[07:07] ;) If anyone from the Ubuntu One team has a moment I have a question about support for integrating U1 into OpenPhoto [09:07] morning all! [09:08] bkerensa, ping! [11:45] hola hola hola! [11:45] good morning everyone! [11:51] alecu, hello! [11:52] alecu, I updated the branch which you comments and though of a perfect name : TidyServiceRunner :) [11:52] mandel, awesome! [11:57] mandel, can you run the script I added to the end of this bug? https://bugs.launchpad.net/ubuntu/+source/python-qt4/+bug/334757 [11:57] Ubuntu bug 334757 in KDE Bindings "update-notifier-kde.py crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] [11:57] alecu, sure, which OS? [11:57] mandel, ubuntu [11:57] alecu, ok, on it! [11:58] mandel, that script crashes aprox 4 out of 10 times with a SIGSEGV [11:58] mandel, I want to see if it happens elsewhere [11:58] alecu, http://paste.ubuntu.com/927810/ [11:59] mandel, awesome. [11:59] alecu, I'll run ti several times for you [11:59] mandel, "Segmentation fault" is what I wanted to get :-) [12:00] alecu, http://paste.ubuntu.com/927812/ [12:00] alecu, seems to fail more here, I'm an x64 for what ever is worth [12:00] * mandel reads the script [12:01] mandel, so, it segfaults 9 out of 12. [12:01] mandel, awesomer. [12:02] alecu, yes, seems very easy to reproduce here [12:02] mandel, the script is to show that a bug we see in ubuntu-sso-client is not really there, but it is in pyqt instead. [12:02] alecu, makes sense.. [12:04] alecu, can I get the reviews asap so I can try to get jenkins working asap, I don't want to have windows in this state, the more we wait.. [12:04] alecu, also, it should fix the issues with the sso tests on mac, so if you have the mini setup it would be great feed back to have [12:07] mandel, I'll do the re-review next [12:07] mandel, in the macmini I've got the dev env halfway thru [12:08] mandel, I may ask you for some help finishing the setup, in half an hour or so. Or after your lunch [12:08] alecu, ok, let me know :) [12:09] mandel, btw: I'm reproducing the pyqt segfault in Lucid!!! [12:10] alecu, WTF! so it has been broken for a looong time! [12:10] and on maverick too. [12:10] awesome. [12:10] alecu, I wonder if we use pyside.. let me try that :) [12:10] mandel, that would be great, because then we can see if the problem is with pyqt or qt itself. [12:11] alecu, yes, that is exactly what I though [12:16] mandel, so: where are you guys installing the windows instance of jenkins? [12:16] mandel, it's a jenkins "slave" or called something like that, right? [12:16] * alecu knows about buildbot's master and slaves. [12:18] alecu, yes, there is a windows ec2 image that jenkins connects to do work wit the tests on windows [12:19] mandel, on EC2 or on canonistack? [12:19] alecu, in the script, you are using a QApplication as in QtGui.QApplication, right? [12:19] alecu, AFAIK EC2 [12:20] mandel, great. And I guess there's no way to do that with osx... so we'll have to use the macmini. [12:21] alecu, look at the script: http://paste.ubuntu.com/927834/ (look at the bus that is used, is the GLib one!!!) and the results: http://paste.ubuntu.com/927833/ [12:21] alecu, can you reproduce it? [12:24] mandel, oh, pyside! [12:24] mandel, I don't have pyside installed, but I don't think we'll manage to get that change into precise :-) [12:25] alecu, no, but it reduces the tech to blame :) [12:25] mandel, perhaps we should try pyqt + the dbus glib mainloop. [12:25] mandel, but who knows what else it will break :P [12:26] good morning! [12:26] alecu, well, as long as we don't put in trunk we are save.. [12:26] ralsina, morning! [12:27] mandel, also: the awful thing is that pyqt has deprecated the pydbus integration, in favour of a different dbus module, so... :-( [12:27] ralsina, I think I found the problem with the "SIGSEGV in QSocketNotifier::setEnabled()" error in sso. [12:27] alecu, ouch! [12:27] alecu: awesome, what was it? [12:27] you! => ralsina hehehe [12:28] ralsina, pyqt being broken since -at least- lucid. [12:28] alecu: we *may* be able to do smething similar for a jenkins slave: http://www.macminicolo.net/jvmx_secure_signup.html [12:28] After we get it all configured in our mini, of coure [12:29] alecu: really? [12:29] alecu: pyqt in general or pyqt-in-ubuntu? [12:29] ralsina, our sso bug is #943046. LP will probably timeout opening that page, since the bug has *way too many* duplicates. [12:29] alecu: the bot is not able to open it. Whoa. [12:30] ralsina, so, I've debugged a bunch, and many times I got segfaults, so I traced it to this other bug: https://bugs.launchpad.net/ubuntu/+source/python-qt4/+bug/334757 [12:30] Ubuntu bug 334757 in KDE Bindings "update-notifier-kde.py crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] [12:30] ralsina, and I added a small script to that bug that shows the problem. [12:31] ralsina, the bug is that pyqt (or qt) segfaults when the connection to the dbus-daemon is broken. [12:31] alecu: looking at the script now [12:32] alecu: but why is dbus daemon dying? [12:32] ralsina, but it fails randomly. On some vms it fails immediately. (like my lucid vm). On some other vms, it never fails (like my precise vm). But on my precise computer, it fails very often. [12:32] alecu: on real life, I mean :-) [12:32] always with a SEGFAULT. [12:32] alecu: it's a race condition, obviously [12:32] ralsina, why? probably when it's going down. [12:33] alecu: so, it has no real consequences, it's just a crash on session closing? [12:33] alecu: because if it is, we could install a handler for sigsegv and hide it [12:33] ralsina, a few of the duplicates say that the bug didn't affect their normal operation at all. [12:34] alecu: exctly. If dbus-daemon dies, none of our stuff works anyway [12:34] ralsina, and I remember reading one that the apport dialog showed up on first starting a just updated precise [12:34] alecu: so it's not something we can recover from in any case [12:34] ralsina, right [12:34] alecu: probably updated dbus daemon restarted itself? [12:35] ralsina, if I try to kill a dbus daemon all my session dies. To debug this I had to resort to twm and a few xterms :-) [12:35] alecu: haha [12:36] alecu: so, can you try installing a signal handler and see ifyour script dies silently? [12:36] I just forgot how much fun is having to find a place for each window as they are mapped. [12:36] alecu: at least twm doesn't ask you to resize on mapping anymore (or does it?) [12:37] ralsina, is python able to handle sigsegvs? I'll give that a try. [12:37] alecu: it should, IIRC [12:37] I mean, is any process able to handle sigsegvs? [12:38] awesome. [12:38] "Because the C signal handler always returns, it makes little sense to catch synchronous errors like SIGFPE or SIGSEGV." [12:38] alecu: so maybe not ^ [12:39] hmm.... http://pypi.python.org/pypi/faulthandler/1.0 [12:39] alecu: looks like you *may* be able to do it from C, but that's painful right now [12:40] alecu: another idea: make apport ignore that specific error / all segfaults? [12:40] I'm doing: apt-cache install python-faulthandler [12:41] alecu: hey, if that works, cool. [12:41] ralsina, I believe when we are out of the beta period, apport gets a lot silenter [12:42] alecu: we could do a bug pattern so apport keeps quiet about this bug and stops reporting it [12:42] ralsina, python-faulthandler is in universe, not in main. [12:42] alecu: not sure if that means it doesn't show the "some app crashed" dialog [12:44] * alecu brbs [12:48] ok, I'm off to have lunch :) [12:49] buen provecho mandel [12:50] this is awesome. Faulthandler comes standard with py3.3. [12:51] ralsina, alecu: what are you guys trying to do? [12:52] dobey, we are trying to find a way to silence some segfaults in pyqt [12:52] dobey: we get a segfault id dbus-daemon crashes [12:53] dobey: or closes (as in end of session) [12:53] alecu: a bug pattern should do that [12:53] dobey: and we got a bazillion dups for that [12:53] or well, it really shouldn't be necessary now. apport should say "This bug has already been reported." [12:53] dobey, awesome. Where should we put that pattern? [12:53] dobey: patters make apport be silent? (the docs are not that clear) [12:53] alecu: i'm not sure exactly [12:54] rye: ^^^ can we have a bugpattern for this bug? [12:54] alecu: it's a launchpad side thing, though [12:54] not on the client [12:54] dobey, even better! [12:54] rye, the bug is https://bugs.launchpad.net/ubuntu/+source/ubuntu-sso-client/+bug/943046 [12:54] Ubuntu bug 943046 in ubuntu-sso-client (Ubuntu Precise) "ubuntu-sso-login crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] [12:54] dobey: we can propose via https://code.launchpad.net/~ubuntu-bugcontrol/apport/ubuntu-bugpatterns maybe [12:55] rye, but LP keeps throwing a timeout, probably because of the many bugs marked as duplicated of that one. [12:55] perhaps, i have no idea. afaik, rye has done most of the ones for our stuff [12:55] dobey: ack [12:55] alecu: yes, because of the dupes, it sometimes gives a timeout on the db query [12:56] ralsina: also, can we perhaps just fix the bug? :) [12:56] dobey: it's a bug in Qt dbus mainloop [12:56] yes [12:56] dobey: a biiiiit too deep in the stack. And in C++ [12:59] alecu: btw, your test script is broken and can leave a bunch of dbus-daemon processes running :) [12:59] dobey, yes.... I didn't care about that much :-) [13:02] gah it's a 3.5 year old bug [13:07] dobey: it's a race condition, happens only in some cases, and if you don't have apport noone wuld notice :-/ [13:08] yes well [13:11] alecu: can you bug on http://www.riverbankcomputing.co.uk/mailman/listinfo/pyqt about it, and send your test script there? [13:13] dobey, sure. [13:13] dobey, and according to this, we should be using QtDBus instead: http://www.riverbankcomputing.co.uk/static/Docs/PyQt4/html/dbus.html [13:16] alecu: we *are* using that [13:16] alecu: or maybe not, that documentation is very uh, "incomplete" i guess [13:17] dobey, we are using dbus-python, we should be using QtDBus [13:18] ralsina, do you know where I can find a repository browser for pyqt? I'm unable to in the riverbank site. [13:19] alecu, sounds about right, we are using the deprecated one.. which is going to be interesting in sso since we need to support gtk and qt.. [13:19] alecu: not really [13:20] alecu: interesting never heard of QtDBusbefore [13:20] alecu: https://code.launchpad.net/python-qt4 ? [13:21] alecu: or /ubuntu/+source/python-qt4/ if not? [13:23] dobey, nice. But my point is that riverbank only releases tarballs, like it was 1996. [13:24] oh, well. this is where i would normally insert some comment about how they're using qt. :) [13:26] alecu: there is no public official repo [13:26] alecu: AFAICS [13:57] I took tuesday as review day, though I really do [13:57] n't care which day I have [13:58] so if anyone prefers tuesday, just switch me to a different unoccupied day, and tell me ;) === m_conley_away is now known as m_conley [14:03] i took monday, but i'll do whatever day if someone needs that one [14:03] thisfred: I am appalled, neigh, disturbed deeply, by your insidious behavior. [14:08] * ralsina will review all branches proposed on saturdays [14:09] and that, BTW is to be read "all branches (proposed on saturdays)" and not "(all branches proposed) on saturdays" [14:09] * thisfred writes a cron job [14:10] * mandel back [14:11] for me as long as it is not tuesday of thurs for me I'm ok [14:11] rugby is those days, I can move things around other days [14:11] alecu: ok, looking to create a bugpattern [14:12] OMG [14:12] alecu: so, what should be the master bug? [14:12] rye, it's https://bugs.launchpad.net/ubuntu/+source/ubuntu-sso-client/+bug/943046 [14:13] Ubuntu bug 943046 in ubuntu-sso-client (Ubuntu Precise) "ubuntu-sso-login crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] [14:13] mandel: how about you take monday and i'll switch to thursday? [14:13] alecu: hm, somebody already added that i guess [14:14] briancurtin, sure, that sounds perfect to me [14:14] mandel: oh, well i guess you could do whatever other day you want, i didn't look and thought other days had filled up [14:15] :) [14:15] dobey, I updated the branches following the comments from you and alecu please review when ever you can [14:15] alecu: Robert Roth @ 484.1.2 - Added bugpatterns for two frequent ubuntu-sso-client bugs, #943046 and #940669 [14:15] mandel: ok thanks [14:16] briancurtin, mondays sound good, I have to start later so that I overlap with others, so is good to wake up late on mondays hehe [14:16] ralsina, 1-1 in 5 min? [14:16] rye, awesome. [14:16] mandel: sure [14:17] rye, But our bug in sso (bug 943046) is in fact caused by a bug in pyqt (bug #334757). [14:17] Launchpad bug 943046 in ubuntu-sso-client (Ubuntu Precise) "ubuntu-sso-login crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] https://launchpad.net/bugs/943046 [14:17] Launchpad bug 334757 in KDE Bindings "update-notifier-kde.py crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] https://launchpad.net/bugs/334757 [14:17] rye, should we mark our bug as a duplicate of that pyqt one? [14:17] alecu: and kill that pyqt bug too? :) [14:17] rye, right :-) [14:17] mandel: should be able to review now. someone was wrong on the internet, so i had to correct them first :P [14:19] dobey, lol [14:20] ralsina, I'm ready when you are (mumble) [14:20] alecu: bug #943046 should be marked as a dup of bug #940669 though [14:20] Launchpad bug 943046 in ubuntu-sso-client (Ubuntu Precise) "ubuntu-sso-login crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] https://launchpad.net/bugs/943046 [14:20] mandel: going [14:20] Launchpad bug 940669 in ubuntu-sso-client (Ubuntu) "ubuntu-sso-login crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] https://launchpad.net/bugs/940669 [14:23] rye, I'm marking it like that, and adding some comments regarding the root issue. [14:24] alecu: okay, and then we will need LP people to fix their query :) so that it works for our enormous dbus missing bug [14:25] rye, it seems to be a bug in pyqt. I've reproduced it in Lucid all the way up to Precise. [14:26] alecu: yes, that report against pyqt was actually filed from Jaunty :) [14:26] awesome :-) [14:27] rye, I've added a script to bug #334757 that shows the segfault only using pyqt. [14:27] Launchpad bug 334757 in KDE Bindings "update-notifier-kde.py crashed with SIGSEGV in QSocketNotifier::setEnabled()" [High,Confirmed] https://launchpad.net/bugs/334757 [14:27] alecu: fix is easy try: ... except SegmentationViolationError: pass [14:27] rye, nice! but now I get an Undefined error! [14:28] for whatever generation reading my comment above - NO, you can't trap this error that way [14:28] * rye knows he will be searching for something like this in the future [14:29] alecu: basically i think it is the same thing as with our python dbus bindings - when dbus disappears prior to our process [14:29] oh [14:29] no [14:30] alecu: g_dbus_connection_real_closed: Remote peer vanished with error: Underlying GIOStream returned 0 bytes on an async read (g-io-error-quark, 0). Exiting. [14:30] * alecu waves to all the Archaeologist aliens and/or cockroaches [14:30] ^ is that an expected response ? [14:31] aaah, you launch your own dbus [14:31] rye, that means that the process did not die with a segfault [14:31] ralsina, bug 907511 [14:31] Launchpad bug 907511 in Ubuntu One Client "Windows: events from directories where no watch has been added should be ignored" [High,In progress] https://launchpad.net/bugs/907511 [14:31] rye, the segfault does not happen every time, just often. [14:32] on the other hand, on my lucid vm it happens every time. [14:32] alecu: ah, you are killing the daemon, right [14:32] alecu: we had the same bug but with native python bindings. but at least they were not segfaulting, only raising an exception [14:32] alecu: yay segfault [14:33] got it? [14:34] alecu: yep [14:39] * rye wants a core file.. [14:39] ulimit -c unlimited, right [14:40] * rye stops debug output to IRC [14:48] ralsina: since i'm not sure of their usual conventions, what do i do with the latest message on https://rt.admin.canonical.com/Ticket/Display.html?id=51590 - i'm guessing i could scp that folder, or is that something else? [14:48] briancurtin: looking [14:49] briancurtin: get those files, put them in dist, do an installer, gpg-sign it, upload the installer and the signature, mention it on the RT, ask for it to be signed [14:49] briancurtin: you should be able to scp from chinstrap, yes [14:49] mandel: do you have a lucid chroot/vm available? [14:50] dobey, I do, yes [14:50] dobey, tell me :) [14:50] getting a link for you :) [14:51] mandel: can you fix devtools to build on it again? it seems to fail to start squid [14:51] mandel: https://launchpadlibrarian.net/100003195/buildlog_ubuntu-lucid-i386.ubuntuone-dev-tools_3.1%2Br60-15~lucid1_FAILEDTOBUILD.txt.gz [14:51] mandel: so the nightlies for it aren't building on lucid :-/ [14:51] dobey, uh! cute, on it! [14:51] thanks! [14:52] dobey, no problem, I might take a little time because I have to update th vm but besides that I'm ready to rumble :) [14:54] mandel: ok. i'm filing a bug for it [14:55] thx [14:56] dobey, I'll try to see if there is an issue with the way squid is started, and other problem is that it might be dependent on the machine used to build it [14:59] dobey, mandel, thisfred, briancurtin, alecu, HARRY!: standup [14:59] in 30" [15:00] gatox and urbanape are off [15:00] me [15:00] me [15:00] harry? [15:00] me [15:01] mandel: bug #980880 is all for you :) [15:01] Launchpad bug 980880 in ubuntuone-dev-tools trunk "Build fails on Lucid - Fails to start squid process" [Undecided,Triaged] https://launchpad.net/bugs/980880 [15:01] dobey: he's our imaginary friend [15:01] dobey, yes sir! [15:01] mw [15:01] meh [15:01] me [15:02] sorry my imaginary friend is i [15:02] hehehe [15:02] me [15:02] terrible joke [15:02] mandel: it's j in python [15:02] DONE: bureaucracy. Felt sick (allergic to bureaucracy?) Saw doc, team call, 1-1 with gatox and mandel. Got drugs. TODO: help around, lots of tiny stuff, too smal and numerous to describe BLOCKED: no [15:02] NEXT: briancurtin [15:02] DONE: signing, testing my broken releases, fixing buildout, rebuilding and testing, actually got signed binaries [15:02] TODO: currently getting the signed bins, rebuilding the installer, sending that off to be signed. figuring out how to get buildout to work nicely with the installer automation [15:02] BLOCKED: none [15:02] NEXT: mandel [15:02] DONE: 1-1. Updated all ubuntuone-dev-tools branches and the affected branched by that (sso and control panel) [15:02] TODO: Beg fore reviews, find out what is going on in L with ubuntu one dev tools bug 980880 [15:02] BLOCKED: no [15:02] dobey, please [15:02] Launchpad bug 980880 in ubuntuone-dev-tools trunk "Build fails on Lucid - Fails to start squid process" [Undecided,Triaged] https://launchpad.net/bugs/980880 [15:02] .DONE: test improvement, u1db reviewing TODO: <-- BLOCKED: no NEXT: dobey [15:03] λ DONE: reviews, u1db packaging investigation [15:03] λ TODO: bug #682954, bug #956407, reviews, SRUs, u1db packaging [15:03] λ BLCK: none. [15:03] Launchpad bug 682954 in Ubuntu One Client stable-2-0 "Typo in man page for u1sdtool, "path's" to "paths"" [Medium,Triaged] https://launchpad.net/bugs/682954 [15:03] Launchpad bug 956407 in Ubuntu One Client stable-2-0 "Apport recipe attaches old/useless log files" [Medium,Triaged] https://launchpad.net/bugs/956407 [15:03] alecu [15:03] thisfred: you have NOTHING to do? [15:03] DONE: reviews, debugged the hell out of the pyqt dbus error [15:03] TODO: finish patches for the secur^W very secret bug, more reviews [15:03] BLOCKED: no [15:03] thisfred: I can help you with that! :-) [15:04] ralsina: that was meant to signify same as DONE, but I'm not overloaded, so if there's something urgent [15:04] thisfred: oh, well, no nothing all that urgent [15:04] briancurtin, can you take a look to my MP [15:04] ?? [15:04] thisfred: you could do some of mandel's reviews, maybe [15:04] thisfred, oh, yes, please :) [15:05] is it tuesday? I didn't think so! [15:05] :P [15:05] mandel: yep, i have them open, am just tied up with getting the signed binaries out, repackaging, and sending the installer off to be signed. i'll be reviewing soon (ran all tests yesterday - all pass) [15:05] briancurtin, ok, thx [15:05] thisfred, no worried, brian got them! [15:05] mandel: I'll get on it after I review the u1db oauth branch [15:06] oh or not :) [15:06] thisfred, ok, I'll grab a coffee and I'll let you know which ones they are [15:06] * mandel coffee [15:06] kcool [15:07] mandel: so, i see "save" is still used a bunch in the tests themselves, which test the testcase [15:18] ok, i need to get some lunch. will continue reviews after, and do sru work [15:18] bbiab [15:20] ralsina: so what i got back from them is, e.g., "python27.dll" and "python27.dll.signed". do i now take all of the *.signed files and rename them to the appropriate exe/dll inside /dist? [15:20] or is *.signed something to verify, and *.exe is the actual signed exe? [15:20] (seems obvious by name, but im new to GPG stuff) [15:21] briancurtin: yes [15:21] briancurtin: I think "yes the second" :-) [15:22] briancurtin: depends on who did the signing ;-) [15:22] Matthias Arnason [15:24] briancurtin: I meant, they sometimes zip everything and sign the zip, or provide a folder with the signed exes and gpg signatures. Don't recall who does which [15:24] briancurtin: in any case, gpg sigs are a couple hundred bytes, so anything larger is the real binary === salgado is now known as salgado-lunch [15:25] ah, i have a folder with file.exe and then file.exe.signed where *signed files are an extra 6-8k, so i'll take the *signed ones [15:42] dobey, really? I though I changed all of them, whihc revno? [15:43] dobey, ok, you mean the var names, let me change that [15:44] briancurtin: if the .exe are the ones you sent, and the others are larger, yes, that's the thing [15:45] ralsina: that's what i figured, just wanted to be sure. when i renamed the foo.exe.signed to foo.exe, i can see in the "digital signature" tab of the file properties that it's signed, so that's it [15:46] dobey, sorted, new version ready with no save [15:49] briancurtin: cool [16:03] dobey, I have run the tests several times in an L vm and it passes all the tests correctly [16:04] dobey, squid gets started with no problems [16:15] mandel: any idea why it would fail in nightlies then? [16:18] dobey, hm.. the way it works i by calling 'squid -k check -f config_file' to check the pid file of the process, what might be happening is that squid is taking too long to start and we run out of tries [16:19] dobey, we have 8 tried to start squid and we use the following waiting periods: 0.4, 0.1, 0.1, 0.2, 0.5, 1, 3, 5 [16:19] dobey, and waiting for 5 and not starting is a very strange thing... [16:22] hrmm [16:25] and why does my liver feel funny [16:27] mandel: still 2 more save, in class names :) [16:27] dobey, where? [16:27] class names in the test file [16:27] dobey, and alcohol [16:27] dobey, let me fix that [16:28] also, a speclling error [16:29] hmm [16:29] dobey, where is the spelling error? [16:30] commented on the mp with it [16:30] +class MultipleSercicesTestCase(TestCase): [16:30] and also queried about the use of "service" as i thought we'd agreed to use "server" instead [16:32] dobey, you are right, I'll change everything to server and not service [16:33] ok, thanks [16:33] sorry this is being a bit extraneous of a review :) [16:37] dobey, no problem :) [16:39] dobey, grep "[S|s]ervice" . -Rn returns no results, so it should be ok now [16:41] ok [16:42] i also commented on the lucid failure bug with another way you might be able to reproduce the problem [16:45] dobey, great, I'll move to that bug then [16:46] dobey, I pushed the new revno for the test cases, let me know so that I make all the changes in the other 4 branches that depend on this :) [16:46] ok. waiting for launchpad to catc hup :) [16:47] mandel: what revno is it, btw? [16:48] dobey, 75 [16:48] ok [16:55] totally going to wear this new t-shirt to the yuri's night party tomorrow [16:56] mandel: i think you forgot to push. launchpad knows nothing about r75 and it's not waiting to rescan [16:56] dobey, really, let me do it again [16:57] ralsina: care to review https://code.launchpad.net/~dobey/ubuntuone-client/man-typo-2-0/+merge/101935 ? :) [16:58] dobey: sure [16:59] dobey: the u1sdtool help was sort of a big lie, last I checked :-) [17:00] dobey, ha, I'm stupid I pushed it to the wrong project [17:00] dobey, should be ok now [17:00] OMFG, a grocer's apostrophe! Shameful! [17:00] ralsina: well, even if it's a big lie, it should be proper english :) [17:01] also this branch is already released in 3.0/trunk. just backporting it for a 2.0.x release for SRU :) [17:01] dobey, FYI I managed to reproduce the issue [17:01] dobey: +1 and global approve [17:02] thanks [17:02] mandel: cool. now fix it ;) [17:02] dobey, I wonder what is wrong, from the apt-get source it does not work, when pulling from trunk it works [17:03] looks like I'm going to learn something.. [17:03] mandel: no idea [17:03] ok, lunch time for me. [17:04] ralsina: also just proposed https://code.launchpad.net/~dobey/ubuntuone-client/apport-newonly-2-0/+merge/101936 if you would please :) === alecu is now known as alecu-lunch [17:05] dobey, what uses is used to build the packages? [17:05] dobey: gona take a bit longer on that one. I am hungry. [17:06] ralsina: buen provecho then :) [17:06] mandel: eh? [17:06] mandel: the packages are built with fakeroot [17:07] mandel: but that doesn't explain why it would only fail on lucid [17:07] so, looks like QSocketNotifier gets somehow dereferenced, i.e. d_ptr becomes 0x0, and then d->threadData->eventDispatcher just blows off [17:07] dobey, the problem might be that the proxy cannot start because the folder most be accessible by the user 'proxy' [17:08] mandel: but that should fail on all versions then no? [17:08] dobey, would be nice to see if we can run 'squid -N -X -f _trial_temp/squid.conf' [17:09] mandel: and also would fail from trunk? [17:09] dobey, if we run that command form the folder we can see if squid does start [17:09] dobey, which probably is not the case [17:10] dobey, let me walk the dog and I'll keep digging === salgado-lunch is now known as salgado [17:22] ok, fun. Private object got dereferenced, but its parent object was not realy notified about that. Fuuun [17:23] rye: what are you debugging? [17:25] dobey: the qt failure when dbus goes away and client gets mad about it [17:25] oh [17:34] * dobey makes a lucid pbuilder chrootball [18:00] dobey +1 on the apport branch [18:02] yay thanks === alecu-lunch is now known as alecu [18:11] rye, I've tracked it to this file: http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/precise/python-qt4/precise/view/head:/dbus/dbus.cpp#L34 [18:12] rye, my guess is that remove_watch should disconnect both "activated" signals that are connected on add_watch. [18:13] rye, since it's either readSocket or writeSocket the one that's failing. (or perhaps it's both) [18:19] * mandel back [18:23] alecu: can you re-review https://code.launchpad.net/~mandel/ubuntuone-dev-tools/tcp-testcases/+merge/99759 now please? [18:23] mandel: can you remove nessita as a reviewer from that branch? or should we bug her to do a review? :) [18:23] dobey, certainly [18:24] dobey, I can reassign not remove, is that correct? [18:24] dobey, I could assign it to me :) [18:24] hmm [18:25] she is on swap today [18:25] mandel, it's ok if you assign it to yourself, then abstain [18:25] dobey, I'm on it [18:25] yeah, change it to you and abstain [18:25]