/srv/irclogs.ubuntu.com/2011/12/02/#ubuntu-desktop.txt

desrtsmspilla1: hey04:54
desrtsmspilla1: mfisch is trying to track down a bug whereby on ubuntu, gdk_beep() doesn't do anything04:54
mfischg'day smspilla104:54
desrtsmspilla1: mutter has code that intercepts the beep and either displays it as a visual flash and/or causes libcanberra to play a beep sound04:54
desrthttp://git.gnome.org/browse/mutter/tree/src/core/bell.c04:54
desrtsmspilla1: does compiz have a plugin to try to do something similar?04:55
mfischsmspilla1: under ubuntu, the beep only works if a) pcspkr module is loaded and b) xset b on was run04:55
desrthe may not quite be awake yet.  he's on the west coast :)04:55
desrtor he may be at lunch.  i guess it's noon there.04:56
mfischdesrt: timezone website says 1pm04:57
mfischdesrt: that's about beer-o-clock in .au on a friday ;)04:57
desrtmfisch: it's actually 7:30 in that case04:58
desrtthey hang their clocks upside down in australia04:58
desrtmfisch: could also be that he actually decided to go to school today :)04:58
pittiGood morning05:01
RAOFdesrt: There's a longstanding bug on compiz for that.05:04
smspilla1desrt: yes05:04
smspilla1desrt: its not on by default05:04
smspilla1desrt: I also need to ask you something05:04
smspilla1desrt: is there any reason why, after an execvp (), writing to a pipe that is O_CLOEXEC (that I set up at the start of my main loop) on a signal causes the relevant gmain source not to have its prepare () function called ?05:05
smspilla1(I'm using glibmm if there are any known issues there)05:05
smspilla1desrt: school is also finished :p05:06
desrtsmspilla1: hold on a second.  it's taking me a little while to parse that sentence :)05:06
smspilla1mfisch: but yeah, you're looking for ..05:06
desrtsmspilla1: so if you execvp() then you no longer get to write to your O_CLOEXEC pipe...05:06
desrtthat's sort of the point05:06
desrtso i think i'm misunderstanding the question05:07
smspilla1desrt: right, well, I call pipe () as soon as I set up my main loop05:07
smspilla1so it  gets set up again no problem05:07
desrtthe mainloop no longer exists after the exec...05:07
desrtthe entire old process image gets blown away05:08
smspilla1oh, sorry I forgot to add05:08
smspilla1we are execvp()'ing ourselves05:08
smspilla1(apparantly you're supposed to do it on SIGHUP)05:08
smspilla1(minor detail ;-))05:08
smspilla1and if I test it with something like write (myPipeWriteEnd, "h", 1); and then do poll (myPipeWriteEnd) poll () returns and I can read data from the pipe05:08
desrtso if all your fds are O_CLOEXEC and you're exec()ing, it doesn't matter05:08
desrtit's as if you started fresh05:08
smspilla1but glib never calls the prepare function after execvp05:08
desrtsmspilla1: can i see some code?05:09
smspilla1desrt: yeah, maybe I included too much irrelevant information05:09
smspilla1desrt: sure05:09
desrtsmaller is better, as always :)05:09
* smspilla1 snickers05:09
* desrt prefers the exit-on-SIGHUP approach (and wait until dbus reactivates you)05:10
smspilla1mfisch: https://github.com/hypodermia/compiz-bell-plugin05:10
smspilla1desrt: ok, hang on05:10
smspilla1desrt: I think the SIGHUP probably comes from the X Server in this case05:11
smspilla1compiz was written before dbus really existed05:11
desrtsmspilla1: you should get that plugin in the distro...05:11
smspilla1desrt: yeah I should, it keeps dropping off my stack05:11
desrtsmspilla1: waiting for dbus activation isn't really appropriate for a window manager, either :)05:11
desrtmfisch: problem solved, then.... (as soon as smspillaz finds time)05:11
smspilla1desrt: and weirdly enough david was opposed to it because it was "putting too much stuff in the compiz process"05:11
smspilla1mfisch: oh on that05:11
smspilla1mfisch: you're probably going to have to remind me every day05:12
smspilla1:)05:12
smspilla1this thing has a tendency to drop off my stack05:12
desrtsmspilla1: hey05:12
desrtsmspilla1: did you remember about the compiz bell plugin?05:12
smspilla1desrt: what's a bell ?05:12
smspilla1successful troll is successful05:12
desrt+1 counter-troll05:13
=== smspilla1 is now known as smspillaz
desrtsmspillaz: you have until i'm done brushing my teeth and flossing05:14
desrtthen i'm going to bed :p05:14
smspillazkk05:14
smspillazthe awkward moment where you can't find your own code05:15
smspillazah here we go05:17
smspillazdesrt: https://code.launchpad.net/~smspillaz/compiz-core/compiz-core.fix_894639-test05:18
smspillaz(its still pushing to it sadly)05:18
smspillazdesrt: http://bazaar.launchpad.net/~smspillaz/compiz-core/compiz-core.fix_894639-test/view/head:/src/screen.cpp#L16605:19
smspillazso if I do poll on signalPipeFds[0] it returns and says there's stuff that's ready05:19
smspillazbut the relevant source for it which is05:20
smspillazhttp://bazaar.launchpad.net/~smspillaz/compiz-core/compiz-core.fix_894639-test/view/head:/src/screen.cpp#L28805:20
smspillaznever actually gets called05:20
smspillazbut *only* after an execvp05:20
smspillazso execvp -> set up pipe -> set up source -> signal -> closeDown -> write -> (prepare not called)05:21
smspillazas opposed to the normal startup case where prepare is called05:21
* mfisch is catching up...05:21
smspillaz(I am writing to it from the signal handler btw, since that's the only safe way I think think of delivering the fact that the signal happened to the main loop)05:22
smspillaz(while the main loop is still polling for events)05:22
smspillaz(whihc never actually wakes up in the case I'm testing it in because we're using a fake X server to do compiz testing)05:22
desrtcloseDown() is in the signal handler?05:23
smspillazyeah05:23
desrtasprintf supposedly because it's signal-safe?05:23
smspillazfor debugging05:23
mfischRAOF & smspillaz: do you have a bug number for the beep issue?05:23
smspillazmfisch: not one off the top of my head05:24
mfischsmspillaz: would it be against compiz?  I'll go look around05:24
desrtsmspillaz: okay... let me get this straight05:24
desrtsmspillaz: you have a pipe pair marked as O_CLOEXEC05:24
smspillazyes05:24
desrtand you write to the write end05:24
desrtthen immediately call execvp()05:25
smspillazuh, no05:25
desrtokay.  good.05:25
smspillazits more like05:25
smspillazpipe -> mainloop -> SIGHUP -> closeDown -> write -> mainloop -> (signal pipe wakes up) -> g_main_loop_quit () -> execvp (ourselves) -> repeat -> SIGTERM -> write -> mainloop (source never gets prepare called)05:26
RAOFmfisch: https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/30117405:26
ubot2Launchpad bug 301174 in pulseaudio "Use proper sound event instead of system beep" [Wishlist,Fix released]05:26
smspillaz(I have another question about g_main_loop_quit but I'll save that for later)05:27
mfischRAOF: thanks.05:27
* mfisch goes to bed satisfied05:27
desrtmfisch: nite05:27
smspillazdesrt: I guess if there's no obvious "no you can't do that" I should probably single step gmain.c to see what the problem is05:28
smspillazI'm guessing its probably a problem with my code05:29
smspillazin some very weird way05:29
desrti just don't understand what could possibly be the difference between a normal exec and the re-exec05:29
desrtunless you foul up the argv or the environment05:29
desrtexec is pretty much a clean-start05:29
smspillazyeah, that's whats confusing me too05:30
desrtargv, the environment or leaked file descriptors should be the only thing that goes across05:30
smspillazok, single stepping it is then05:30
smspillazdesrt: also meant to ask05:30
smspillazdesrt: If you have sources A, B and C, and B and C are stuck in poll () and have never been dispatched before and then A gets an event and calls g_main_loop_quit ()05:31
smspillazis it normal that B and C remain stuck in poll and the mainloop doesn't actually quit ?05:31
desrtsmspillaz: you may be interested to know that glib has unix signal watches these days05:31
desrtsmspillaz: see g_unix_signal_add()05:31
smspillazdesrt: cool, I'll look into that05:31
smspillaz(the documentation says "if there are sources still waiting to be dispatched, g_main_loop waits on them)05:32
smspillaz(so right now I have this awkward workaround which immediately dispatches and force fails all sources as soon as a g_main_loop_quit happens)05:32
desrtsmspillaz: it doesn't really work like that...05:33
desrtassuming you have no threads, poll() is only ever running when nothing else is running05:33
smspillazs/stuck in poll/returns to poll/05:33
desrtso if you're calling g_main_loop_quit() then, by definition, nothing is stuck in poll05:33
desrtnah..05:33
smspillazhm ok05:33
desrtthe loop won't poll again after g_main_loop_quit()05:33
smspillazthats at least what I've obvserved05:34
desrtthere was a race about that before05:34
desrtbut only when threads were involved05:34
desrtand i think the race is fixed now05:34
smspillazhmm05:34
smspillazok, I'll have another look05:34
smspillazdesrt: It could be that glibmm's Glib::MainLoop::quit () is broken05:34
mfischdesrt & smspillaz: thx for the assistance05:35
desrtmfisch: no prob05:35
desrtsmspillaz: imho, glibmm is not worth the trouble05:35
desrti'd just use glib's C API directly from C++05:35
smspillazdesrt: yeah I've noticed some ... weird stuff about it05:35
smspillazdesrt: though the C API is awkward to use directly05:35
smspillazor at least, it doesn't fit our object model that well05:35
desrtsmspillaz: the C API is awkward to use from C too :)05:35
smspillazdesrt: I think glib is just awkward really05:36
smspillazor maybe I'm awkward05:36
smspillazand glib is really cool05:36
desrtglib is pretty good for what it is05:36
smspillaz:p05:36
desrtit's pretty hard to make C look like an object-oriented language though05:36
desrtso some awkwardness is necessarily implied :)05:36
smspillazdesrt: http://images.cheezburger.com/completestore/2011/12/1/d5fb6962-c9c0-4095-8bcb-9a9140340b66.jpg05:38
smspillaz:305:38
smspillazdesrt: I dunno, I'd like to think that we can get nice C++ bindings to G*, but being able to do that is really hard05:39
smspillazdesrt: anyways, g'night :)05:39
smspillazI read through the documentation for gmmproc this morning and05:40
smspillazyeah05:40
smspillazmy brain hurt05:40
broderugh, but gmmproc is doing so much more work than it needs to these days05:41
broderi really, really wish somebody had a tool that generated C++ bindings using GI05:41
smspillazoh there isn't one ?05:41
smspillazhmmmmmmmmmmm05:41
* smspillaz plots05:41
broderwell, there's smoke-gobject/gobject-consume/whatever it's called05:41
broderbut that's not really functional05:41
broderand gmmproc requires pre-writing .hg and .ccg input files that are basically pre-processor food05:42
smspillazyeah  Isaw05:42
broder(so with gmmproc you're basically binding everything by hand)05:42
smspillazwhen I was reading the gmmproc documentation I was kind of like05:42
smspillazoh taht05:42
smspillaz*that05:42
smspillaz"haha"05:42
brodersmspillaz: anyway, have to run, but if you were to write a tool that built gobjectmm-based bindings using just GI information, that would be *so* *cool*05:44
broderand i'd be happy to try and help05:44
smspillazI can see desrt sitting somewhere in Canada screaming "use a real language!" :p05:44
broderC++ is a real language!05:45
broderC++11 even more so!05:46
pittibroder: even more so, it's like the union of ten languages05:50
pittiif only it had a sensible way of handling ABI stability05:50
smspillaz_ZICanHaveAGoodABITooC8bp9_R3w05:53
pittismspillaz: or the really nasty habit of putting constructors in .h files, it's insane05:57
pitti*cough* apt *cough*05:57
RAOFWell, where else are you going to put them? :)05:58
pittiRAOF: that's an obvious answer05:58
pitti"in your face" of course, everytime it blows up05:58
RAOFIt's always fun when the .h file has a mixture of class declarations and inline functions.  Hello, nux!05:58
pittihaving a ctor like "MyClass(): member1(0), member2(NULL)" might look nice and convenient, but it's utter insanity whenever you have to add a new private member to a class05:59
pittiwith that, changing private members effectively breaks ABI06:00
RAOFWell, yes.06:00
smspillazpitti: yeah, I usually try to keep code out of headers in compiz too06:02
smspillazthough there are still some places where it is there (left by other people that are not yours truly of course)06:02
smspillazpitti: code out of headers and PIMPL is how I roll06:03
smspillazpitti: I think C++11 brings a standard ABI though doesn't it ?06:04
smspillazalthough, it still really easy break the ABI06:04
smspillazpitti: C++ would also be nice if template error messages were not hundreds of lines long06:06
RAOFHeh.06:07
RAOFIt would be nice if templates were runtime generated code, too :)06:08
smspillaznah06:08
pittierr, what?06:08
smspillazpitti: nope, apparantly it doesn't, must have misread something06:09
didrocksgood morning06:51
pittihey didrocks07:05
didrocksbonjour pitti, how are you?07:05
pittididrocks: quite fine, thanks! had some fun this morning unbreaking the archive07:10
didrocksoh? the archive broke after alpha1 unfreeze?07:11
pittisoftware-center was uninstallable, causing all CD builds to fail, plus nautilus-python breaking edubuntu07:11
pittiyes07:11
pittiwell, that's why we have a stable+1 team :)07:11
didrocksheh, from day 2, pitti for rescussing the archive! :)07:13
* didrocks hugs pitti07:13
* pitti hugs back didrocks07:16
=== tkamppeter_ is now known as tkamppeter
tkamppeterpitti, hi07:46
pittihey tkamppeter08:18
tkamppeterpitti, hi08:19
tkamppeterpitti, it is about CUPS 1.5.x. The IPP backend has a lot of regressions.08:20
tkamppeterpitti, bug 881843, bug 877958, bug 879625, bug 88358508:21
ubot2Launchpad bug 881843 in cups "CUPS IPP print to Novell servers error since 11.10 upgrade" [Undecided,Confirmed] https://launchpad.net/bugs/88184308:21
ubot2Launchpad bug 877958 in cups "After every single page on ipp: Unable to get print job status" [Undecided,Confirmed] https://launchpad.net/bugs/87795808:21
ubot2Launchpad bug 879625 in cups "CUPS fails to request authentication when printing over IPP w/ TLS" [Undecided,Incomplete] https://launchpad.net/bugs/87962508:21
ubot2Launchpad bug 883585 in cups "Kubuntu 11.10 -- Network/Local Printers found but cannot print - Unable to get printer status" [Undecided,Confirmed] https://launchpad.net/bugs/88358508:21
tkamppeterpitti, a volunteer, Robert Bradley, tried to fix these and worked with the reporters to get more info.08:22
tkamppeterpitti, he also suggested to downgrade the IPP backend to the one of CUPS 1.4.x and tells that this solves the problems.08:23
tkamppeterpitti, now I am thinking about doing this in our CUPS package, especially also as an SRU for Oneiric.08:24
tkamppeterpitti, on Precise I would then return to the current backend when the bugs are fixed upstream.08:24
tkamppeterpitti WDYT?08:25
pittitkamppeter: hm, I thought the new one implied some change in the names/URIs of the printer08:25
pittitkamppeter: if we would go back, wouldn't we change them back and break them again?08:25
=== jibel_ is now known as jibel
tkamppeterpitti, I did not know about that. The release notes http://www.cups.org/documentation.php/doc-1.5/whatsnew.html do not tell about that.08:27
tkamppeterpitti, from where do you know that.08:27
pittitkamppeter: oh, I'm mixing that up with the USB backend08:28
pittithat was libusb vs. /dev/lpUSBnn08:28
tkamppeterpitti, the USB backend is fine now after 2 SRUs.08:28
tkamppeterpitti, I could quickly prepare a new CUPS SRU package and we could replace the current not-yet-approved one.08:29
pittitkamppeter: would be interesting to see a diff of the USB backend between 1.4.6 and 1.5.0 to see the scope of teh changes08:30
tkamppeterpitti, you mean s/USB/IPP/08:31
chrisccoulsonniiiiiiiiiiiiiiice http://www.youtube.com/watch?v=RuIQskGD3u008:31
chrisccoulsongood morning everyone08:31
pittitkamppeter: yes, I have it all backwards this morning :(08:32
pittihey chrisccoulson08:32
chrisccoulsonhi pitti, how are you?08:33
tkamppeterpitti, diff is attached to bug 881843, comment #6.08:33
ubot2Launchpad bug 881843 in cups "CUPS IPP print to Novell servers error since 11.10 upgrade" [Undecided,Confirmed] https://launchpad.net/bugs/88184308:33
=== gema_afk is now known as gema
pittitkamppeter: uh, this looks like a whole rewrite indeed08:41
pittitkamppeter: so, if you think that's best, fine for me; I don't know whether downgrading would potentially break anyone's configuration or setup08:41
pittii. e. if they are using new features of the 1.5 backend08:42
tkamppeterpitti, as the release notes do not mention anything, it seems that this is merely only stuff to improve on Mac OS X.08:43
tkamppeterpitti, so I will apply it and tell you when the SRU package is ready for approving.08:44
tkamppeterpitti, can you already approve the HPLIP SRU? Thanks.08:44
pittitkamppeter: will do an SRU run later today (have some urgent stuff to fix right now)08:45
pittitkamppeter: so you want me to ignore or reject the current cups SRU upload?08:45
tkamppeterpitti, please reject the current CUPS SRU upload, I will re-upload it with the IPP backend patch.08:46
=== jodh is now known as jodh_
pittitkamppeter: roger, done08:53
chrisccoulsonhmmm, this is depressing - https://bugs.launchpad.net/ubuntu/+source/firefox/+bugs?field.tag=test-suite08:58
pittichrisccoulson: are these filed by a bot of your's?09:01
chrisccoulsonpitti - no, i filed them manually09:02
chrisccoulsoni should probably figure out a way to do that automatically though09:02
rodrigo_hello09:13
pittihey rodrigo_, how are you?09:20
didrocksgood morning rodrigo_09:20
Sweetsharkmoin desktop world!09:26
rodrigo_hi pitti, didrocks09:27
rodrigo_and Sweetshark :)09:27
didrocksgood morning Sweetshark!09:27
pittihey Sweetshark09:30
pittiSweetshark: I have a bug to report09:30
pittiopenoffice.org 1:3.3.0-709:30
pittiSweetshark: that is correctly spelled "3:1.3.3-7"!09:31
tkamppeterpitti, about bug 897309, some proprietary printer driver packages need gs-esp. Should I simply add a "Provides: gs-es" to the "ghostscript" binary package?09:32
ubot2Launchpad bug 897309 in ghostscript "Package «gs-esp» is still needed in Oneric and higher" [Undecided,New] https://launchpad.net/bugs/89730909:32
pittitkamppeter: depends; do they have a versioned dependency? if it's just "gs-esp", a Provides: will work fine09:33
pitti(meh, gs-esp was many, many years ago)09:33
seb128hey09:44
Sweetsharkpitti: huh?09:45
pittiSweetshark: just kidding09:50
pittihey seb12809:50
seb128hey pitti, how are you?09:51
pittiseb128: quite fine :)09:52
pitticleaned up after mvo and myself breaking the archive and daily builds yesterday09:52
pittiand now working on bug 897680 (ugh)09:53
ubot2Launchpad bug 897680 in ubiquity "Precise Desktop 64Bit: libc6 fails to install if "install 3rd party software" is selected" [High,In progress] https://launchpad.net/bugs/89768009:53
pittistable team FTW :)09:53
seb128pitti, ;-)09:55
tkamppeterpitti, the bug does not say anything about versioned or not, so seems to be better to introduce a dummy package to assure that it works with all drivers.09:55
pittitkamppeter: I'd start with a Provides:09:55
pittitkamppeter: I don't like reintroducing 5 year old clutter for that :/09:56
pittiseb128: indicator-me looks obsolete, right?09:56
seb128pitti, it is, I though we dropped it in Oneiric?09:56
seb128pitti, it got merged in indicator-session09:56
pittiseb128: that was replaced with the new session plugin09:56
pittiseb128: yes, but apparently re-synced to precise; I'll remove again and blacklist09:57
seb128pitti, right, did it come back by some way?09:57
seb128oh ok09:57
seb128yes, seems about right ;-)09:57
mvopitti: *eeeh* what did I do ?09:58
pittimvo: software-center depended on the new plugins which were in binNEW and universe09:58
mvoaha, ok09:58
tkamppeterpitti, CUPS SRU re-uploaded.10:00
tkamppeterpitti, patch also pushed to BZR.10:01
pittitkamppeter: current version is in testing, so I guess I better upload this10:05
pittitkamppeter: thanks!10:05
Sweetsharkpitti: /me is in deer-in-headlight-mode. Monday is release branch off.10:06
Sweetshark;)10:06
pittiSweetshark: oh, that's kind of "release time crunch" for LibO then?10:06
Sweetsharkpitti: yes, the scheduling of LO and Ubuntu never leaves me without a 'release time crunch' ;)10:08
chrisccoulsonhah, i wish i could say the same for firefox10:12
chrisccoulson24th April - Firefox 12, 26th April - Precise ;)10:12
seb128chrisccoulson, it's like the anti-perfect schedule, end of year holidays, hard freeze...10:15
rodrigo_brb10:19
seb128pitti, help :p10:56
seb128bug #89897310:56
ubot2Launchpad bug 898973 in gtk+2.0 "package gtk2-engines-pixbuf 2.24.6-0ubuntu6 failed to install/upgrade: './usr/share/doc/gtk2-engines-pixbuf/NEWS.gz' is different from the same file on the system" [Undecided,Confirmed] https://launchpad.net/bugs/89897310:56
seb128seems like lool broke precise as well :p10:56
seb128that got 2 bug reports today so I guess it's a real bug somewhere10:56
pittioh dear, that sounds like bug     87108310:58
ubot2Launchpad bug 871083 in gzip "gzip -9n sometimes generates a different output file on different architectures" [High,Triaged] https://launchpad.net/bugs/87108310:58
pittiseb128: only known workaround for that right now is to upload a package with dh_compress -XNEWS, I'm afraid11:01
seb128:-(11:01
pittiseb128: do you have time to dupe and upload the workaround, or shall I? (I'm a bit tight on time today, need to catch a train later on and finishing up something)11:02
seb128pitti, I can do it11:03
* pitti hugs seb128, thanks11:03
* seb128 hugs pitti back, you're welcome11:03
seb128pitti, "dupe" the bug you mean? I will rather just close it with the workaround with a note the real issue is the gzip one11:03
pittiseb128: as you prefer11:04
seb128ok11:04
seb128I'm on it ;-)11:04
pittiI'd dupe it so that we collect all the tasks on the master bug11:04
pittiand know what to revert once the real bug is fixed11:04
pittibut we can add a task without duping, too11:04
seb128pitti, ok, makes sense,  I can do that11:05
htorquepitti: hello! about above mentioned bug: i reported it too and it got duped - shouldn't apport have told me that?11:05
pittihtorque: I'm afraid we can't do that yet for package install failures11:06
pittithe log parsing isn't reliable enough yet to do that11:06
htorqueah ok, so just for crashers.11:06
pittias many errors are followups, some are noise, etc.11:06
pittiit's not easy even as a human to spot the real root cause11:06
pittitkamppeter: nice, Robert is already sending patches for the regressions to upstream11:09
loolseb128: you want me to upload with -XNEWS?11:17
seb128lool, if you want to feel free, I'm finishing something and was going to do that next but if you want to do it that's welcome ;-)11:18
htorquewill -XNEWS also take care of README.gz and changelog.Debian.gz? because they seem to cause trouble too.11:19
seb128no, they should probably be added to the -X list11:20
seb128lool, ^11:20
pittihtorque: is that confirmed?11:20
pittihtorque: in libtasn, only NEWS was affected11:20
loolthe bug and the dup only have NEWS listed11:21
looloh it's in gtk+2.011:22
pittiyou can download both .debs, dpkg-deb -x them, and compare md5sums11:22
htorquepitti: i removed NEWS.gz and did 'install -f' and it complained about changelog.Debian.gz. removed that too, then it showed the same for README.gz. removed that one and the update worked.11:23
loolseb128: So why does it relate to me?  :-)11:23
loolhmm E: gtk2-engines-udeb udeb: udeb-contains-documentation-file usr/share/doc/gtk2-engines-udeb/11:24
loolgrabing gtk+2.011:24
seb128lool, I overlooked the -pixbuf part of the name, or rather confused it with your gtk2-engines upload from yesterday :p11:24
seb128lool, sorry :-(11:24
loolseb128: see how you blame me for everything!!1!11:25
seb128lool, (or in fact it's my secret plan to make you do desktop work ;-)11:25
seb128lool, (but sussssh)11:25
loolpitti: It's a bit ugly to add these -Xs if it also affects changelog and README  :-/11:26
loolAh cjwatson was looking after it11:28
pittilool: but at least we exactly know what to revert once it's fixed11:28
loolpitti: You mean we can grep Contents to find packages to revert?11:29
pittilool: no, I think we should add tasks to that master bug11:29
looloh wow, we moved from gzip 1.3.12 from *2007* to 1.4.0 from 201011:30
pittilool: but that wasn't it11:30
loolpitti: Ok; I hope it's not going to affect too many packages11:30
pittilool: we just never noticed beefore multiarch11:30
pittiotherwise I had reverted it already11:30
loolpitti: Ok11:30
pittii. e. it's not a regression11:30
loolpitti: It's just certain contents?11:31
pittiyes11:31
pittiand I never managed to reproduce it locally11:31
pittimust be something spethial on the buildds11:31
pittidifferent kernel, different file system, I don't know11:31
pittiPPAs are fine, local builds are fine with both lucid and precise kernel,e tc.11:31
looloh wow11:31
pittisome buildds allegedly run the hardy kernel, I didn't try that locally11:31
pittilool: I suppose it's something like "truncate() doesn't zero blocks under some circumstances" or "some memory pages remain uninitialized when you write the tail" or whatnot11:32
loolwe don't seem to log kernel version with build logs11:37
pittiright, so I could only guess11:37
loolproposed https://code.launchpad.net/~lool/launchpad-buildd/log-kernel-version/+merge/8424711:38
pittilool: merci11:40
pittiseb128: retracer assertion failure> will have a look on monday, -ENOTIME any more today11:42
seb128pitti, ok11:42
pittiprobably a weird bug which needs to be duped, etc., but need to figure out the logic11:42
tkamppeterpitti, SRU for bug 897309 uploaded, precise package will come soon.11:45
ubot2Launchpad bug 897309 in ghostscript "Package «gs-esp» is still needed in Oneric and higher" [Undecided,New] https://launchpad.net/bugs/89730911:45
pittiseb128: ok, worked around for now, they _should_ be back up until the next gksu crash comes in11:49
seb128pitti, danke11:49
seb128pitti, what is wrong with gksu?11:49
seb128(just curious)11:49
pittibug 89887411:50
ubot2Launchpad bug 898874 in gksu "gksu crashed with SIGSEGV in __strlen_ia32()" [Medium,Confirmed] https://launchpad.net/bugs/89887411:50
pittithe traces all look slightly differently11:50
pittiI don't yet have a firm idea where the duplication logic fails11:50
pittiit's rather conservative right now with lots of asserts11:50
pittiso I'd rather just let it crash again when it happens11:50
seb128ok11:50
pittiI need to run now, cu later on the train (maybe, I'll mostly work offline)11:56
pittihave a nice weekend!11:56
loolpitti: Good WE!11:57
=== MacSlow is now known as MacSlow|lunch
tkamppeterpitti, all SRUs for this week are uploaded now and ready for approval: CUPS, HPLIP, Ghostscript (10 bugs).12:22
seb128ricotz, hey12:39
seb128lool, so will you handle gtk or should I have a look? it seems you started on it?12:39
seb128pedro_, ola, thanks for the sru verification ;-)12:40
pedro_seb128, hello! np, i'm reviewing the ones that need verification today12:42
seb128great12:43
ricotzseb128, hey12:45
seb128ricotz, how are you?12:45
ricotzseb128, a bit busy, but fine12:46
ricotzseb128, hoping you are fine too12:46
seb128ricotz, it would be nice to get rhythmbox updated in precise, do you want to do it? I can do the sponsoring for you12:46
seb128ricotz, I'm fine thanks ;-)12:46
ricotzseb128, i see, i think you can grab the ppa package, i dont remember making packaging changes though12:47
seb128ricotz, ok, are you fine if I sponsor that to precise then?12:47
seb128ricotz, is that version working fine for you? ;-)12:48
ricotzseb128, i am using it all day, i think the crashes i am getting arent related to rhythmbox, so feel free to sponsor it12:48
seb128ricotz, great, thank you12:49
ricotzseb128, as usual the tarball contains some patches to have a working "make dist"12:50
seb128ricotz, that's fine, did you send those in bugzilla or something?12:50
seb128ricotz, btw you should use "snapshot" rather than "release" in your changelog for git snapshots ;-)12:51
ricotzseb128, if i recall it correctly fta had an plugin issue, but nothing serious12:51
seb128ok12:52
ricotzseb128, havent sent them upstream since it is only some uncommented stuff in makefiles and a vala sample plugin problem12:53
ricotzseb128, yeah, the version string doesnt contain git so my script puts release in there12:54
* ricotz needs to strip the "git" to keep the update path with the repo version ;)12:58
loolseb128: I had started it, but had started researching the gzip issue instead; doing it now12:58
seb128lool, thanks ;-)12:59
ricotzseb128, btw, how is it going with glib 2.31.x since you mentioned updating clutter/cogl which would need this newer glib13:01
seb128ricotz, today is friday so not the best day for a new glib serie, I think I will upload it early next week13:02
seb128ricotz, is the current version working fine for you?13:02
seb128i.e no surprise?13:02
seb128out of the eog patch we will ned13:03
seb128need13:03
ricotzmhh, can't think of anything currently13:03
seb128ok, that's a good sign ;-)13:03
ricotzthere were mostly cairo/gtk issues13:03
ricotzseb128, did you see problems with the newer tp-mission-control version?13:04
seb128no, but I didn't watch closely, kenvandine is probably a better person to ask when he will be online13:04
ricotzah, gnome-system-monitor has a similiar issue like eog13:05
ricotzok13:05
ricotzif there arent any reports this tp problem might be caused by glib13:05
seb128not that I noticed but I'm not sure many people run precise yet13:06
=== MacSlow|lunch is now known as MacSlow
seb128but anyway some bugs is not the end of the world for precise, I just wanted to make sure there is no known stoppers13:06
seb128we will figure what to do with issues when we hit them ;-)13:06
ricotzseb128, right13:06
ricotzbetter to find them early in the cycle rather than waiting with the upload13:07
* rodrigo_ lunch13:33
chrisccoulsonseb128, you should upload glib today. precise is boring right now ;)13:33
chrisccoulsoni like living life on the edge!13:33
seb128chrisccoulson, be my guest and do it!13:33
chrisccoulsonlol13:34
chrisccoulsoni can't ;)13:34
chrisccoulsonprobably a good job!13:34
seb128you still didn't apply for upload rights?!13:34
chrisccoulsonnot yet. i haven't really needed to recently13:34
seb128you need them today see! ;-)13:34
chrisccoulsonlol13:34
* desrt yawns13:48
desrtsmspillaz: get your situation sorted?13:48
ockham_hi, anyone feel like reviewing http://revu.ubuntuwire.com/p/unity-lens-bliss ?14:16
mdeslaurThought of the day(tm): instead of deciding the default desktop applications based on toolkits and programming languages, we should have a valgrind faceoff14:18
HOHOHaneyL14:19
HOHOHaneyoops14:19
seb128didrocks, can you explain r211 in the nautilus vcs?14:22
* didrocks looks14:22
seb128didrocks, it's used to get the strings translated?14:23
seb128or I'm overlooking something?14:23
didrocksseb128: hum, I would have thought that, let me check again, but I think we don't preprocess it at all14:23
seb128didrocks, we have a patch that add it to the POTFILES.in14:24
seb128debian/patches/13_translate_unity_launcher.patch:+debian/nautilus-home.desktop.in14:24
seb128didrocks, ^14:24
didrocksseb128: oh, ok, my bad then, please revert it14:24
didrocksseb128: I tried to look at debian/rules mainly14:24
desrtgood morning, didrocks, seb12814:24
didrockshey desrt :)14:24
seb128didrocks, ok, good, I prefered to check if the .desktop got deprecated and I was overlooking something14:24
seb128didrocks, thanks ;-)14:24
seb128desrt, hey, good friday to you!14:24
didrocksseb128: no no, just didn't see where it was used and thinking that we got the translation from another desktop file. Thanks for looking!14:25
desrthappy friday!14:25
pittidesrt: TGIF!14:31
* desrt was quite enjoying this week, actually14:31
dobeypitti: hey14:35
dobeypitti: you wouldn't happen to have any idea why quilt would just randomly exit in the middle of applying a patch, would you?14:35
pittidobey: urgh14:36
=== HOHOHaney is now known as Laney
pittidobey: reproducible?14:36
dobeyand exit cleanly, that is14:36
pittiI see how dpatch could, but not quilt14:36
dobeypitti: sort of; it happens in the recipe build i'm trying to set up for rhythmbox daily builds14:36
pittidobey: that's with "quilt push", or "quilt push -a", or package build?14:36
pittiah, so during dpkg-source -x ?14:37
pittieither way, perhaps stracing reveals an error?14:37
loolseb128: Fighting local gtk+ build issue; the first one is fixed (docbook), I'm now getting another one, the first error in the log is from the testsuite, but it seems ignored; anyway, I'd rather fix it, are the new ubuntu_* functions supposed to be in the ABI or not?  they aren't in headers14:37
dobeyit's quilt push -a -v i tihnk14:37
dobeythink14:37
loolubuntu_gtk_get_use_overlay_scrollbar, ubuntu_gtk_menu_shell_activate_first etc.14:37
seb128kenvandine, ^14:37
dobeyhrmm, actually, maybe it is being applied, but then is failing to unapply for some reason14:37
loolkenvandine: Hey Ken, search for "FAIL: abicheck.sh" in https://launchpadlibrarian.net/86349189/buildlog_ubuntu-precise-amd64.gtk%2B2.0_2.24.8-0ubuntu1_BUILDING.txt.gz14:38
seb128lool, they are used only by overlay-scrollbar I think so I'm not sure how public we want them14:38
loolkenvandine: the lines with +es are new public symbols; either we want these to be public, then we should have them in a .h and update gtk/gtk.symbols, or we want them to be private, naming them _something should be enough to hide them14:38
loolseb128: Ok; will leave it to Ken14:39
dobeypitti: unfortunately, i can't strace, as it's on the LP server; i can't reproduce the problem locally :(14:39
seb128lool, I would not bother with cleaning for gtk2, those are probably fine to let as it14:39
seb128lool, but I'm less perfectionist than you ;-)14:40
* kenvandine looks14:40
dobeypitti: https://launchpadlibrarian.net/86377740/buildlog.txt.gz14:40
loolseb128: I'll be uploading now, but I didn't manage to complete a build; I get an error and didn't verify the binaries: dpkg-buildpackage: erreur: fakeroot debian/rules binary a produit une erreur de sortie de type 214:42
loolI think it wont happen on the buildds14:42
seb128lool, yeah, if you just changed the rules to add the -X there is no reason runtime should break, just upload ;-)14:43
pittidobey: but I supose 04_pause_button.patch actually does apply cleanly locally?14:43
pittidobey: did you try with bzr-builder?14:43
dobeypitti: bzr dailydeb recipe isn't working for me locally, because for some reason it seems to want the rhythmbox branch to be a source package branch with the upstream-foo tags in it :(14:44
pittidobey: uh? well, any of the branch needs to provide a debian/ dir14:45
pittidobey: what does your recipe look like?14:45
pittidobey: it should work locally, otherwise there's little chance of working in LP14:45
dobeyhttps://code.launchpad.net/~ubuntuone-hackers/+recipe/rhythmbox-dailies14:45
pittidobey: are you trying to merge against ubuntu:rhythmbox?14:45
dobeyno14:45
dobeylp:rhythmbox14:45
dobeywhich is the import of upstream git master14:46
kenvandinelool, those have bounced around a bit between being in headers and not... agateau cleaned that stuff up quite a bit last cycle and there was some reasoning for this...14:46
kenvandineagateau, do you remember?14:46
loolkenvandine: It's inconsistent with other patches where gtk.symbols is correctly patches14:46
lool*patched14:46
dobeypitti: building manually locally works fine; i can dpkg -S -sa and then pbuilder-oneiric the .dsc and it builds14:46
loolseb128: uploaded14:46
james_wPatch 04_pause_button.patch does not remove cleanly (refresh it or enforce with -f)14:46
james_wmake: *** [reverse-patches] Error 114:46
kenvandinelool, yeah... maybe we just need to add them there14:46
seb128lool, thanks!14:47
didrocksok, all unity test cases in checkbox now :) Will have to debug checkbox now though :/14:47
loolkenvandine: I had a quick look, and given these aren't in headers, I'd rather have them renamed from ubuntu_foo to _ubuntu_foo14:47
pittidobey: can you try with: nest-part packaging lp:~ubuntuone-control-tower/rhythmbox/packaging-dailies debian debian14:47
pittidobey: otherwise you put the *whole* branch into debian/14:47
loolkenvandine: If you look at gtk.symbols, it maps to header files14:47
pittidobey: i. e. you'll have a debian/debian/14:47
james_wdobey, that doesn't sound like it randomly exiting cleanly in the middle of applying a patch?14:47
dobeypitti: the whole branch is only the stuff that's inside debian/14:48
pittidobey: oh, I see14:48
pitti(that's unusual)14:48
pittidobey: so, I'm in a train, so I can't try that recipe right now (too little bandwidth)14:48
pittidobey: do you have the error message that it produces locally?14:48
dobeypitti: sure14:48
kenvandinelool, yeah... i want agateau to weigh in on that, i think there was a reason they need to be public14:48
dobeybzr: ERROR: No such tag: upstream-2.90.1+r7750dding file 20/2514:48
dobeythat last bit is where it screwed up writing to the terminal and bzr branch output got overwritten in the shell14:49
dobeythe dding.* bit14:49
james_wdobey, are you running with --allow-fallback-to-native?14:49
pittidobey: I'm out of my wisdom then :/ but look, james_w to the rescue!14:50
dobeyjames_w: ah no14:50
james_wRUN ['bzr', 'dailydeb', '--safe', '--no-build', '/home/buildd/build-47ee96de33267c60c108b9c3ed958637e9ffc494/chroot-autobuild/home/buildd/work/recipe', '/home/buildd/build-47ee96de33267c60c108b9c3ed958637e9ffc494/chroot-autobuild/home/buildd/work/tree', '--manifest', '/home/buildd/build-47ee96de33267c60c108b9c3ed958637e9ffc494/chroot-autobuild/home/buildd/work/tree/manifest', '--allow-fallback-to-native', '--append-version', '~precise1']14:51
dobeytrying that now14:51
dobeybut just running the quilt push -a -v then quilt pop -a -R in the tree works fine for me; which is why i'm confused14:52
james_wdobey, what about with "fakeroot debian/rules clean" in between?15:04
dobeyi'll try that in a minute; i've commented out all the patches in series except for one now15:05
dobeyand it still fails on that one patch; which is a one-line change :(15:06
dobeyjames_w: debian/rules clean rmoves the patches15:08
dobeyjames_w: so i suppose if i run quilt pop now it will fail15:09
dobeyit seems to be what is runs quilt pop anyway15:10
james_wI guess that could be the cause of quilt failing15:10
dobeywell, it's failing *inside* the "fakeroot debian/rules clean" in the recipe build15:11
pittihave a nice weekend everyone!15:17
dobeycheers pitti15:18
mterrypitti, can you set https://blueprints.launchpad.net/ubuntu/+spec/desktop-p-unity-greeter to be a precise goal?  it doesn't show up in work item tracker right now15:23
seb128haha15:23
seb128mterry, welcome to the club of the people who tried to get .symbols for cpp and gave up on it using -V ;-)15:23
mterryseb128, heh.  :)15:24
didrockswe should create a launchpad team :)15:24
charlesseb128, my inbox says you are a bug triaging *machine* :)15:24
mterryseb128, I thought I had it.  I used pkg-kde to write a symbols file that had some tags in it, but not enough.  still failed on amd6415:24
seb128charles, lol15:24
seb128charles, I wish others who do a bit extra triage so I would have less to do :p15:24
seb128mterry, btw while you are here, do you want me to take over the gedit 3.2.3 sru?15:25
* mterry bugs didrocks about reviewing his quickly-gtk3 branch15:25
seb128mterry, i've time today for it15:25
mterryseb128, did I sign up for that?  I didn't write it down...  sorry  :-/15:25
didrocksmterry: s/bugs/hugs rather? :)15:26
mterrydidrocks, ah yes... typo...  :)15:26
seb128mterry, I'm slightly disappointed that you didn't do the compiz sru sponsoring yesterday btw :p maybe pitti will get to it on monday during his turn ;-)15:26
didrocksmterry: as you suffer from marshmalling the C++ symbols file as well, I guess it can take some time for looking at your branch!15:26
mterryseb128, oh no I do have the gedit sru in my queue15:26
mterryseb128, I'll get to it in a bit15:27
seb128mterry, no, I signed you up for it in a but comment you probably didn't read :p in fact they worked around the pygobject,snippet segfault15:27
seb128mterry, you had that pygobject bug assigned to you15:27
seb128mterry, ok, thanks15:27
mterryyup, bug15:27
mterry86377315:27
seb128mterry, if you have other things to do that's fine, I'm not too busy today so I can do it as well, just tell me15:28
mterryseb128, no problem, you keep triaging  :)15:28
seb128lol15:28
asacheyho15:33
asacat some point there was the system -> preferences -> preferred applications thing15:33
asacnowadays that doesnt exist anymore :(15:34
asacat least i cannot find it15:34
asacwhere can i now switch mailto: and http: mime handler?15:34
asacme opens gnome-control-center again in hope to find it15:36
didrocksasac: it is in gnome-control-center15:36
didrocksasac: the system info15:36
didrocksthen, there is preferred apps15:36
asacdidrocks: guys that make absolutely no sense :)15:37
didrocksasac: I didn't say it did make sense :)15:37
asaci tried _every_ other button in control center15:37
asacthis one i left out15:37
didrocksno every, see, don't lie! :)15:37
* didrocks hugs asac15:37
asacoh no15:37
asaci cannot set a custom command there anymore15:37
asaci want to make desktop-webmail package work again15:37
asacguess it needs a .desktop file15:37
didrockshum, not sure about that one apart from editing the mimeapps file :/15:37
asachmm15:38
didrocksor yeah, creating a desktop, should be better15:38
asacit has a .desktop file15:38
asacah15:38
didrockshum, and the mimetype is matched?15:38
asacMimeType=x-scheme-handler/mailto15:38
asacis missing15:39
asacodd15:39
asacdo i need to run something to pick this up if i try it manually?15:39
didrocksgood question, I would say just put in the standard XDG directory and maybe rerun g-c-c, never tried TBH15:40
didrocksmterry: so, gesttings pick the new schemas in XDG_DATA_DIR ?15:40
asaclet me see15:41
dobeyfor a .desktop file?15:43
mterrydidrocks, yeah15:44
dobeyjust put the .desktop in /usr/share/applications, and i think there is a trigger that runs the desktop file cache thing15:44
mterry(at least that's my memory, it's been a while since I wrote that branch)15:44
didrocksmterry: waow, great, I was thinking it was compulsory to add the filename in /etc/dconf/profile/15:44
dobeyto change the default, just put an override in the gsettings overrides dir i guess15:45
asacdidrocks: just restarting it didnt help15:47
asacdidrocks: i changed the .desktop file in applications15:47
asaccat /usr/share/applications/desktop-webmail.desktop  | pastebinit15:48
asachttp://paste.ubuntu.com/757192/15:48
dobeyasac: you need to run update-mime-database15:48
didrocksdobey: this is not about override, this is about trunk testing15:48
asackk15:48
asacthanks dobey15:48
asacdobey: hmm. wasnte there a more convenient scrip that we call in postinst?15:49
asacthis one tells me:15:49
asacsudo update-mime-database15:49
asac[sudo] password for asac:15:49
asacUsage: update-mime-database.real [-hvV] MIME-DIR15:49
dobeyasac: i think there's a trigger that runs if your package has .desktop file in /usr/share/applications, so you don't have to do anything in postinst actually15:49
* asac tries to think hard15:49
dobeyasac: update-mime-database /usr/share/applications15:49
dobeysudo of course :)15:50
asachmm. i am sure in the past we had a convenient script hat didnt need the dir15:50
asacbut ok15:50
asaci am old anyway15:50
* ogra_ agrees15:50
ogra_:P15:50
asachehe15:50
asacsudo update-mime-database /usr/share/applications15:50
asacDirectory '/usr/share/applications/packages' does not exist!15:50
asacso see15:50
asaci tell you ... we dont use that :)15:50
dobeyoh15:50
dobeyupdate-dsektop-database15:50
dobeysorry15:51
dobeygot confused with fdo mixing mime and .desktop stuff :)15:51
asacyay15:51
asacand that is not in PATH15:51
asacthat sounds more like what i remember15:51
dobey[dobey@lunatari:~]: which update-desktop-database15:51
dobey/usr/bin/update-desktop-database15:51
asachmm15:51
asacwell its in path15:52
asacstrange15:52
asacok thanks15:52
asaclets see15:52
asaccool15:52
asacthanks dobey15:53
dobeysure :)15:53
seb128hey asac15:55
seb128asac, did you get it working?15:55
asacseb128: yes. though send link from browser doesnt work (guess a bug)15:58
asacbut thats on desktop-webmail side ... so thanks15:58
asacok off15:58
* asac sick15:58
seb128asac, take care!15:58
asacwill do. thanks!15:59
asacseb128: oh ... can you tell me what wrapper/api is used when i right click on gnome-terminal and use send mail ?16:00
asacxdg-open i guess16:00
seb128chrisccoulson, ^ do you know?16:01
seb128asac, but I would guess it's gio16:01
seb128like it probably triggers the default handler for those urls16:01
seb128same story than for the browser16:01
chrisccoulsonyeah, i'd imagine it's just using gio16:01
chrisccoulsonnothing sane uses xdg-open16:01
chrisccoulsononly chrome uses that ;)16:02
seb128chrisccoulson, not even firefox? ;-)16:02
chrisccoulsonseb128, nope :)16:02
seb128good ;)16:02
asacis there a command line wrapper or just api?16:02
asacin any case xdg-open shows the same problem16:02
asacas firefox send mail16:03
asacand gnome-terminal send mail16:03
asacanyway ... will dig a bit deeper16:03
chrisccoulsonasac, what's the issue?16:03
asacchrisccoulson: i am sick ... will ping you over weekend if i feel better :/16:04
asacthanks for your helP!16:04
chrisccoulsonasac, ok, no worries. hope you feel better soon :)16:04
asacit is about desktop-webmail16:04
asacso most likely nothing important fo ryou :)16:04
asacbye16:04
seb128lool, it seems your DB2HTML=non-existent broke the i386 gtk2 build17:49
seb128well I assume it's that, I don't see what else would have done it17:50
seb128chrisccoulson, congrats, you won bug #86549018:23
ubot2Launchpad bug 865490 in nautilus-sendto "nautilus-sendto crashed with SIGSEGV in _start()" [Medium,Confirmed] https://launchpad.net/bugs/86549018:23
seb128;-)18:23
chrisccoulsonyay \o/18:24
chrisccoulsonright, i've got to disappear. travelling round to jo's parents house this evening18:25
chrisccoulsonsigh :/18:25
* didrocks waves good night and good week-end!18:25
didrockssee you on Monday everyone!18:25
Sweetsharkchrisccoulson: I was tought it that way: action items always go to the guy who is not in the meeting ;)18:26
Sweetshark... too late18:26
* Sweetshark is gone for the weekend too.18:26

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