[04:00] * TheMuso finally works out that the unity panel is drawn by unity/nux, and the panel service only takes care of showing gtk elements when the correct region is clicked for an indicator/menu... [06:23] Joey should get that papercuts post out fast [06:48] morning [07:40] good morning [07:42] didrocks: good morning. how are you? [07:43] kvalo: I'm good, thanks, and you? [07:43] didrocks: I'm good as well. waiting for the long weekend :) === njpatel is now known as njpatel_ [08:13] morning folks [08:26] MacSlow: morning [08:26] hey kamstrup === ssj6akshat is now known as ssj6akshat|away [09:14] didrocks: is there a known issue with the default browser setting? [09:15] dbarth_: hum, not AFAIK, what's up? [09:15] didrocks: my irc client keeps opening firefox whereas chrome is set as my default browser [09:15] dbarth_: oh you mean, it doesn't pick the right default browser for you? [09:15] yeah [09:15] that changed [09:15] ah [09:15] and someone need to do a chromium (and chrome?) [09:15] it's bugging me as well [09:15] you can fix that manually IIRC [09:16] it's using a mimetype, let me look at my logs [09:20] dbarth_: /usr/share/applications/defaults.list [09:20] dbarth_: x-scheme-handler/http=firefox.desktop [09:20] you probably want to change that [09:21] (same for https) [09:21] there is the local equivalent IIRC [09:29] didrocks: thanks, it works; i assume there is a bug against the default application selector in gnome? [09:29] dbarth_: not sure there is a bug, but it's tracked === ssj6akshat|away is now known as ssj6akshat [09:51] kamstrup: still have time for review? :) https://code.launchpad.net/~kvalo/indicator-network/libconnman-sort-services/+merge/42597 [10:02] kvalo: approved, although I was slightly less thorough than usually since I have quite a lot to do today [10:04] kamstrup: oh, I'm disappointed now ;) but seriously, thanks a lot [10:49] didrocks, u there? [10:51] seiflotfy: yes [12:02] ronoc: hi. can you review this one, please: https://code.launchpad.net/~kvalo/indicator-network/libconnman-sort-services/+merge/42611 [12:03] kvalo, sure [12:07] kvalo, the fixme comment in mock-service.c is a little odd ? [12:08] also cmp_services returns a mixture of booleans and ints, just a semantic thing but shouldn't all be just ints [12:09] kvalo, which could be an enum, which would reader nicer [12:10] kvalo, sorry reader the cmp_services incorrectly [12:13] ronoc: the comment in the mock means that it really should be a subclass for dbusmenu menuitem, but for now I just subclassed it from gobject [12:15] kvalo, sure, it might be nice to typedef the return value so as it is explicity what is being returned, not just a mere int but 'rating' [12:15] ronoc: yeah, cmp_services() should return gints only [12:16] just a thought [12:16] ronoc: ah, but that's defined in glib [12:16] will approve with that comment [12:16] kvalo, which is defined in glib ? [12:17] ronoc: http://library.gnome.org/devel/glib/unstable/glib-Doubly-Linked-Lists.html#GCompareFunc [12:17] ronoc: what cmp_services() need to return === MacSlow is now known as MacSlow|lunch [12:17] kvalo, okay, approving [12:18] ronoc: thanks! === smorar_ is now known as smorar === smorar is now known as smorar_ === smorar_ is now known as smorar === MacSlow|lunch is now known as MacSlow [14:02] you know what? one day, I'll learn to type != when I want != and not == :/ [14:14] dbarth_: when do you want to do the get-involved unity page overview? [14:15] dbarth_: jono would like me to announce it today so if we can find a few more bugs that would be great [14:15] didrocks: ^ [14:15] jcastro: I'm finishing fixing one and get work with you just after that [14:15] if you have time :) [14:15] maybe dbarth_ should join as well [14:16] ok rock! [14:22] morning everyone [14:22] tedg: Hello, out of you cody and conor, who is the best to talk about creating a custom indicator menu item widget? [14:23] doctormo, It depends on which part of me you ask. My ego definitely says me, but other parts of my psyche may disagree. [14:23] doctormo, Is this for an app indicator? [14:23] tedg: We'll go with your ego, it's the best part of you ;-) [14:24] jcastro: give me 20 mins, and then i'm your man [14:24] This isn't for an app indictator, this is for experiments with a progress indicator. Novacut are interested and there a few designs on the wiki already. [14:25] doctormo, Ah, interesting. [14:25] tedg: First a simple question, would it be possible to do in python? [14:25] hey lamalex [14:26] doctormo, Partially. System indicators have a loadable module part (which is typically very small, visualization only) and that would need to be in a compiled language. But the service that implements all the logic just exists over dbus, so it can be in any language you wish. [14:26] doctormo, We've stayed away from Python specifically because of the large memory footprint. Currently, all the services in their base state use less RAM than one copy of the Python interpreter. [14:27] doctormo, That's why we've been pushing Vala in that space. [14:27] makes sense, will have to push that here too then. [14:28] ted is vala fan #1 [14:28] Heh, except that I personally haven't written anything in Vala yet :) [14:28] Though, I do think it's a good idea. [14:29] Ted says "that's good in principle, do that" :-D [14:30] Do as I say not as I do. :) [14:31] I find it hard to justify the memory cost of an interpreter on small things that are running continually. The percentage cost of the interpreter is too high. [14:32] That sort of talk is anti-perl [14:32] hi tedg! :) [14:33] Good morning jderose [14:33] tedg: wait, what, indicator-applet uses less memory than one python interpreter? [14:33] doctormo, I'm not anti-anything. I just remember what it was like to have 16MB of RAM :) [14:34] hyperair, The various services -- indicator-messages-service, etc. [14:34] tedg: if you remember then fix the indicators to stop leaking memory >_> [14:34] tedg: do you know i need to kill indicator-applet every 6 hours or so? [14:34] tedg: doctormo was relaying your wisdom to me on #novacut, so wanted to say hi, and thanks [14:34] tedg: that reclaims around 100M of memory each time. [14:34] Things like that applet grow a bunch because they start caching all the application memus. [14:34] tedg: which applications? [14:35] hyperair, For the application menu bar. indicator-applet-appmenu [14:35] hyperair, Hmm, I don't see a leak like that.... [14:35] tedg: i have no fewer applications in the indicator after restarting, and i don't use the menu bar. this is just the old indicator-applet in maverick [14:35] tedg: try keeping your system on for 6 days at a time, and you'll see it. [14:35] hyperair: Interesting case, can you debug? [14:35] indicator-applet is my #1 restarted process [14:35] doctormo: i don't know where to start [14:36] tedg: in fact, i don't know why ubuntu seems so heavyweight recently. [14:36] tedg: i used to be able to run a VM on 64-bit ubuntu on 2G of RAM [14:36] hyperair, Hmm, I literally just rebooted this morning so I don't have anything to compare -- but I typically do and hadn't noticed :-/ [14:36] with firefox and thunderbird [14:36] tedg: now with 2G of ram i need to choose between firefox and thunderbird. [14:36] which to have open. open one, close the other, and vice versa [14:36] >_> [14:37] machine goes nuts otherwise [14:37] Hmm, I have 2G and no swap.... [14:37] try using 64-bit ubuntu [14:37] I am using 64-bit [14:37] what [14:37] that's so unfair =_=" [14:37] hyperair: Could this help you? http://munir.wordpress.com/2006/08/05/finding-memory-leaks-using-mtrace/ [14:39] doctormo: these days we use valgrind [14:39] doctormo: but stuff like mtrace and valgrind make things go really really slowly, and consume loads of CPU. [14:39] doctormo: it's not something you want to run for 6 hours [14:40] hyperair: You may not have to? I thought you could tell a leaky pipe without waiting for the flooded apartment. [14:41] doctormo: another thing is gtk+/glib is really valgrind unfriendly. [14:42] tedg: what's your memory usage like usually? [14:44] tedg: this is purely guesswork, but i'm thinking that the messaging indicator is the one causing problems. something between indicator-applet and indicator-messages-service (indicator-messages-service doesn't need killing, but indicator-applet does) [14:44] hyperair, I'm getting old -- I can't remember anything anymore ;) [14:45] hahah [14:45] I don't have anything indicator* over 14M [14:45] =\ [14:45] try again in a few hours [14:45] maybe you need something active in your messaging indicator [14:45] e.g. liferea with lucidfox's indicator patch [14:46] hyperair, I have 5 apps life in my messaging menu, including liferea :) [14:46] live [14:46] hmm =\ [14:46] i have 4 [14:46] tedg: which version of liferea? [14:47] hyperair, 1.6.3 [14:47] hmm maybe i should drop the version [14:48] i think this one has an outdated patch [14:48] i'm on 1.7 [14:53] Are there any indicator service plugins that use vala? the sound plugin is c and I assume it's that way for a reason? [14:55] htorque, present? [14:55] doctormo, The .so's are all C. Sound uses a bunch of Vala for the service though. [14:56] doctormo, There's not specifically a *reason* that the .so's are C (it's GObject) but I've not tried to do that before. [14:56] DBO, yes [14:56] tedg: Yes I see that part, just curious because my vala is better than my c and I didn't want to learn c in order to experiment with this. [14:56] htorque, I am really trying to reproduce your crash on close bug [14:57] can you give me any help? [14:57] DBO, i simply opened an application and closed it (doesn't matter if i use the window border "x", the quit in the quicklist or ctrl-c from the terminal) [14:58] doctormo, In theory, the only thing in the .so is a single symbol that instanticates the type of the object subclass that you've made. It *should* be a single line of C.... [14:59] DBO, i'll try it on a second system [14:59] htorque, can you give me as much information about your current unity/compiz/bamf setup as possible [14:59] doctormo: vala compiles into C, so i don't think there are any issues with writing services in vala really. [15:00] tedg: Can you give me a minute to walk me through some basics of compiling and running for debugging your datetime indicator? I'll play with that. [15:00] * jderose will follow along [15:01] doctormo, bzr branch lp:indicator-datetime ; cd indicator-datetime ; ./autogen.sh ; make :) [15:02] Oh, you should probably put a "sudo apt-get build-dep indicator-datetime" in there :) [15:02] tedg: Half the story, now how about running, inserting, removing? don't I have to restart the computer every time I compile? [15:02] jcastro: pong [15:02] doctormo, No install libindicator-tools and then you'll have /usr/lib/libindicator/indicator-loader [15:02] DBO, using unity r659, bamf r365 and compiz 0.9.2.1+glibmainloop2-0ubuntu4, my compiz profile should be the default "ubuntu" one [15:03] tedg: Ah there's the key info. thanks. [15:03] doctormo, So then you can, in src, do /usr/lib/libindicator/indicator-loader .libs/libdatetime.so [15:04] brilliant, and kill to murder it if it's a bastard child right? [15:04] htorque, when did the crashing start precisely? [15:04] tedg: ah, so does trunk only run on natty? i'm on maverick ATM [15:04] DBO, right after installing bamf r365 (after a session restart) [15:05] DBO that was before updating unity [15:05] jderose, No, I haven't moved it over to Natty yet. [15:05] jderose, If I do, I'll branch it. [15:05] jderose, I don't think it's going to need an GTK 3 features, so it should run on Maverick through the cycle. [15:05] hmm, i'm still getting errors from ./autogen.sh even after i atp-get build-dep [15:06] htorque, I am deeply confused them by how this is crashing... [15:06] jderose, pastebin? [15:06] No package 'geoclue' found [15:06] No package 'liboobs-1' found [15:06] jderose, Oh, sorry, I didn't realize those were in trunk already. [15:07] tedg: well see, i just found a bug in debian/ then :) [15:07] jderose, You'll need libgeoclue-dev and liboobs-1-dev [15:07] DBO, i'm compiling bamf on a second system now, maybe this one is bad from too much testing (but it shouldn't be) [15:07] htorque, mikkel solved it [15:07] jderose, Well, there's no debian directory in this as it's an upstream branch. And build-dep works off the version in the Maverick archive. [15:08] tedg: well, sure... should have assumed it works in natty [15:09] jderose, Actually, it doesn't work in Natty yet either as the new trunk hasn't been released yet into the Natty archive :) [15:09] I take it geoclue and liboobs-1 are new deps to the datetime indicator and so not covered by the build-dep. I wonder if geoclue is available in maverick. [15:10] doctormo, It is. [15:10] doctormo, Universe in Maverick [15:10] doctormo, You won't get ubuntu-geoip, but that's not required. [15:10] tedg: bash: /usr/lib/libindicator/indicator-loader: No such file or directory [15:10] htorque, please try rev 366 [15:10] jderose, Did you install libindicator-tools? [15:10] ah, nope [15:10] DBO, on it [15:12] tedg: hi. I ported indicator-network to gdbus and now I have a problem of stale entries in the menu after network-service has crashed. any ideas where to start debugging it? [15:12] kvalo, Uhm... gosh. Not really... they should be getting unref'd the same. [15:13] kvalo, You know, you could try my GDBus port of Dbusmenu ;) [15:14] what is this GDbus? it scares me. [15:14] doctormo, DBus bindings that are now in GLib. [15:14] dbus-glib is now deprecated [15:14] Ah, not what I feared it might be (a gnome replacement for dbus) ;-) [15:15] tedg: I think I did a mistake somewhere, I just need to find where [15:15] DBO, seems to work fine (and i can start apps more than four times ;)) [15:15] doctormo, If you ask the GNOME folks they wrote dbus and everything else, so dbus is effectively a core component of GNOME OS :) [15:15] htorque, lovely [15:15] tedg: I saw your huge patch :) [15:15] htorque, fix a bug, reveal another it was hiding [15:16] huzzah [15:16] dbarth__: I'm set whenever you are [15:16] tedg: shhhh! it doesn't matter _who_ wrote it, the important thing is that it doesn't _feel_ gnomish. [15:16] tedg: so when i run /usr/lib/libindicator/indicator-loader .libs/libdatetime.so, what is supposed to happen? nothing seems to appear in panel [15:17] jderose, It opens a new window with the indicator in it. [15:17] ah, okay... window was like 1px wide or something, didn't see [15:17] doctormo: I don't think anyone in the KDE world is confused about how useful FDO is for cross-DE work that's fair and balanced. [15:18] DBO, yep, well done! :) [15:18] tedg: so is there an easy way to stick it into panel, replace one currently running? what's your testing workflow? [15:18] thank you so much for your wonderful never ending stream of high quality bug reports :) [15:19] jderose, Well, if you install it you can kill the indicator-applet and it will reload with the new indicators. But, I use the loader for almost everything. I'll only put it in the system directory after building a package. [15:20] I'm probably more to the anal side of not having anything on my system that isn't from a package. [15:20] DBO, i'm just writing stuff down, but yw! :) [15:20] tedg: so are there any differences (for debugging and testing) between running in a window and running in panel? [15:21] * jderose thinks that's because packages rock, tedg :) [15:22] jderose, Not really. It matters for the appmenus because they deal with issues like window focus. But all the other indicators don't really care. [15:23] We should probably poll ronoc and kvalo for their development style as well. They might be different. [15:23] tedg: okay, awesome... thanks for tutorial [15:23] jderose: can you private message me the history? my computer crashed. [15:23] OT: is #ayatana logged somewhere publicly? [15:24] OT tedg: that would be against the evil internal policy wouldn't it? [15:24] this registration mechanism is proving more difficult than I expected [15:24] tedg, it's on http://irclogs.ubuntu.com/ [15:24] doctormo, I thought that all #ubuntu channels were logged, no? [15:25] jcastro: grr, missed the ping back sorry [15:25] tedg: they are, here where htorque has shown. But I thought it would only apply to #ubuntu-ayatana [15:25] htorque, Cool, thanks! [15:25] ronoc, :( Yeah, it seems like you might have to use custom filters? [15:26] tedg, yeah I have been trying to use the bus.watch pattern but I'm blocked by a pretty major vala bug [15:27] tedg, have been speaking phako on #vala and he has pointed me towards rygels design [15:27] doctormo: whoops, that sure didn't work. Excess Flood. [15:29] jderose: I'll just have to wait for the logs to update [15:30] doctormo: i wonder at how many lines/bytes the Excess Flood triggers... i didn't paste that much! [15:31] 5 perhaps [15:31] jderose: You could pastebin it [15:31] doctormo, I wonder if it's by time or data size... we can keep chatting to push them ;) [15:32] see tedg, you're just full of good ideas and advice. doctormo: http://paste.ubuntu.com/539453/ [15:36] jderose: I always consider ted to have these on his shoulders: http://www.youtube.com/watch?v=Fv-sKP17xTw [15:38] tedg, i think I will go this route and implement something similar, not ideal but I'm going to connect to the name owner changed signal and react accordingly [15:40] kenvandine, hi, do you have five minutes spare to look at a Brainstorm idea? [15:41] doctormo: hehe... you ever watch the venture brothers? the voice actor for the devil is the same as the moppet twins, hero is same as Brock Samson (aka The Tick) [15:41] compiz alt-tab is really slow [15:41] drives me nuts [15:41] lamalex, Nvidia? [15:42] tedg, intel [15:42] I believe didrocks was saying that to smspillaz yesterday, but his was nvidia. [15:43] yeah, I get that too [15:43] when I have a lot of window in a ws [15:43] I think it's when staticswitcher try to create the thumbs or whatever [15:48] mpt, sure [15:48] kenvandine, , which has been implemented as [15:49] Is it a reasonable architecture, and if so, what's the next step for getting it into Ubuntu? (packaging it, I guess) [15:50] * kenvandine is looking [15:51] mpt: I would say those kind of things should be reviewed by mvo [15:51] hm, that makes sense [15:55] yeah, have mvo look at it [15:56] but it looks like it would be just getting the package included [15:56] although installing it didn't actually make it work for me [15:56] looks like a cool solution, and minimal change to accomplish it [15:57] ok, thanks kenvandine [15:57] anytime mpt [16:15] tedg, going to have a dbusmenu and indicator-datetime upload today? [16:16] kenvandine, dbusmenu, no. The review probably won't be done until next week -- but I'd like to talk about packaging it. [16:16] kenvandine, It's a 9000 line diff :-/ [16:16] ok [16:17] i am uploading libindicator now [16:17] then i'll get back to g-ir-scanner and appindicator, ewww [16:17] kenvandine, Okay, I have patches to split up indicator-application and libappindicator [16:17] tedg, how about indicator-datetime? [16:18] tedg, split? [16:18] kenvandine, Yeah, i can do that one for sure. [16:18] kenvandine, yeah, into two tarballs. So hopefully it makes the library easier to adopt for other distros. [16:18] ah [16:18] yeah [16:18] doesn't help with g-ir-scanner though :) [16:19] kenvandine, Heh, yeah. And the repos are branched, so any changes should apply correctly to the various repos. [16:19] * tedg loves using a DVCS with proper file IDs :) [16:33] didrocks: dbarth: lamalex: ok guys I've got a blog post ready to go and we've done a once over on the get-involved unity page, but I still only have like 4 bitesize bugs [16:34] surely you guys can get me at least 10 or 15? [16:34] jcastro: I'm currently reviewing them [16:34] \m/ [16:34] jcastro: do you have the link to the bitesize btw? [16:34] (will be easier for me :)) [16:59] https://bugs.launchpad.net/unity/+bugs?field.tag=bitesize [16:59] didrocks: ^ [16:59] didrocks: hey apt-get build-dep unity should pull in all the right stuff for people to get started right? [16:59] I am adding a developer section to the usb key install stuff [17:00] so people can just make a USB key, install stuff, and then start hacking [17:00] jcastro: right [17:00] jcastro: awesome idea [17:01] jcastro: sorry, still discussing bugs right now and opening some, will get a list soon :) [17:01] no worries [17:07] https://wiki.ubuntu.com/Unity/InstallUSBKey [17:07] ok there are the docs for installing the dev stuff [17:11] jcastro, why is the text in the new wiki theme so small? (sorry if that's OT here) [17:20] htorque: not sure. :-/ [17:21] htorque: I have it set on large in my browser [17:21] htorque: feel free to update/fix that page if you find broken stuff if you try it [17:26] jcastro, nothing's broken there - looks fine. :) as for the fonts: it's across all browsers and other pages show bigger fonts, don't know... http://img.xrmb2.net/images/237664.png (chromium, opera, firefox) [17:26] good weekend everybody... see you monday [17:54] ivanka: my phone just crashed [17:55] ivanka: I'll be back on in a minute [17:55] seg: no worries, I will be here [17:56] good day all [17:57] ivanka: doesn't seem to be working [17:57] seg: I have enough to be getting on with [17:57] seg: shall bug you again if I need more [17:57] seg: and then you can get on with your day! [17:57] seg: thank you so much for your time [17:58] ivanka: great! Let me know if you need more. I can also answer questions via e-mail, which would save you the difficulty of transcribing [17:58] seg: ah, but then I would lose the tone :-) [17:59] hah [17:59] seg: have a great friday and I will catch you another time if I have more specific questions, will email them [17:59] cool [17:59] have a nice weekend [18:01] didrocks: john has acked the list of quicklist things and I'm ready to publish it [18:01] didrocks: so what's the plan as far as the .desktop files? [18:01] or is that a dbarth thing? [18:11] jcastro: it's not implement in unity compiz yet [18:11] jcastro: so, will have to wait for a week [18:11] anyone on dual monitor with unity? [18:33] didrocks: ok no problem [18:34] coz_: yep, nvidia twinview over here [18:34] jcastro, cool...how is it working [18:34] the only bug left for me is the top bar spans 2 monitors, but neil is on it and should be fixed soon [18:34] jcastro: I went through all bugs [18:34] jcastro, ah ok [18:34] I added some: Unity [18:35] https://bugs.launchpad.net/unity/+bugs?field.tag=bitesize [18:35] there are 10 now [18:35] yeah!!!! [18:35] jcastro, mm then I guess I will attempt it on this machine with dual monitors...thanks.... [18:35] jcastro: need to go on week-end now :) [18:36] thanks man! [18:36] coz_: everything else works though, it maximizes to the right monitor, etc. [18:36] jcastro, is neil going to have an option for both span both monitors and not ? [18:36] jcastro: yw :) [18:37] * spikeb tests a1 [18:41] brb on A1 [18:47] htorque: put your unity launcher question update as a seperate answer, since it's the correct way to do it now and then mark it as the accepted one. [18:55] jcastro, alright! [19:09] * spikeb-natty considers himself impressed [19:47] klattimer1, doctormo, FYI, I branch indicator-datetime so lp:indicator-datetime/0.1 is the Maverick branch (no geoclue, etc.) and lp:indicator-datetime has all the goodies. I don't think it'll effect much, but FYI. [19:56] smspillaz, present? [21:13] tedg, ok, i think i have a sane GIR and vapi for appindicator [21:13] mind taking a quick look at them to see if they make sense? [21:13] * tedg high fives kenvandine! [21:13] Sure! [21:20] kenvandine, Make sense to me. [21:20] woot [22:56] bratsche, Can I get you to review this *long* patch? https://code.edge.launchpad.net/~ted/indicator-appmenu/alt-d-problem/+merge/42699 === smorar is now known as smorar_