| === Ursinha_ is now known as Ursinha | ||
| xnox | Oohhh wrapper support in autopilot \o/ | 00:10 |
|---|---|---|
| thomi | xnox: yup :) | 00:11 |
| xnox | thomi: does it work with ubiquity as well now? | 00:11 |
| thomi | dunno, but it's tested with other bash scripts | 00:11 |
| xnox | cool. | 00:13 |
| balloons | wrapper support? | 05:12 |
| balloons | yum! | 05:12 |
| balloons | I will try autopilot with ubiquity in the morning | 05:16 |
| thomi | balloons: :) | 05:17 |
| pitti | Good morning | 05:41 |
| dholbach | good morning | 07:12 |
| jibel | good morning | 08:25 |
| dholbach | salut jibel | 08:28 |
| jibel | Hey dholbach | 08:28 |
| pitti | dholbach: guten Morgen! | 08:29 |
| dholbach | salut pitti | 08:29 |
| pitti | jibel: FYI, I filed an upstream bug about the e-d-s build failure | 08:29 |
| dholbach | comment ça va? | 08:29 |
| pitti | dholbach: nous avons eu une bonne we, merci! et toi? | 08:29 |
| dholbach | oui, moi aussi :) | 08:30 |
| pitti | oh, perdon, "un bon we" | 08:31 |
| dholbach | cette semaine je vais avoir un membre de CouchSurfing qui est Française ici - comme ça je peut pratiquer un peu | 08:32 |
| pitti | dholbach: oh, très bien ! plaisir à parler français | 08:35 |
| dholbach | :-) | 08:36 |
| jibel | pitti, I looked into this WebKit failure but couldn't find where the problem is | 08:37 |
| jibel | pitti, list tells me 1.11.90 is available | 08:38 |
| pitti | jibel: perhaps because WebKit is missing from jhbuild.modules? | 08:38 |
| pitti | that seems the main oddity | 08:38 |
| jibel | when I build automatically it says "not in module list" | 08:38 |
| jibel | when I build manually it builds 1.11.4 | 08:38 |
| jibel | it doesn't make sense | 08:38 |
| pitti | .4, really? I built it automatically and got 1.11.5 | 08:39 |
| jibel | and ~/gnome/checkout/webkitgtk-1.11.90 has been fetched automatically | 08:39 |
| jibel | pitti, "buildone" built 1.11.5 but "build" built 1.11.4 | 08:40 |
| pitti | uuh | 08:40 |
| jibel | ah $ jhbuild run pkg-config --modversion webkitgtk-3.0 | 08:41 |
| jibel | 1.11.90 | 08:41 |
| jibel | after buildone | 08:41 |
| jibel | I'll remove the checkout directories for versions <1.11.90 | 08:42 |
| smartboyhw | Hey phillw | 09:36 |
| jibel | pitti, I rebuilt WebKit with buildone and got 1.11.90 | 10:03 |
| jibel | pitti, but "jhbuild build" fails with http://paste.ubuntu.com/5564298/ | 10:04 |
| pitti | jibel: it still thinks that this is the current checkout dir and wants to check its status? | 10:05 |
| jibel | pitti, actually buildone -n fails too | 10:05 |
| jibel | so jhubild uses a different moduleset with -n | 10:05 |
| jibel | pitti, I pushed a version of our auto-jhbuild scripts that pulls sources in parallel which make update much faster | 10:25 |
| pitti | jibel: awesome, thank you! | 10:26 |
| jibel | I wished I could pushed the version that also build in parallel but given the problems in tha lab this morning it's compromised | 10:26 |
| jibel | pitti, jhbuild was using the moduleset from jhbuild source tree when nonetwork is set, which was not the latest version | 11:23 |
| pitti | jibel: ah, so that just needs a git pull in jhbuild/ ? | 11:23 |
| jibel | pitti, I set moduleset_dir to ~/.cache/jhbuild instead | 11:23 |
| jibel | and did that too | 11:23 |
| jibel | pitti, also parallel builds is back for WebKit (it was disabled until 5 days ago) and this package now builds in less than 10 minutes :) | 11:27 |
| xnox | Did I get this right that utah has reboot support now? | 11:55 |
| * xnox wants to write a few tests as a recent package update failed mounting ntfs-3g partitions on boot. | 11:56 | |
| === _salem is now known as salem_ | ||
| === salem_ is now known as _salem | ||
| === _salem is now known as salem_ | ||
| balloons | howdy xnox.. i see you have assigned yourself this bug, and it's marked as high: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1080701. Any comments on the current status? | 15:06 |
| ubot5 | Launchpad bug 1080701 in ubiquity (Ubuntu Raring) "After 'Preparing to install Ubuntu' screen, raring installation hangs" [High,Confirmed] | 15:06 |
| xnox | balloons: i am thinking to purge replace/reuse recipes and make the upload as it has been confirmed by my housemate as the cause. | 15:06 |
| xnox | and then continue troubleshooting why it doesn't work. | 15:07 |
| xnox | as that will remove the option to inline upgrade (replace old ubuntu with new ubuntu without removing all files, documents and settings) | 15:07 |
| balloons | sounds like a plan. the Qa community folks would be happy to help continue iterating on it, but many are stuck until something is put into place for it | 15:07 |
| balloons | so, fix it or break it.. as long as it's broken differently eh? ;-p | 15:08 |
| balloons | hmm, so xnox also btw, ubiquity isn't working with the new autopilot yet. it launches but still grabs the wrong pid | 15:15 |
| xnox | i haven't looked at the code, but i am fearing it will not be so simple with ubiquity: it forks, it re-execs, it escalates privileges. | 15:18 |
| balloons | right.. i'll just await thomi again and see what happens | 15:18 |
| balloons | i will check out some of the other script starting apps and see what happens | 15:18 |
| balloons | xnox, I want to highlight your blog entry on the nexus7 preseed work.. let me know when it lands | 16:04 |
| xnox | Ack. | 16:06 |
| === tgm4883_ is now known as tgm4883 | ||
| balloons | xnox, ohh before I forget, can you actually leave the same comment you left me on https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1080701? | 17:01 |
| ubot5 | Launchpad bug 1080701 in ubiquity (Ubuntu Raring) "After 'Preparing to install Ubuntu' screen, raring installation hangs" [High,Confirmed] | 17:01 |
| SergioMeneses | hi all! | 17:07 |
| balloons | hi SergioMeneses | 17:12 |
| SergioMeneses | balloons, \o how are you? | 17:13 |
| balloons | not too bad.. how about yourself | 17:14 |
| balloons | ? | 17:14 |
| SergioMeneses | balloons, fine! I found this bug 1132569 yesterday | 17:15 |
| ubot5 | Error: Launchpad bug 1132569 could not be found | 17:15 |
| balloons | SergioMeneses, private bug? | 17:15 |
| SergioMeneses | jejeje my bad it is a private bug | 17:15 |
| SergioMeneses | balloons, yes, it is | 17:15 |
| balloons | ahh. can't see it then | 17:27 |
| balloons | I'm trying to follow-up on all my pending bugs | 17:30 |
| SergioMeneses | balloons, that sounds fun | 17:30 |
| balloons | SergioMeneses, lol.. yes of course | 17:46 |
| SergioMeneses | balloons, jejeje | 17:54 |
| balloons | so, we actually have a manual testcase we need to get updated | 17:54 |
| SergioMeneses | bug 1132144 | 17:54 |
| ubot5 | bug 1132144 in Ubuntu Manual Tests "Totem Tests (1429), there is no "Youtube" dropdown available" [High,Confirmed] https://launchpad.net/bugs/1132144 | 17:54 |
| balloons | looks like the totem test is outdated | 17:54 |
| balloons | yep | 17:55 |
| SergioMeneses | balloons, I was reading about it | 17:55 |
| SergioMeneses | balloons, I can work on it tonight | 17:55 |
| balloons | SergioMeneses, excellent, ty | 18:01 |
| SergioMeneses | it doesnt look hard to make :) | 18:02 |
| SergioMeneses | ok I have to go! see you later balloons | 18:09 |
| SergioMeneses | take care | 18:09 |
| balloons | u2 | 18:09 |
| ssbob | filing bugs is fun :) | 18:10 |
| balloons | hey ssbob ! | 18:10 |
| balloons | how's things? | 18:10 |
| ssbob | howdy balloons | 18:10 |
| ssbob | good, you? | 18:10 |
| balloons | pretty good.. had the munchies so i got a little snack ;-0 | 18:10 |
| ssbob | ah yeah it is lunchtime here in Texas ... pizza in the oven :) | 18:11 |
| balloons | doing an upgrdae install on real hw | 18:11 |
| balloons | mm.. pizza | 18:11 |
| SergioMeneses | jajaja ssbob you're right about bugs :) | 18:12 |
| ssbob | balloons, if you get some time in the near future, I would like to pick your brain about working life at canonical. I am hunting for a new job. | 18:14 |
| balloons | ssbob, sure.. | 18:15 |
| ssbob | thanks, just poke me when you have the time. | 18:16 |
| balloons | ssbob, k, we shall see. | 18:16 |
| balloons | i'm going to be on vaca , so i'll try and squeeze you in today in a few mins ;-0 | 18:17 |
| ssbob | no rush, I know you have to be crazy busy this week getting ready for vaca | 18:17 |
| balloons | Noskcaj, I just used testdrive on raring with vbox 4.2 | 19:09 |
| Noskcaj | balloons, yay, i'm just using Unit's original fix | 19:09 |
| balloons | ahh | 19:09 |
| thomi | Hi everyone | 19:28 |
| thomi | elopio: alesage tells me you had an autopilot question for me? | 19:28 |
| elopio | thomi: yes, I do. | 19:29 |
| elopio | thomi: can you give a quick look at my branch? https://code.launchpad.net/~elopio/ubuntu-sso-client/autopilot/+merge/150260 | 19:29 |
| thomi | sure thing | 19:29 |
| elopio | thomi: my first problem is that I don't know how to do an emulator for an application that doesn't have a well-known name in the bus. | 19:30 |
| balloons | hey thomi | 19:30 |
| balloons | I'll let you and elopio talk first | 19:30 |
| elopio | hey balloons. I am now close to undertanding autopilot! :) | 19:31 |
| njin | balloons, hallo are you on raring ' | 19:32 |
| njin | ? | 19:32 |
| thomi | elopio: the TESTABILITY environment variable - is that built into Qt, or is it specific to the SSO client? | 19:35 |
| elopio | thomi: that's specific for SSO | 19:36 |
| thomi | elopio: ok | 19:37 |
| elopio | the qt window is not started through command line on a normal way, so they added that to inject the -testability to the argv. | 19:37 |
| thomi | elopio: so it looks good to me, I'd recommend splitting the emulators and test code into separate files, and I'm not suer I'd make the emulators switch to the current page when you construct them... but apart from that, it looks good to me | 19:37 |
| elopio | thomi: but I was trying to build the emulator as the unity one, from DBusIntrospectionObject. | 19:39 |
| balloons | elopio, excellent :_) | 19:39 |
| balloons | njin, hey hey! | 19:39 |
| thomi | elopio: yeah, that's do-able, but probable more trouble than it's worth | 19:39 |
| balloons | yes, I am on raring, and in fact just upgraded the other machine to raring as a test | 19:39 |
| njin | balloons, great, can you play an mp3 file on Videos ? mine stop after one second | 19:40 |
| elopio | thomi: the other problem is that I shouldn't start the ubuntu-sso-login-qt directly, but from a call to com.ubuntu.sso bus. They I won't have the process id, and it's a pain to find the autopilot proxy object. Doable, but really weird. I thought that with an emulator, it wouldn't matter how I started the window. | 19:41 |
| balloons | njin, just tried | 19:41 |
| balloons | njin, yep, plays fine | 19:41 |
| elopio | s/They/Then | 19:42 |
| njin | balloons, thanks, so I investigate in mine... | 19:42 |
| thomi | elopio: well, creating the autopilot proxy object yourself isn't too hard - from memory the code you want to look at is in the autopilot.introspection module. You could perhaps write the code to launch the client via DBus, and then create the proxy object yourself? | 19:43 |
| thomi | the convenience classes for Gtk and Qt and there because theyt're the most common application types | 19:43 |
| thomi | but it seems like you need something a bit more advanced | 19:43 |
| thomi | luckily, this is pretty easily extensible | 19:44 |
| elopio | thomi: I did that, but maybe I missed a nice API, because I had to get the pid of the process called by dbus, then get the pid of the child launched by that process, then get the list of names on the bus, filter the ones connected by my process pid, then create the autopilot proxy object and then set the pid to that proxy. | 19:46 |
| Letozaf_ | balloons, Hi | 19:46 |
| elopio | thomi: was that stupid, or is it the only way? | 19:46 |
| thomi | elopio: there's an easier way - you can just pass the pid to..... (looks up API...) | 19:47 |
| balloons | Letozaf_, hello! | 19:47 |
| thomi | Letozaf_: get_autopilot_proxy_object_for_process | 19:48 |
| thomi | oops | 19:48 |
| thomi | elopio: ^ | 19:48 |
| thomi | elopio: it expects a subprocess.Popen-like object, essentially a class witha 'pid' attribute | 19:48 |
| thomi | so you can duck-type your way into that reasonably easily :) | 19:48 |
| Letozaf_ | balloons, for bug 1122162 I will now try to reproduce it but it's a weired one, does not always happen | 19:48 |
| ubot5 | bug 1122162 in ubiquity (Ubuntu) "cannot enter personal information on ubiquity's "Who are you" screen" [Undecided,New] https://launchpad.net/bugs/1122162 | 19:48 |
| xnox | Letozaf_: it's usually reported against nexus7 with/without onboard and it has been reported against panda-board as well. | 19:49 |
| Letozaf_ | balloons, anyway it happened to me when I selected Italian language, I will try also with English | 19:49 |
| xnox | but i didn't test desktop panda image for a while now. | 19:49 |
| Letozaf_ | xnox, oh! didn't know this | 19:49 |
| balloons | xnox, ohh, you know of the bug eh?> | 19:50 |
| Letozaf_ | xnox, well I am downloading arm image now and I will see if I can find out what triggers it | 19:50 |
| xnox | it's not even specific to Who are you screen, but rather any input field, e.g. the wifi network password as well. | 19:50 |
| Letozaf_ | xnox, yes that happens also to me | 19:51 |
| xnox | there is a better bug report about it as well. | 19:51 |
| Letozaf_ | xnox, I mean the wifi network password thing | 19:51 |
| balloons | xnox, ahh ok.. let's dupe that bug then? | 19:51 |
| * xnox can't remember what it was. Possibly it was linked to ubuntu-nexus7 project as well. | 19:51 | |
| balloons | howdy chilicuil | 19:52 |
| elopio | thomi: well, yeah, that function does all I mentioned. But now looking it more carefully, that's what's done by launch_application anyway. | 19:52 |
| chilicuil | hey balloons o/ | 19:52 |
| thomi | elopio: yup | 19:53 |
| elopio | thomi: so, yeah, what's different on my case is the way to get the pid. Sounds ok. | 19:53 |
| thomi | elopio: yeah.. | 19:53 |
| elopio | thomi: and just to compare, what would be needed to make a DBusIntrospectionObject for my not-well-known-named app? | 19:54 |
| thomi | elopio: you mean for an emulator? | 19:55 |
| elopio | thomi: yes. To make it work like the one in Unity. | 19:55 |
| thomi | ok, you'd need to: | 19:56 |
| thomi | 1) make your proxy class, and make it derive from DBusINtrospexctionObject. | 19:56 |
| thomi | 2) Before instantiating it, set the class attrs DBUS_SERVICE & DBUS_OBJECT to the service name & object path respectively | 19:56 |
| thomi | I'm pretty sure that's all you need.... | 19:56 |
| elopio | thomi: yes, but I don't have a dbus_service name. I'm going to ask the developers how hard would it be to add that, and maybe they do all the hard work :) | 20:00 |
| elopio | thanks thomi. I'm enjoying this tool :D | 20:00 |
| thomi | elopio: yeah, you need that. The launch code sets that up for you using the connection name it finds on the session bus | 20:00 |
| thomi | elopio: thanks - if you think anything is missing, feel free to make a MP against lp:autopilot :) | 20:01 |
| elopio | thomi: yes, I will propose the get_child_by_type. I'm used to selenium, and there are functions for many and for one everywhere. | 20:03 |
| elopio | so I ended calling get_children_by_type(...)[0] like five times. | 20:04 |
| Letozaf_ | balloons, for instance now I was trying out today's ARM image, and on the keyboard window, I selected LAO, then clicked on detect keyboard, answered the questions and when my Italian keyboard got detected the forward button is grey and cannot select it to carry on with install | 20:05 |
| Letozaf_ | balloons, this already happend to me but did not report the bug as it did not happen again on reboot, now it happend | 20:06 |
| thomi | elopio: you propose it, I'll approve it :) | 20:13 |
| balloons | Letozaf_, sorry was meetin' | 20:18 |
| balloons | Letozaf_, it sounds like xnox knows about this.. xnox do you have the"better" bug you mentioned? | 20:18 |
| balloons | Letozaf_, so if you reboot and do it again, it happens again | 20:19 |
| balloons | ? | 20:19 |
| Letozaf_ | balloons, I will try now, I was waiting to see what you wanted me to do :D | 20:19 |
| xnox | it's random and not clearly reproducible, sometimes it's there and sometimes it isn't, rebooting "helps" | 20:19 |
| xnox | so it is racy. | 20:19 |
| xnox | I didn't look for that bug. let me ask ogra. | 20:20 |
| Letozaf_ | xnox, yes I can confirm its non quite easily reproducible | 20:20 |
| Letozaf_ | xnox, it had already happend but could not reproduce it | 20:20 |
| Letozaf_ | balloons, xnox buy the way now I have no cursor when entering WiFi network password :D | 20:23 |
| balloons | hmm.. well xnox would you want to invest time in trying to reproduce or no? | 20:23 |
| xnox | balloons: not really, as it's beyond my knoweledge this is a bug somewhere between X, gtk, kernel(?!) as there is nothing funny about ubiquity's code here. We kind of rely and expect for toolkit to work. | 20:26 |
| Letozaf_ | balloons, xnox other bug: when I get to the timezone window, I click somewere on the map, then try to enter Rome in the input field but cannot write anything | 20:27 |
| xnox | Letozaf_: the bug is broken keyboard input - so yeah you'd see it on any input field, if the bug is present in a given boot session. | 20:28 |
| Letozaf_ | xnox, ok | 20:28 |
| Letozaf_ | xnox, balloons now the keyboard bug did not occur, I selected LAO, detected my keyboard and carried on normally :( | 20:29 |
| Letozaf_ | xnox, balloons I did the exact same thing, I took not of them | 20:29 |
| xnox | anyway ogra marked bug 1122162 as a dupe now | 20:30 |
| ubot5 | bug 1093050 in ubuntu-nexus7 "duplicate for #1122162 compiz dies during oem-config and steals the focus so that input fields are unusable" [High,Confirmed] https://launchpad.net/bugs/1093050 | 20:30 |
| Letozaf_ | xnox, so do you want me to do anything about these bugs or should I let go ? | 20:32 |
| xnox | it is known and hard =) | 20:32 |
| xnox | Letozaf_: but don't let me stop you from trying to figure out why compiz crashes. E.g. can you see / access / upload files in /var/crash/* ? | 20:33 |
| Letozaf_ | xnox, in this install my /var/crash folder is empty | 20:34 |
| Letozaf_ | xnox, it's only got a .lock file | 20:35 |
| Letozaf_ | balloons, regarding umockdev Martin Pitt answered my mail | 20:40 |
| balloons | Letozaf_, awesome | 20:42 |
| balloons | did you get it working? | 20:43 |
| balloons | thomi,freed up now? so autopilot still introspects the wrong pid for ubiquity | 20:43 |
| balloons | tho it's different than before maybe? | 20:43 |
| Letozaf_ | balloons, I will try now Martin fixed a libudev bug in raring | 20:46 |
| Letozaf_ | balloons, If you want I can send you his email | 20:46 |
| balloons | Letozaf_, ohh.. you found a bug eg/ | 20:49 |
| balloons | nice | 20:49 |
| balloons | ;-0 | 20:49 |
| balloons | your just too good at that ;-0 | 20:49 |
| Letozaf_ | balloons, :") | 20:55 |
| thomi | balloons: really? | 21:00 |
| thomi | balloons: what makes you think that? | 21:01 |
| balloons | thomi, yea.. I can say it doesn't work.. unless I didn't have the fix in the autopilot I ran | 21:01 |
| balloons | I used the ppa, and I did it this morning | 21:01 |
| balloons | I'm happy to repeat the experiment | 21:01 |
| Letozaf_ | balloons, it worked :D | 21:19 |
| balloons | Letozaf_, sweet! | 21:21 |
| Letozaf_ | balloons, I've got shotwell working with a "virtual" camera :D | 21:22 |
| balloons | Letozaf_, that is awesome | 21:23 |
| balloons | good work | 21:23 |
| balloons | I'm excitied | 21:23 |
| Letozaf_ | balloons, well almost ... I can see a fake camera with photos in it but I get an error when I import them :( | 21:24 |
| Letozaf_ | balloons, I will try to work this out, but at least I got a fake camera with photos in it :) | 21:25 |
| balloons | indeed.. that's really cool | 21:25 |
| thomi | balloons: how did it not work? Did you get an error message of some kind? | 21:28 |
| balloons | thomi, yea.. it gave me the error about not loading autopilot | 21:28 |
| thomi | hmmm | 21:28 |
| thomi | probably the sudo thing, rather than autopilot itself | 21:28 |
| balloons | ohh right | 21:29 |
| balloons | hmm | 21:29 |
| balloons | was I doing it properly... | 21:29 |
| * thomi writes a quick test | 21:29 | |
| * balloons reloads the test environment | 21:29 | |
| thomi | balloons: to test this, I wrote a bash script that just does 'sudo gedit $*' | 21:31 |
| * thomi will try and launch that | 21:31 | |
| thomi | ...as soon as my dist-upgrade finishes | 21:31 |
| balloons | thomi, ahh.. kk | 21:32 |
| SergioMeneses | hi guys | 21:33 |
| thomi | balloons: yeah, it seems that when you launch something with sudo, autopilot can't find it on the session bus | 21:38 |
| thomi | probably it connects to a different session bus? But then, launching autopilot as root (with sudo) doesn't seem to fix it eiter | 21:38 |
| thomi | *either | 21:38 |
| balloons | thomi, yes, we went through that the other day in determining things | 21:39 |
| thomi | so, I think we need to find someone who understands the interaction between sudo and dbus a bit better | 21:39 |
| balloons | so, what do you do to fix the bash issue? | 21:39 |
| balloons | for instance, I still can't seem to introspect say skype or vlc | 21:39 |
| balloons | hmm.. or firefox.. but launch works properly now in those cases | 21:40 |
| balloons | it | 21:40 |
| balloons | it just fails to load the autopilot module :-( | 21:40 |
| thomi | balloons: autopilot now searches through child pids. So try making a bash script that launched gedit, then launch it like this: | 21:40 |
| thomi | autopilot launch -i Gtk ./launch_gedit.sh | 21:40 |
| thomi | and it "just works" :) | 21:40 |
| balloons | thomi, will it try and introspect every child pid? | 21:40 |
| thomi | balloons: yes | 21:40 |
| balloons | aka, until it latches onto one? or every pid? | 21:41 |
| thomi | balloons: it stops at the first one it finds with the autopilot plugin is loaded | 21:41 |
| balloons | it doesn't print that it is doing that | 21:41 |
| thomi | just the first one | 21:41 |
| balloons | thomi, ahh.. that sounds like it should work then | 21:41 |
| balloons | at least much better than what i'm seeing | 21:41 |
| balloons | is the build for that up? | 21:41 |
| balloons | did it get into the ppa? | 21:41 |
| balloons | maybe it didn't... | 21:41 |
| thomi | balloons: it's in raring | 21:42 |
| thomi | balloons: 1.2daily13.02.25-0ubuntu1 | 21:42 |
| * balloons looks | 21:42 | |
| balloons | 1.2daily13.02.14-0ubuntu1+bzr139pkg0raring1 | 21:42 |
| balloons | sure enough | 21:43 |
| balloons | no good | 21:43 |
| thomi | balloons: time to do an update then :) | 21:43 |
| balloons | might have a versioning error | 21:43 |
| balloons | k, the one in the vm is correct ;-0 | 21:44 |
| balloons | weird.. i updated so many times.. lol.. i see it now.. | 21:44 |
| * balloons rattles brain | 21:44 | |
| balloons | so who would be a good dbus/sudo expert candidate? | 21:46 |
| balloons | thomi, so does autopilot log every pid it attempts to connect to? | 21:47 |
| balloons | that would be helpful in understanding what's happening when we're messing with ubiquty | 21:47 |
| balloons | ohh I see this now: Searching for eligible PIDs: | 21:47 |
| balloons | i wasn't getting that in the vm loading ubiquity | 21:48 |
| thomi | balloons: it doesn't. It uses 'ps' to get a list of pids to look for. THen it trawls the session bus for the current user looking for interfaces that are owned by a pid in that list. Then it looks for the autopilot introspecxtion interface | 21:48 |
| balloons | hmm.. does it wait before doing so? | 21:49 |
| balloons | ubiquity does lots of fun stuff.. if you ps too soon, you might not be seeing it | 21:49 |
| balloons | cool.. so looking at this again, autopilot does list the proper pid in one of its "Searching for eligible PIDs" lines | 21:55 |
| thomi | balloons: it re-executes ps every second to get the new PIDs | 21:55 |
| balloons | let me give you the log thomi | 21:55 |
| thomi | so as long as the correct process is launched within 10 seconds, you should be fine | 21:55 |
| balloons | thomi, ahh... gotcha | 21:58 |
| balloons | gonna play around with launching as root, etc | 21:58 |
| thomi | balloons: you need to find a dbus expert :) | 21:59 |
| balloons | thomi, yes.. indeed, any suggestions? ping the tech list? | 22:00 |
| thomi | balloons: hmmmm perhaps someone like ted? | 22:00 |
| thomi | he seems to know everything :) | 22:00 |
| balloons | thomi, lol.. kk | 22:01 |
| tedg | Howdy | 22:03 |
| balloons | thomi, tedg not only knows everything, he's willing to share that "everything" with us | 22:04 |
| tedg | thomi, balloons told me you broke the Internet | 22:04 |
| balloons | so I'll let thomi give you the further details ted but here's the gist. We're using autopilot to launch ubiquity. Ubiquity loads and forks then re-exec's it's final main process. This is the process we want to introspect. To help with that thomi modified autopilot to watch ps and grab the children pids of the initial process and try and introspect them, stropping once it's got one it can introspect | 22:09 |
| balloons | what's happening now is that it's ultimately failing to load the autopilot module, and hence, fails to introspect | 22:09 |
| thomi | tedg: so my question would be: if a process connects to the session bus, but is launched vis 'sudo', it seems to connect to a different session bus? | 22:12 |
| thomi | tedg: perhaps a session bus for root or something? | 22:12 |
| xnox | if there is no session bus, there is nothing to talk over to dbus-activate/advertise anything. | 22:13 |
| xnox | i thought we were agreeing to preserve the environment and hence pass the same `currentu-users'` DBUS_SESSION_ADDRESS such that final ubiquity process appears on the same one. | 22:14 |
| thomi | xnox: autopilot doesn't modify the environment at all... but if I launch gedit with the ap plugin enabled with sudo then it doesn't appear on the session bus, so *something* is breaking it | 22:15 |
| xnox | thomi: how about lauchning it with: sudo -E autopilot-wrapper gedit ? | 22:16 |
| * thomi tries that | 22:16 | |
| xnox | cause ubiquity is launched with environment preservation passed to any/all priviledge escalation helpers. | 22:16 |
| thomi | xnox: ahhhh, that works | 22:17 |
| xnox | awesome. So in theory it all should work with ubiquity as well ;-) | 22:17 |
| thomi | balloons: so.... autopilot can in fact connect to a dbus interface that's running as root | 22:17 |
| thomi | yeah | 22:17 |
| balloons | lol | 22:17 |
| tedg | Glad I could help. | 22:18 |
| xnox | thomi: so how do i best try this myself? e.g. should I use autopilot from some daily ppa or just the archive? | 22:18 |
| tedg | :-) | 22:18 |
| === salem_ is now known as _salem | ||
| thomi | xnox: the version in raring is fine | 22:19 |
| tedg | I'm a bit surprised that ubiquity is connecting to a session bus at all though. | 22:19 |
| tedg | Or, only with the module. | 22:19 |
| thomi | tedg: well, that's the autopilot introspection plugin | 22:20 |
| xnox | tedg: to inhibit disk-autolaunching, monitoring networking connectivity from network-manager, populate custom indicators (if running in ubiquity-only mode), maybe other things. | 22:21 |
| xnox | we also twiddle with gsettings during installation (to turn off screensaver and things like that, while installation is in progress) and that needs session-dbus behind our back as well. | 22:21 |
| tedg | Huh, interesting. I thought ubiquity just killed everything when you started to install :-) | 22:22 |
| tedg | I should probably install more than once every 3 years ;-) | 22:22 |
| xnox | tedg: thanks to dbus-activation it would just respawn everything back these days =))))) | 22:22 |
| tedg | xnox, Oh, wait for upstart based user session! | 22:23 |
| balloons | xnox, python-autopilot is in universe, so enable it fy | 22:26 |
| xnox | tedg: it works =)))) from the ppa, but many things are still interesting. E.g. at the moment we have a race where both gnome-session & upstart are fighting over each other to respawn compiz. | 22:27 |
| * xnox is running his desktop in upstart based user session ;-) | 22:28 | |
| balloons | ok, so thomi, xnox.. how do we launch ubiquity / autopilot in order to have this work now? | 22:28 |
| balloons | sudo -E? | 22:28 |
| thomi | balloons: should be as simple as: autopilot launch -i Gtk ubiquity | 22:28 |
| balloons | http://paste.ubuntu.com/5566147/ | 22:32 |
| thomi | balloons: it looks to me like it's only searching one pid? | 22:33 |
| balloons | that one was weird.. let me do another | 22:34 |
| balloons | http://paste.ubuntu.com/5566147/ | 22:34 |
| balloons | that's what i've seen | 22:34 |
| balloons | so it sees pid 5967 in there, and claims to search it | 22:35 |
| balloons | but I also see the failed to load module autopilot before it gets that far | 22:36 |
| thomi | balloons: that's not right, what does this give you: | 22:36 |
| thomi | ps -o pid --noheaders --ppid <ubiquity pid> | 22:36 |
| balloons | syntax error :-) let's see | 22:37 |
| balloons | ahh | 22:38 |
| balloons | 5952 | 22:38 |
| balloons | which is the gksudo --preserve-env call | 22:38 |
| balloons | but xnox says we want the python3 call, pid 5967 in this case | 22:39 |
| thomi | balloons: ok, so autopilot does this recursivelky | 22:39 |
| thomi | balloons: the 'Searching for eligible PIDs' line should contain a list that shows the entire tree of PIDs | 22:40 |
| balloons | why fail on 5952 tho? it claims to search the next 4, but it seems it already failed beofre that | 22:40 |
| thomi | balloons: which log are you looking at now? | 22:42 |
| balloons | ohh | 22:42 |
| balloons | sorry.. copy/paste error :-( | 22:42 |
| thomi | nw :) | 22:42 |
| balloons | i think.. | 22:43 |
| thomi | looking at the code, it logs the errors, but keeps on searching | 22:43 |
| balloons | http://paste.ubuntu.com/5566156/ | 22:43 |
| balloons | that's what I'm looking at | 22:43 |
| thomi | balloons: ahhh, that looks better | 22:43 |
| balloons | so in that example pid 5967 is what we want | 22:44 |
| balloons | so according to the log, it tried and got no reply? | 22:44 |
| balloons | so despite the earlier failure, it's still looking | 22:44 |
| balloons | hmm weird | 22:46 |
| thomi | balloons: I have a lunch appt in town, but when I get back I'll fix the logging in autopilot so you can see which pid produces which error | 22:49 |
| balloons | thomi, see this line: 22:33:34.353 DEBUG proxies:413 - Executing introspect queue due to error | 22:49 |
| thomi | balloons: yes, that comes from dbus itself | 22:49 |
| thomi | not autopilot | 22:49 |
| balloons | thomi, sure.. fixing up the logging there as to what's going on would be good | 22:49 |
| thomi | I was looking at that as well :) | 22:49 |
| balloons | thomi, so that means it tried to connect then to our proper pid? | 22:49 |
| balloons | thomi, well, I am out the rest of the week now.. so heh, I hate to leave things like so, but that may be how it lies | 22:50 |
| thomi | balloons: well, you don't know, since it doesn't log out which pid it's trying to connect to | 22:51 |
| balloons | thomi, ahh.. yes, but that's my guess.. it comes in line with the pid it should connect.. I think it's really close now | 22:53 |
| phillw | have fun balloons, don't forget to de-op yourself :P | 22:53 |
| balloons | anyways.. enjoy lunch | 22:53 |
| * balloons asks, am i op'd/ | 22:53 | |
| phillw | show as here :) | 22:53 |
| phillw | lol | 22:54 |
| balloons | tks phillw | 22:54 |
| phillw | I'll drop you an email about some thoughts on the global jam. Reply at your leisure. | 22:54 |
| balloons | phillw, yes please | 22:55 |
| balloons | feel free to email.. i won't have irc presence.. (not active anyway0 | 22:55 |
| balloons | ciao everyone | 22:56 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!