[01:50] eek [01:50] i've been told the hardcoded value for switching workspaces has been reverted [01:50] but its in the unity ppa.. not in the beta :( [05:40] anyone feel like doing a review? https://code.launchpad.net/~brandontschaefer/nux/nux.fix-input_method_text_change/+merge/95508 [07:18] is it possible to disable the hug menu when i press the alt-key? or change the key-binding? i need the alt-key for emacs. [08:08] didrocks: ping [08:08] didrocks: I almost done the implementations, I think I make a merge proposal in a few hours, the last thing to solve is how call nautilus gracefully and in a way that doesn't raise ubuntu developers anger! [08:09] angeloc: I think kamstrup will surely review you work, so let's check with him :) [08:09] didrocks: thank you! [08:10] yw :) thanks to *you* [08:10] kamstrup: can you help me? === smb` is now known as smb [08:24] angeloc: sure thing [08:24] what do you need help with? [08:24] I'm trying to solve bug 773841, it's bitesize, it's my second contribution. [08:24] Launchpad bug 773841 in unity-place-files (Ubuntu) "\\192.168.1.x opens http:\\192.168.1.x in firefox as opposed to smb://192.168.1.x in nautilus" [Low,Confirmed] https://launchpad.net/bugs/773841 [08:25] I almost done, unity file lens now understands uri like ssh://, sftp://, ftp:// [08:26] kamstrup: but I cannot use launch_default_for_uri method because it cannot hadle not mounted volumes [08:27] indeed [08:27] I think we had a similar problem before [08:27] I found that in trashlaunchericon.cpp in unity, it spawned a process with xdg-open [08:28] oh... I've been trying to eradicate xdg-open for a while now [08:28] somehow people keep using it :-) [08:28] kamstrup: yes, sure [08:29] kamstrup: I have to call directly nautilus, because xdg-open cannot handles not mounted volums itself (it calls gvfs-open that uses the same function of launch_default_for_uri) [08:29] angeloc: I think the approach we've used elsewhere is to fetch the default app for "inode/directory" mimetype [08:29] which will be nautilus, but this allows users to replace the default file manager [08:30] kamstrup: so, I have to get the application for the mimetype "inode/directory", than I can spawn a process with such an application? [08:31] var file_manager = AppInfo.get_default_for_type("inode/directory", true) [08:31] file_manager.launch_uris(...) [08:32] launch_uris is the method called by launch_default_for_uri, it cannot handle unmounted volumes, I chatted about that with gnome guys in gnome devel irc channel [08:33] kamstrup: right now the problem is exactly this! gvfs-open calls launch_uris opening mountable location but it throws this error [08:34] kamstrup: sftp://192.168.1.10/: error opening location: The specified location is not mounted [08:34] angeloc: but I thought the problem that "default for uri" gave you firefox? [08:35] when you get the appinfo for inode/directory that should not run xdg-open or anything like that, but the nautilus executable directly [08:35] kamstrup: this is ok now, as I stated before, I almost done, unity file lens now understands uri like ssh://, sftp://, ftp:// and behaves correctly [08:38] angeloc: at least the inode/directory approach works here. I just tried it in Python: [08:38] >>> from gi.repository import Gio [08:38] >>> a = Gio.AppInfo.get_default_for_type ("inode/directory", True) [08:38] >>> a.get_id() [08:38] 'nautilus.desktop' [08:38] >>> uris = ["sftp://grillbar@grillbar.org"] [08:38] >>> a.launch_uris(uris, None) [08:38] this works well [08:39] assuming you have credentials to grillbar@grillbar.org :-) [08:39] kamstrup: wow! There sould be something wrong with launch_default_for_uri in glib because it doesn't works and should produce the same result [08:40] angeloc: well, yes and no [08:40] angeloc: it just means that the default uri handler is not what we expect [08:40] not necessarily that there is a bug in the function [08:41] it's been a long standing issue that Ubuntu opens random network drives in Firefox [08:41] kamstrup: ok, I understand, so I can narrowing down because sftp:// is handled incorrectly instead modify unity file lens code [08:41] as with all bugs out of my scopeI blame didrocks ;-) [08:42] kamstrup: AhHAhhAHhA [08:42] that's soooooooooo easy :) [08:43] kamstrup: ok, so I proceed with AppInfo.get_default_for_type ("inode/directory", True), you agree? [08:43] angeloc: +1 [08:43] angeloc: please add a comment in the code explaining why you're doing it like this [08:44] otherwise some unsuspecting hacker will revert it to the more easy (and buggy) code path [08:44] kamstrup: yes! I think I can push a merge proposal in an hour, are you here to review it? [08:44] angeloc: yep [08:44] kamstrup: I'm exited! [08:45] kamstrup: excited, sorry! [08:46] angeloc: yeah please don't exit() before the patch is ready ! ;-) [08:47] kamstrup: yep ;) ! [09:14] kamstrup: It works, It works! [09:14] woohoo! nice work angeloc! :-) [09:31] weird, reproduced 943851 again with compose-enabled layout and boy that bug report has got some activity :) [09:35] didrocks, is it possible to disable the hug? it's in the way when i use emacs (alt key is the problem). [09:38] toabctl, yes, you can disable hud via ccsm [09:39] rye, is it an own package which i can remove with apt-get? [09:39] toabctl, no, that's compiz plugin setting [09:41] rye, what's the plugin name in ccsm? [09:41] toabctl, unity [09:42] rye, ah :) [09:42] rye, thanks! [09:42] toabctl, you are very welcome :) [09:48] angeloc: ping me when you have the branch ready for review [09:48] kamstrup: ok, I'm solving last glitches, I'll push it as soon as possible [09:49] great [10:33] hi! i tried to post a unity mockup to the ayatana mailing list, but it looks like it no longer exists! could you please tell me where i should post my mockup? [10:38] nilux, i guess unity-design? [10:38] https://lists.launchpad.net/unity-design/ [10:39] rye: thanks! [10:49] didrocks, hey, I'm investigating the hide-mode issue in the iso, looks like the user has the value overridden [10:49] didrocks, if I `gsettings reset` the key it's fine [10:50] Saviq: well, I know that [10:50] didrocks, do you know where that might come from? [10:50] Saviq: I'm wondering if the trigger didn't work on schema changed [10:50] Saviq: because it's a beta1 install [10:50] kamstrup: I'm in stuck with this regexp [10:50] regexes.prepend (new Regex ("[a-zA-Z0-9\\-]+\\...+")); [10:50] didrocks, that's not it, if the user has manually selected a different value, we won't be resetting it to 0 [10:51] didrocks, the default user simply shouldn't have a value selected [10:51] kamstrup: that overwrites the old one Regex (".+\\...+") [10:51] Saviq: well, I know how desktop components work, but thanks again :) [10:51] the 2 is stored in the user's dconf [10:51] didrocks, sorry [10:51] Saviq: what I am telling you is: [10:51] 1. I installed beta1 yesterday [10:52] 2. I then, *without changing anything* installed the new schema from the new unity-2d package [10:52] and the key wasn't changed [10:52] it won't [10:52] my bet is that the gsettings compile schema isn't triggered [10:52] which is valid in fact [10:52] as nothing changed [10:52] it's a dconf issue [10:52] kamstrup: the one I wrote should match only domain without third level, domains in wich the first dot is prepended with alphanumeric chars, like google.com [10:52] didrocks, are you saying that if a user has manually set the hide-mode to 2 [10:53] or whatever, to 1, 0, whatever [10:53] then installing the new package should reset it to 0? [10:53] Saviq: I'm not [10:53] Saviq: see before, on the 1. 2., did I say that i changed the value? [10:53] this regex ensures me that \\192.168.1.12 or ssh://192.168.1.2 will be matched, they are matched elewhere [10:53] didrocks, that's the thing [10:53] didrocks, the user's database on the ISO has the value changed [10:53] kamstrup: but this regexp doesn't match [10:54] Saviq: ok, as it's not clear, again *I didn't change anything* [10:54] didrocks, I'm not telling you did [10:54] I didn't change the hide-mode value [10:54] didrocks, I'm telling it's preinstalled changed [10:54] right [10:54] that's why the default isn't taken into account [10:54] the default value for the schema [10:54] no not the default [10:55] ? [10:55] the default is fine [10:55] kamstrup: sorry! rephrasing -> this regex ensures me that \\192.168.1.12 or ssh://192.168.1.2 will NOT be matched, they are matched elewhere [10:55] if I go "gsettings reset", the value is taken from the default in the schema [10:55] which means it wasn't reset initially [10:55] which means the user's dconf db has the 2 saved [10:55] Saviq: which is an issue, as 2 was the default [10:56] when you change a default [10:56] by a new default [10:56] you should get the new default [10:56] yes [10:56] of course [10:56] that's how gconf is working for years [10:56] and how dconf is supposed to work [10:56] and that is how dconf is working [10:56] but for some reason the value is saved in the user's dconf db [10:57] Saviq: that's why I need to talk to desrt [10:57] Saviq: do you confirm that's the case for you as well? [10:58] didrocks, yes [10:58] if you change a default in the schema [10:58] get the compiler running [10:58] you still have the previous value? [10:58] angeloc: ah, now I understand what you were trying :-) [10:59] didrocks, you need to trigger the gsettings daemon to refresh schemata, right? [10:59] didrocks, and it should happen [10:59] Saviq: the packaging does that [11:00] yues [11:00] -u [11:00] that's my point [11:00] Saviq: but you mean, it doesn't work? [11:00] I don't get you :) [11:00] everything works fine [11:00] but the user on the CD has the value overriden [11:00] so no defaults are taken into account [11:01] the user's dconf needs to get cleaned up on the CD [11:01] angeloc: how about bcc.co.uk? [11:01] hum [11:02] Saviq: do you know what's overidde it? [11:02] didrocks, the user's dconf database is shipped on the cd [11:02] AFAIU there shouldn't be a .config/dconf/user at all [11:02] Saviq: urgh, ok, I start to understand now [11:03] * didrocks need to start a live [11:03] that's clearly not good [11:03] nope [11:03] Saviq: that's why I didn't understand you first [11:03] :) [11:03] ok good [11:03] thanks for spotting it, so yeah, the issue makes totally sense now [11:03] angeloc: I think the check needs to be if it starts with http?: or otherwise is an alphanumeric string without : or \ [11:03] Saviq: btw, what do you use to read it? [11:04] didrocks, just go 'gsettings get com.canonical.Unity2d.launcher hide-mode' as soon as you launch [11:04] kamstrup: there is something wrong woth my build env, I recompiled with the original regexp but it doesn't works, trying to understand .. [11:04] it will give you 2 even though 0 is the default [11:04] make that Launcher [11:04] not launcher [11:04] then if you go 'gsettings reset com.canonical.Unity2d.Launcher hide-mode' [11:04] Saviq: ok, I though you saw the value in the binary blob :) [11:04] and 'gsettings get' again, it will give you 0 [11:04] Saviq: yeah, that's what I did here [11:04] Saviq: let me boot a live [11:04] angeloc: and you did a pkill -f unity-files-daemon and is running out of the build tree ala src/unity-files-daemon? [11:05] didrocks, installing dconf-tools now to take a look from dconf-editor [11:05] kamstrup: yes! [11:05] odd! [11:05] Saviq: well, it won't help more [11:05] Saviq: it's basically using the same lib than gsettings [11:05] and parsing the schema by hand to get the default [11:06] didrocks, yes, but it will show whether the value is overridden [11:06] by bolding the key [11:06] kamstrup: i messed up something... [11:06] I just want to confirm [11:06] Saviq: right, but it clearly is and if you are sure than on the live there is this user db [11:06] yes [11:06] I am [11:06] * didrocks boots boots boots :) [11:08] so yeah, there is this file [11:08] not sure it's generated on first boot [11:08] or not [11:08] anyone happening to have a local copy of the xembed spec lying around? the oneline one on freedesktop.org is corrupted... [11:08] I don't see a "2" in it, though [11:09] kamstrup: ok, found! I'm a perfect idiot! I'm building a local package with dpkg-buildpackage to test it, in the meantime I changed debian changelog with dhc. After recompilation I was installing the old deb... [11:09] didrocks: any idea why https://jenkins.qa.ubuntu.com/job/automerge-unity-2d/152/console failed? [11:09] angeloc: ah, even testing the packaging... committed you are :-) [11:09] tsdgeos: because some people are using daily build in the staging ppa for compiz which breaks unity instead of the merger… [11:10] tsdgeos: ah not for this one though :) [11:10] didrocks: Saviq blamed himself for this one [11:10] tsdgeos: the link is https://jenkins.qa.ubuntu.com/job/automerge-unity-2d/153/console, isn't it? [11:10] didrocks, my fault [11:10] didrocks: both 152 and 153 [11:10] fixed already [11:11] great :) [11:11] kamstrup: yes, for sure, this way the patch should work and a merge proposal is more likely to be accepted, or I'm wrong? [11:15] Saviq: ok, I have no time for looking at this right now, I confirm though. Can you please open a bug and assign it to me so that I can look at this later? [11:15] didrocks, I will [11:15] thanks [11:16] kamstrup: it works! [11:29] didrocks, can't assign to you, here's the bug https://bugs.launchpad.net/ubuntu/+source/unity-2d/+bug/942772 [11:29] Ubuntu bug 942772 in unity-2d (Ubuntu) "Precise unity-2d launcher auto-hiding out-of-box" [Undecided,Confirmed] [11:30] assigning myself then. Thanks [11:35] mhr3, hey [11:35] sup seb128 [11:35] mhr3, what was the issue again with systemtap on Ubuntu? what is not working? [11:35] mhr3, discussing with some people on #ubuntu-devel but I don't remember the specifics [11:36] seb128, our kernel doesn't have required component [11:36] mhr3, was some kernel side missing? [11:36] mhr3, which ones? [11:36] U... something :) [11:36] let me check [11:36] thanks [11:36] seb128, right, UTRACE [11:36] it's not merged in upstream kernel [11:37] all most of the other distros are shipping it [11:37] almost* [11:37] almost all*... :P [11:38] mhr3: thanks (i originally asked in #ubuntu-devel) - let's say i can make that work, i'd still need to recompile glib2.0 with --enable-systemtap, right? [11:38] mhr3, thanks [11:38] htorque, yes [11:39] thanks :) [11:39] mhr3, just spotted your locate branch! This is wonderful :) [11:39] htorque, moreover there's some weirdness in stap, it didn't work for me with my standard user account, needed to run it with root (although i did add myself to the relevant user groups) [11:40] htorque, and being root doesn't play nice with rest of desktop... (dbus etc) [11:41] mhr3: thanks for the hint. i'll give it a go anyway. :-) [11:41] htorque, bottom line if you're shooting at using it to track down memory issues or ref counting issues, patch glib itself instead :) [11:42] or... you know use fedora :) [11:44] davidcalle, our view of "wonderful" differs :P [11:44] mhr3, wonderful in the sense that it will make the vala port of the videos lens much easier :P [11:45] ah, yea in that case it's a masterpiece ;) [11:45] hehe ;) [11:48] why not use valgrind to trace memory issues? [11:51] cking: it's mainly to satisfy my curiosity :-) [11:53] kamstrup: sorry, I'm late, there were more problems than what I tought! I'll make a merge proposal in a few hours, I cannot make it now, but it works! [11:53] angeloc: no worries, I'll get notifications by mail when you submit it any way. I check it many times per day [11:54] kamstrup: ok [11:57] htorque, just saying that valgrind is a pretty easy to use and powerful too for identifying memory issues [12:00] cking, and unable to track down ref counting issues [12:00] mhr3, now understood. === danilo_ is now known as danilos [12:09] om26er: i opened a bug report about the white pixel at (0, 0): bug 944701 - afaic you've seen that too? if so, please confirm (and if you've found a way to reproduce it, pretty please add it). :-) [12:09] Launchpad bug 944701 in unity (Ubuntu) "Sometimes a single white pixel shows in the top-left corner" [Undecided,New] https://launchpad.net/bugs/944701 [12:11] mhr3: did you see my last comment on the locate mp? [12:14] kamstrup, reading [12:18] kamstrup, anytime we yield in the async method, the cancellable is passed to the async operation, there is no place where we'd let the mainloop spin and the cancellable wouldn't be "ready to get cancelled" (like a timeout), and since we can get cancelled only from the main thread (courtesy of libunity) it's fine [12:18] kamstrup: done, Imade it! [12:19] kamstrup, is that clear? === _salem is now known as salem_ [12:41] mhr3: is the cancellable not passed down from libunity? let me check... [12:42] looks like it is [12:43] mhr3: in that case we are waiting for an idle call to dispatch. We can not know if something else has been scheduled before the idle [12:43] or [12:44] hmmm maybe I am confused :-) [12:44] angeloc: do you have a link for the merge request? [12:46] mhr3: maybe you're right. But in any case I think you need to check the cancellable before you spawn locate [12:47] mhr3: we'll have waited 500ms before you spawn it === greyback is now known as greyback|away [13:21] sorry about repetitive question - i am trying ton understand whether light menu backgrounds in windows (e.g. context menus) are a feature or a bug (925895) [13:24] Cimi, ^ [13:25] rye: try this in gnome-shell, as it may be a unity bug, as Cimi has commented [13:28] roignac, good idea, however i don't know what to expect - in oneiric all menus had dark background - indicator and gtk2/gtk3 and firefox xul, openofffice. Now indicator menus are black, sometimes indicator submenus are white, context menus are white (well, light background, not completely white) [13:29] I'm sure this should be consistent - black in Ambiance, white in Radiance [13:29] I mean, all black [13:32] works fine for me in gnome-shell, except FF - menus are white [13:35] angeloc: did you see my comment on your mp? === yofel_ is now known as yofel [13:37] seb128, isn't it fixed? [13:38] FF is a gtk2 apps [13:38] it will have white menus [13:38] because I only care of unity [13:38] kamstrup, yea, i do check it with the set_error_if_cancelled() [13:39] and in unity their menus are in the panel thus black [13:39] kamstrup, since this is vala it's more like throw_error_if_cancelled() === greyback|away is now known as greyback [13:43] Cimi, rye: it works for me on current precise in a guest session [13:43] well tested with indicator menus in unity [13:45] Saviq: hey, I don't remember if you have a pbuider-like environment on your system? [13:45] Saviq: that will maybe be easier for you to reproduce the issue [13:45] didrocks, not yet [13:45] didrocks, you mean the failing test? [13:46] Saviq: yeah, I can help you setting that up if you want [13:46] erm, Cimi, are context menus in gtk3apps e.g. gnome-terminal supposed to have dark backgrounds? [13:46] Saviq: are you familiar with pbuilder? [13:46] didrocks, not yet, but I think I did touch it once [13:46] didrocks, but I'm sure I can find stuff on one of the wikis [13:47] didrocks, if you have a link handy that would be cool [13:47] Saviq: yeah, but I have some tweaked config :) [13:47] Saviq: let me help you on that [13:47] Saviq: so: sudo apt-get install pbuilder debootstrap devscripts [13:47] didrocks, did you see the latest result http://localhost:8070/job/automerge-unity-2d/lastBuild/consoleFull ? [13:47] grr [13:47] https://jenkins.qa.ubuntu.com/job/automerge-unity-2d/158/console [13:48] didrocks, I fixed the broken test [13:48] Saviq: ah? [13:48] but there's another one that suddenly started failing [13:48] The following tests FAILED: [13:48] 6 - focuspathtest (Failed) [13:48] argh :/ [13:48] yes, that is old [13:48] yeah, will be easier for you to reproduce the environment [13:48] and has been passing for ages now [13:48] probably a racy dbus call [13:49] rye, no [13:49] didrocks, what's more all the tests passed in another jenkins install [13:49] Saviq: yeah, but you are not using a chroot [13:49] Saviq: contrary to the merger setup [13:49] didrocks, might be [13:49] hence the fact I can show you how to have one locally [13:49] didrocks, ok got pbuilder [13:49] ok, so then, you need a .pbuiderrc file and a pbuilder-hooks one [13:50] one sec, I put those in a branch [13:50] Saviq: ~unity-merger/unity-merger-tool/trunk [13:51] Saviq: from this branch, takes .pbuilderrc and .pbuilder-hooks/ directory [13:51] in .pbuilderrc, change the HOOKDIR to point to you .pbuilder-hooks [13:51] Saviq: are you on precise? [13:52] didrocks, yes [13:52] Saviq: ok, so then, just create the pbuidler: sudo pbuilder create --debootstrapopts --variant=buildd [13:53] Saviq: an finally, drop a B10shell script in your pbuilder-hooks directory containing: http://paste.ubuntu.com/865100/ [13:53] Saviq: that will drop you into a shell once the build finished [13:53] Saviq: make it executable for pbuilder to take it into account [13:54] from there, you can run the make check :) [13:54] didrocks, ok that's great [13:54] Saviq: ping me if you need any help [13:56] kamstrup: I'm back! === MacSlow is now known as MacSlow|lunch [14:00] hi, should the spread workspaces function include minimised windows? [14:03] kamstrup: I deleted old merge proposal and made a new correct one ... big fail... [14:04] kamstrup: epic fail ... [14:07] angeloc: you have an URL for the mp? I don't have it in my inbox yet [14:07] bug 773841, is on the page [14:08] Launchpad bug 773841 in unity-place-files (Ubuntu) "\\192.168.1.x opens http:\\192.168.1.x in firefox as opposed to smb://192.168.1.x in nautilus" [Low,Confirmed] https://launchpad.net/bugs/773841 [14:14] angeloc: looks good at a glance, but I need to take it for a test spin later [14:15] kamstrup: yes! I'm reeeally happy! [14:20] angeloc: it's awesome dude; you'll make a lot of other people happy by fixing this :-) [14:21] kamstrup: I have a vm ready that I use for testing, if you want I can make you an ssh to test it out [14:24] angeloc: no no :-) I have plenty servers to test against :-) [14:24] angeloc: off the top of my head I can see that you've included updates to the debian changelog, we don't usually do that; we should clear that with didrocks [14:24] kamstrup: ok! [14:25] ok, the latest bug I corrected, ubuntu devs did make me the chage [14:25] hum, maybe you took the wrong branch? [14:25] as normally, the upstream branch don't containg the debian/changelog [14:28] didrocks: I think you are right ... [14:28] didrocks: bzr branch https://code.launchpad.net/~ubuntu-desktop/unity-lens-files/ubuntu [14:29] didrocks: instead it should have been bzr branch lp:unity-lens-files [14:30] right :) [14:30] the good news [14:31] is that you can just import your commit [14:31] * smb needs to express his utter disgust with the decision of having alt tap bringing up the hud [14:31] with bzr merge .. the branch -r [14:31] didrocks: nooooo, i'm crying [14:32] didrocks, ok, wait a few minutes and I remake a new merge proposal with the correct branch [14:32] didrocks: i delete the current mp and the current fix branch [14:32] didrocks: ok? [14:34] didrocks, a question, apt-get source unity-lens-files says me that I have to retrive the latest version from https://code.launchpad.net/~ubuntu-desktop/unity-lens-files/ubuntu, why? === chaoticuk is now known as chaotic [14:38] angeloc, because that's the vcs used for the packaging [14:38] it might have changes not uploaded yet and that's where the work goes [14:39] thanks seb128, I understand [14:40] seb128 before software becames a package is in a limbo, when packaged you should report changes versus the current package. I suppose this is true after an ubuntu release [14:41] didrocks, can't find that info, (how) can I tell pbuilder to install xvfb? [14:42] Saviq: so, look at D10specifictests hook [14:42] htorque, i see it somewhere [14:42] *sometimes [14:42] confirmed the bug [14:42] Saviq: I apt-get install it there [14:42] didrocks, ok thanks [14:42] Saviq: you need to make it executable [14:43] hmm it is [14:44] and I didn't get the shell [14:44] must've messed up the hooks [14:45] Saviq: did you changed your HOOKDIR in .pbuidedrrc? [14:45] yes [14:45] to the right dir, isn't it? :) [14:45] D09custompool got ran [14:45] interesting [14:46] let me scan through the log again [14:46] check for D10specifictests [14:46] there is no reason one is ran and not the other [14:46] apart from different executable state [14:47] ok so it did install xvfb but the tests failed nevertheless [14:47] ah, more interesting :) [14:47] so remove B10runmakecheck [14:47] or make it not executable [14:48] you have B10shell? [14:48] didrocks, ah so it didn't drop me into a shell 'cause tests failed [14:48] ?/ [14:48] yeah ;) [14:48] ok that makes sense [14:48] the script is B10shell? [14:48]