patches back
[00:47] <directhex> let me try rebooting, and if it's buggered, you can talk me through that
[00:49] <Keybuk> it involves compiling kernels atm
[00:50] <directhex> hm, i get x now. but no compiz, sadly
[00:50] <directhex> and boot time is still pretty slow... probably 15-20 secs between grub and plymouth starting
[00:51] <Keybuk> oh, you won't have compiz with nouveau
[00:51] <Keybuk> it's 2D only
[00:54] <Keybuk> coo
[00:54] <Keybuk> my mad optimised VGA code is *much* faster
[00:55] <Keybuk> http://pastebin.ubuntu.com/398090/
[00:56] <directhex> well, at least it displays something... but i value compiz more than pretty plymouth
[00:56] <Keybuk> heh
[00:57] <Keybuk> you value compiz more than you value FREEDOM!
[00:59] <directhex> yes. correct. free software always has the *potential* to be the best there is. how true that may be in actual terms varies somewhat per project
[01:02] <directhex> there we go, nvidia-glx is back
[01:02] <directhex> and now, bedtime
[01:14] <smoser> Keybuk, are you actually around
[01:14] <smoser> ?
[01:14] <Keybuk> smoser: yeah
[01:15] <smoser> you mentioned getting a /bin/sh shell on console early in boot to help debug.
[01:15] <smoser> how would i do that ?
[01:15] <Keybuk> I've no idea
[01:15] <Keybuk> it's your system
[01:15] <Keybuk> I know nothing about cloud
[01:15] <smoser> no. from upstart/boot perspective
[01:15] <smoser> how would i enable it.
[01:15] <smoser> i tried:
[01:16] <smoser> start on starting
[01:16] <smoser> task
[01:16] <smoser> console output
[01:16] <smoser> exec /bin/sh
[01:16] <Keybuk> init=/bin/bash on the kernel command-line is always favourite
[01:16] <smoser> (which i realize is busted -- the exec)
[01:16] <Keybuk> the exec is fine
[01:16] <Keybuk> nothing wrong with that
[01:16] <Keybuk> though you mean "start on startup" I think ;-)
[01:16] <smoser> i figured the fact that it was task would mean it wouldnt come back
[01:16] <smoser> hmm... whats startup versus staring ?
[01:16] <smoser> i've used starting before
[01:16] <Keybuk> task makes no difference
[01:16] <Keybuk> man 7 starting
[01:16] <Keybuk> man 7 startup
[01:17] <Keybuk> starting is to do with jobs
[01:17] <Keybuk> startup is the first event emitted by upstart
[01:17] <smoser> ok
[01:18] <smoser> so, the init=/bin/bash route. i can get there, but then dont know hwere to go from there.
[01:18] <smoser> exec /sbin/init
[01:18] <smoser> is hardly going to be helpful :)
[01:18] <Keybuk> right, set up debugging, then exec init
[01:31] <Keybuk> oh, wow
[01:31] <smoser> Keybuk, so with the above script in place, to launch 'sh'
[01:31] <Keybuk> the "solar" theme on a 16-color framebuffer is simply glorious
[01:31] <smoser> will it block the rest of boot ?
[01:32] <Keybuk> no
[01:36] <smoser> Keybuk, so if i got you a console on such a system with init=/bin/bash, could yo upoke around ?
[01:36] <Keybuk> yes
[01:36] <smoser> answering "dude, its friday night" is ok
[01:36] <smoser> :)
[01:36] <Keybuk> if you can give me a console on a system that's "hung" is great
[01:36] <Keybuk> but it won't be right now ;)
[01:36] <Keybuk> cause I'm literally off to bed in minutes
[01:37] <smoser> ok. I'll ping you monday.
[01:37] <genii> dude. It's Friday evening!
[01:37] <Keybuk> but set one up for me on monday!
[01:37] <smoser> you're UK now ?
[01:37] <Keybuk> I'm usually UK
[01:37] <smoser> yeah. i was just not expecting you here at 1:40 am
[01:37] <Keybuk> of course, my sleep pattern is according to the timezone of fairy land
[01:37] <Keybuk> my mother is snoring too loud, can't sleep
[01:37] <smoser> right. thanks. will bother you monday.
[01:45] <happyaron> jcastro: ping
[01:46] <jcastro> happyaron: pong
[01:53] <Caesar> jcastro: congrats
[01:53] <jcastro> Caesar: thanks!
[02:00] <jcastro> robbiew: awake?
[02:05] <jcastro> anyone awake? nautilus and gnome-panel are broken on login
[02:06] <jcastro> should probably fire off a warning on the twitter thing
[02:06] <kklimonda> jcastro: actually a lot of things is broken as all .desktop files which have X-Ubuntu-Gettext-Domain are malformed
[02:07] <kklimonda> at least those uploaded after beta1 freeze
[02:12] <jcastro> kklimonda: can you help me find the package that caused this? And if there's a bug?
[02:12] <LaserJock> is it a problem with gnome-pkg-tools?
[02:12] <jcastro> I called robbie, so he's aware
[02:13] <jcastro> so if we can help find out where it is exactly
[02:13] <LaserJock> somehow when X-Ubuntu-Gettext-Domain is added it's getting added in the wrong place
[02:14] <LaserJock> the .desktop files in the source packages are fine for the limited cases I checked
[02:15] <robbiew> jcastro: done
[02:15] <jcastro> https://bugs.edge.launchpad.net/ubuntu/+source/gnome-panel/+bug/525240
[02:15] <jcastro> robbiew: this seems to be the bug
[02:15] <robbiew> ack
[02:17] <jcastro> hmm, no, that can't be right
[02:17] <jcastro> mike saw the bug but I wonder if it's the same, he attached his stuff to an older bug
[02:19] <LaserJock> seems like it
[02:19] <LaserJock> that is a crash, right?
[02:19] <kklimonda> jcastro: 525240 is too old for this being the same issue - bug 542343 is more like it but it has no real data
[02:19] <jcastro> https://bugs.edge.launchpad.net/ubuntu/+source/gnome-panel/+bug/542343 too
[02:19] <jcastro> kklimonda: *nod*
[02:26] <Caesar> cjwatson: you still around?
[02:27] <LaserJock> jcastro: somebody maybe want to update the topic in ubuntu+1?
[02:27] <happyaron> well, and a desktop short cut to gnome-terminal, which helped me out temporary
[02:27] <jcastro> alt-t
[02:28] <happyaron> oh, I trid alt+f2 and it doesn't work
[02:28] <jcastro> yeah, alt-f2 is provided by the panel
[02:31] <kenvandine> yo yo
[02:31] <jcastro> hi kenvandine
[02:34] <jcastro> kenvandine: any place I can look to help find the problem?
[02:34] <kenvandine> gnome-session[1341]: WARNING: Could not parse desktop file /etc/xdg/autostart/indicator-applet.desktop: Key file does not start with a group
[02:34] <kenvandine> that looks weird
[02:35] <jcastro> I get that too
[02:35] <chrisccoulson> jcastro - it's a problem with the new langpack.mk
[02:35] <kenvandine> hey chrisccoulson
[02:35] <chrisccoulson> i reassigned bug 542343 accordingly
[02:35]  * kenvandine is trying to repro in a VM
[02:36] <LaserJock> chrisccoulson: when did that get introduced? I couldn't find it
[02:36] <chrisccoulson> LaserJock, yesterday just before gnome-panel
[02:40] <kenvandine> i just verified that before updating, i do not have those autostart errors in .xsession-errors
[02:42] <kenvandine> ok, i am seeing parse errors during the upgrade
[02:44] <kenvandine> http://pastebin.ubuntu.com/398118/
[02:44] <kenvandine> chrisccoulson, ^^
[02:44] <kenvandine> chrisccoulson, they have the entry
[02:46] <chrisccoulson> right
[02:46] <chrisccoulson> basically every package built with the new cdbs version
[02:47] <kenvandine> ugh!
[02:48] <kenvandine> this is exactly what seb128 was worried about... with the archive opening up on a friday evening :/
[02:48] <chrisccoulson> yeah, i know :(
[02:48] <jcastro> that's exactly what he told us
[02:48] <chrisccoulson> all of these packages will need a rebuild once cdbs is fixed
[02:48] <kenvandine> seb128 is ALWAYS right
[02:48] <chrisccoulson> plus any other package which is uploaded
[02:48] <kenvandine> chrisccoulson, ok... is someone working on that?
[02:48] <chrisccoulson> kenvandine, everyone is asleep now i think
[02:49] <jcastro> kenvandine: don't think so, you two are the only ones here
[02:49] <kenvandine> the baby just got sick again... i need to go :(
[02:49] <chrisccoulson> pitti did the initial change, so i subscribed him to the bug
[02:49] <jcastro> go take care of your kid
[02:49] <wgrant> Is it worth flipping the buildds onto manual to stop any more damage?
[02:49] <kenvandine> i couldn't fix cdbs anyway... i don't think
[02:49] <kenvandine> wgrant, yes
[02:50] <jcastro> robbiew_: what should we be doing?
[02:50] <kenvandine> sigh... well i'll try to check back in later
[02:50] <ScottK> wgrant: It doesn't affect everything.
[02:52] <chrisccoulson> only things that ship a desktop file in main
[02:52] <wgrant> ScottK: No, but it can be fixed within a publisher cycle if people are quick.
[02:52] <ScottK> chrisccoulson: Not all packages use CDBS.
[02:52] <chrisccoulson> yeah, that's true
[02:54] <ScottK> The gettext stuff is, IIRC done differentl for KDE pacakges that still use CBDS, so that narrows the field further I think.
[02:54] <ScottK> Toss a "y" in there somewhere for better spelling.
[03:04] <emet> does the software store have an API/commandline args
[03:05] <emet> I notice apt-url uses synaptic command line to issue package installation
[03:08] <LaserJock> good luck guys, I'm off to bed
[03:25] <psusi> hrm... seems plymouth still doesn't say press enter to reboot once the cd is ejected on beta 1 cd... also the dmraid bug still really needs fixed before release
[03:29] <crypt-0> the latest kernel update broke grub it sets "	set root=(hd3,1)" instead of "	"set root=(hd2,1)" manual change no luck...
[03:29] <crypt-0> error message cant find /dev/mapper/cryptroot
[03:48] <crypt-0> anyone alive?
[03:48] <persia> Most folk.
[03:49] <persia> Some aren't, but that's just a matter of semantics regarding electronic entities.
[04:01] <crypt-0> persia, any idea to the above problem?
[04:03] <persia> None, but I'd suggest that #ubuntu+1 was a more appropriate forum to ask about issues with updates.
[04:03] <persia> (or #ubuntu if you're running supported releases)
[04:07] <crypt-0> yeah, no one has  a clue there...
[04:07] <persia> Doesn't make this a support channel though.  Not sure where to direct you.
[04:11] <crypt-0> i know
[04:11] <crypt-0> sometimes people answer me though
[04:12] <crypt-0> ...ive never haday help from #ubuntu
[04:12] <crypt-0> just blank stares
[04:19] <mneptok> well, to be fair, you haven't asked a question.
[04:19] <crypt-0> the latest kernel update broke grub it sets "	set root=(hd3,1)" instead of "	"set root=(hd2,1)"
[04:19] <crypt-0> error message cant find /dev/mapper/cryptroot
[04:19] <crypt-0> any idea to the above problem?
[04:19] <mneptok> that's a statement. not a question. and you should ask questions in #ubuntu or #ubuntu+1, as persia said
[04:19] <persia> No, it's a question.
[04:20] <persia> it's just an off-topic question, so there's a good chance nobody will answer.
[04:20] <crypt-0> manual change of grub.cfg didn't work
[04:20] <mneptok> crypt-0: please take the issue to one of the official support channels. this is the wrong channel to discuss this.
[04:20] <crypt-0> there is a god chance no one will  answer in #ubuntu been over  an hour
[04:21] <mneptok> that does not make this a support channel.
[04:33] <persia> So, the GNOME desktop is broken.  I've seen discussion in backscroll, and in other channels.  #ubuntu-desktop logs show silence on the issue.  Am I correct that the fix is to revert the changes to address bug #535650, and reupload some stuff?  Has anyone already written a script to determine affected packages in launchpad?
[04:39] <wgrant> It seems odd that everybody seems to have gone to sleep with the world broken and lots of users affected.
[04:40] <persia> Indeed.  At this point, I'm willing to just fix it, but I want to make sure I'm actually fixing it.
[04:40] <wgrant> Anything built in main in the last 8 hours depending on CDBS needs to be considered.
[04:40]  * wgrant has a look.
[04:43]  * happyaron perhaps we can send a mail to mailing lit
[04:43] <happyaron> list
[04:43] <persia> ajmitch, TheMuso, Hobbsee, StevenK: any of you currently about?
[04:43] <wgrant> The Europe and the US are asleep. That's unlikely to help.
[04:43] <happyaron> wgrant: we they are awake they may see the alert
[04:44] <wgrant> That's a long time away yet.
[04:44] <persia> happyaron: It will be fixed by then,.
[04:44] <wgrant> Particularly as we are entering a weekend.
[04:44] <happyaron> oh
[04:44] <persia> In the worst case, I'll temporarily make myself core-dev (although I'd much rather some actual core-dev was around)
[04:44] <happyaron> :)
[04:46] <happyaron> people in Asia have been awake, and reports are rushing into our LoCo forum...
[04:46] <persia> Indeed.  And lots of people in the Americas have gone to bed frustrated with broken envionments.
[04:47] <persia> It mostly affects GNOME (login fails), but it also affects all the CDBS hacked .desktop files in the archives, so lots of applications are disappearing from menus.
[04:47] <lifeless> persia: you're not core-dev?
[04:47] <persia> lifeless: I'm not.
[04:48] <lifeless> persia: but you can make yourself core-dev ?
[04:48] <persia> Yes.
[04:48] <lifeless> *blink*
[04:48] <wgrant> DMB?
[04:48] <persia> I'm not supposed to do that, but I'm willing to take the risk of losing my DMB appointment for this.
[04:48] <persia> wgrant: Yes.
[04:49] <lifeless> well, I'm in the US right now
[04:49] <wgrant> We may have a core-dev or two incoming.
[04:49] <persia> That would be good.
[04:49] <lifeless> but if this is urgent, I'd escalate
[04:49] <lifeless> e.g. by ringing stevenk's mobile ;)
[04:49] <lifeless> speaking of which, gnight.
[04:49] <persia> gnight.
[04:49] <mneptok> it's not *that* late in parts of the US (2249 here in MDT)
[04:50] <lifeless> mneptok: I'm in boston
[04:50] <lifeless> nearly 1am
[04:50] <persia> mneptok: Do you know some core-devs in that timezone?
[04:50] <mneptok> lifeless: sorry to hear it. :P
[04:50]  * happyaron sorry, what's DMB?
[04:50]  * StevenK_ looks in
[04:50] <persia> mneptok: Can you make them show up?
[04:50] <dmb> who me?
[04:50] <persia> StevenK: \o/
[04:50] <persia> dmb: No.  Developer Membership Board.
[04:50] <mneptok> persia: not sure if there are core devs in MDT, but there sure are in PDT
[04:50] <wgrant> Ah good, my poking worked.
[04:50] <persia> StevenK: GNOME is broken.  Please fix :)
[04:51] <wgrant> StevenK: Bug #542343
[04:51] <dmb> :)
[04:51] <persia> wgrant: Did you get the list?
[04:51] <wgrant> CDBS needs reverting, and I'm working on a list of stuff that needs rebuilding.
[04:52] <StevenK> So pitti's fix that was uploaded 8 hours ago is wrong?
[04:52] <persia> At best, incomplete.
[04:52] <wgrant> It's the one that broke it, it seems.
[04:52] <persia> It fixes that one case, but breaks GNOME login.
[04:55]  * StevenK looks at 1/rules/langpack.mk.in and then wishes he hadn't
[04:55] <wgrant> +1
[04:56] <StevenK> So I should just revert the ubuntu6 changes or do I need to do more?
[04:56] <persia> Reversion is the easiest way out.
[04:57] <persia> Causes a regression for bug #535650, but in this case the cure is worse than the disease.
[05:00] <StevenK> persia, wgrant: cdbs uploaded
[05:00] <wgrant> StevenK: Thanks.
[05:00] <wgrant> LP is being stupid.
[05:00] <persia> StevenK: Thank you.
[05:02] <StevenK> Do I need to re-upload evolution in a while?
[05:03] <wgrant> And lots of other stuff...
[05:03] <StevenK> Give me a list of stuff in main and I'll sort it out
[05:03] <persia> But we have to wait for the new CDBS to be on ftp.internal (or whatever it's called) first.
[05:04] <StevenK> Absolutely, I know
[05:04] <wgrant> Which will be around 80 minutes.
[05:16] <wgrant> Oh, there's no actually that much broken.
[05:16] <wgrant> I just grepped every binary built in the last 9 hours.
[05:17] <persia> From my quick review of -changes, it looked like 6-7 packages.
[05:20] <StevenK> How many are in main?
[05:20] <wgrant> All the affected binaries are in main.
[05:20] <wgrant> Since langpack.mk should only affect main.
[05:20] <StevenK> Ah. What's the list?
[05:21] <wgrant> Just rechecking.
[05:21] <wgrant> http://paste.ubuntu.com/398158/ is the list of binaries that could be affected (they have the X-Ubuntu-Gettext-Domain mangling)
[05:21] <wgrant> Some of them aren't actually broken, at least on i386, which is odd. Maybe they were built early enough.
[05:22] <StevenK> Ooh, and ubiquity. Nice.
[05:22] <wgrant> Arch skew is going to be annoying, since cdbs is arch-indep.
[05:24] <wgrant> ubiquity is OK, at least on i386.
[05:25] <wgrant> Oh, and it's arch-all, so it's fine.
[05:27] <wgrant> http://paste.ubuntu.com/398160/ appears to be the full broken list for i386. My grep on other archs was flawed (they were rather behind, so more may be affected).
[05:29] <StevenK> That's source or binary?
[05:29] <wgrant> And the LP API sucks, so I can't filter on builds completed since X, nor can I get files for binaries published since X.
[05:29] <persia> wgrant: Is that grep against built-binaries, or against the build logs?
[05:29] <wgrant> That's against built binaries.
[05:29] <wgrant> But those are source names.
[05:33] <persia> Even if that isn't a complete list, it's likely enough to stop the noise.
[05:33] <wgrant> Right.
[05:33] <persia> And the rest can be hunted later.
[05:33] <wgrant> gnome-panel and nautilus are the really important ones.
[05:33] <persia> I've heard a bunch of complaints about rythmbox and totem because they disappear from the menus.
[05:33] <wgrant> Right.
[05:34] <wgrant> evolution too, but since everyone except me seems to use Gmail that shouldn't be less of a problem.
[05:34] <wgrant> Er, should be less of a problem.
[05:36] <persia> Rather, I think the noisest complainers are those who want their movies and music now.
[05:36] <persia> I probably use gedit more than anything else on that list, personally.
[05:49] <wgrant> Ah, activity on the bug.
[05:49] <wgrant> Rick has assigned it.
[05:50] <wgrant> But I suspect that won't be acted upon for quite a few hours yet.
[05:51] <StevenK> I suspect I should update it
[05:51] <wgrant> I've just commented.
[05:51] <persia> That might be good.
[05:55] <wgrant> If only the publisher was not so slow.
[05:55] <persia> Isn't there some manual mode that lets it be triggered on-demand?
[05:55] <persia> Or wouldn't that help in this case.
[05:56] <wgrant> It would have only helped if we had disabled it just before cdbs was uploaded (thus avoiding a publisher run), and run it manually once cdbs had built.
[05:56] <wgrant> It would have saved around half an hour.
[06:16] <wgrant> StevenK: ftpmaster.internal is cocoplum itself, isn't it?
[06:23] <ajmitch> persia: did I miss something?
[06:24] <joe_> After upgrading to Lucid, I was unable to access my network drives with a message that some obsolete pkg version were installed. I then used the update manager to see if any updates were available and found the ones previously mentioned were among those recommended. I then found I could only do a Partial update and after doing so I could open my Home Folder icon, so I rebooted and now only...
[06:24] <joe_> ...have an empty Desktop. Any help available to resolve this?
[06:24] <wgrant> joe_: #ubuntu+1 for support please.
[06:24] <wgrant> But this bug is known, and should be mostly fixed in about two hours.
[06:24] <wgrant> joe_: For now, press Alt+T, and run gnome-panel.
[06:24] <wgrant> That should get things mostly working.
[06:24] <persia> ajmitch: bug #542343 : I was calling core-devs to find a volunteer to fix.  StevenK was faster than you.
[06:25] <ajmitch> persia: ah good, I'm glad he got to it :)
[06:25] <joe_> wgrant; Thanks
[06:26] <joe_> After upgrading to Lucid, I was unable to access my network drives with a message that some obsolete pkg version were installed. I then used the update manager to see if any updates were available and found the ones previously mentioned were among those recommended. I then found I could only do a Partial update and after doing so I could open my Home Folder icon, so I rebooted and now only...
[06:26] <joe_> ...have an empty Desktop. Any help available to resolve this?
[06:26] <persia> joe_: ?  misplaced paste?
[06:27] <joe_> persia: Yes, sorry.
[06:50] <wgrant> persia, StevenK: Mirrors should be syncing in a minute or two, and except for possibly builds completing in the last hour, the source package list I gave is complete across all architectures.
[06:50]  * persia looks at recent builds
[06:54] <persia> I thik update-notifier on sparc is likely also munged, but since last I heard no lucid kernel booted on any sparc anyone had, I don't think it matters by comparison.
[06:55] <wgrant> I'll recalculate the list later, including everything that finished between two hours ago and whenever the buildds catch up after now.
[06:56] <wgrant> Odd -- a.u.c hasn't been updated.
[06:56] <wgrant> It should have been around five minutes ago.
[06:57] <wgrant> Ah, there we og.
[06:57] <wgrant> Syncing now.
[07:00] <wgrant> And cdbs is there.
[07:00] <wgrant> Does somebody want to rebuild gnome-panel and nautilus noW?
[07:01] <wgrant> StevenK: ^^
[07:01] <persia> ajmitch: Now's your chance to be faster, if you like.
[07:02]  * persia reopens bug #535650
[07:03] <wgrant> They both take less than 10 minutes to build, so we shouldn't lose anything on i386 unless they're not uploaded in the next half hour.
[07:03] <wgrant> amd64 is less lucky, though :/
[07:03] <persia> Why so?
[07:04] <wgrant> It has one buildd stuck on linux, and 3 packages in the queue.
[07:04] <persia> That just needs a buildd admin.
[07:04] <wgrant> Still, we have a conveniently placed buildd admin who can twiddle scores for us.
[07:06] <wgrant> Hm, this is suboptimal.
[07:06] <persia> Which?
[07:06] <wgrant> Everyone seems to have run away.
[07:07] <persia> Hobbsee TheMuso StevenK ajmitch : Would someone please rebuild the packages listed at http://paste.ubuntu.com/398160/ to address bug #542343?
[07:08] <wgrant> Only gnome-panel and nautilus for now.
[07:08] <wgrant> They are critical, and the others will compete for buildd time.
[07:08] <persia> Why?
[07:08] <persia> Ah.
[07:08] <wgrant> And without Hobbsee, we cannot guarantee build order.
[07:08] <persia> So we essentially need Hobbsee, really.
[07:12] <wgrant> Attempting to contact by several means...
[07:13] <wgrant> amd64 is probably a lost cause.
[07:13] <wgrant> It's tied up for another 20 minutes. Unless we're very lucky.
[07:14] <persia> Indeed.  amd64 will be very difficult to fix this publisher run.
[07:17] <persia> Lesson for next time: when you corner a core-dev for timing sensitive stuff, get signed source packages available somewhere for independent upload later.
[07:17] <wgrant> Indeed.
[07:22] <StevenK> Sorry, I was adk
[07:22] <StevenK> *afk
[07:22]  * Hobbsee comes back too
[07:22] <wgrant> Yay.
[07:22] <StevenK> How about I just put a copy of my private key up on people.ubuntu.com? :-P
[07:22] <wgrant> You now have around 20 minutes to rebuild gnome-panel and nautilus.
[07:22] <wgrant> Or we miss another publisher run :/
[07:23]  * StevenK works on rebuilds
[07:23]  * Hobbsee fishes out buildd.py
[07:23] <wgrant> Thanks.
[07:23] <wgrant> We probably won't need rescoring.
[07:24] <wgrant> But it would be a handy facility to have just in case.
[07:24] <persia> No.  i386 has an empty queue, and we missed the window for amd64
[07:24] <wgrant> amd64 still might just happen.
[07:24] <persia> But not because of rescoring.
[07:24] <wgrant> Right.
[07:24] <wgrant> mysql-cluster should finish in a few minutes.
[07:25] <persia> Oh, excellent.
[07:25] <wgrant> Then we need about 20 minutes of buildd time from it.
[07:25] <wgrant> In the next 37 minutes.
[07:25]  * StevenK uploads gcc
[07:25]  * wgrant stabs.
[07:25]  * Hobbsee deflects wgrant's stabbing
[07:25]  * wgrant hastily patches OOo.
[07:25] <persia> actually, we could use a rescore on armel : we really don't want to have to build chromium-browser twice.
[07:25] <StevenK> I have to upload chromium?
[07:25] <Hobbsee> ok, let me know what you want rescored, and when
[07:26] <wgrant> StevenK: No.
[07:26] <persia> StevenK: We have a Hobbsee that can defend you from uploading chromium-browser
[07:26] <persia> Yeah, sparc and armel both need rescoring support for the rebuilds.
[07:27] <wgrant> They're less critical.
[07:27] <persia> If I could boot a lucid kernel on any of my armel devices, I'd disagree with you :)
[07:27] <wgrant> So, remaining steps in the world domination plan:
[07:27] <wgrant>  1) Get those two rebuilt.
[07:27] <wgrant>  2) Rebuild the rest of my list.
[07:28] <wgrant>  3) Once the queues clear, I'll rerun the check and we rebuild the stuff that is newly broken.
[07:28] <wgrant> Then all is fixed, except the original bug the fix for which broke everything.
[07:28] <persia> Which was a workaround patch for a different bugfix that broke everything.
[07:28] <StevenK> Right, uploading everything.
[07:29] <wgrant> StevenK: Just those two, I hope?
[07:29] <StevenK> No, we can rescore
[07:29] <wgrant> No.
[07:29] <StevenK> We can't?
[07:29] <wgrant> We can't terminate a build once it's building.
[07:29] <StevenK> Bleh
[07:29] <wgrant> So we need to get those two accepted first.
[07:30]  * StevenK sighs at * for not expanding in the order wgrant wanted
[07:30] <persia> and then those two get rescored on amd64 / armel / sparc, and then everything else gets uploaded.
[07:30] <wgrant> Right.
[07:30] <StevenK> Pity I've already destroyed that plan
[07:30] <wgrant> How much made it in?
[07:30] <wgrant> All of them?
[07:30] <StevenK> Yeah
[07:31]  * persia checks
[07:31] <wgrant> Hobbsee: Can you rescore the two critical ones on all archs, please?
[07:31] <wgrant> We still may have time, if the first dispatched builds are short.
[07:31] <persia> Hobbsee: Can you rescore nautilus and gnome-panel on amd64/armel/i386/sparc please?
[07:31] <persia> Oh, right.  All architectures :(
[07:31] <wgrant> Crap.
[07:31] <Hobbsee> rightio
[07:31] <wgrant> gedit/empathy/evolution
[07:31] <wgrant> At least one of those doesn't sound too long.
[07:31] <persia> empathy isn't so bad.
[07:31]  * wgrant checks.
[07:32] <wgrant> 5 jobs (16 minutes)
[07:32] <wgrant> That looks OK.
[07:32]  * persia has "3 jobs (eight minutes)
[07:32] <StevenK> Remember that i386 is blazingly quick, when it wants to be
[07:32] <StevenK> The amd64 builders, less so
[07:32] <Hobbsee> The source version for 'nautilus' in Lucid (main) is at 1:2.29.92.1-0ubuntu2. <-- old, i'm guessing?
[07:32] <wgrant> gedit takes 6 minutes.
[07:33] <wgrant> empathy 12
[07:33] <StevenK> I uploaded -0ubuntu3
[07:33] <wgrant> evolution half an hour.
[07:33] <Hobbsee> bah
[07:33] <Hobbsee> how long does it take for launchpad to update so buildd picks it up?
[07:33] <persia> That still gives us a chance at two buildes.
[07:33] <wgrant> So we should be able to make it on i386 with about 10 minutes to spare.
[07:33] <persia> powerpc will miss.
[07:34] <persia> StevenK: Can we pause the publisher to get a few extra minutes?
[07:34] <wgrant> Hobbsee: You might have to do it manually. If it's not working now, it probably won't work for half an hour.
[07:34] <Hobbsee> well, so launchpad lib picks it up
[07:34] <persia> (just in case)
[07:34] <wgrant> We probably won't need it, but would be nice to know if it's an option.
[07:34] <StevenK> I can, but I'd rather not hand-hold the publisher if I don't have to.
[07:35] <persia> OK.  Let's hope it doesn't come to that.
[07:35] <Hobbsee> nautilus prodded
[07:36] <wgrant> OK, gedit finally finished installing build-deps, so we could have an i386 builder in a couple of minutes.
[07:36] <Hobbsee> wonder why i can't modify the build scores inline anymore
[07:37] <StevenK> wgrant: palmer is the quick one, too, remember
[07:37] <wgrant> StevenK: It's not acting like it :/
[07:38] <wgrant> Ah, finishing now.
[07:38] <Hobbsee> all prodded
[07:38] <persia> Hobbsee: Thank you.
[07:38] <wgrant> Hobbsee: Thanks!
[07:38] <Hobbsee> np
[07:39] <wgrant> Oh I really with buildd-manager sucked less.
[07:39] <StevenK> s/t/s/
[07:39] <wgrant> That.
[07:40]  * persia grumbles about debug symbols being nifty and all, but...
[07:41] <wgrant> Oh.
[07:41] <wgrant> empathy's done.
[07:41] <wgrant> And gedit will be in a few seconds.
[07:41] <persia> Is.
[07:41] <persia> We're building the target packages
[07:41] <wgrant> Ah, yes, DB replication lag.
[07:41] <wgrant> Excellent.
[07:41] <wgrant> amd64 has grabbed nautilus, too.
[07:42] <wgrant> i386 should make it by 10 minutes, and amd64 might by a couple.
[07:42] <persia> and amel
[07:42] <persia> (but it won't make it : those are *slow*)
[07:42] <wgrant> Heh.
[07:46] <StevenK> gnome-panel is very nearly done
[07:48] <wgrant> It's done.
[07:48] <wgrant> And taken by a damn private build.
[07:48] <wgrant> Fortunately we got what we wanted.
[07:48] <StevenK> Will you two stop stressing now? :-P
[07:48] <persia> Almost.
[07:48] <persia> waiting on nautilus and publisher.
[07:49] <StevenK> nautilus is just cleaning up
[07:49] <persia> So should be fine.
[07:49] <wgrant> nautilus is about to finish on amd64 too.
[07:50] <wgrant> So gnome-panel might juuuuust make it.
[07:50] <wgrant> i386 is safe.
[07:50] <wgrant> All built.
[07:51] <wgrant> yellow has gnome-panel... the race is on...
[07:52] <persia> huito has gnome-panel too, not that it really matters, again.
[07:55]  * wgrant sends more hamsters into yellow.
[07:56] <persia> crested has reached pkgmaintainermangler, so we might get another buildd there, for the less critical stuff.
[07:57] <wgrant> Looks like most of i386 will be done.
[07:57] <persia> not that it matters in the next 5 minutes
[07:58] <wgrant> gnome-panel on amd64 is finishing up.
[07:58] <wgrant> It might just make it.
[07:59] <wgrant> Now we can blame buildd-manager if gnome-panel misses it.
[08:00] <persia> The extra nice thing about schroot is that it throws away the changes when used with sbuild, which is faster.
[08:00] <wgrant> persia: Well, it extracts a fresh tarball anyway.
[08:00] <wgrant> So removing everything is pointless.
[08:00] <wgrant> Unless it makes use of it to stop services cleanly.
[08:00] <wgrant> OK.
[08:00] <wgrant> Finished and uploaded.
[08:00] <StevenK> With 3 minutes to spare
[08:00] <persia> StevenK: Thanks a lot for pushing those.
[08:00] <wgrant> StevenK: It's :03, or :04?
[08:01] <wgrant> So everything is good now.
[08:01]  * StevenK checks
[08:01] <wgrant> Thanks.
[08:01] <StevenK> wgrant: The former.
[08:02] <StevenK> So you two can buy me beer, even though wgrant is too young to buy it :-P
[08:02] <wgrant> StevenK: Oy, I'm 18 now.
[08:02] <wgrant> So unless we go to the US...
[08:02] <StevenK> Pity you look 12 :-P
[08:03] <persia> heh
[08:03] <persia> 535650 all reopened.  Is someone else closing 542343?
[08:04] <persia> Or do we wait for publisher to complete for this?
[08:05] <wgrant> We'd best wait, I think.
[08:05] <wgrant> Given that closing it is going to make everyone upgrade, inevitably :/
[08:05] <persia> Good point.
[08:23] <ajmitch> looks like I missed all the fun again
[08:58] <wgrant> Looks like everything's done.
[09:01] <persia> Excellent.
[09:31] <lifeless> wgrant: persia: is it good now?
[09:32] <wgrant> lifeless: Yes.
[09:32] <wgrant> Somebody should probably tell ubuntustatus.
[09:32] <wgrant> And force a mirror push.
[09:32] <persia> mirror push happened post-publisher.
[09:33] <wgrant> I believe a full external push only happens once every six hours.
[09:33] <persia> But lots of mirrors are pull-only, which can be as rare as once each 8 hours.
[09:33] <lifeless> wgrant: hourly for push mirrors
[09:33] <wgrant> lifeless: I see some push mirrors out of date, though.
[09:33] <persia> Or even once a day for a couple particularly special cases.
[09:33] <wgrant> At least I thought they were push mirorrs.
[09:33] <lifeless> pull mirrors can be arbitrarily far apart
[09:33]  * wgrant checks the graphs.
[09:33] <lifeless> wgrant: check their trace file
[09:34] <wgrant> lifeless: That's what I'm looking at.
[09:34] <persia> wgrant: I've been going by http://people.canonical.com/~jpds/mirrors/push-archive-mirrors-dot.png
[09:34] <wgrant> persia: So have I.
[09:35] <wgrant> ftp.acc.umu.se just updated a few minutes ago.
[09:35] <wgrant> It had been unupdated for 5 hours.
[09:35] <lifeless> ubuntustatus is a twitter/identica thing?
[09:35] <wgrant> lifeless: Yeah.
[09:36] <wgrant> It was updated last night by $UNKNOWN with a bad bug number for this issue.
[09:36] <persia> Isn't that mostly a robbiew thing?
[09:36] <wgrant> Probably.
[09:37] <wgrant> It's sort of like launchpadstatus.
[09:37] <wgrant> It's only updated when people are around to fix the issue quickly.
[09:37] <wgrant> So it's not actually all that useful.
[09:38]  * persia idly wonders why us.archive.ubuntu.com is none of the US Push Mirrors
[09:38] <wgrant> persia: I believe discussions are ongoing.
[09:38] <wgrant> But IIRC nobody wanted to take the load.
[09:38] <persia> Ah.
[09:38] <lifeless> because rsync is not a good fit for mirroring 400K files
[09:38] <persia> Works for me.
[09:38] <wgrant> lifeless: Yes. lmirror looks good.
[09:38] <lifeless> and no one US mirror can handle us.archive.ubuntu.come load
[09:38] <lifeless> wgrant: thanks
[09:39] <lifeless> wgrant: did I mention it to you?
[09:39] <persia> And they can't be round-robined?  There are current 4 servers claiming to be us.a.u.c
[09:39] <wgrant> lifeless: No.
[09:39] <wgrant> persia: round-robin with apt is unsafe unless they're in perfect sync.
[09:39] <lifeless> wgrant: I'm curious (I haven't been hiding it, just not had it at the point to make noise till like, last night), how you ran across it
[09:40] <wgrant> (like, say, prat, lithium, drescher, and jackass)
[09:40] <persia> Isn't that mostly safe for push mirrors though?  I thought that the only guarantee of that was standard push-mirroring for drescher/lithium/prat/jackass
[09:40] <wgrant> persia: Push mirrors are tiered, and each tier takes a few minutes.
[09:40] <wgrant> (Until everybody moves to lmirror)
[09:40] <lifeless> persia: its not safe unless the mirror update script pauses and sync, or they have exactl ythe same bw
[09:41] <persia> lifeless: How do you mean pause and sync?  You mean something like the algorithm used in ubumirror?
[09:42] <lifeless> wgrant: we did a scaling test last night; or rather jpds did and I wafted encouragement and bug fixes
[09:42] <wgrant> Yeah, I heard it was pretty nice and quick.
[09:42] <lifeless> wgrant: 0.24 seconds to mirror a change, 24K file repository
[09:42] <wgrant> Awesome.
[09:43] <lifeless> yeah
[09:43] <wgrant> How long does journaling take?
[09:43] <lifeless> ~ the same as an rsync run today.
[09:43] <lifeless> but you should be able to write journals directly from LP, with appropriate thought.
[09:43] <wgrant> Excellent.
[09:43] <wgrant> Right.
[09:44] <lifeless> persia: So, apt archives are tolerant of some forms of 'not-perfect':
[09:44] <lifeless> you can have extra files
[09:44] <wgrant> Although that's harder with apt-ftparchive, I guess we could easily generate it internally for the pool and then diff dists/.
[09:44] <persia> lifeless: which I why I thought the sync pool/ sync dists/ clean pool/ model ought just work.
[09:45] <wgrant> So you sync pool first, then wait until everybody is in sync, then quickly sync dists across all of them.
[09:45] <lifeless> as jpds said to me when I apologised for only have a prototype - its a personal time thing - rome wasn't built in a day
[09:45] <wgrant> But you'd need to two it in two phases.
[09:45] <lifeless> persia: its not tolerant of missing files.
[09:45] <wgrant> rsync pool without deleting, rsync dists, rsync pool with deleting?
[09:45] <persia> also, when my mirror is out of sync, I just get a download failure, and bump to the next least distant mirror.
[09:45] <lifeless> persia: and apt doesn't pin to a single mirror server in a rotation alias.
[09:45] <persia> It's perfectly tolerant as long as there are backup servers in sources.list.
[09:46] <persia> It just stops trusting that source for that file during that run.
[09:46] <wgrant> persia: No, you get sig failures and alarms everywhere if dists is slightly out of sync.
[09:46] <lifeless> persia: so, folk configured to point as us.archive.ubuntu.com generally don't have a backup 'server'
[09:46] <persia> Ah, I'm thinking about pool/ out of sync.  dists/ sync is quick-like.
[09:47] <persia> lifeless: Anyway, how does this lmirror thing work?  Should I try using it already, or isn't it ready yet?
[09:47] <lifeless> persia: so the point is to never have a situation that lasts more than a few seconds where any of the servers in a rotation alias have missing files relative to a different servers Packages
[09:47] <persia> Yeah, I can see that.
[09:47] <lifeless> persia: early alpha; if you're interested checkout lp:lmirror and have a read
[09:47] <hyperair> i generally sync the dists/ first to a separate temporary location, sync pool, and then sync dists with the aforementioned temporary location
[09:48] <persia> I've been reading the source.  Just curious if you wanted users.
[09:48] <lifeless> hyperair: thats only enough to be internally consistent, not consistent with other mirrors, which is the issue
[09:48] <hyperair> lifeless: what do you mean consistent with other mirrors?
[09:48] <lifeless> so you need to do the pool non-delete mirroring to all the other mirrors aliased to whatever rotation you're in.
[09:48] <persia> hyperair: So if you have, say, 12 machines at sg.archive.ubuntu.com, they all need to be in sync.
[09:49] <hyperair> persia: ah i see what you mean
[09:49] <lifeless> and then do the dists + pool delete-mirroring to all the other mirros at the same time
[09:49] <lifeless> persia: I'd be delighted if you wanted to play with it, file bugs, write docs, whatever.
[09:50] <persia> lifeless: I'll need guidance for setup, etc.  I'm happy to run stuff, and try to script it to require less thought.
[09:50] <lifeless> persia: it is not feature complete; specifically it doesn't have sanitised journals, signed journals, enough diagnostics nor signalling points to do cross-machine-update synchronisation yet.
[09:50] <lifeless> persia: read the MANUAL.txt
[09:50] <lifeless> persia: if its not guidance enough, file a bug ;)
[09:51] <lifeless> (I'm serious)
[09:52] <persia> lifeless: What target servers have lmirror enabled so far?
[09:52] <persia> (even as dumb senders)
[09:54] <lifeless> none
[09:54] <lifeless> we only found out last night UK/east-coast US time zone that it could successfully mirror on the right scale.
[09:57] <persia> lifeless: heh.  I'll need to get another server up first then.  If you enable it somewhere, and want to play with a peer, let me know.
[09:58] <lifeless> persia: persure
[09:58] <lifeless> what will probably happen is that jpds, elmo and I will become confident enough in it that we'll add it to the main mirror set at some point
[09:58] <lifeless> at which point it can be used :)
[10:00]  * wgrant slurps down the final chunk of binaries to check for more brokenness.
[10:00] <persia> Makes sense.  I just have a feeling that timezone skew being what it is (why are you awake again), there's a chance I might be able to play peer before that happens.
[10:01] <lifeless> persia: EJETLAG
[10:02] <persia> heh
[10:03] <persia> well it's evening now, so you should be all set to go to sleep for breakfast.
[10:05] <lifeless> persia: :P
[10:05] <lifeless> 6am is my normal rising time, so I was only ouy by 30 minutes or so anyway
[10:11] <wgrant> Unless there's a build in the wild that I've missed, the archive is clean.
[10:11] <wgrant> All finished broken builds have been superseded.
[10:14] <persia> wgrant: sparc and armel aren't done yet.
[10:15] <wgrant> persia: Right.
[10:15] <wgrant> But those builds are good.
[10:16] <persia> Oh, you mean that we no longer have any sources that need rebuilding?
[10:16] <wgrant> Right. By 'the archive is clean' I meant 'it will clean itself up without intervention once the obscure slow archs finish'
[10:18] <persia> sparc doesn't have to be slow, but hardware is annoyingly expensive.
[10:18] <wgrant> Right.
[10:19] <persia> armel isn't obscure, but there's *three* devices available in retail that can run lucid, and we don't have kernels for any of them.
[10:19] <wgrant> Ow.
[10:19] <persia> (and it is slow)
[10:27] <lifeless>  isn't android armel?
[10:32] <persia> lifeless: I thought it was available for both armel and i386.  I may be mistaken.
[10:32]  * persia jumps on the "We have more platforms than you" trampoline some more, and then gets off ashamed, noticing Debian
[10:33] <wgrant> We were close for a while.
[10:34] <persia> Well, yeah, but we didn't have enough hppa users to sustain it (6 people does not a userbase make).
[10:34] <persia> And lpia should never have existed in the first place.
[10:34] <wgrant> Right.
[10:34] <persia> MIPS would be nice, but runs into the "buildds are hard to get" issue.
[10:34] <wgrant> Yeah...
[10:35] <persia> I'm unsure how much longer we'll have ia64, because there aren't that many users.
[10:35] <persia> (and only 3 developers I know about)
[10:35] <lifeless> persia: I mean the android hardware platform that people buy, e.g. nexus one
[10:35] <wgrant> That's ARM of some variety.
[10:35] <wgrant> Probably armel.
[10:36] <qense> Sometimes -- when Plymouth doesn't display the error message "init: ureadahead main process terminated with status 4 -- Plymouth seems having so much fun at changing the colours of the four dots under the text "Ubuntu 10.04" that it seems to forget that it needs to boot, it goes on forever. Is there already a bug report for this issue, if not, against what package should I report it? ureadahead, or plymouth?
[10:36] <persia> lifeless: That's indeed armel.  That can't run lucid because 1) we don't have that kind of kenel, and 2) there's some silliness in the way the kernel loads taht means we can't easily replace it with a custom kernel if we wanted.
[10:36] <lifeless> wgrant: as I thought :P
[10:36] <wgrant> persia: Are they really ARMv7+thumb2?
[10:37] <persia> qense: The 4 dots are independent of anything else.  File the bug against ureadahead and the plymouth theme.
[10:37] <persia> wgrant: Supposedly.  I've not actually fiddled with one.
[10:37] <qense> persia: OK, I'll do.
[10:37] <qense> thanks
[10:39] <persia> wgrant: More specifically, every ARMv7(a) solution is expected to support Thumb2 : it's part of the specification.
[10:40] <wgrant> persia: Ah, I see.
[10:40] <persia> The choice of instruction set (-marm vs. -mthumb2) is just a matter of code density.  -mthumb2 saves about 40% code size at about a 10% performance loss, which usually ends up at equivalent or better performance because of l2 cache sizes.
[10:41]  * wgrant has just an ARMv5 device running Debian.
[10:44] <persia> wgrant: As much as I'm happy with my NetWalker, until the kernel situation fixes itself (Go DeviceTree!), I can't recommend more unless you really want to hack on stuff.
[10:47]  * persia kinda wishes someone would either port wine to powerpc/sparc/armel *OR* make the builds fail sooner
[10:47] <wgrant> and/or P-a-s it.
[10:48] <persia> I don't much like P-a-s.  Most of my experience with it has been finding it obnoxiously hard to get stuff *off* the list when it was overly precise to make new ports happen.
[10:49] <persia> But wine has a specific call to check the platform that runs well into the build and dies if it doesn't match a set.  Such a check should happen *before* compiling everything.
[10:49] <persia> Alternately, it should be dropped, and if people actually have powerpc or sparc Windows binaries, they can report bugs.
[10:49] <persia> wine/armel is kinda special, because there never was a Windows port to armel.
[11:48] <chrisccoulson> persia / wgrant / StevenK - thanks for fixing the GNOME breakage
[11:48] <persia> chrisccoulson: Thank Hobbsee also, but it needed doing.
[11:49] <chrisccoulson> oh, thank you Hobbsee too
[11:49] <wgrant> chrisccoulson: No problem.
[11:49] <chrisccoulson> (sorry, i thought i'd got everyone from the scrollback)
[11:49] <wgrant> Somebody might want to poke ubuntustatus about it.
[11:51]  * chrisccoulson thinks that we probably shouldn't unfreeze on a friday afternoon again
[11:51] <wgrant> Possibly not.
[11:51] <wgrant> But it wasn't too troublesome to fix, once we found a core-dev or two.
[11:52] <chrisccoulson> yeah, i wish i could have helped a bit. i can't upload cdbs, but i could have uploaded most of the components that were broken
[11:52] <wgrant> Oh, right, packagesets. Handy.
[11:53] <chrisccoulson> yeah, i can upload stuff in the ubuntu-desktop packageset
[11:53] <chrisccoulson> but cdbs isn't in that
[11:56] <didrocks> chrisccoulson: +1 on not unfreezing on Friday
[11:56] <chrisccoulson> didrocks - yeah, i think we should probably raise that on monday. i know seb128 was concerned too
[11:57] <wgrant> I'm a little concerned that a whole lot of people knew it was broken, apparently pinged relevant people about it, then all went to bed.
[11:57] <cjwatson> well, having the beta slip wasn't in anyone's original plan
[11:57] <didrocks> chrisccoulson: I was too. Also, on my previous company, I have very very bad examples on "releasing" on Friday (unfreeze == release, especially after a beta/alpha when everyone will upgrade after installing them…)
[11:58] <didrocks> sure
[11:58] <cjwatson> once it did slip, what else were we supposed to do?  people were screaming for an unfreeze
[11:59] <chrisccoulson> cjwatson - do we really gain much by unfreezing on friday? i don't think it would have hurt to wait until monday...
[11:59] <didrocks> sure, but at least, we have good examples know to show how wrong it happens and make people wait a little more (just some thoughts there)
[11:59] <didrocks> chrisccoulson: the only thing is that people are testing on week-end, so, we basically waste a week
[12:00] <cjwatson> chrisccoulson: hindsight is 20:20; yesterday, lots of people were clamouring for unfreezing
[12:00] <persia> I think we gain a *lot* from friday unfreeze.
[12:00] <persia> That said, in the rare event it happens, more folk should be around to cover until the next timezone becomes available.
[12:01] <cjwatson> fwiw, from what I can see from scrollback, this got fixed nearly as quickly as it would have done on a Monday night
[12:01] <cjwatson> and I can say that without particular defensiveness since I wasn't involved
[12:02] <persia> Well, not quite.
[12:02] <wgrant> We only decided to fix it nearly three hours after it was first brought up.
[12:03] <persia> There were a couple core-dev who saw the problem and didn't jump on it shortly after discovery, and the heavy push didn't start until about 4 hours after discovery.
[12:03] <didrocks> sure, nobody's to blame here and that's not the point. It's just that it's risky and can be in some cases more critical
[12:03] <cjwatson> the timezone thing would have bitten the same way on Monday, wouldn't it?
[12:03] <persia> didrocks: Agreed. My suggestion is that some of those folk in the far western timezones might stay a little later on a Friday night that we unfreeze until those in the far eastern timezones are around, rather than delaying the freeze another 3-4 days.
[12:04] <kklimonda> ww/w 10
[12:04] <persia> cjwatson: Depends.  Often people in the Americas stay late enough in their evenings to catch Asia/Oceania through to mnidday or so.
[12:04] <kklimonda> persia: do you know how it sounds? "you guys have to stay for few more hours on the friday night"? ;)
[12:05] <persia> kklimonda: Only on a release delay.
[12:06] <persia> And we caught it anyway, just I think it's better than not unfreezing.
[12:06] <persia> This happens once a year or so, so it oughtn't usually matter.
[12:07] <chrisccoulson> i did stay around quite late last night, but i was powerless to do anything :/
[12:08] <persia> Understood.
[12:10] <wgrant> The privilege problem was easily solved with enough determination. Things might have been set in motion a bit quicker if it hadn't looked like the right people had been poked a few hours earlier.
[12:11] <persia> Indeed.  It was explicit notification that some folks were going to bed without fixing it that helped get me noisy about it.
[12:11] <cjwatson> Did anyone in Canonical invoke DealingWithCrisis for this?
[12:11] <persia> (not that those folk had permission to fix, but the sense "the day is over now: it will get solved in the morning")
[12:11] <persia> Not that I saw.
[12:11] <cjwatson> That's perhaps a shame ...
[12:12] <cjwatson> It doesn't look as though chmod 0 on the archive would have helped anything, but there are other measures there that can be helpful in this kind of situation
[12:12] <cjwatson> particularly waking appropriate people
[12:13] <wgrant> It didn't really need people to be woken. It just needed the people who were awake to be aware that it wasn't being handled by the normal people.
[12:13] <persia> And once they were made aware, it got sorted about as fast as publisher could churn
[12:13] <cjwatson> wgrant: one of the measures in DealingWithCrisis is an explicit statement that, if you're awake and know about a crisis, you have to explicitly hand it off to someone before going to sleep
[12:14] <wgrant> cjwatson: Ah, good.
[12:14] <cjwatson> it's designed as a recipe for remembering the things you always forget when things are going wrong
[12:14] <wgrant> We could have saved some time with some judicious publisher hand-holding, but it went OK.
[12:14] <cjwatson> so I think, in retrospect, it should have been used here
[12:15] <cjwatson> perhaps we should make it, or some parts of it, public so that people not in Canonical can more easily DTRT; they may not have all the necessary phone numbers, but ...
[12:15] <cjwatson> I'll follow up on that on Monday
[12:15] <wgrant> Thanks.
[12:16] <cjwatson> Caesar: let me know what you want, we're not necessarily available at the same time ...
[12:16] <cjwatson> Caesar: (answering your ping from 02:26 UTC)
[12:21]  * wgrant sleeps.
[12:43] <asac> directhex: why didnt we get mono 2.6? ;)
[12:43] <asac> is that a intrusive thing requiring all rdepends to get adjusted etc. ?
[13:13] <ogra_cmpc> asac, x-loader sits in NEW, bug 542662 is filed, we just need a FFe and MIR for it now
[13:13] <ogra_cmpc> (i wasnt sure if i should put all three in the same bug)
[13:14] <persia> Conserve bugs today!  Reduce, Reuse, Recycle!
[13:15]  * persia finds it very frustrating to have to page back and forth through 3-4 bugs to track one issue
[13:20] <ogra_cmpc> ok, i merged FFe and needs-packaging ... i suspect MIR needs to go separately though
[13:20] <persia> Why?
[13:20] <ogra_cmpc> because the upload will close the needs-packaging one :)
[13:21] <persia> Well, it won't, because of a bug, but yeah, I see your point, and accept it.
[13:21] <ogra_cmpc> heh, thanks
[13:21] <persia> And there's stuff to be done post-upload pre-MIR (like setting a bug contact, etc.)
[13:22] <ogra_cmpc> right
[13:27] <emgent> good evening.
[13:49] <jelmer> Laney, hi
[14:13] <asac> ogra_cmpc: great ;)
[14:19] <directhex> asac, because 2.4 is upstream's LTS branch. but yes, there'd be some fiddling involved in changing now. lots of fiddling.
[14:22] <asac> directhex: asking because its claimed that all armel test failures are fixed there ;)
[14:22] <asac> directhex: do we have a package somewhere so i could verify?
[14:22] <directhex> i've heard that kind of assurance pretty frequently
[14:22] <asac> maybe debian experimental?
[14:22] <asac> hehe
[14:23] <directhex> it's not in experimental yet. meebey's started on it, i believe
[14:23] <asac> directhex: well, we vargaz backported one more patch to 2.4 ad that fixed 6 test cases ;) .... however, he said the rest is tough to backport because it relies on other changes
[14:23] <asac> directhex: is it straight forward to build 2.6 from svnlocally?
[14:23] <asac> directhex: or do i need a bunch of other libs to be updated etc.?`any idea?
[14:23] <directhex> make sure it goes into a non-$PATH prefix
[14:24] <asac> directhex: yeah. i can do that. do i need to run make install before running tests?
[14:24] <directhex> and keep a close eye on library dependencies (i.e. you may need to manually call gacutil -i to install system mono libs into the svn mono GAC)
[14:24] <directhex> tests, no, make test should be fine just from building
[14:25] <asac> right. thats basically what i want to try only ;)
[14:25] <asac> any important configure options i should use?
[14:25] <directhex> nah, not really
[14:25] <asac> cool
[14:25] <directhex> brb, need to get some lunch
[14:26] <asac> directhex: svn co svn://anonsvn.mono-project.com/source/trunk/mono  ?
[14:26] <asac> or do i need more modules?
[14:26] <directhex> asac, you need the mcs module too
[14:26] <asac> kk
[14:26] <asac> so that needs to get built first?
[14:26] <asac> hmm. guess i need to install then
[14:27] <directhex> nah, put mcs and mono in the same place, and mono configure will look in ../mcs/
[14:27] <directhex> iirc
[14:27] <asac> cool
[14:27] <asac> thanks
[14:27] <asac> enjoy lunch ;)
[14:57] <directhex> asac, short version, if you're serious about 2.6 in lucid, you need to get meebey in here immediately and find some way to bribe him into reallocating all his "new gaming pc and game" time into "debian" time to have even the slightest hope of something being ready in time
[15:06] <pitwalker> in lucid beta the runlevel 1's menu first entry why not work - "resume    Resume normal boot"?
[15:39] <doko> pitti, kees: any idea about http://launchpadlibrarian.net/41422691/buildlog_ubuntu-lucid-amd64.gdb_7.1-1ubuntu1_FAILEDTOBUILD.txt.gz ?
[15:39] <doko> Searching for duplicated docs in dependency gdb...
[15:39] <doko> cd: 17: can't cd to debian/libgdb-dev/usr/share/doc/libgdb-dev
[15:39] <doko> make: *** [binary-makedeb-IMPL/libgdb-dev] Error 2
[15:39] <mnemo> how can I verify that a sync from Debian is safe? i.e. how can I fetch the Debian package and build it on Ubuntu ?
[15:39] <persia> mnemo: `pull-debian-source foo; sbuild lucid foo.dsc`
[15:42] <mnemo> persia: that gave me the squeeze (testing) version, but I want the sid (unstable) version of the Debian package
[15:46] <mnemo> found it now.. "pull-debian-source blah unstable"
[15:46] <lifeless> bzr branch lp:debian/unstable/packagename foo
[15:52] <mnemo> the bzr way didn't work actually
[15:52] <mnemo> http://paste.ubuntu.com/398367/
[15:53] <mnemo> and if I use "pull-debian-source blah unstable", then sbuild fails :(
[15:53] <mnemo> http://paste.ubuntu.com/398366/
[15:53] <mnemo> I suppose I need to run "sbuild-createchroot" but I don't know what params to pass to it
[15:58] <persia> mnemo: `mk-sbuild lucid` should do it, if you're running lucid.
[16:16] <LaserJock> persia: does mk-sbuild only work with lvm?
[16:17] <mnemo> "sbuild lucid foo.dsc" didn't work but "sbuild -d lucid-amd64 foo.dsc" worked perfectly
[16:17] <mnemo> thanks persia and lifeless
[18:18] <lifeless> anyone remember the website/script that reports on patches for packages across multiple distros?
[18:18] <jcastro> harvest
[18:18] <jcastro> lifeless: this what you're looking for? http://daniel.holba.ch/harvest/
[18:19] <mneptok> jcastro: http://bit.ly/9phQJZ
[18:19] <jcastro> mneptok: I am now aware of those, thanks!
[18:19]  * mneptok bows
[18:20] <lifeless> jcastro: no
[18:21] <lifeless> jcastro: though it is, similar.
[20:21] <pa__> hi
[20:21] <pa__> is it normal that i have an ntop running since mar 03?
[20:21] <pa__> that i cannot see on consoles?
[20:21] <pa__>  /usr/sbin/ntop -d -L -u ntop -P /var/lib/ntop --access-log-file /var/log/ntop/access.log -i eth0 -p /etc/ntop/protocol.list -O /var/log/ntop
[20:51] <kees> doko: I imagine it's from whatever change was made for "* Do not duplicate documentation in gdb64, gdb-source, and libgdb-dev.
[20:52] <kees> "
[20:52] <kees> but I've not see that error before
[21:40] <lamont> so how come we still regenerate the font cache on every font install?
[21:54] <lamont> seriously.  was it necessary to throw the max/min/kill buttons to the other side of the window?
[21:55] <wgrant> lamont: That's what the Internet said.
[21:55] <wgrant> But DX says yes.
[21:55] <mnemo> (and also re-order them)
[21:55] <Tm_T> lamont: I keep menu button in left side (double/tripleclick closes) and maximise in right side
[21:55] <lamont> because yeah, 20 years of muscle memory is nothing
[21:55] <xnox> To have room for zeitgeist button
[21:56] <lamont> it was bad enough when karmic made the window borders 1 pixel so that I couldn't grab-n-drag to resize with my poor laptop built-in mouse, but really?
[21:56] <wgrant> lamont: Oh, I thought I was the only one who noticed the thin window border thing.
[21:56] <wgrant> That annoys the crap out of me.
[21:56]  * lamont goes looking for a background that is less harsh on his eyes
[22:01] <cjwatson> lamont: I like Cosmos
[22:02] <lamont> cjwatson: it's very pretty.  and not very ergonomic
[22:02] <lamont> "indicator-applet" closed unexpectedly, and apport doesn't support reporting that kind of crash.
[22:02] <cjwatson> how can a wallpaper be ergonomic?
[22:02] <lamont> I wonder how known that is
[22:02] <cjwatson> (Cosmos is one of the default background selections.  Actually it's a slide-show)
[22:02] <lamont> if it enhances eye-strain, it's not ergo
[22:03] <wgrant> lamont: Was it an assertion failure?
[22:03] <lamont> ISTR yes
[22:03] <wgrant> I've had it did like that once.
[22:03] <cjwatson> oh, I haven't noticed eyestrain from it
[22:03] <cjwatson> of course, normally my wallpaper is hidden by maximised windows, and I only see it on occasion
[22:06] <lamont> shift + super == ??
[22:06] <lamont> yeah - I don't maximize windows, and get annoyed everytime an app (hello oo.o, evince, etc) believes that it should maximize on opening
[22:10] <lamont> is it just me, or are the + and - header in xchat etc taking more spaces than the right/down arrows did?
[22:15] <lamont>   devicekit-disks libparted1.8-12 <-- I wonder, do I need those, or does libvirt-bin need a good beating?
[22:19] <lamont> so is there any easy config option in gnome to get the min/max/kill buttons back where my fingers think they are?
[22:19] <tjaalton> yes
[22:19] <tjaalton> /apps/metacity/general/button_layout
[22:20] <tjaalton> from gconf-editor
[22:20] <cjwatson> lamont: both devicekit-disks and libparted1.8-12 are obsolete
[22:20] <cjwatson> unfortunately due to a conflicts in the *old* libparted runtime package (thus not deletable without time travel) the upgrade is more complicated than it ought to be
[22:21] <tjaalton> move the colon in front
[22:21] <Laney> and swap min/max
[22:21] <lamont> yeah
[22:21] <tjaalton> oh that too
[22:22] <lamont> interesting... when the stupid typing monitor tells me to take a break now, I have to move the mouse before I can hit alt-p to delay the break...  this is not how I want that to work
[22:22] <tjaalton> 'maximize,minimize:,close' seems best for me
[22:24] <Laney> what's the comma?
[22:24] <lamont> actually have to mouse over the button before the button decides to exist.
[22:24] <lamont> Laney: a separator
[22:24] <tjaalton> Laney: heh, the other one was extra in this case
[22:25] <Laney> oh, I thought it might mean something in that context
[22:30] <lamont> ok. so after I accidentally break a tab out of a firefox window by dragging it up and out, how the &*(%^)*&_%^_)  do I get it back into the tabbed window?
[22:31] <Laney> you can just drag tabs between windows
[22:32] <crimsun> kees: could you parse the second bullet in the initramfs-tools 0.92bubuntu69 changelog for me, please?
[22:32] <lamont> and when there are no tabs?
[22:32] <Laney> open a blank one so you get the tab bar, I guess
[22:32] <lamont> AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\
[22:33] <lamont> so... this release is causing me to dig into more and more arcane bits of gnome that I've never cared enough to tweak.
[22:33] <lamont> how do I change that nice slate-grey background inside the app menus, so that I can actually read the text that comes up.
[22:33] <lamont> that is, where is the *^( color theme defined/modified?
[22:34] <wgrant> lamont: Most apps should have light text in their menus.
[22:34] <wgrant> Except strange creatures like Evolution which override the text colour in parts.
[22:34] <lamont> firefox, giving me ideas of what I'm typing, gives me a nice dark blue on grey which is completely  unreadabkle
[22:34] <lamont> and it seems to have adopted the panel color
[22:35] <lamont> and once I get far enough on this, I'll check that buildd
[22:36] <geser> doko, kees: see http://launchpadlibrarian.net/41456559/gdb_7.1-1ubuntu1_7.1-1ubuntu2~ppa1.diff.gz for a gdb FTBFS fix. The problem was that the build symlinked the whole /usr/share/doc directory and that the symlink can only be resolved for the installed package and our symlinking code tried to access it. Disabling the doc symlinking for those two packages fixes it.
[22:39] <cjwatson> lamont: IMO that firefox thing is just a bug and should be reported as such if it hasn't been already
[22:39] <cjwatson> there's no way that can be a desirable default colour combination
[22:41] <lamont> cjwatson: against firefox?
[22:42] <doko> geser, kees: well, I'd rather disable the symlinking code if the docdir is a symlink
[22:43] <cjwatson> lamont: dunno, that would be a start I guess ...
[22:43] <cjwatson> or light-themes?  dunno
[22:43] <doko> geser: there are more packages using symlinks for the doc dirs
[22:43] <geser> doko: true
[22:45] <doko> geser: I just can't see which change in cdbs did change this. it did work 10 days ago with the cdbs snapshot
[22:45] <doko> gdb snapshot
[22:50] <geser> doko: the rules file from the snapshot didn't symlink the /usr/share/doc directories, this change came with the merge/last upload (http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/lucid/gdb/lucid/revision/46#debian/rules)
[22:52] <doko> geser: I'd rather fix cdbs to do nothing if a symlink is found
[23:17] <geser> doko: I might have an idea how to fix it but want to talk to pitti first as I'm not fully sure about it