=== yofel_ is now known as yofel [09:28] morning all [10:13] Happy Monday, everyone! :D [11:10] o/ [11:13] hi! [11:35] gatox, duanedesign buenos dias! [11:35] mandel, buenas [11:39] gatox, I have a qt question, qt the style in gnome is grabbed from the theme, right? [11:40] gatox, I have done the dialog and the links show blue (I did no styling what so ever) I suppose we are going to be settings a global theme for the qt port of sso for that, right? [11:40] mandel, yes....... we are going to set a style for that.... and the default style is taken from the system [11:41] gatox, ok, so atm links are blue, yikes! [11:41] in a very veryyyyy low level in the qt classes..... when qt start running, takes the default windows manager widgets depending on which system is running........ is aa big IFDEF [11:42] gatox, so, when ever you set the style for that you have to remember that i should be able to import it (I suppose you are going to use a resources file) since the auth dialog for sso is a diff process [11:42] gatox, and it will be a diff process even for sso [11:43] gatox, how is the port of sso going? got anything running under qt? [11:43] mandel, yes....... we were talking about that the last week, we are going to see how to do it.... because the installer is going to need those styles..... but you can not apply 2 styles to something unless you join those strings..... so, we are going to see which should be the cleaner way to do it [11:44] mandel, was running, some minor failures.... but now i'm making a huge refactoring for the controllers and the pages [11:44] gatox, do something similar to what cofigglue does with config files, simple add a loader thaqn returns a string and forget about what the loader does :P [11:45] gatox, uh, do not put the logic in the ui, if you do I'll hate you! [11:45] :P [11:46] mandel, i'm not doing that.... i'm going to work defining some models or controllers..... but using inheritance instead of creating the controllers by its own, because that was causing a lot of troubles [11:46] gatox, what kind of troubles? [11:49] mandel, several problems trying to create the controllers and asociating them to the pages, in several cases depends iin the order that you do it, because the controllers try to obtain data that doesn't exist until the page is added to the wizard..... i fix that..... but you have to keep in mind those kind of things.... but then another issues appear, so i began the refactoring with controllers [11:50] gatox, makes sense, as long as the controller logic has no knowledge of the ui perse, that is get_text or text etc it should be ok [11:50] gatox, ideally you will want to use the controller in gtk or qt, otherwise you are duplicating logic, which is wrong [11:54] hello everyone! [11:54] nessita, hi [11:54] hola gatox [11:56] nessita, morning! [11:56] hola mandel! [11:57] nessita, I did my first iteration with design regarding the style of the auth dialog and we should have a second one later today [11:57] nessita, there is only one implementation thing that I'd like to know how to do which is getting the css style from sso to be used by the dialog, but It think that can be solved later, right? [11:58] mandel: yes, ATM we're not loading css in sso, so we can setup that later [11:59] nessita, ok, shall I use a css for that dialog, so that it lands with the correct style, that way we won't have problems with the ui freeze, right? [11:59] nessita, or do we I simply hardcode the colors etc (orange links for example) and we file a bug to move to a qss later? [11:59] mandel: no css yet, we have 2 weeks for IU freeze, but only one for feature freeze [12:00] mandel: no css at all, ie no colors for now [12:00] mandel: gatox will take care of that since he's already adapting the windows installer qss to sso [12:00] nessita, ack [12:00] nessita, when is feature freeze? this week? [12:01] mandel: we can land stuff for FF until this Friday [12:01] mon is "testing" day for us [12:01] tue is packaging day [12:01] (actual freeze is on the 16th) [12:02] ack [12:04] nessita, so, what I'll plan is the following, get creds dialog for landing tom morning after the second chat with lisettte, then move to the ssl certificate one asap so that we can get two reviews from here by friday and try to land that one then [12:04] all this with no qss [12:05] mandel: what about leaving the cert dialog for next week, and having this week the simplest whole "flow" working for this Friday? by the simplest flow I mean: [12:05] nessita, all that will include the code to store the creds in the keyring and fake 'launching' the dialogs when needed, later we can look into css and the spawning of the process, or is that done already? [12:05] mandel: what's "all that"? (not sure what you mean [12:06] nessita, all that, means the previous message [12:07] mandel: let me clarify what I mean with "the simplest flow": [12:07] nessita, but knowing the deadlines, lets do the simple process where we have, webclient tries to connect, gets a 407, we show the dialog and store the creds in the keyring and uses them [12:07] mandel: WAIT :-) [12:07] nessita, sorry, I'll wait, I think we are stepping on each others conversation :) [12:07] * mandel hates irc race conditions :P [12:08] mandel: so, I'm proposing to have, ASAP, the following scenario working: [12:09] * a user has a proxy setting of the form someuser:somepassword@foo.com [12:09] * the user opens the controlpanel and everything works, no prompt and no nothing, but the whole proxy experience works with at least the settings above [12:10] * next step, the same as before ^ but querying the user for the credentials from the UI [12:10] * next step, the above plus cert validation handling [12:11] nessita, sounds reasonable for this week, does that fit in the overall deadlines? [12:12] mandel: just to understand you... is reasonable the first scenario? the first 2? the whole 3? [12:13] nessita, I think following those steps and ignoring freaky friday to get the first 2, last one is more of a pain [12:13] nessita, that will be implementing it for libsoup and qnetwork [12:14] mandel: right. So cert invalid handling can be a exception, for example. Can you focus on having the first two by this Friday? and yes, I should give you the spawner tomorrow [12:15] nessita, yest, I will put 120% of time on making sure we got to friday with those 2 and correct integration tests with squid [12:16] mandel: nice! anyways, please let's validate this with alecu, though I already talked about this with him [12:16] nessita, sure, lets talk with him after the stand-up that way we can align everyones work [12:17] mandel: right [12:19] hello mandel ☺ [12:19] were you at FOSDEM last weekend? [12:19] JanC, no, I did not managed to buy a flight ticket to brussels :( [12:20] not enough time to booked it and find someone to keep the dog.. [12:21] well, there was one U1 developer from Italy I think [12:22] JanC, hm.. vds or maybe teknico [12:24] working or worked on mobile-related stuff? (I only catched part of a conversation of him with Allison at the dinner) [12:25] JanC, yes, one of those two :) [12:25] ☺ [12:27] JanC, how did it go? lots of work? [12:28] FOSDEM was great as always, and the availability of the new "K" building made it even better [12:28] good morning! [12:28] no more stands in the narrow corridor where also everybody has to pass to attend talks in the Janson auditorium [12:30] and a brand new 2nd large auditorium for main track talks [12:31] JanC, makes sense.. I guess it was a lot of work, is the ubuntu booth getting bigger? [12:31] JanC, I know czajkowski was around there, did you guys catch up? [12:31] ralsina, morning! [12:31] good morning mandel! [12:31] the booth wasn't big, but we had some interesting conversations with other people etc. [12:31] mandel: any reviews you need? [12:32] ralsina, not at the moment :) [12:33] ralsina, although if you have time, can you find out what changed that vroke the tests of the windows isntaller in jenkins? [12:33] ralsina, the set up did not last long before we broke it :( [12:33] mandel: will do [12:33] mandel: e.g. http://coworking.betagroup.be/ offering us free meeting space in Brussels ☺ [12:33] JanC, nice! I guess is hard for ubuntu, other distros have more presence in fosdem [12:34] and asking us for Ubuntu posters! [12:34] JanC, are there ubuntu posters? [12:34] mandel: we have about 1000 left I think (but they are in dutch!) [12:35] JanC, haha I'm not surprised :) [12:35] really, the design is open and available somewhere, you can get some translated & printed yourself if you want ;) [12:36] (that's what we did) [12:36] JanC, I'm glad you guys go over there.. we should pay more attention to fosdem.. but well we do what we have time to do ;) [12:39] also, several Ubuntu developers are also Debian developers, e.g. there was a Debian guy giving a talk about multiarch, and he referred to the fact that it was easier/faster to get that implemented in Ubuntu than in Debian, so Ubuntu did get some good presence in such ways (thanks to Steve Langasek who did a lot of work on multiarch and is both a DD & an UD) [12:40] mandel: there is a guy from the Launchpad team who attends FOSDEM almost every year, and comes with us to the dinner ☺ [12:40] JanC, nice, I did not know that.. [12:44] and yes, I met Laura (and also Lydia from Kubuntu/KDE) [12:44] and Dav iey [12:45] to be honest, I think Canonical could do more to raise presence at FOSDEM ☺ [12:47] JanC, I think is a matter of asking for it, explain the use, etc.. also if you mention other ocmpanies are making the effort.. [12:48] * mandel hates his ISP [12:48] mandel: the roundtable about community development-related things that Laura was part of opened some eyes for some other distros ;) [12:49] locoteams, ambassador programs etc. [12:50] gatox: added another comment to https://code.launchpad.net/~diegosarmentero/ubuntuone-client/network-detect/+merge/88545 (found more leftovers of NM) [12:50] apparently OpenSuse & Fedora came to the (for them) surprising conclusion that they have more funding than Ubuntu locoteams organising a booth ;) [12:50] nessita, ack [12:50] nessita, when on N trying to run the gtk tests I'm getting the following: *** Running GTK test suite for ubuntu_sso *** [12:50] The specified reactor is not supported. [12:51] nessita, did we change something, I though that on N we could use the glib reactor to run the tests.. [12:51] (and I'm not asking for more funding for the locoteam/booth; I think having more developers actually present is much better) [12:51] mandel: now we depend on the gi reactor, which is only available for Linux P [12:52] mandel: so you should run the tests in P. Good news is that the whole suites are run with a single command: ./run-tests [12:52] nessita, and.. how are we going to run-tests for the old distros? [12:52] JanC, yeah, I think it too, getting developers to be more reachable is a better practice [12:53] mandel: we're not for now, after FF we will analyze that [12:53] mandel: we had to move to gi in other layers... so we'll analyze backport after FF [12:54] nessita, ack [12:54] Hello. Seems as if Banshee or Ubuntu One liks to misbehave once again. I have bought music before via Banshee, but today I got the message "Your purchased music folder is not subscribed. New purches will not download to this computer." What the heck does that mean and how can I solve this? [12:54] nessita, I get the same message with P and I have the ppa present, is there a new package that needs installing? [12:55] rye, can you give a hand to Mrokii [12:58] Oh, and I should mention that the "subscribe"-button in Banshee does nothing. [12:58] mandel: if you have our nightlies ppa, and you have everything up to date, it should work. Could you please paste the output of: apt-cache policy python-twisted-core ? [12:59] nessita: python-twisted-core: [12:59] Installed: 11.1.0-1ubuntu1~precise1 [12:59] Candidate: 11.1.0-1ubuntu1~precise1 [12:59] Version table: [12:59] *** 11.1.0-1ubuntu1~precise1 0 [12:59] 500 http://ppa.launchpad.net/ubuntuone/nightlies/ubuntu/ precise/main amd64 Packages [12:59] 100 /var/lib/dpkg/status [12:59] 11.1.0-1 0 [12:59] 500 http://archive.ubuntu.com/ubuntu/ precise/main amd64 Packages [12:59] mandel: that sounds about right. What's the error you're having? [13:00] nessita, the exact command and output: [13:00] mandel@ironman:~/Projects/Canonical/ubuntu-sso-client/auth-proxy$ ./run-tests [13:00] *** Running GTK test suite for ubuntu_sso *** [13:00] The specified reactor is not supported. [13:01] mandel: what does apt-cache policy ubuntuone-dev-tools says? [13:02] mandel, looking [13:03] nessita: ubuntuone-dev-tools: [13:03] Installed: 3.1+r56-15~precise1 [13:03] Candidate: 3.1+r56-15~precise1 [13:03] Version table: [13:03] *** 3.1+r56-15~precise1 0 [13:03] 500 http://ppa.launchpad.net/ubuntuone/nightlies/ubuntu/ precise/main amd64 Packages [13:03] 100 /var/lib/dpkg/status [13:03] 2.99.2-0ubuntu1 0 [13:03] 500 http://archive.ubuntu.com/ubuntu/ precise/universe amd64 Packages [13:03] mandel: you do not have any env variable set specifically for PYTHONPATH? [13:04] mandel: looks like the devtools in the PYTHONPATH may not be the one released in our PPA [13:05] nessita, looking into that, maybe at some point I changed that, thx for the info I'll look into the system setttings [13:05] ack [13:05] nessita, probably at some point I installed manually the package and that is the issue [13:05] mandel: sounds like it [13:06] nessita, and it is, I just check /usr/local :) [13:06] ;-) [13:06] Mrokii, sorry for the delay, please open ubuntuone-control panel and see whether the music folder is subscribed [13:07] mandel: I advice never do a local installation. Either use a special PYTHONPATH=something ./run-tests, or wait for the official package (if you can wait, of course) [13:08] nessita, is probably me being stupid and testing the setup.py of devtools wrong [13:08] mandel: ah, to test a setup.py, you can use: ./setup.py install --prefix=/tmp/foo [13:09] rye: Where in the control panel do I see which music folders are subscribed? [13:09] nessita, that why I said be being stupid hehe I have some bad days, you know me ;) [13:09] ah, je [13:09] Mrokii, in folders tab [13:10] rye: I only a tab called "Cloud Folders", but that only shows a dbus-error. [13:11] Mrokii, oh, that may be the reason why it does not work. Could you please run the following in the terminal -- u1sdtool --status ? [13:12] rye: That command gave me "Failure: dbus.exceptions.DBusException: org.freedesktop.DBus.Error.NoReply: Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken." [13:14] Mrokii, ok, could you please check what is inside ~/.config/ubuntuone/syncdaemon.conf file? [13:14] nessita: need any reviews yet? You mentioned on friday you may need some early today :-) [13:15] ralsina: not yet, sorry, doing some reviews myself :-) [13:15] nessita: cool. Just wanted to fill some slots ;-) [13:15] Rye: This is in the file: [13:15] [bandwidth_throttling] [13:15] read_limit = 2097152 [13:15] write_limit = 2097152 [13:15] on = False [13:15] [__main__] [13:15] files_sync_enabled = False [13:15] ralsina, nessita do you have squid installed? can you confirm where is the bin located? [13:16] mandel: nopes, I don't [13:16] mandel: not yet,can have it in 2' [13:16] Mrokii, could you please remove files_sync_enabled line - that's what causing the syncdaemon to shut down [13:16] ralsina, are you running P or O? [13:16] rye. Sure. [13:16] incidentally, we were discussing this bug with nessita just 5 minutes ago [13:16] mandel: O [13:16] nessita, no worries, I'll look at this with ralsina :) [13:16] ack! [13:17] ralsina, superb! can you install squid and late me know the path [13:17] mandel: downloading... [13:17] ralsina, 'cause if it is what it is I'm going to cry [13:17] rye: Do I need to restart UbuntuOne? And if so, how? [13:18] rye: Hm, okay, it seems "Cloud Folders" works now. Showing my the "purchased music"-folder. [13:18] Mrokii, close the control panel and reopen it - sd should start working again [13:19] mandel: squid install in process. Slooooooooow process [13:19] duanedesign, ^ possible reason when users get "folder is not subscribed" and subscribing does not work in banshee - filesync is disabled [13:19] mandel: /usr/sbin/squid [13:20] ralsina, putos!!!! so, in P from alpha2 it is in /usr/sbin [13:20] ralsina, wait, what? [13:20] mandel: same in O [13:20] ralsina, can you run squid without sudo? [13:21] mandel: checking [13:21] mandel: fail because it can't read squid.con [13:21] rye: thanks for the heads up [13:21] conf [13:22] ralsina, oh, but you can call the bin, on P you can't [13:22] mandel: note: on P at least, squid starts by default when installed [13:22] Sorry,on O [13:22] mandel: permissions of the binary? [13:22] ralsina, yeah, it looks like that, they changed them in one of the updates.. [13:23] ralsina, which fracks up the integration tests [13:23] mandel: oh, fun [13:23] mandel: maybe you can add yourselfto a group or something [13:23] ralsina, oh, no fun no fun [13:23] mandel, hi, lisettte just came to me asking about the design of a proxy dialog [13:24] mpt, morning! yes, tell me :) [13:24] mandel, so I was wondering, why is Ubuntu One dealing with proxies at all? [13:24] mandel, is this a Windows-only thing? [13:25] rye: Thank you. It seems to work now, even though Banshee still shows that error-message. [13:25] mpt: making u1 work with proxies is one of the most requested features wehave [13:25] mpt, uh.. you should talk with ralsina or nessita about this :) [13:25] rye: Yep, the album I purchased is definitely downloaded. [13:25] mpt: hi there [13:25] mpt, I'm too low in the chain :) [13:25] ralsina, I'm not disputing that it should work with proxies. I'm wondering why Ubuntu One needs any UI for it. [13:25] mpt: we're implementing proxy support on Linux and Ubuntu [13:25] Mrokii, you might want to restart banshee for it to re-connect properly, and by the way, is Purchased Music in cloud folders enabled? [13:26] mpt: to request credentials if the user requires authentication [13:26] mpt: sorry, i the proxy requires it [13:26] ralsina, do you think that every program on Ubuntu that connects to the Internet should have its own proxy authentication dialog? [13:26] rye: Yep, it is. And I did restart Banshee after that. From looking at my download-usage it's clear that something is downloading the music to the local folder.. [13:26] mpt: as of right now, every one does [13:27] really? [13:27] mpt, I'm tested it with diff browsers [13:27] mpt: chrome and firefox have each one [13:27] ralsina, Chrome and Firefox are cross-platform and (afaik) don't assume the existence of NetworkManager [13:27] mpt, some gnome apps do work with out it if you provide the proxy url as username:password@domain [13:27] mpt: so is u1 [13:28] mpt, but when the url does not have does, either they do no support auth proxies or they have their own dialog [13:28] hi nessita, sorry, I'm not ignoring you, my mind is just temporarily boggled :-) [13:28] rye: Though I have a feeling as if it's the UbuntuOne-client itself rather than Banshee, downloading the music, as the album I purchased doesn't show up in Banshee while the tracks are downloaded. [13:29] mpt: is ok, I'm just letting you and ralsina have the conversation [13:29] nessita, what do you mean by "Linux and Ubuntu"? Are you targeting Slackware or something where NetworkManager isn't necessarily present? [13:29] mpt: my mistake (is my subconscious denying windows). I meant Windows and Ubuntu [13:29] ah [13:29] mpt: there are users that don't have NM installed, or are connecting thrugh something else (for example wvdial + 3g modem) [13:30] Holy crap. [13:30] mpt: for example, when we assumed NM was always there, we had lots of bug reports about u1 saying "you are not connected to the internet" [13:30] mpt: we do hit a ton of corner cases ;-) [13:31] ralsina, we have those bug reports for Ubuntu Software Center too ... we haven't paid much attention to them [13:31] mpt: besides what ralsina says, from the conversations I've read (but I haven't been participating actively in those), seems like in the Ubuntu desktop there is no generic UI to handle entering proxy auth info [13:32] mpt: since proxy support is our big new feature, we didn't wat something as simple as a missing proxy auth dialog to block users [13:32] nessita, mpt, ralsina FYI https://bugzilla.gnome.org/show_bug.cgi?id=646354 [13:32] mpt: if/when there is a generic proxy credentials dialog, we can just remove ours [13:33] Gnome bug 646354 in Network "Network proxy has no proxy authorization" [Normal,New] [13:34] nessita, ralsina: Could you be tempted into implementing it in NetworkManager (and depending on it) instead of rolling your own? :-) [13:34] i.e. fixing that Gnome bug [13:34] mpt: not this week ;-) [13:34] mpt: and we need it by friday [13:34] Does Evolution have its own proxy authentication dialog? [13:34] mpt, ralsina: since we need it to also work on windows, I'm not sure is a possibility [13:35] nessita: we may make it a windows-only thing [13:35] nessita: but not now [13:35] ralsina: yes, but that would put a burden in our tasks that will make us drift from our roadmap [13:36] nessita: indeed. I amnot talking about this cycle but about some vague foggy future into which my vision doesn't reach ;-) [13:36] Okay, I'm just going to throw something at a wall, then I'll come back and look at the design problem [13:36] ralsina: also, I understand that we want to prioritize that users have the same experience in both OS's, so not sure how this impact that [13:37] nessita: well, in proxy this is relativesince the config is alreadydone very differently (it's system dependent) [13:37] rye: Anyhow, even though Banshee doesn't Banshee doesn't recognise the downloads (yet), at least getting my music-files work, so Thank you for your help. [13:37] mpt: sorry if this upsets you, was not the intention [13:38] ralsina: yes the config is different from system to system, but having the same dialog in both will hide that from users, no? [13:38] nessita: the dialog is just for the credentials [13:38] yes [13:38] and for ssl cert validation failure [13:39] nessita: right [13:40] I think having the same proxy authentication dialog across platforms would pale into insignificance compared with not having to authenticate at all because you had already authenticated using the system-supplied dialog in another app :-) [13:40] mpt: agreed. [13:40] But until we have a system-supplied dialog, I guess you need a layout for it, yes? [13:40] mpt, mandel: I was under the assumption we already had a layout? [13:41] If so, then I've forgotten what the design question was [13:41] mpt, ralsina I'm working with lisettte about it, that is why mpt found out :) [13:41] mpt: ah, ok [13:41] ralsina, mpt: yes, we already have the UI designed and implemented [13:41] It was something about errors, perhaps [13:41] e.g. if you enter the wrong password [13:41] mpt, yes, it was exactly that [13:42] ok [13:42] mpt, more if there was a correct way to notify the user, take into account people with disabilities etc.. [13:42] mpt: I see now that I should have talked about this more with you guys before starting the feature work. [13:42] mandel, Ubuntu, Windows, and OS X all have a way for a window to request attention. [13:43] mpt: will try to be more cooperative in the next one :-/ [13:43] ralsina, no worries, I'm annoyed at the power company, not the electricians [13:44] mpt, hm, we where looking at something like this: http://ubuntuone.com/7NOpuQsbRnpAgb1OXfgxFr [13:44] mpt, so we wanted to know if there was/is a policy about doing this, if the error color is in the theme etc.. but I think is something you should discuss with lifeless [13:44] or lisettte ;-) [13:44] mpt, sorry lisettte [13:45] ok [13:45] mpt, autocomplete failure hehe [13:45] when she's back from lunch [13:45] (I have long wondered why IRC clients aren't smart enough to autocomplete to the most recent speaker, rather than the alphabetically first) [13:46] mpt, because humans are lazy and implement a good idea in a lazy manner [13:46] because alphabetic is deterministic [13:48] ralsina, if there is an authentication failure, can you tell whether it's the login name or the password? [13:49] mpt: No [13:49] ok [13:52] mpt, I'm off to lunch, but it you need any examples of what other apps do let me know because I have everything setup for this tests [13:52] * mandel lunch [14:00] mpt, I think it makes a lot of sense to have proxy settings and credentials to be directly associated with each different network configured in network manager. [14:01] mpt, but looking at NetworkManager trunk I see that "proxies" is a big item in the projects TODO file. [14:03] alecu: agreed, only nutcases like my have network-independent proxy settings ;-) [14:03] s/my/me/ [14:04] ralsina, there seems to be some add-on script for NM to have different proxy settings... but it's not standard, and it seems that it needs manual poking in a config file. [14:05] mpt, ralsina: http://cgit.freedesktop.org/NetworkManager/NetworkManager/tree/TODO#n371 [14:05] ^ that's the Proxies item in NM TODO [14:07] Now I appreciate a little more [14:08] and here's the script to set the gnome/kde proxy according to the NM network: http://marin.jb.free.fr/proxydriver/ [14:09] proxies are frightening [14:11] (and inherently broken) [14:16] hola team [14:20] hi there briancurtin! [14:20] hoooola briancurtin! how was the flight back home? [14:20] hola dobey! [14:20] dobey: shall we disable the test suite run in sso nightlies builds? they are failing due lack of gi reactor [14:21] it was a very long day, but i made it home in one piece so it wasnt that bad [14:21] briancurtin: did you manage to get alfajores? [14:22] nessita: yes! i chose chammas for this time. i haven't eaten any yet [14:22] nessita: they are failing on everything but precise? [14:23] dobey: they are failing on everything, including precise... they are built in a machine without our PPA, no? [14:23] (let me check( [14:23] )) [14:23] no [14:24] ah, a dep is missing on precise. [14:24] dobey: hum, you're right, failures are for not-precise, but somehow my precise sso package is not updated [14:24] dobey: python-zope.interface? [14:25] no. gir1.2-gtk-3.0 [14:25] ah no [14:25] right === dduffey_afk is now known as dduffey [14:25] bu :/ [14:25] dobey: you fix or shall I? [14:25] i'll fix [14:26] ack [14:28] and we'll have to just disable it building on some of the older distros now :( [14:29] dobey: building or test suite check? [14:29] building [14:30] dobey: aren't the gi deps available in < maverick? [14:30] no [14:31] dobey: no? (every time you say no without an explanation of why not a kitten dies :-P) [14:31] and there's no guarantee it will work on maverick or natty [14:31] nessita: you asked a yes or no question. i don't think "no they aren't there" needs more explanation does it? :) [14:32] dobey: perhaps you could mention which package is not there, I know gi.Soup is, for example [14:32] at least in < natty [14:32] soup is probably fine [14:32] dobey: Gtk is the issue? [14:32] but gtk/webkit will be problematic [14:32] I see [14:32] ok, let's put time in that after FF/UIF [14:34] well, maybe we could ship just the qt ui or something. not sure. all this stuff will need lots of compatibility testing [14:35] yes [14:36] at least sso is it's own thing [14:36] the music store stuff is going to be really painful [14:44] alecu: would you know if I can check how many clients are connected to a MockWebServer? [14:45] nessita, you may be able to count them from the factory that creates protocol instances [14:46] nessita, I'm not sure if the web server sites that we use expose that in a clear way. [14:48] alecu, mandel, dobey, nessita, gatox, briancurtin: standup in 12 minutes! [14:48] ralsina, ack [14:48] uh [14:51] ralsina, ack [14:51] * nessita wrote notes [14:57] * mandel also has notes [14:58] it is cold. [15:00] me [15:00] me [15:00] me [15:00] me [15:00] me [15:00] alecu, mandel? [15:00] me [15:01] me too [15:01] gatox: go! [15:01] DONE: [15:01] Sprint week, learn about syncdaemoon, propose some branches about cp, sso and the installer. Start with sso refactoring, and play a lot of board games with alecu and briancurtin :P [15:01] TODO: [15:01] Finish with SSO Refactoring [15:01] BLOCKED: [15:01] No [15:01] nessita, go [15:01] DONE: sprint! awesomeness of awesome, very productive, briancurtin is a SD master now (!) [15:01] TODO: make controlpanel use the sso webclient to remove dependency on qt4reactor and the be able to provide an ubuntu package. Reviews for gatox. [15:01] BLOCKED: nopes [15:01] NEXT: briancurtin [15:01] DONE: sprint all last week, learned a lot about syncdaemon, proposed a fix for bug #824252 [15:01] TODO: more investigation on bug #820350 to get the readonly/readwrite situation corrected [15:01] BLOCKED: no [15:01] NEXT: ralsina [15:01] Launchpad bug 824252 in ubuntuone-control-panel (and 1 other project) "Should export an interface to validate udf paths (affects: 1) (heat: 6)" [Medium,Triaged] https://launchpad.net/bugs/824252 [15:01] Launchpad bug 820350 in ubuntuone-client "WIndows: either set_dir_readwrite and/or set_dir_readonly are not doing what they should (affects: 1) (heat: 6)" [High,In progress] https://launchpad.net/bugs/820350 [15:01] DONE: mac interview, video lens poking and asking, reviews, lots of tiny things. TODO: u1-installer call (currently in it), fix stderr (hopefully), tech leads call BLOCKED: no NEXT dobey [15:01] λ DONE: releases [15:01] λ TODO: gwibber again finally [15:01] λ BLCK: None. [15:01] mandel [15:01] DONE: First implementation of the creds dialog (from friday). Chatted with lisette about it and ways to improve it. Started doing some integration tests for it and turns out you need sudo to run squid3. I'll be looking on how to solve this in u1-devtools, if adding the user toa userwgroup works I'll delay the fix so that we meet the deadlines. [15:01] TODO: Look into the squi3 issue. Make web-client work with user:password@domain. [15:01] BLOCKED: no [15:01] alecu, go go [15:01] DONE: SD sprint in Córdoba [15:01] TODO: debug ping url authentication issue, work on SD proxy tunnel [15:01] BLOCKED: no [15:01] NEXT: None [15:01] COMMENTS anyone? [15:02] nop [15:02] apparently, it is cold :P [15:02] everyone: this Friday is the last day to land features into our trunks [15:02] it is bloody cold [15:02] (after that, we'll need a FFE) [15:03] so, code your features and seek reviews! [15:08] nessita, dobey,alecu, mandel: if any of you needs a FFE, now's a god time to start asking for it ;-) [15:08] all time is god time for me [15:09] ralsina: hum... I thought FFE are asked after FF? [15:09] ralsina: can we do it before that? [15:09] nessita: I think it's before, but I may be wrong, since I never asked for any [15:09] nessita: you can ask before if you know you will slip. or at least, you can prepare everything to ask on day of freeze i guess [15:09] right [15:11] ralsina: not sure if you have read this: https://wiki.ubuntu.com/FreezeExceptionProcess (I haven't read it completely :-)) [15:17] trivial review: https://code.launchpad.net/~dobey/ubuntuone-dev-tools/fix-lint/+merge/91667 [15:21] * gatox lunch === gatox is now known as gatox_lunch [15:22] dobey: approved [15:26] grazi [15:26] prego [15:30] alecu: would you remind me *again* if we'll need the qt4reactor for controlpanel in windows? (once the webclient is replaced by the sso's) [15:31] nessita: yes, for IPC, IIRC [15:31] nessita, we will need it, yes. Under windows we'll need a reactor as long as we use PB for IPC [15:32] nessita: btw, i set my u1client gi branch back to needs review on friday, as it seems tarmac is hitting some issue with it still; and i'm not sure what it is exactly :( [15:32] ralsina, alecu: right, right, right, righ [15:32] t :-) [15:32] I need that tatoo-ed in my arm [15:32] dobey: have an error we can debug? [15:32] nessita: it's causing tarmac to exit with a Unicode error [15:33] booooooooo [15:33] dobey: and without gi reactor is not failing? :-/ [15:33] is scary [15:33] dobey: unicode error somehwere in our source files? [15:33] i don't think it's an issue with the reactor itself [15:33] nessita: no, a unicode error inside tarmac [15:34] oh, yum [15:34] dobey: do we have a plan to "fix" that? [15:35] fix the u1client branch, or fix tarmac somehow to avoid whatever's causing this? [15:35] dobey: right, but if the unicode error is not in u1client, how can we fix that in there? [15:36] nessita: the unicode error is caused by u1client's terminal output probably [15:36] gatox_lunch, nessita please take a read about links in QLabel and gtk themes: https://pastebin.canonical.com/59494/ [15:36] dobey, also, if you have an idea about the above, your input is welcome [15:37] mandel: may I ask what for? [15:37] mandel, i'll read it in a few minutes..... finishing lunch [15:37] nessita, I was working with the auth dialog and noticed that a link was showing blue and not orange which is what the theme was giving [15:38] nessita, so lisettte ask who could indeed confirm that this was due to qt miss behaving a little [15:38] nessita, there is nothing I'm going to do related to this, but certainly you need to know about it for the work you are doing in sso [15:39] ugh [15:39] indeed [15:41] mandel: I would not worry about coloring for now... I'm pretty sure we can override that with a custom qss [15:42] hrmmmmm [15:42] nessita, yes, I'm not worrying, I'm fowarding you the info so that you know what you will have to deal with, I'm not writing a single code for that I expect the work from gatox to deal with it [15:43] mandel: perfect :-) [15:43] mandel: thanks [15:43] you know [15:43] maybe it's not my branch that's the issue [15:47] try one more time [15:49] dobey: +1 === gatox_lunch is now known as gatox [15:51] bugger, a lock [15:52] nessita, your branch that uses build_signed_iri has not landed yet, right? [15:52] nessita, your u1cp branch I mean. [15:52] nessita, can I take a look at that? [15:52] alecu: no, I'm cleaning it up, will propose in 0.5 hr [15:52] alecu: yes, let me push (only one test case missing) [15:52] functionality is all in place [15:52] nessita, great. [15:53] alecu, the squid integration tests broke with the last alpha because squid3 is present in /usr/sbin which is a pita. It can be executed as a non root user, so just make sure that you can find it in the path [15:53] nessita, I'm interested in the part that passes the "parameters" arguments for oauth signing, and whether we should also pass the parameters in the iri [15:53] alecu, I don't think we should focus on fixing that atm [15:53] mandel, cool [15:53] ugh [15:54] ubuntuone-client stable-2-0 is spewing log messages to console during tests :( [15:55] alecu: lp:~nataliabidart/ubuntuone-control-panel/use-webclient (all tests are GREEN, but one is a TODO) [15:55] hrmm, unity alt+tab is showing evolution with 2 windows, but they are the same window (and both are selected at the same time in alt+tab) [15:55] fun [15:55] nessita, great, thanks. [15:57] alecu, are you working on adding support to webclient for user:password@domain urls? I'd like to add that as the first step to see auth proxies done [15:57] alecu, I think that webclient should be smart enough to parse that [15:57] mandel, I'm not working on that, no. It's not on my plans. [15:58] alecu, ok, then that is mine! [15:58] alecu, don't touch it! [15:58] mandel, you mean "proxy" urls or "destination" urls? [15:58] mandel, right now we have no use for "destination" urls like that. [15:58] alecu, proxy urls, aka I go to system settings and set the jost as user:password@domain [16:12] ugh, double fail this time [16:16] alecu, it looks like ubuntu_sso/utils/webclient/gsettings.py should be smart enough to parse the case in which we have user:password@domain, having setup a fake auth proxy in my system when running gsettings list-recursively org.gnome.system.proxy I get the wrong info [16:16] alecu, host is not split and therefore we don't get the right info, do you agree that I touch that code to do the correct thing? [16:17] alecu, example of the wrong output: http://paste.ubuntu.com/831541/ [16:19] mandel, yes, I agree with fixing that. Please add me as reviewer if you make a branch for it. [16:20] alecu, doing it atm [16:20] greeat [16:25] alecu: is the branch helpful? I'm pushing the latest fixes [16:28] nessita, it was useful, thanks. [16:36] alecu, ralsina: may I have a review? https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/use-webclient/+merge/91676 [16:36] will have lunch now! [16:36] brb [16:36] nessita: por supus [16:36] * nessita -> lunch [16:37] queued [16:40] webm0nk3y just showed me this: http://www.stgraber.org/2011/09/25/using-arkose-for-development-and-packaging/ looks like a great tool for setting up temporary dev environments [16:43] alright, need to grab some lunch; bbiab === beuno is now known as beuno-lunch === beuno-lunch is now known as beuno [16:58] * nessita is back [16:58] dobey: is merged! yey! [17:00] davidcalle: ping [17:00] Chipaca, hey [17:01] davidcalle: hi! [17:01] davidcalle: I don't know if ralsina told you about the v0 video search api at all? [17:02] Chipaca, davidcalle: I mentioned the searches you mentioned before the last time [17:02] ok, let's chat a little :) [17:02] Chipaca, if it's the one named "pyvore" I'm already using it. [17:02] davidcalle: PM coming up [17:06] * alecu notices that pyvore is Chipaca's blog [17:06] alecu: Chipaca is cool and modern and postson tumbler now ;-) [17:07] also, chipaca.flavors.me (but not using that for anything, it might be broken and all :) ) [17:08] oh, it's http://flavors.me/chipaca now [17:21] nessita, FYI I sent you a public url with the unattachable image [17:21] looking [17:24] mandel, ralsina, alecu, lisettte: shall we mumble about this proxy UI issue (multiple mocks)? [17:24] nessita: which issue? [17:24] I'm available if needed [17:25] lisettte: well, perhaps I'm missing a piece of info, and context, but we now have another mock for the dialog sent from matthew... so I would like to talk about that, to clarify a bit [17:26] nessita, wait, there is a diff one? [17:26] mandel: I thought you mentioned that? [17:27] mandel: perhaps I misunderstand, but from your email I read that we have new mocks different from the originals [17:27] nessita: i am doing the correct mock now; i can either finish it today, or jump on mumble and finish it tomorrow [17:27] lisettte: I prefer mumble now, since we already have the UI implemented with the originals mocks [17:28] nessita, I'm in mumble, but lets be quick since it nearly EOD for lisettte and I [17:28] ok [17:28] * mandel hates timezones.. [17:29] ralsina: if you can, we're in U1 - Desktop [17:29] nessita: going [17:31] nessita: well it's merged. i don't know if 'yay' is the right expression ;) [17:37] nessita, fyi all my branches are ready for review........ [17:37] nessita, i'm working in the refactoring now [17:44] nessita: it merged, because i sort of cheated :-/ [17:45] gatox: nice [17:45] dobey: oh yes? how you cheated? [17:47] nessita: i changed the last bit of the test command to "; make maintainer-clean" instead of "&& make clean" to see if tarmac would still crash; but if tests failed and it didn't crash, it would still land. though the tests shouldn't fail as they passed everywhere i've run them so far (including by hand on the tarmac instance itself, as the tarmac user) === mpt_ is now known as mpt [17:53] well, at least we have a working music store in ubuntu precise now. :) [17:54] dobey: yay! [17:54] yay! [17:55] dobey: yey! [17:55] ok, EOD for manuel! [17:55] catch you all tom! [17:57] bye mandel! === mpt_ is now known as mpt [19:06] alecu: ping (I mean here, not there) [19:09] nessita, pong [19:09] whee, come on banshee. [19:09] alecu: any idea why we still have a SyncTimestampChecker in ubuntu_sso/utils/__init__? [19:11] nessita, it's still used by syncdaemon [19:11] ah, ok [19:11] nessita, when we move syncdaemon to use the webclient for webservice calls we will get rid of that too. [19:11] ack [19:12] alecu: any advice how to properly fix bug #927788? [19:12] Launchpad bug 927788 in ubuntu-sso-client "When import ubuntu_sso.utils.webclient, a reactor gets installed (affects: 1) (heat: 6)" [High,In progress] https://launchpad.net/bugs/927788 [19:12] alecu: twisted.web.* is installing a reactor, so is from twisted.internet import reactor [19:15] nessita, I'm able to reproduce it, but I don't understand just yet why it's happening [19:15] alecu: we have imports (at module level) of the form: [19:15] from twisted.internet import defer, reactor [19:15] from twisted.web import http [19:16] all those install a reactor [19:16] (if there isn't one installed) [19:16] nessita: looks like sso is failing on precise now due to failing tests as the proxy bits seem to need something which isn't installed [19:16] dobey: hum... proxy bits should be ignored if deps are not there, even for tests. Have a trace? [19:16] https://launchpadlibrarian.net/92119605/buildlog_ubuntu-precise-i386.ubuntu-sso-client_3.1%2Br846-33~precise1_FAILEDTOBUILD.txt.gz [19:16] nessita, in trunk I don't see any "from twisted.internet import defer" [19:16] nessita, this one I see: "from twisted.web import http" [19:17] sorry [19:17] alecu: this is sso, no? [19:17] I don't see any "from twisted.internet import reactor" [19:17] there are plenty reactor imports [19:17] alecu: grep for "from twisted.internet import defer, reactor" [19:17] ;-) [19:18] alecu: ^^ i guess whatever tries to run gsettings with subprocess.Popen, should be trapping OSError and raising a more useful exception, as well :) [19:18] nessita, ok, I see some of those, but I don't see it imported in the webclient. [19:18] dobey, good point, thanks. I'll take a look. [19:19] alecu: well, the simple fact of importing "from ubuntu_sso.utils.foo" willl load the whole utils modules in mem? [19:19] so tcpactivation will be "read", afaict [19:19] alecu: anyways, the http imports may also be not-helping [19:19] alecu: anyways, besides moving those imports to not-module-level, do you see a cleaner solution? [19:20] nessita, I don't think the whole utils module should be loaded... let me do some IRL [19:22] ok, have a working branch [19:22] is aweful [19:23] nessita, so: "from ubuntu_sso import utils" does not load tcpactivation in mem, as I expected. [19:23] alecu: how are you testing that? inspecting sys.modules? [19:24] nessita, I've put a import pdb; pdb.set_trace() in tcpactivation.py just before the "from twisted.internet import ..." [19:25] alecu: what if you do "from ubuntu_sso.utils.webclient import WebClientError" [19:25] alecu: right, I just confirmed inspecting sys,modules [19:26] alecu: anyways, if someone import utils.ipc or utils.tcpactivation from other project, we'll get the same issue [19:26] nessita, right. And tcpactivation needs twisted, so it's fine there. [19:26] alecu: and the reactor is imported also in utils/webclient/txweb.py [19:27] nessita, and txweb needs twisted too [19:27] alecu: well, is not fine to have the default reactor installed when we need to install the qt4 one [19:27] nessita, we should find some other way to do it in ipc.py only [19:27] alecu: my point is that if you install the qt4reactor after just loading some modules in mem, you will get reactoralreadyimported [19:28] which, I think, is a problem. A module, just because is being imported, should not install a reactor [19:28] nessita, that's why the qt4reactor is always installed before importing other modules. [19:28] alecu: well, not always [19:28] nessita, it's the same with other customs reactors afaik [19:28] nessita, "must be always installed" [19:28] alecu: for sure, I use qt4reactor as example [19:28] alecu: did you review my branch in the controlpanel? that shows the issue [19:29] alecu: try executing my branch on windows, you will get a reactor already imported (and I changed nothing regarding that) [19:29] nessita, so, in all the examples of other custom reactors they get installed before importing other modules. [19:29] alecu: what examples are you referring to? [19:29] nessita, I've not reviewed it yet, but I'll trust that it breaks. [19:29] nessita, the examples in the twisted docs [19:31] alecu: so, when would you say we should fix this? in particular for my controlpanel branch. From my POV, a python library should have no side effects for just importing it. I do trust you about the twisted doc, but I don't think that's clean === vds` is now known as vds [19:32] dobey: did I miss the traceback in this reactor talk? [19:33] nessita: i guess so [19:33] https://launchpadlibrarian.net/92119605/buildlog_ubuntu-precise-i386.ubuntu-sso-client_3.1%2Br846-33~precise1_FAILEDTOBUILD.txt.gz [19:34] nessita, so: we either import the reactor inside each function that uses it, or... [19:35] nessita, we change our main script not to import any stuff before the right reactor is selected. [19:35] alecu: that's my original question to you. The "or" part, I don't know what or [19:35] alecu: if we choose the latter, we need to do that for every client from sso.webclient [19:35] alecu: I prefer (though not love) delaying imports inside sso [19:36] oh, there, a mguillech, look [19:36] nessita, it seems that the only part of the webclient that needs this fix is the http import. And that's fixable. [19:36] facundobatista: I'm here [19:36] alecu: and the reactor in txweb, no? [19:36] mguillech, so, you have two machines [19:37] nessita, txweb is going away soon [19:37] alecu: I'm proposing branch, but wanted to confirm with you the "solution" [19:37] yes, one desktop and one laptop [19:37] mguillech, you put a file in one machine, it uploaded, still didn't appear in the other one? [19:37] alecu: remember we said we would live it JIC? [19:37] alecu: leave* it :-) [19:37] nessita, right. And it's not being used right now at all, right? [19:37] dobey: so tarmac has not gsettings stuff? [19:38] facundobatista: right, it only syncs when I upload a file from a specific machine. It then starts up the sync process, uploading the recently created file and removing/renaming/whatever other pending files [19:38] I can see all the changes on the web correctly [19:38] as far as I can tell [19:38] mguillech, mmm... what? sync process isn't "started up", unless you starting the client everytime [19:39] mguillech, can you reproduce it? if yes, let's put logs in DEBUG mode and rocanrol [19:39] facundobatista: sync process as 'sync action' rather than 'a sync daemon being inserted on the process table' :-) [19:39] facundobatista: did you check he does not have an unsuubscribed UDF in the laptop? [19:39] facundobatista: can you point me to where that debug flag is set, please? [19:40] nessita, so: leave txweb as is, since we are not using. And let's move the "from twisted.web import http" inside each method that uses them [19:40] nessita, (with a comment, explaining why we've moved them there because otherwise the reactor would be imported) [19:40] alecu: doing that now, also moving the rector from tcpactivation and ipc since that will collide with u1client when we make it use those helpers [19:41] nessita, is the above good for you? [19:41] alecu: yes, already done [19:41] nessita, great. [19:41] mguillech, http://pastebin.lugmen.org.ar/7245 [19:41] nessita, btw: any idea how we should prevent this from happening again? It's something that bites us each cycle :P [19:42] alecu: from my POV, is a bug in twisted reactor management. I would suggest as "fix" for us (more than fix is a workaround), to never import reactor (and .web) at module level [19:42] nessita: this isn't tarmac. it's nightlies. [19:43] * alecu is thinking of a script that gets run after all tests, that tries to import each module in our code, and makes sure that the reactor is not imported [19:43] nessita: i added the dependency and started another build now. but i guess we need better error messages in the case it's not available [19:43] dobey: oh right (expensive context swicth). So, no gsetting in the nigthlies machines? is that ok or we need another source depends? [19:43] dobey: agreed [19:43] alecu: +5! :-P [19:44] facundobatista: got my suggestion? [19:44] nessita, yes, thanks [19:44] nessita: ideas for alecu's suggestion? [19:44] oh [19:45] i ready that wrong :P [19:45] dobey: your question made me shiver! [19:45] err, read that wrong [19:45] i need a new keyboard. [19:46] alecu: that would be easy to do, assuming there's an easy way to get a list of all the modules [19:46] alecu: it's also a hard thing to fix, when it does fail. :( [19:47] facundobatista: alright, client is logging on both machines [19:47] mguillech, in DEBUG mode? great! put a file in one side, and if after ~10m it didn't replicate on the other side, grab both logs and send them to me [19:49] facundobatista: yes, in DEBUG mode. Log file is $HOME/.cache/ubuntuone/log/syncdaemon.log, right? [19:49] mguillech, yeap [19:50] mguillech: It seems that u1cp opens nautilus directly. I would make a symlink called ~/bin/nautilus pointing to your faourite file manager ;-) [19:51] ralsina: dirty workaround :-D [19:51] ralsina: thanks [19:51] ralsina: it does? [19:51] mguillech: will be fixed in a later release once we switch to the QT UI [19:51] dobey: looks like it. I am in KDE and it stillopens nautilus, while xdg-open uses dolphin [19:52] ralsina: at least in Qt I know of a QDesktopServices that will open a file/folder in a cross platform way :-) [19:52] mguillech: could you file a bug and assign to me? username is ralsina [19:52] ok, will do [19:52] ralsina: that's weird [19:53] ralsina: I guess that urls can't be seen due to the same issue, or is it something different? [19:53] mguillech: could be [19:53] ralsina: ie, when clicking on 'Official Support' link within a button [19:53] mguillech: that opens firefox here [19:54] ralsina: code suggests it does gtk_show_uri(), which should open dolphin or whatever in kde. but maybe the mime types config is messed up? [19:54] ralsina: I use firefox sometimes on this machine [19:54] dobey: I suspect that's not heavily teted outside GNOME/Unity [19:55] facundobatista: okay, I got the file downloaded after ~ 4 minutes [19:55] ralsina: given we don't properly support anything else, sure :) [19:55] dobey: exactly [19:56] mguillech, great [19:56] it should be faster, though [19:56] facundobatista: I've deleted one file in my laptop to see if it takes the same amount of time to be deleted on the other side [19:56] facundobatista: I guess :-) [19:57] mguillech, normally delays are in the notifications... in the internal servers, and client to server to client, not mattering much about the operation [19:58] facundobatista: file was removed in less than a minute. Odd enough, last time I tested that it won't delete the file for a couple of minutes. I wonder whether bandwith here is screwing all these tests up [19:58] mguillech, it shouldn't... bandwith is not really used for unlink... the message is really slow [20:03] EOD for me! see you tomorrow! [20:04] nessita: looks like tests pass now on precise. but lint is failing. looks like another missing dep perhaps :) [20:04] dobey: we're closer, want me help you? [20:05] nessita: where does "dbus.mainloop.qt" come from exactly? [20:05] ralsina: how do I reference you in a ticket I'm filing? I'm on http://askubuntu.com/questions/ask [20:06] dobey: python-qt4-dbus (from the top of my head, let me confirm) [20:06] mguillech: I don't thik there is a shortcut there. Just put a regular link [20:06] dobey: yes [20:06] ah, it's an so extension [20:06] ok, so i got it right then [20:07] dobey: also, could you review when you have some time? https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/gicrazyness/+merge/91708 [20:07] ralsina: I meant, you told me that the bug should be assigned to you (and you gave me your username for that purpose). Where in the page am I able to do that? Perhaps I misunderstood what you said [20:07] nessita: it says "crazyness" in the branch name! [20:07] dobey: YEAP [20:07] mguillech: just give me the bug # I will assign it to me :-) [20:07] k, got you [20:07] nessita: i fear. [20:08] dobey: you'll love it! [20:09] *darn*. askubuntu won't let me log in with my SSO account [20:09] mguillech: it should... [20:09] mguillech: what error are you having? [20:10] nessita: no errors, just that SSO is not a Log In option within http://askubuntu.com/users/login :-) [20:10] nessita: I have stack exchange, google, facebook, etc. No SSO though :-( [20:11] mguillech: click on more options [20:11] hum, no [20:11] mguillech: choose "login with launchpad" [20:11] nessita: well, I'm using my LP account now [20:12] nessita: thanks! [20:13] mguillech: you're welcome. Yes, that UI is not good... [20:13] ralsina: http://askubuntu.com/questions/101976/links-wont-open-in-dolphin-using-u1-client-on-kde [20:13] mguillech: thx [20:14] nessita: not a big deal, is just that I'm used to having the SSO log in link inn every Ubuntu page out there :-) [20:15] alright; will take a quick break === briancurtin is now known as briancurtin-lunc [20:19] ralsina: can I have a trivialish review, please? https://code.launchpad.net/~nataliabidart/ubuntu-sso-client/delay-reactor-import/+merge/91712 [20:21] have a good day everybody, thanks for your help [20:29] nessita: looking [20:29] uh.... it's very late! [20:29] * alecu runs to the kinder === alecu is now known as alecu_away [20:32] nessita: and sso package built on precise finally! [20:35] man, having copyright year in individual source files is annoying [20:35] nessita: +1 on the trivialish [20:35] dobey: you rock! next, conquer the wold [20:35] world* [20:35] working on it. [20:35] dobey: can we do something better re: copyright? [20:36] ralsina: thanks! [20:36] probably. i'll think about how to do it better. of course, it's something that will probably require legal/ceo approval [20:37] nessita: i also +1ed the reactor delay branch [20:37] dobey: ack and thanks [20:37] dobey: didn't you love gicrazyness? [20:37] i am weeping [20:37] looking at it now :) [20:38] ugh. we were actually instantiating empty objects to pass around in signals?! [20:38] dobey: what do you mean with "empty objects"? [20:39] dobey: the tests? [20:39] in* the tests [20:39] oh, those are tests aren't they [20:39] but yes, still :) [20:39] yeap [20:39] yes, using object() is dirty [20:39] why did it have a gobject.TYPE_PYOBJECT [20:40] dobey: that arg was a dict, what would you use instead? [20:40] dobey: how can you tell a gtk signal that a param is a python dict? [20:40] oh. TYPE_POINTER [20:40] hum [20:41] TYPE_POINTER???? sheesh [20:41] but it looks like you removed that from the signal [20:41] perhaps that's a simpler fix? (though I like removing an used param) [20:41] dobey: the param was not being used [20:41] so I chopped it off [20:41] * nessita likes to chop things off [20:42] nessita: hrmm; that breaks the API though; but i suppose this isn't used outside of control panel, so not a huge issue [20:42] dobey: not used, and soon to be chop off as well :-P [20:42] well, not that soon [20:43] what's with the xvfb -a? [20:43] dobey: xvfb -a will use another port if the default one is used, which may be the case when running two xvfb [20:44] dobey: I like that better instead of adding a fixed delay in between [20:44] ok [20:45] i would avoid the if check for XVFB though [20:45] dobey: why? === briancurtin-lunc is now known as briancurtin [20:45] dobey: I think we want to support people running the suite without it? [20:46] nessita: do we? doesn't it pop up a million windows without it? [20:46] dobey: after the fix that was breaking the gtk suite in P, no :-) [20:46] did you apply that fix to cp already as well? [20:46] dobey: a single windows is open and closed extremely fast [20:47] or do you need to add that fix here as well? [20:47] dobey: not yet... I have to. Anyways, I was trying to mimic what we had before: the `which xvfb-run` will give you "the empty string" if not installed [20:47] dobey: so, today we "support" running tests without xvfb, I wanted to maintain that [20:48] I was not planning add the fix in this branch... shall I? [20:48] nessita: ok. can you set XVFB_CMDLINE="" as default then perhaps? [20:48] sure! [20:49] nessita: how big of a change is the fix? 1 line? [20:49] dobey: I haven't analyze it, but should not be that much [20:50] nessita: if it's only a couple lines, it's probably better done in here [20:50] dobey: XVFB_CMDLINE="" added and pushed [20:50] let me analyze the other fix [20:50] nessita: so that it's done, and we don't forget, and don't break peoples' machines if they don't have xvfb :) [20:51] dobey: good news: the fix is 0 lines :-) (remove the use of xvfb and run the suite, no windows were invading the desktop) [20:52] lol [20:52] ok [20:53] dobey: confirmed by looking at the code that the ui is hide and deleted. Will improve that code though, is ugly [20:53] *very* ugly [20:53] * nessita hits herself [20:53] nessita: so; do we need the -r gi for the non-gtk, non-qt, non-dbus tests? [20:54] dobey: yes, because the webclient is imported from sso, and it uses soup [20:54] dobey, nessita, alecu_away: I am taking a break, I will work a couple of hours late tonight, so feel free to tell me about reviews here (I have logs) [20:54] gi.Soup [20:54] gatox: you too [20:54] So bye! [20:54] nessita: ok; and the dbus tests? do they need to be run without gi for some reason? [20:55] ralsina, ok! [20:56] nessita: hrmm; also you're not doing --gui for the gtk+ tests in this branch. should you? and is that perhaps why it's not opening windows? [20:56] dobey: yes, that code has not been ported to gi... it uses the old gobject libs... and since there are cross imports between the GTK ui and the dbus_service module, I explictely put it to run with the glib reactor (even though is the default, may not be in the future) [20:56] dobey: let me add the --gui [20:56] nessita: hrmm; so the dbus tests import the static gtk or gobject bindings? [20:56] dobey: yes (the static gobject) [20:57] ah ok [20:57] nessita: btw; you do know about the short form of the argument, right? (-r instead of --reactor=) [20:57] dobey: yes [20:57] I like long forms betters, I tend to forget what the short one mean [20:57] it seems odd to me to mix the two styles when they aren't necessary :) [20:58] I can change that [20:58] where did I mix it? [20:58] it's not a big issue. just curious why you aren't use -r :) [20:58] dobey: wait, where should I be using --gui? for the GTK_TESTS_PATH? [20:58] well -i and -p are being used instead of --ignore-paths and --ignore-files [20:58] right [20:59] nessita: yes, no? should the gtk tests not use --gui? [20:59] dobey: we never used --gui in the past for gtk tests [20:59] ok [20:59] I can add it... [20:59] well if it's not necessary, that's fine [20:59] ack [20:59] i just thought it was :) [21:03] nessita: well +1 for me, even though it doesn't replace the control panel with a nice gtk3 version ;) [21:03] dobey: I consider doing that as a freaky friday. May do it after freezes... [21:04] dobey: all pushed in that branch fyi [21:05] aww, but i'd want it in precise ;) [21:08] dobey: is that considered a feature or a bug fix? :- [21:08] :-D [21:08] yes :) [21:09] yes? [21:09] it was an "or" question [21:11] but not exclusive or :) [21:11] it's more like binary or [21:12] dobey: so, can a change like that be landed after FF without a FFE? [21:13] probably not [21:13] who decides? or who shall I ask about that when in doubt? (for this case and for any other) [21:13] and definitely not, after ui/string freezes [21:13] release team would decide i guess [21:14] and i am pretty sure for precise, it'd be a "sorry suckers, you have to wait for quetzal" [21:14] right [21:14] another question, not related: [21:15] dobey: so, in an incoming branch the controlpanel suite will import, for a given test suite, the following: [21:15] from ubuntu_sso.utils.webclient.tests import BaseMockWebServer [21:15] hmmm [21:16] dobey: I need to install that tests module with setup.py and make packaging-dailies recipe generate two binaries, one regular and one -dev that will install the tests/ module? [21:16] and then do the same with the "real" sso package [21:16] and make the sso-dev a source dep on controlpanel? [21:16] python-bzrlib.tests - distributed version control system - testsuite [21:16] i guess we should do like that [21:17] dobey: any rationale to choose .tests instead of -dev? [21:18] it has precedence? [21:18] dobey: precedence meaning... something special in packaging jargon? [21:19] meaning it has been packaged that way before [21:19] ack, fair enough [21:19] precedence means "previous occurrance as example of proof" [21:20] dobey: right, I call that "jurisprudencia" :-p [21:20] i think all the python-foo-dev packages are for development files for compiled extensions [21:20] heh [21:20] which according to wikipedia, in english that is http://en.wikipedia.org/wiki/Case_law [21:21] google translates precedence to prioridad [21:21] dobey: right, that is why I asked [21:21] (I also read precendence as priority) [21:21] heh [21:21] it also lists "precedencia" [21:22] :P [21:22] which in spanish is kinda priority as well :-P [21:22] so, do i get a second monitor [21:24] hmm [21:25] doh! [21:26] how did i miss control panel. [21:26] sigh; and you know, i didn't get to look at gwibber today either. meh [21:26] and i am hungry [21:32] dobey: nooooo [21:32] look at gwibber! :-) [21:35] hmm; but should i get a second monitor [21:45] ok, I'm off [21:45] bye all! === alecu_away is now known as alecu [22:32] cheers all === dduffey is now known as dduffey_afk === dduffey_afk is now known as dduffey === dduffey is now known as dduffey_afk