evRiddell: not yet, but I think I had the timing slightly off when I tested.  I'll try again in the morning.00:18
cjwatsonev: bug 627663 is kind of nasty00:38
ubot2Launchpad bug 627663 in ubiquity (Ubuntu) "Ubiquity-kde starts partitioning and formatting without verifying (affects: 1) (heat: 8)" [Undecided,Confirmed] https://launchpad.net/bugs/62766300:38
cjwatsondesign issue?00:38
CIA-71ubiquity: superm1 * r4258 ubiquity/ (bin/ubiquity-dm debian/changelog): When using xfsettingsd, set the background to black first.00:58
Riddellcjwatson: I thought that was the whole point of this new ubiquity design (bug 627663)02:19
ubot2Launchpad bug 627663 in ubiquity (Ubuntu) "Ubiquity-kde starts partitioning and formatting without verifying (affects: 1) (heat: 8)" [High,Confirmed] https://launchpad.net/bugs/62766302:19
ScottKsuperm1: jockey-txt -a worked, so I'll have to investigate further next time I try an install.  Thanks again.03:22
superm1ScottK, so if that worked, maybe what you should check is before rebooting, was that applied to the current session automagically?03:22
superm1because it's executed in the live environment, not the chroot03:23
ScottKWill do.03:23
superm1and if it worked in the live environment, then the logic that makes it happen during the post install step might be where it's busted03:23
ScottKI need to nuke my current install and start over once we have a respin, so I'll look then.03:23
CIA-71ubiquity: superm1 * r4259 ubiquity/ (debian/changelog ubiquity/plugins/ubi-language.py):03:45
CIA-71ubiquity: Set release_notes_found to False initially to prevent a scenario it gets03:45
CIA-71ubiquity: accessed before being set when offline.03:45
ScottKIf it's relevant to determining what process is triggering the oom killer, /bin/sh /lib/partman/commit.d/30parted has the pid immediately before swapoff /dev/sda5 (pretty reliably).04:03
ScottKev, Riddell: Not sure if it's good news or badnews, but after the respin with 2.3.12, I don't get the OOM condition anymore using the same options on the same system.05:27
=== ara_ is now known as ara
evcjwatson: I'm not convinced on bug 627663.  You have to click through two screens before it does anything.  Perhaps the existing text that says something like, "X additional partitions are not shown.  Use the <link>advanced partitioner</link> to see them", should be changed to "X additional partitions are not shown and will be destroyed..."08:19
ubot2Launchpad bug 627663 in ubiquity (Ubuntu) "Ubiquity-kde starts partitioning and formatting without verifying (affects: 1) (heat: 8)" [High,Incomplete] https://launchpad.net/bugs/62766308:19
evgood morning, all08:19
evScottK: very interesting08:20
evall that changed was fixing a bug on the user setup page that should've just crashed ubiquity (rather than using up all of your memory)08:21
evand ugh to IRC bouncers that drop the connection, then discard their buffer before your client automatically reconnects.  ZNC is awful.08:23
superm1ev, re the whole jockey-text not working, it looks to me like it actually tries, but the postinst fails on the man-db trigger08:29
superm1which i'd guess is due to ubiquity hogging debconf08:29
evsuperm1: is there a bug for this?  (and dear lord, do you sleep?)08:30
evah, hm, that's a tricky one08:30
superm1i was just about to hit the hay :)08:30
superm1i made the mistake of working out way too late, and then i wasn't tired08:31
evah, been there08:31
evdoes jockey actually need to install these packages?08:31
evor can we rely on it telling ubiquity what to install later on, as it already does08:32
superm1during install?  I suppose that depends on whether that wireless widget really needs to be functional that early08:32
superm1it's probably better to not actually install at that point since moving forward to the next page is then dependent upon waiting for bcmwl to compile potentially08:32
evjust thinking through the code on this one08:34
ev(but do go to bed if you're already off, we can chat about when you get back)08:34
superm1okay, g'night08:35
evyeah, now that install.py and plugininstall.py are separate and all network access is done in the latter, which doesn't start until all the pages are finished, we could shove the actual jockey call later in the pages.08:37
evAh, wait no, it would mean no Internet for the timezone page.08:37
evWhich is fairly important, I think.  The geonames service (which is still being deployed) wont work without it.08:37
evI suspect the order is correct, we just need to negotiate how we manage sharing debconf or the work done with debconf (so maybe ubiquity installs the packages for jockey?), and that we ensure that the UI is updated with a spinning cursor and all that jazz.08:39
evI'd hope compiling bcmwl doesn't take too long on modern hardware.08:39
aramorning ev09:13
araa couple of ubiquity bugs you might be interested in having a look09:14
evhi ara09:14
arabug 62789909:14
ubot2Launchpad bug 627899 in ubiquity (Ubuntu) "Wrong disk size in the partion manager (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62789909:14
evcurrently trying to work through the wubi one09:14
arabug 62790209:14
ubot2Launchpad bug 627902 in ubiquity (Ubuntu) "The list of available networks is blank (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62790209:14
arasure, thanks!09:14
cjwatsonRiddell,ev: right, the bug seemed to be that there was a warning in certain modes which was not present in this one09:15
cjwatsonjockey installing packages> can we set it up to do passthrough?09:15
evjockey> oo, that hadn't occurred to me.  Perhaps we can, assuming it doesn't input.09:16
cjwatsonman-db seems to be something of a canary for debconf not being set up quite right09:17
evheh, indeed09:21
evara: updated, thanks09:24
araev, thanks09:25
cjwatsonev: bug 625258 has popped up on the iso-testing list - is it fixable for beta?09:59
ubot2Launchpad bug 625258 in ubiquity (Ubuntu) "ubiquity crashed with UnboundLocalError in part_ask_option_changed() (affects: 8) (heat: 40)" [High,Confirmed] https://launchpad.net/bugs/62525809:59
cjwatsonlooks like it ought to be reasonably simple09:59
evlooking into it now09:59
cjwatsonev: also, is there anything I can help with?09:59
cjwatsonall other things being equal, I was planning on digging out the Windows CD and having a look at Wubi09:59
evah, I was working on that already but help is always welcome10:00
evlooks like a busted config file10:00
evI quickly see comments about fi (I think) being an invalid command fly past before it gets to the grub shell10:00
cjwatsonnot seeing anything obviously wrong10:04
evme either10:04
evand the script checker checks out10:04
cjwatsonmm, well, for whatever that signifies :)10:05
evhaha, yeah10:06
cjwatsonbacking up existing contents of that system, will take a while10:16
evI'm sure it will still go faster than my windows VM over NFS over wireless separated by a ceiling and some distance.10:17
cjwatsonwireless here is not exactly stellar10:18
cjwatsonthe backup is laptop -> wireless -> ceiling -> wireless -> wired back up to other laptop -> USB disk10:18
evhaha, impressive10:19
cjwatsonit would probably be faster to find another wire but then I'd have to brave the nest of cables10:19
evbeen there10:19
cjwatsonis bug 627899 a dup of bug 626299?10:19
ubot2Launchpad bug 627899 in ubiquity (Ubuntu) "Wrong disk size in the partion manager (affects: 1) (heat: 6)" [High,Confirmed] https://launchpad.net/bugs/62789910:20
ubot2Launchpad bug 626299 in ubiquity (Ubuntu) "During install reports size of new partition as 0.0 B (affects: 1) (heat: 8)" [High,Confirmed] https://launchpad.net/bugs/62629910:20
evthere might be a Virgin employee skeleton or two behind our TV10:20
* ev checks10:20
evfixed.  I thought I had seen that before.  I looked when ara pointed me at that bug, but google failed.10:21
araev, I have reported now a bug that prevents installation: bug 62793710:22
ubot2Launchpad bug 627937 in ubiquity (Ubuntu) "The installation stalls when copying files (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62793710:22
arammm, coffee is calling10:23
evara: can you give that another try with ubiquity -d, then post new logs?  The problem is not evident from these.10:26
araev, sure, will do10:28
evmuch appreciated10:28
CIA-71ubiquity: evand * r4260 trunk/ (debian/changelog ubiquity/plugins/ubi-partman.py):10:31
CIA-71ubiquity: Only set the label on the 'use entire disk' button if we're not10:31
CIA-71ubiquity: going into the advanced partitioner (LP: #625258).10:31
CIA-71ubiquity: evand * r4261 trunk/debian/changelog: releasing version 2.3.1310:46
araev, I wasn't able to reproduce it with -d option :(10:56
araev, anyway, I will upload the logs when finished, in case you see something weird there11:00
arammm, it is also stalling now (in another point)11:01
araso, yes, I will upload the logs11:01
evokay, thanks11:05
evcjwatson: http://people.canonical.com/~evand/tmp/wubi-failure.flv - stepping through that with mplayer is probably the easiest viewing experience.  Perplexing stuff.11:15
evI've build wubi with set debug=all in the config file here, if it's of help: http://people.canonical.com/~evand/wubi/maverick/wubi-r192-maverick.exe11:16
evin a relatively up-to-date maverick chroot11:16
cjwatsonI wonder if this is a change in grub11:22
cjwatsonso, the basic problem is that wubildr.cfg is being parsed using the rescue parser and not the normal parser11:23
RiddellI'm still getting a million plugininstall.py instances with today's ubiquity11:24
evRiddell:  damn, okay.  Can you put up new logs?  Also, can you tell me a bit about what you're doing to trigger this?  Real hardware or VM?  How much memory?11:27
Riddelli386, real hardware, 512MB memory, yesterday's CD with updated ubiquity11:30
Riddellwill try with today's CD now11:30
cjwatsonev: could you try http://paste.ubuntu.com/486688/ ?11:33
cjwatsonI've tested it with 'make winboot2 && kvm -kernel build/winboot/wubildr', which at least lets me see it producing rather different error messages11:33
evcjwatson: sure11:33
evRiddell: can you apply this patch in the livefs and try: http://paste.ubuntu.com/486690/11:34
evthen post new logs11:34
evjust a bit of extra instrumentation11:34
Riddelllots http://muse.19inch.net/~jr/tmp/ubiquity1/11:35
Riddelllogs rather11:35
Riddellev: ok will do11:35
cjwatsondear lazyirc: where oh where has my windows product key gone?  sigh11:39
cjwatsonsilly proprietary software11:39
ScottKev: Bug 627822 is the one I filed for the jockey issue.11:47
ubot2Launchpad bug 627822 in jockey (Ubuntu) "bcmwl driver not installed during install even when selected (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62782211:47
araev, uploaded the debug syslog to bug 62793711:53
ubot2Launchpad bug 627937 in ubiquity (Ubuntu) "The installation stalls when copying files (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62793711:53
ara(sorry for the delay)11:53
evara: that seems to be missing /var/log/installer/debug11:55
evthe tarball just has your syslog11:55
araev, I thought that in debug mode that was enough... OK, never mind, I'll upload it again (as soon as I get those logs again...)11:56
evah, no worries actually11:56
evit's the media change bug11:56
ev(bug 627672)11:56
ubot2Launchpad bug 627672 in ubiquity (Ubuntu) (and 1 other project) "[Maverick Beta] install from USB stuck retrieving files 2/6 Hp Mini (affects: 1) (heat: 8)" [Undecided,New] https://launchpad.net/bugs/62767211:56
evthis was from a USB disk, right?11:56
araev, right11:56
araev, do you want me to mark it as duplicate of that one?11:57
evplease do11:57
evI started digging through apt on that one last night before bed11:57
evI'll pick back up momentarily11:58
evara: what did you create the USB disk with?11:58
arausb-creator, in my up-to-date maverick laptop11:58
Riddellhttp://kubuntu.pastebin.com/138PLvjz  what's wrong with that?  two %s, two arguments11:59
evah, whoops12:00
evenclose the arguments in a tuple12:01
evsyslog.syslog('debconffilter_done: %s %s' % (str(dbfilter.status), dbfilter.__module__))12:01
evsorry about that12:01
cjwatsonev: any luck with that wubi patch?12:21
evstill waiting on it12:23
evin retrospect I should've just copied the new wubildr, instead of reinstalling wubi12:24
Riddellev: http://muse.19inch.net/~jr/tmp/ubiquity2/12:24
evRiddell: thanks12:27
araev, I get a crasher in ubiquity after selecting partitions (ubuntu netbook),  is that something you're already working on?12:30
evhard to say without seeing the traceback12:31
araone sec12:32
araev, something like PageGtk has no attr 'release_notes_found'12:34
evara: superm1 fixed that in the pending ubiquity12:34
aracan you point me to the bug number, please?12:35
evara: I don't believe there is one, I think he just caught it while testing himself.  Feel free to create one with the logs and I'll link to it in the changelog.12:37
araev, sure, will do, thanks12:39
araev, bug 62801112:44
ubot2Launchpad bug 628011 in ubiquity (Ubuntu) "Ubiquity crashes on Ubuntu Netbook (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/62801112:44
evara: updated, thanks12:49
CIA-71ubiquity: evand * r4262 trunk/debian/changelog: Add LP reference12:49
araev, thanks to you12:50
evcjwatson: confirmed fixed12:54
cjwatsonev: yay12:56
evI'll build and upload a new wubi12:57
evthanks a lot!12:57
cjwatsonabout to commit it here12:58
evah, go ahead12:58
evgetting ahead of myself :)12:58
cjwatsonlet me just double-check that I'm using the right command to enter normal mode12:58
cjwatsonthere are a few subtly different variations12:58
cjwatsonyeah, I am.  good12:59
cjwatsonev: ok, r19313:00
cjwatsoncan 628011 be marked fix released, since 2.3.13 is in the archive?13:04
evright, built, uploaded, and symlink updated.13:05
evon it13:05
cjwatsonoh, 2.3.13 isn't in the archive yet, my bad13:07
cjwatsonnot sure what happened there, but it's accepted now13:07
cjwatsonanything else I can do?13:09
evRiddell: updated patch, if you have a moment to test it: http://paste.ubuntu.com/486726/13:11
cjwatsonexception in traceback processing> !13:12
evcjwatson: if you have the spare cycles, any of the ubiquity bugs you deem critical.  I'm working on Riddell's at the moment, and was going to see if I can reproduce the 'replace apt media' bug on the netbook shortly.13:12
evcjwatson: yeah, craziness.  If you have any thoughts on that (http://muse.19inch.net/~jr/tmp/ubiquity2/debug, http://muse.19inch.net/~jr/tmp/ubiquity2/syslog), I'm all ears :)13:13
evand actually13:15
evthat's fairly ugly code13:15
evI probably should've rewritten that to not read the entire output into memory while I was over there13:15
evoh, nevermind.  I'm misreading.13:16
Riddellev: testing13:27
cjwatsontaking bug 62754913:29
ubot2Launchpad bug 627549 in ubiquity (Ubuntu) "ubiquity crashed in kubuntu oem mode (affects: 1) (heat: 6)" [High,Confirmed] https://launchpad.net/bugs/62754913:29
Riddellev: http://muse.19inch.net/~jr/tmp/ubiquity1/13:57
evRiddell: thanks14:31
evI'm so far unable to reproduce this 'media change' when installing from USB bug15:02
evah, I wonder if cancel_handler throws us right back into find_next_step.15:32
evRiddell: did the finished dialog pop up before this crash occurred?15:32
Riddellev: it's not a crash, the machine grinds to a halt15:46
Riddelland now, over an hour later, there is a finished dialogue15:46
Riddellit's still got several plugininstaller.py instances running and it's still swapping like mad15:48
evyay, I can reproduce it on my netbook!15:56
ev(in the loosest sense of yay)15:56
evI think I fixed it, but I haven't entirely wrapped my head around the problem yet17:03
CIA-71ubiquity: cjwatson * r4263 ubiquity/ (debian/changelog ubiquity/frontend/kde_ui.py):17:15
CIA-71ubiquity: * KDE frontend:17:15
CIA-71ubiquity:  - If the slideshow doesn't exist, hide self.ui.pageMode rather than17:15
CIA-71ubiquity:  self.page_mode (LP: #627549).17:15
CIA-71ubiquity: superm1 * r4264 ubiquity/debian/ (changelog control): Provides indicator-renderer (LP: #616165)17:51
evcjwatson, superm1: I'm very tied up in this KDE bug (though I think I just found the root of it), I don't suppose either of you have any free cycles to look over bug 627672 , or can reproduce it, as I haven't been able to do so yet.18:13
ubot2Launchpad bug 627672 in ubiquity (Ubuntu) (and 1 other project) "[Maverick Beta] install from USB stuck retrieving files 2/6 Hp Mini (affects: 2) (dups: 1) (heat: 16)" [Undecided,New] https://launchpad.net/bugs/62767218:13
cjwatsonI'm about to head out ...18:15
evno worries, I'll still get to it when I'm done here18:16
superm1ev, i haven't been able to reproduce it myself, but i'm also unable to test with web access easily and that might be a key factor here18:36
evah, I've been testing offline.  I'll give web access a go after I'm done here.18:36
superm1according to manoj's logs, he did have an IP from NM, it would be good to know all the selections he made though in ubiquity18:38
evOn the KDE bug, debconffilter_done is getting called with ubiquity.components.install as an argument well past it returning 0.19:47
evThis causes the loop of plugininstall objects, which can be avoided if you wrap the call to find_next_step with a dbfilter.status is not None check, and remove the spurious call to self.app.processEvents (which causes recursive calls to watch_debconf_fd_helper_read -> process_input -> debconffilter_done).19:47
evHowever, I'm still not sure why debconffilter_done with ubiquity.components.install is getting called after plugininstall is already running.  My current theory is that we're not properly tearing things down after each dbfilter finishes, but I'll have to investigate that after I walk the dog as I need an eye break.19:47
evI did notice that the KDE frontend seemingly doesn't handle HUP19:47
evright, back in a bit19:47
cjwatsonev: hate to hassle, but any progress?  I don't know what to do about Kubuntu images for tomorrow right now23:06
cjwatsonis the workaround you outline above sufficient for us to get by for beta?23:08
evcjwatson: I'm a bit stuck.  I think we need to QSocketNotifier.setEnabled(false) at debconffilter_done, or better handle EBADF23:12
evThe workarounds I mentioned above will still result in an OSError being raised when process_line is called (again, due to the gone-away fd)23:14
evbut the install will complete successfully23:14
evjust with apport dialogs23:14
evI presume23:14
evQSocketNotifier.setEnabled> The connection between a dbfilter and its socket notifier is not exposed in the current program flow23:16
cjwatsonhm, needing to handle EBADF generally strikes me as a sign of misdesign23:17
cjwatsonbut I may be out of line23:17
evI equally may be entirely wrong23:17
evbut ubiquity.components.install is sticking around and its QSocketNotifier is firing seemingly at random, despite having a no longer existent fd.23:18
cjwatsonfiddling with the socket notifiers seems plausible ...23:18
cjwatsonI wonder if you get socket notifier events when the last writer on a read fd closes (e.g.)23:19
cjwatsonseems like the sort of thing you might get a POLLHUP event for, or equivalent23:19
cjwatsonand I guess if the fd has actually been closed then selecting on it is undefined23:19
evthis is seemingly the disconnect between it and gobject23:19
evwhile gobject has an enum for hup, qt does not23:20
evas far as I can tell from http://doc.qt.nokia.com/4.6/qsocketnotifier.html23:20
evI need to step out for about half an hour23:20
evdo you think it's worth it for me to pull an all nighter on this when I get back?  Or are we already past what's reasonable for beta.23:20
cjwatsonit does seem that maybe the frontend ought to be informed in FilteredCommand.wait23:21
cjwatsonI don't like telling you to pull all-nighters; I'm likely to be up for a while and think that either we fix this or we delay Kubuntu beta, FWIW23:22
evokay, I'll be back in 30 and will get back on it then23:23
evI've been entirely focused on this, and unfortunately haven't had any time for the USB apt bug, beyond not being able to reproduce it after a number of attempts23:24
cjwatsondownloading a kubuntu desktop image now just in case23:24
cjwatsonmanjo said he just posted debug logs to 627672; I'll see if I can zen anything out of them23:25
cjwatsonev: reading through this, I think it would be entirely correct for debconffilter_done to detach (even destroy?) the socket notifiers23:28
* cjwatson reads through the Qt signals docs as a refresher23:33
cjwatsonev: slightly different idea.  In the GTK frontend, the way this works is that watch_debconf_fd_helper returns False, which happens after FilteredCommand.process_input returns False to indicate that it shouldn't be called again23:42
cjwatsonev: in GTK, returning False from an io_add_watch signal means that the watch is removed from the list of event sources and won't be called again23:42
cjwatsonev: this doesn't seem so straightforward in Qt, but I think we can achieve a similar effect using something like this (entirely untested, might not be idiomatic, etc.): http://paste.ubuntu.com/486993/23:43
cjwatsonhm, I'm not convinced about the logic for multiple dbfilters in parallel though23:45
cjwatsonev: maybe http://paste.ubuntu.com/486995/?  self.debconf_fd_counter and self.current_debconf_fd were unused and would have had to be adjusted so I removed them23:49
cjwatsonI'm not familiar with this object.signal[int].connect() notation so I trust that I have got it right with disconnect()23:49
evright, back23:55
Riddellev: if you do pull an all nighter I'm happy to test at any hours of the day or night for beta23:57
evtrying that now23:57
evRiddell: cool, thanks23:57
Riddellequally if we don't get kubuntu beta done on time that's life23:57
cjwatsonI don't want us to kill ourselves, but nor do I want to slack23:57

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