[07:44] <CFHowlett> does ubuntuone for iphone also run on the ipad?
[09:28] <mandel> dobey, ralsina will update the proposals
[09:30] <mandel> morning all!
[09:40] <JamesTait> Good morning all! :D
[10:08] <ralsina> mandel: no I won't
[10:08] <ralsina> and good morning!
[10:13] <mandel> ralsina, I forgot the I :P
[10:13] <mandel> ralsina, ups hehe
[10:13] <ralsina> mandel: hehe
[10:14] <ralsina> mandel: nice work on the creds-qt script!
[10:14] <mandel> dobey, I meant 'dobey, ralsina I will update..'
[10:14] <mandel> ralsina, I need to fix that one, there is some refactoring needed so that we have a windows and linux and provide the correct qss on windows
[10:14] <mandel> ralsina, I've ran it a looks ugly.. very ugly, also, we need an icon from lisettte
[10:15] <ralsina> yes
[10:15] <ralsina> however, it's not terribly crucial that the proxy credentials dialog is pretty
[10:15] <mandel> ralsina, also I founf bug 961027 with the release you sent
[10:15] <ralsina> and it's a two-line fix, copied from the usoc gui main
[10:16] <ralsina> mandel: yes
[10:16] <mandel> ralsina, yeah, is simple, but I like things to be nice :)
[10:16] <ralsina> I also got a more misteryous one about using "_ " as a formatting character
[10:16] <ralsina> which I have not reported yet
[10:17] <mandel> ralsina, he, it removes the _ because it takes them like an action?
[10:17] <ralsina> In any case, I need to hack on the qss loading so it becomes parametric, so we can do things like $separator_color in it instead of having 60 #abcdef in it
[10:17] <ralsina> mandel: I suspect we are sending %_ somehow to the logger
[10:18] <ralsina> mandel: or something similar
[10:18] <mandel> ralsina, also, I've set up a auth proxy on windows and I'm not getting the creds auth asked.. so I think that QNetworkProxyFactory.setUseSystemConfiguration is not really working..
[10:19] <ralsina> weird, I did get that before
[10:19] <mandel> ralsina, I guess I have to set the proxy in the NetworkAccessManager to use QNetworkProxy.Default
[10:19] <ralsina> I will do some IRL and reporting today I guess
[10:20] <mandel> ralsina, can you test that because afaik in my system it is no using the proxy, and it is a 1 liner :)
[10:20] <ralsina> mandel: not right now, but yes
[10:21] <mandel> ralsina, can you give me some background about this bug 958938
[10:22] <ralsina> mandel: I think that's why the creds are not requested. See https://pastebin.canonical.com/62693/
[10:22] <ralsina> mandel: I just got that with an authed proxy
[10:24] <mandel> ralsina, hm.. let me take a look
[10:53] <mandel> ralsina, keyring is buggy.. I'm really considering doing my own lib for that called secrets with a sync and async api
[10:54] <ralsina> mandel: sure. But is there a way around that for today? ;-)
[10:54] <mandel> ralsina, I think so :)
[10:54] <ralsina> mandel: yay
[10:54] <ralsina> ok, time to be a dad for bit, then walk, then breakfast. Will be back in about 90 minutes
[11:29] <gatox> good morning!
[11:29] <mandel> gatox, morning!
[11:29] <mandel> gatox, did you fix the bug?
[11:30] <gatox> mandel, yes! :D i couldn't be more happy :P
[11:30] <mandel> gatox, what was the issue>?
[11:30] <urbanape> Hey, all. Will probably be out this morning. Lex still has pink-eye, and I need to run him to the doctor. He'll likely stay home, and hopefully Amber can take the afternoon off.
[11:31] <gatox> urbanape, ack
[11:32] <gatox> mandel, it was some problem between some part of the code not clean and the package not being generated correctly.... and mmy brain completely dry to look at the same thing for a long time :P
[11:33] <mandel> gatox, weird..
[11:33] <gatox> mandel, yes....... the good news, is that it was an important bug and now it's fixed
[11:33] <gatox> mandel, i'm reading your blog..... good post
[11:34] <gatox> about pyqt twisted reactor
[11:34] <mandel> gatox, evil qtreactor..
[11:34] <gatox> agree
[11:46] <nessita> hello everyone!
[11:48] <mandel> nessita, morning!
[11:49] <gatox> nessita, hi
[11:50] <nessita> hola mandel, gatox
[11:50] <nessita> ralsina: hello! you around?
[11:51] <gatox> ralsina, when you are around, look at this: http://ubuntuone.com/4ae2jDxfp5LKhpXYKeYTIT :D \o/
[11:52] <nessita> gatox: what did you install to get performance indicators?
[11:53] <gatox> nessita, let me remember
[11:54] <gatox> nessita, i look for "indicator" in the software center..... and install the system load indicator
[11:55] <nessita> crap, I still need the reviews to move forward with the releases
[11:56] <nessita> ralsina: I need your reviews ASAP, please :-)
[12:10] <gatox> nessita, mandel do you know which is the dbus api to publish a file and get the link? it wasn't really clear for me yesterday
[12:10] <mandel> gatox, from dbus or the rest api?
[12:11] <gatox> mandel, i was using dbus for everything
[12:11] <mandel> gatox, no idea.. wask in #chicharra they should know
[12:11] <gatox> i tihnk that might be change_public_access..... but i have to check that
[12:12] <gatox> mmm.... yes.... it seems to be that....... and then you have to listen the signal PublicAccessChanged if i'm correct
[12:12]  * gatox answering himself
[12:15] <mandel> gatox, if you hear voices, get worried
[12:15] <alecu> hello #ubuntuone!
[12:15] <gatox> mandel, crap...... :P
[12:15] <gatox> alecu, hi! o/
[12:15] <mandel> alecu, morning!
[12:17] <nessita> gatox: have you check on d-feet? that's the easiest way to digg that
[12:17] <gatox> nessita, yes..... i was checking that.... but i wasn't sure..... but i think that it is the one i mention
[12:18] <nessita> gatox: yes, that's it (I didn't read that :-D)
[12:18] <gatox> nessita, :P thanks
[12:19] <nessita> alecu: would you please help me with one review? is the update of the stable-3-0 branch for u1client, and I guess you can check if it makes sense
[12:19] <alecu> nessita, gatox: I heard you guys were needing a replacement for deferToThread for windows
[12:19] <alecu> nessita, sure, I can make reviews
[12:19] <nessita> alecu: https://code.launchpad.net/~nataliabidart/ubuntuone-client/stable-3-0-update-2.99.91/+merge/98535
[12:19] <nessita> alecu: re deferToThread, I think there is no need... right mandel?
[12:19] <alecu> nessita, gatox: did you end up putting the qt4reactor back just in order to use deferToThread?
[12:20] <gatox> alecu, yess
[12:20] <mandel> nessita, no, I fixed it :)
[12:20] <nessita> alecu: on windows we agreed we never could let the qt4reactor go since we depend on that for IPC
[12:20] <nessita> alecu: so we did not "put it back", apparently we just weren't using it right
[12:21] <alecu> nessita, ok.
[12:21] <alecu> mandel, so, how did you fix it? do you need a review on that branch?
[12:21] <mandel> alecu, and the deferToThread issue was due to this: http://www.themacaque.com/?p=1067
[12:21] <mandel> alecu, already landed
[12:23] <nessita> mandel: you know that we hve the same issue in linux, without any reactor involved? (now that you mention that)
[12:24] <mandel> nessita, what? deferToThread no working?
[12:25] <nessita> mandel: no, doing an exec on a dialog and eveything freezing
[12:26] <mandel> nessita, dammed, that is annoying, did you try to set WA_ShowModal to false?
[12:26] <alecu> mandel, the solution in your blogpost sounds right
[12:27] <nessita> mandel: but I want the dialog to be modal... and no did not try that
[12:27] <alecu> mandel, but I see a small issue that can lurk in there.
[12:27] <mandel> nessita, then we are stuck with show, I wonder why did the qt people add it
[12:27] <mandel> alecu, shoot!
[12:28] <mandel> ralsina, how did you set the proxy in you system? which version of windows is it?
[12:28] <alecu> mandel, since ._exec freezes every event going to the parent window, we *have* to make sure that nothing can be clicked on the parent window of the dialog when we use your solution.
[12:28] <alecu> mandel, otherwise breakage can occur
[12:29] <alecu> mandel, since the code that was doing the ._exec may rely on nothing changing on the parent.
[12:29] <alecu> does it make sense?
[12:30] <mandel> alecu, yes it does, in the exampe case is for the proxy_creds dialog so we have no problem, but for control panel etc.. it is indeed something to consider
[12:30] <alecu> (I've been thru similar situations in other unspeakable frameworks, and changing from modal to modeless dialogs usually brings this issue.)
[12:30] <mandel> alecu, one question, have to tested proxy support on windows? how did you do it?
[12:31] <alecu> mandel, if it is for the proxy credentials then we *might* have that same issue.
[12:31] <alecu> mandel, think of this:
[12:31] <alecu> you pop up the credentials dialog.
[12:31] <alecu> mandel, but the user clicks again on the "ok" button in the parent.
[12:31] <mandel> alecu, the credentials dialog is a diff process from any other
[12:31] <alecu> mandel, or in the "cancel"
[12:32] <alecu> mandel, oh, ok.
[12:32] <alecu> mandel, anyway: are the parent buttons being disabled while the child is showing?
[12:32] <alecu> mandel, (nevermind if they are in different processes or different mainloops)
[12:33] <alecu> I'll test it right now.
[12:33] <mandel> alecu, exactly, the problem you are stating I think is already fixed by gatox and nessita :)
[12:33] <alecu> nessita, gatox: awesome!
[12:33] <mandel> alecu, I'm setting the proxy in the windows control panel things and I'm not getting a 407
[12:33] <mandel> alecu, eventhough I do not have the creds and is a auth proxy
[12:34] <alecu> mandel, but does it connect?
[12:34] <mandel> alecu, yes, control panel gets all the data
[12:34] <nessita> mandel, alecu: the u1cp gets disabled when the u1cp knows is popping up a window, which is *not* the case for credentials
[12:34] <alecu> mandel, anyway: you should ask brian since I remember reading in the IRC logs that he got proxies working on windows.
[12:35] <mandel> alecu, I think ralsina got them working too
[12:35] <alecu> nessita, so we need to let the u1cp know when proxy credentials are being requested
[12:36] <nessita> alecu: well, is very very tricky to do that, since... how would you do that?
[12:37]  * alecu thinks
[12:38] <nessita> ralsina: I need reviews from you :'(
[12:39] <alecu> nessita, I would do it like this:
[12:40] <alecu> nessita, when the control panel creates the webclient it passes an optional parameter. This parameter is a function, that takes a boolean, and we can name the argument something like this: "requesting_proxy_credentials"
[12:41] <alecu> nessita, whenever the webclient needs to show the credentials dialog, it calls that function with True
[12:42] <nessita> alecu: any chance the webclient emits a signal when "is processing"?
[12:42] <alecu> nessita, the control panel already knows this:
[12:43] <alecu> nessita, it's from the time webclient.request is called until the deferred returned by it is fired (with either callback or errback)
[12:44] <alecu> nessita, perhaps the control panel could inherit the webclient, add that specialization to request and use that specialized class to do requests.
[12:44] <nessita> alecu: the thing is that all this happens in the backend... not in the client
[12:44] <nessita> alecu: client -> frontend I mean
[12:45] <alecu> right.
[12:45] <nessita> alecu: and right now the frontend will disable the part that is being loaded
[12:45] <nessita> but not the whole UI
[12:45] <alecu> nessita, then I think it's safe enough.
[12:45] <nessita> so if the account info is being requested to fill in the account tab, the "file sync status" widget and the get storage button are not disabled
[12:45] <nessita> neither are the 'get help button', nor the twitter or facebook button
[12:46] <alecu> nessita, the worst thing that can happen in all these cases is the proxy credentials dialog being thrown to the background
[12:46] <mandel> ralsina, pin
[12:46] <mandel> g
[12:46] <alecu> nessita, I think we should pass the window id so the proxy creds dialog is shown on top too
[12:47] <nessita> alecu: well, following what you mentioned before to mandel, the user could click on the "parent"?
[12:47] <alecu> nessita, yes, the user could click on the parent. And the parent will get focus, but the areas being requested would be grayed out.
[12:48] <alecu> nessita, and if the user clicks on a different area, another creds dialog will pop up.
[12:48] <mandel> alecu, nessita yet, he could close the window, right?
[12:48] <mandel> and the proxy creds dialog will be left there, or am I wrong?
[12:48] <nessita> mandel: yes
[12:48] <nessita> I guess so, yes
[12:48] <alecu> mandel, yes, the proxy creds dialog would be below other windows.
[12:48] <nessita> ideally we need to make the Qt dialog be 'transient for' other window id
[12:49] <alecu> mandel, *you* need to make sure that the process is terminated when the parent dialog is closed.
[12:49] <nessita> alecu: but how can mandel know that?
[12:49] <alecu> nessita, that will help with it not being "below" the parent. But not with it closing when the parent closes.
[12:50] <nessita> alecu: if a window is transient for another, the parent can not be closed (at least in gtk)
[12:50] <alecu> nessita, does this happen when they are in different processes?
[12:51] <nessita> alecu: yes, the former sso worked like that. Sso was a process, and the window id was the control panel one (another process)
[12:51] <nessita> (sso process was the same as the sso UI process back then, but I think it applies the same)
[12:51] <alecu> nessita, great then. It's probably a WM thing, so it should work on Qt too.
[12:52] <nessita> alecu: anyways, I asked ralsina some time ago help for doing this... and unless we go deep in the wm stack, there is no way to do it
[12:53] <alecu> mandel, so: in _launch_proxy_creds_dialog you should keep the reference to the process spawned by spawn_program. And in WebClient.stop you should make sure that the process is killed.
[12:53] <alecu> nessita, I don't understand what is that you want to do.
[12:53] <alecu> nessita, (I mean, with ralsina)
[12:54] <alecu> nessita, closing the child when the user clicks to close the parent?
[12:54] <nessita> alecu: no, setting a window to be transient of another window given its window id
[12:54] <mandel> alecu, nessita  that means that we need to change the runner implementation, which uses glib or qt, to return the process use for the creds dialog
[12:54] <nessita> alecu: that can not be done from qt
[12:54] <ralsina> Hello again, sorry I am late, had to do something.
[12:55] <ralsina> nessita: I did a few of the reviews, did I miss any?
[12:55] <alecu> nessita, I'll take a look at transient windows from qt later today.
[12:56] <ralsina> mandel: pong?
[12:56] <mandel> ralsina, ho are you testing auth proxies on windows?
[12:56] <nessita> ralsina: unless I'm very wrong, I emailed you asking for 3 reviews and got only one done :-/
[12:56] <alecu> mandel, probably we need to change that, yes. I think that the right way would be to return a tuple with the current deferred, and some way to close the process.
[12:56] <ralsina> nessita: ack, I did dobey's too yesterday. Will do them now.
[12:57] <ralsina> mandel: by using my public proxy which I have told everyone about a few times already.
[12:57] <mandel> ralsina, I know that, but how did you tell the system that you are using a proxy?
[12:58] <mandel> alecu, ok, I'll file a bug for that and will take care of it as soon as I find out why I cannot get proxy working on windows
[12:59] <ralsina> mandel: I configured it on chrome
[12:59] <ralsina> mandel: which gave me the system's configuration dialog
[12:59] <alecu> mandel, it sounds like an interesting bug, but let's better look at the bug list and see if there's something more urgent than that.
[12:59] <mandel> alecu, sure
[12:59] <alecu> mandel, I think today is a fine day to make plans rather than dive into code.
[12:59] <ralsina> alecu: doing a widows release soon is not low priority
[13:00] <alecu> btw: I love it when a plan comes together.
[13:00] <ralsina> alecu: no, I need to get windows bugs fixed. Sorry.
[13:00] <mandel> ralsina, I'll try the chrome path, but it seems that doing it from the control panel does not work
[13:00] <mandel> alecu, you are missing the cigar, I'm sure there is unicode for that :)
[13:01] <alecu> mandel, this is it: 烟
[13:02] <ralsina> mandel: it's the same dialog
[13:02] <gatox> ralsina, hi there!! did you see my screenshot?
[13:04] <ralsina> gatox: not yet
[13:04] <gatox> ralsina, here is the link: http://ubuntuone.com/4ae2jDxfp5LKhpXYKeYTIT (when you can)
[13:05] <mandel> ralsina, that is why I'm puzzled
[13:06] <ralsina> mandel: in fact, every app in the system picked it up.
[13:06] <mandel> ralsina, in my case, all apps but u1 are using it
[13:07] <mandel> ralsina, cause the release you made has proxy support, right?
[13:07] <ralsina> mandel: yes
[13:07] <ralsina> mandel: if it didn't have it, why would I get proxy errors on stderr
[13:09] <mandel> ralsina, I know, I'm just discarding possible reasons
[13:09] <alecu> nessita, I'm reviewing this branch of yours: https://code.launchpad.net/~nataliabidart/ubuntuone-client/stable-3-0-update-2.99.91/+merge/98535
[13:10] <alecu> nessita, if I do a diff with trunk, the only thing that changes is the version in AC_INIT in configure.ac
[13:10] <mandel> ralsina, I think is my system since I just for an error telling me it could not lauch a process within my projects dir..
[13:10] <nessita> alecu: yes
[13:10] <alecu> nessita, what kind of review you need?
[13:10] <nessita> alecu: you should diff against stable-3-0
[13:11] <nessita> alecu: is a review that should check that features there have an approved FFe, and that no unwanted changes are there, mainly
[13:11] <cparrino> Chipaca, aquarius, beuno_, joshuahoover - online via dongle + windows machine
[13:11] <alecu> nessita, right. I branch stable-3-0. I merge in your changes. I get trunk minus that small diff.
[13:11] <alecu> nessita, oh, ok.
[13:12] <beuno> cparrino, wrong channel?  :)
[13:12] <alecu> nessita, in that case, I see this, that has no FFe: Fix tunnel spawning code so that it works on windows.
[13:12] <nessita> alecu: yes, but the idea is not compare against trunk, but compare the diff against stable-3-0
[13:12] <cparrino> beuno - only channel available to me atm
[13:12] <beuno> ay
[13:13] <alecu> beuno, should we trust this guy? It looks like a fakeparrino.
[13:14] <beuno> cparrino, what's your favorite christmas pudding?
[13:14] <nessita> beuno: is easier than that, just start to argue with him, and of he plays along, he's he :-D
[13:14] <alecu> lol
[13:15] <nessita> (you can recognize me in the same way ;-))
[13:15] <mandel> cparrino, if manuel was to shit on something.. what would it be?
[13:15] <mandel> :P
[13:16] <beuno> haha
[13:16] <mandel> ralsina, I', stupid, I left sso running in a terminal from a branch I was working on..
[13:17] <cparrino> mandel, there's no such thing as "my favorite english pudding"
[13:17] <ralsina> mandel: ayayayay
[13:17] <cparrino> alecu - no jodas!
[13:18] <cparrino> :-)
[13:18] <beuno> alecu, checks out.
[13:18] <mandel> ralsina, I though it closed like the linux one..
[13:18] <ralsina> mandel: no, it doesn't
[13:18] <alecu> :-)
[13:18] <mandel> ralsina, yep, I've realized that, why is that?
[13:19] <ralsina> mandel: because it starts too slow and because there was a bug where if you start it twice quick it starts twice
[13:19] <mandel> ralsina, oh, cute
[13:19] <ralsina> mandel: and if you start it twice none of the instances reply on IPC
[13:20] <alecu> nessita, I don't see a FFe for bug #956407, yet it's in the MP
[13:21] <ralsina> nessita: reviews finished
[13:21] <nessita> alecu: right, but that bug is not a new feature... just a bug fix
[13:21] <nessita> ralsina: thanks!
[13:21] <alecu> ok
[13:22] <nessita> alecu: we still can land bug fixes until final freeze
[13:22] <nessita> alecu: which is april 12th
[13:25] <alecu> nessita, I have just seen this add_info for apport function. I did not know it even existed. I think we should have it include the proxy.log file.
[13:25] <alecu> nessita, I'll open a new bug for that.
[13:25] <nessita> alecu: awesome
[13:26] <dobey> alecu: does the proxy.log contain any 'private' info?
[13:27] <alecu> dobey, I think not, but I should double check. I just realized that syncdaemon.log is not included, so it might make sense not to include it either.
[13:27] <alecu> I'll create the bug anyway, and we can check if we should include it or invalidate the bug.
[13:27] <dobey> right
[13:28] <dobey> it would be nice if we could automagically redact things in the logs
[13:30] <alecu> dobey, perhaps by including some special sequence while printing it in the logs?
[13:30] <alecu> dobey, like: SECRET("passWd"), and it gets translated into SECRET("XXXXX")
[13:32] <dobey> alecu: depends on how we're currently doing it. but something like that
[13:33] <dobey> it would be nice if we could register regex matches in apport somehow, like r'password="(.*)"' and apport would just replace the (.*) match with "[REDACTED]" or something
[13:33] <dobey> but alas
[13:35] <nessita> alecu: any other thing to note from the branch?
[13:36] <alecu> nessita, I'm halfway thru reviewing the code.
[13:36] <nessita> alecu: nice!
[13:38] <dobey> nessita: are there any left that i need to look at?
[13:38] <nessita> dobey: reviews? nopes :-)
[13:38] <dobey> aye, ok
[13:39] <dobey> i see ralsina approved the u1client one
[13:39] <nessita> dobey: yes, but alecu is also looking at it
[13:39] <dobey> right
[13:39] <nessita> dobey: since he added most of the proxy code, I wanted to make sure the update had his approve
[13:39] <dobey> ping me when it lands, and i'll make the tarball
[13:39] <nessita> dobey: yes
[13:42] <mandel> I'm off to have lunch
[13:42]  * mandel lunch
[13:48] <alecu> nessita, +1
[13:48] <nessita> alecu: thanks!
[13:48] <alecu> nessita, should I change the status too?
[13:49] <nessita> alecu: already done, thanks :-)
[13:49] <gatox> nessita, i have a doubt regarding this issue: https://bugs.launchpad.net/ubuntuone-control-panel/+bug/944256
[13:50] <gatox> nessita, which should be the fix: no limit? a different limit?
[13:53] <briancurtin> i'm not nessita, but i would go with no limit. the great majority of users are likely to be on high-speed connections with bandwidth to spare. if this was 15 years ago i'd probably think differently (but i'd also be 12 years old)
[13:56] <nessita> gatox: is not like you need to change the default :-)
[13:56] <nessita> gatox: there is bug that is missreading the defaults from syncdaemon
[13:56] <nessita> gatox: if you look at /etc/xdg/ubuntuone/syncdaemon.conf
[13:56] <nessita> you will see that the default is:
[13:56] <gatox> nessita, ahhhhhhhh that's the bug!!... i didn't understand that from the bug report
[13:57] <nessita> [bandwidth_throttling]
[13:57] <nessita> on.default = False
[13:57] <nessita> read_limit.default = 2097152
[13:57] <nessita> write_limit.default = 2097152
[13:57] <nessita> gatox: so, in general, controlpanel should not change the defaults from the syncdaemon's default
[13:57] <gatox> nessita, the bug report just say: I installed ubuntuone-control-panel-qt on a clean system and it defaulted to limit bandwidth to 2048 kb/second.
[13:58] <nessita> gatox: rigth, so since controlpanel uses syncdaemon defaults, it means we're showing something wrong
[13:58] <gatox> ok
[14:00] <alecu> nessita, ping
[14:01] <alecu> nessita, I'm trying to install ubuntuone-client-proxy in a VM that has an older version of ubuntu-sso-client.
[14:01] <alecu> nessita, it says that ubuntuone-client will be updated (perfect) but it does not mention ubuntu-sso-client being updated.
[14:02] <nessita> alecu: what version of ussoc do you have?
[14:02] <alecu> nessita, I think that might be an error, since ubuntuone-client now uses the webclient from ubuntu-sso-client
[14:02] <alecu> nessita, 3.1+r922-42~precise1
[14:03] <nessita> wow that's old... well, in nightlies is very messy to put dependencies using revno version
[14:03] <nessita> so we're just requiring ussoc >= 3.1
[14:03] <alecu> nessita, oh, ok.
[14:03] <nessita> alecu: you should update "by hand" as well, but the production package will be requiring ussoc 2.99.91
[14:03] <nessita> so no worries there
[14:04] <alecu> nessita, so, this won't happen in the main repo, right? if it only happens on nightlies I don't care :-)
[14:04] <alecu> great
[14:04] <alecu> thanks!
[14:04] <nessita> :-)
[14:05] <nessita> dobey: did you make the tarball for protocol?
[14:06] <dobey> nessita: yes
[14:06] <nessita> dobey: buuuu you did not let me know :-)
[14:06] <nessita> I was staring at the version=2.99.91 and was confused
[14:07] <dobey> you were off playing with the kids at school :P
[14:08] <nessita> dobey: heh
[14:10] <dobey> i discovered that bzr merge-upstream does uscan for you, yesterday
[14:10] <nessita> dobey: yes it does, that's why we need to fix the watch file every time
[14:11] <nessita> dobey: so you uploaded protocol 2.99.91 to ubuntu as well?
[14:12] <dobey> i haven't uploaded it yet, no
[14:12] <nessita> dobey: will you or shall I?
[14:12] <dobey> i will
[14:13] <alecu> nessita, in the latest nightlies this file is missing: /usr/lib/ubuntu-sso-client/ubuntu-sso-proxy-creds-qt
[14:13] <nessita> alecu: oh... hum... let's see
[14:14] <alecu> nessita, and obviously sso fails when it tries to get the credentials.
[14:14] <alecu> nessita, also, control panel fails in an ugly way
[14:14] <nessita> alecu: how does it fail? when doing what?
[14:15] <alecu> nessita, (it pops up a dialog with that error, I'm trying to reproduce to write the steps)
[14:15] <nessita> alecu: yes, dialog popping up is the expected way
[14:15] <nessita> alecu: the dialog sometimes freezes, and I think that's caused by the same exec_ issue that mandel had
[14:18] <alecu> nessita, btw: the new control panel with the integrated wizard is awesome. I still don't like the cloud picture, and the text that's too bold, but it's awesome nonetheless :-)
[14:18] <alecu> nessita, gatox, ralsina: congrats!
[14:18] <ralsina> alecu: too bold?
[14:19] <ralsina> alecu: change your default font to non-ubuntu-font and it's fixed ;-)
[14:19] <alecu> ralsina, but my font is the default! :-)
[14:19] <nessita> alecu: thanks!!!
[14:20] <ralsina> alecu: ok ok ok and thanks!
[14:20] <nessita> alecu: ok, great catch about the -qt binaries, will tweak packaging dailies now
[14:21] <nessita> alecu: the whole code for those binaries is under the same namespace as the qt UI, right?
[14:21] <nessita> oh no!
[14:22] <nessita> there is a ubuntu_sso/webclient/qtnetwork.py outside ubuntu_sso/qt
[14:22] <nessita> gah this will be complicated
[14:22] <alecu> nessita, don't know about namespaces. mandel?
[14:22] <nessita> alecu: how does the webclient chooses the implementation use?
[14:23] <alecu> nessita, ubuntu_sso/utils/webclient/__init__.py
[14:23] <nessita> alecu: hum, this name is no longer valid, no? "is_qt4reactor_installed"
[14:24] <ralsina> nessita: let's do the 1:1 close to EOD so we can review how the releases went, ok?
[14:24] <nessita> ralsina: perfect
[14:24] <nessita> alecu: also, I think I tested this, and asking for QCoreApplication.instance() will give you not None even if there is a QApplication running
[14:24] <alecu> nessita, right: we should have used a different name there when we replaced the check with PyQt instead of qt4reactor
[14:25] <alecu> nessita, I know that mandel made a branch that took it out, then another to add it.
[14:25] <alecu> *to add it back.
[14:25] <nessita> alecu: trunk has it
[14:25] <alecu> nessita, right: he took it out as you or ralsina requested. Then he put it back
[14:25] <nessita> alecu: also... will you please help me with something? webclient/common.py:247
[14:25] <nessita> alecu: why he put it back?
[14:26] <ralsina> nessita: panic
[14:26] <alecu> nessita, 247, yes.
[14:27] <nessita> alecu: so... what if qt is not available, ergo de SSL_DIALOG program is not availble?
[14:27] <ralsina> nessita: he said he got different objects on both tests, so it kinda scared him into putting it back in. Worst case it's redundant, not important
[14:27] <nessita> alecu: I know we don't need to support a non-qt dialog,  but what will happen in that case? boom? nothing?
[14:28] <alecu> nessita, probably we would get a similar WebClientError
[14:28] <nessita> alecu: is there any chance to test that in the next 10 minutes? or is it too much to ask?
[14:28] <briancurtin> besides general usage testing and some of the non-auth proxy stuff i did the other day, does anyone have any particular areas they want me to look at on windows?
[14:29] <alecu> nessita, oh, wait.
[14:29] <nessita> ralsina: why did he panic? is the same instance https://pastebin.canonical.com/62722/
[14:29] <nessita> alecu: /me waits
[14:30] <alecu> nessita, we disabled ssl overrides. Now any ssl error is a webclienterror.
[14:30] <nessita> alecu: not sure what that means, sorry :-$
[14:30] <alecu> nessita, ubuntu_sso/utils/webclient/qtnetwork.py, line 85
[14:31] <ralsina> nessita: I would have to find logs but he was getting two different objects at some point.
[14:31] <nessita> alecu: seeing that line, but really not sure what that means IRL, when a user is actually using this
[14:31] <nessita> ralsina: ok, no need to do that now
[14:32] <ralsina> nessita: right
[14:33] <nessita> alecu: requesting a new nightlies build that will ship the qt UIs for proxy support only when the ussoc-qt package is installed
[14:33] <nessita> will build in 9 minutes approx
[14:33] <alecu> nessita, sorry, Lucila was telling me about the mirror that we need to replace in our car. :P
[14:33] <alecu> nessita, she got her sideview mirror "crashed" too.
[14:33] <nessita> alecu: $855 is costing me to replace my mirror
[14:33] <alecu> :-(
[14:34] <alecu> nessita, our costs ~ARS 250
[14:34] <nessita> alecu: since my car is under warranty, I can only repair it at the official shop
[14:34] <alecu> we are going to Av. Warnes in a few hours.
[14:34] <nessita> the chevrolet original mirror is $610, and the "mano de obra" $250
[14:34] <nessita> a crazyness
[14:36] <mandel> alecu, yes, I did add it back 'cause I had an issue with it, but can be remove, I can double check
[14:36] <mandel> alecu, what about namespaces?
[14:37] <alecu> mandel, I don't care about removing those bits right now. nessita had some concerns because she requested only checking for the coreapp instance.
[14:37] <alecu> mandel, but in any case, it's harmless.
[14:37] <nessita> yes
[14:37] <alecu> mandel, regarding namespaces...
[14:38] <alecu> mandel, nessita asked: "the whole code for those binaries is under the same namespace as the qt UI, right?"
[14:38] <alecu> mandel, but now she realized the answer is "no"
[14:38] <alecu> because there are some bits that are in the webclient namespace.
[14:38] <mandel> alecu, which are 'those' binaries?
[14:39] <mandel> alecu, and if it is about qt, yes, we have code under webclient
[14:40] <nessita> mandel: so, we "can't" (is complicated) code that requires qt under webclient. Not code that do not support not having qt around...
[14:40] <nessita> alecu, mandel: wanna mumble about this? I need to solve it asap to release to ubuntu, and freeze is today
[14:41] <alecu> dobey, http://pastebin.ubuntu.com/893693/
[14:41] <mandel> nessita, sure, give me 5 mins (updates are taking place)
[14:41] <mandel> nessita, I'll ping you as soon as mumble works
[14:41] <alecu> dobey, should I open a bug for it?
[14:41] <nessita> mandel: I m there
[14:41] <alecu> mandel, nessita: I'm fetching the laptop to mumble.
[14:42] <ralsina> alecu, nessita, mandel: do you need me on that call?
[14:42] <dobey> alecu: yeah. and sigh.
[14:43] <dobey> although i guess that's a bug in pygobject
[14:44] <alecu> dobey, does this mean that no notifications will be shown at all?
[14:45] <dobey> alecu: you get that on precise?
[14:46] <alecu> dobey, up to date precise with nightlies
[14:46] <alecu> dobey, in fact, let me update the rest of precise, because I only updated the nightlies.
[14:46] <alecu> dobey, I'll ping you back.
[14:48] <alecu> nessita, ralsina, gatox: you may want to check this screenshot of the error dialog: http://ubuntuone.com/0gcDwz2oNBMDOrUNUj89R6
[14:48] <ralsina> alecu: yes, I saw that, I need to fix it
[14:48] <gatox> alecu, wow!
[14:48] <nessita> alecu: yes, saw that
[14:48] <alecu> ralsina, gatox: should I open a bug for it?
[14:48] <ralsina> alecu: please and assign to me
[14:48] <alecu> ralsina, ack
[14:48] <nessita> alecu: yes please
[14:49] <gatox> alecu, it's a feature...... we really encourage users to hide the details
[14:49] <nessita> alecu: is the dialog also frozen sometimes?
[14:49] <gatox> :P
[14:51] <dobey> ooh!
[14:51] <mandel> nessita, alecu FYI waiting for updates to  finish to be able to mumble..
[14:51] <dobey> a progress bar!
[14:51]  * mandel hates libre office
[14:52] <alecu> mandel, click it off, dude!
[14:53] <alecu> (Stan: Dude, what the fuck is wrong with German people? )
[14:56] <mandel> lol
[14:56] <mandel> alecu, nessita done, opening mumble, yet should we talk after the stand up?
[14:56] <dobey> alecu: weird. so i didn't get the "Foo is uploading" notification, but no TypeError in logs either
[14:56] <dobey> it's almost done
[14:56] <nessita> mandel: sure
[14:57] <dobey> (or maybe i just missed the notification because it pops up in a bad place)
[15:00] <nessita> me
[15:00] <gatox> me
[15:00] <briancurtin> me
[15:00] <mandel> me
[15:01] <gatox> alecu, dobey ralsina ?
[15:01] <nessita> dobey: with this new ussoc packaging issue, I will not be able to do the u1client packaging and upload to ubuntu today. I'm happy to do it next time, though :-)
[15:01] <ralsina> me
[15:01] <alecu> me
[15:01] <ralsina> DONE
[15:01] <ralsina> oops
[15:01] <nessita> DONE: finished landing branches to complete the u1cp wizard. Proposed update-3-0 branches for ussoc, u1client, u1cp and u1windowsinstaller.
[15:01] <nessita> TODO: releases
[15:01] <nessita> BLOCKED: ussoc packaging may be broken
[15:01] <nessita> NEXT: gatox
[15:02] <gatox> DONE:
[15:02] <gatox> Fixed the reactor issue, couple of reviews, update the state of some issues that were already closed. Started working on Bug #944256.
[15:02] <gatox> TODO:
[15:02] <gatox> Finish with limit bandwidth bug, keep working on some issues in my queue.
[15:02] <gatox> BLOCKED:
[15:02] <nessita> urbanape: standup?
[15:02] <gatox> No
[15:02] <gatox> briancurtin, go
[15:02] <briancurtin> DONE: debugging, call, 1-1, add_to_autostart branch, some testing last night and this morning
[15:02] <briancurtin> TODO: insert add_to_autostart call in case where no wizard runs on startup, whatever else needs to be done for release, probably get back to Mac stuff
[15:02] <briancurtin> BLOCKED: none
[15:02] <briancurtin> NEXT: mandel
[15:02] <mandel> DONE: Fixed creds dialog not working on windows due to using exec_ in the dialog. Looked at proxy on windows and the diff issues we have with the proxy creds. Also found an IPC bug comming from sd in ralsinas last release.
[15:02] <mandel> TODO: Fix proxy support and proxy creds on windows. Move to next bug in the list.
[15:02] <mandel> BLOCKED: no
[15:02] <mandel> ralsina, please
[15:02] <ralsina> no notes yet,sorry
[15:02] <alecu> DONE: broken wifi while traveling meant I had to take a forced freaky tuesday. Been doing IRL of the nightlies today
[15:02] <alecu> TODO: more IRL to get the nightlies in shape
[15:02] <alecu> BLOCKED: not today
[15:02] <alecu> NEXT: ralsina
[15:02] <dobey> meh
[15:03] <alecu> NEXT: dobeeh
[15:03] <ralsina> DONE: started perf reviews, proposed --installer option, reviews, build windows for QA , manager-like-substances handled TODO: fix styling bugs on windows, herd cats, random other stuff BLOCKED: no
[15:03] <dobey> λ DONE: releases, uploads
[15:03] <dobey> λ TODO: u1client release, finish uploads, bug triage/fixing, some closed milestone reorg
[15:03] <dobey> λ BLCK: none.
[15:04] <nessita> any comments anyone?
[15:05] <nessita> ralsina: is urbanape with us today?
[15:05] <ralsina> nessita: he should be, but I am locked out of canonicaladmin until I can get back to linux
[15:05] <ralsina> so maybe I misplaced him
[15:05] <dobey> nessita: you forgot thisfred?
[15:05] <gatox> ralsina, nessita, he said earlier that he was with some problems.... let me check the logs
[15:05] <dobey> oh right
[15:05] <dobey> he's gone
[15:05] <nessita> dobey: he's off today
[15:05] <nessita> dobey: having a happy birthday I hope
[15:05] <ralsina> thisfred is off the rest of the week
[15:05] <nessita> gatox: ack
[15:05] <dobey> yeah
[15:06] <dobey> just realized, when tab complete wasn't working :)
[15:06] <nessita> heh
[15:06] <mandel> dobey, that would have never happened with gatox, he would have pinged someone else :P
[15:07] <gatox> ralsina, nessita urbanape says: Hey, all. Will probably be out this morning. Lex still has pink-eye, and I need to run him to the doctor. He'll likely stay home, and hopefully Amber can take the afternoon off.
[15:07] <dobey> hehe
[15:07] <gatox> mandel, jejje
[15:07] <nessita> alecu, mandel: mumble then?
[15:08] <nessita> alecu: ping?
[15:09] <alecu> I'm there
[15:14] <dobey> uh oh
[15:14] <dobey> ze progress bar hasn't gone away
[15:14] <dobey> and i haven't gotten any notification about the upload
[15:15] <briancurtin> brb, quick run to the coffee shop (forgot to buy coffee at grocery store :/ )
[15:16] <apol> mandel: ping
[15:16] <dobey> ah, now there is a TypeError
[15:16] <dobey> weird
[15:16] <apol> or anyone who can help me with SSO login, it broke when I moved from 11.10 to 12.04 -.-
[15:16] <dobey> broke?
[15:17] <apol> stopped working
[15:17] <dobey> can you be more specific about what isn't working exactly?
[15:18] <apol> ok
[15:18] <apol> I was accessing the dbus interface in com.ubuntu.sso in /credentials
[15:18] <apol> called com.ubuntu.sso.ApplicationCredentials
[15:19] <apol> now I don't seem  to have this service anymore
[15:20] <dobey> nessita: ^^ is this just the "old/deprecated api" getting removed?
[15:20] <mandel> apol, pong!
[15:20] <mandel> apol, how exactly?
[15:20] <nessita> dobey: one sec, mumble
[15:21] <apol> mandel: hmm... I don't know what other info to give :/
[15:22] <mandel> apol, sso has service activation, nessita is the master of this piece of code :)
[15:22] <dobey> mandel: it sounds like the old api being removed; and one needs to use /com/ubuntu/sso/blah/blah as the path now
[15:23] <apol> so... what/where is the new api?
[15:24] <dobey> apol: please be patient, and nessita will help in a few minutes. :)
[15:24] <apol> ah ah
[15:24] <apol> sorry
[15:25] <nessita> apol: https://wiki.ubuntu.com/SingleSignOn/UbuntuSsoClient (will give more details after I finish this call)
[15:26]  * apol waits... :)
[15:27] <apol> nessita: yes, that's what I was using before... maybe I need to install more packages?
[15:28] <dobey> ok, lunch time for me. bbiab
[15:29] <nessita> apol: did you read the paragraph that reads:
[15:29] <nessita> "NOTE: formerly, the ApplicationCredentials interface was implemented under the /credentials object path. That interface is deprecated and should not be used. However, current applications using this interface will be able to do so until the Ubuntu 11.04 release inclusive, since it won't be removed it until 11.10. "
[15:29]  * briancurtin chugs coffee
[15:29] <nessita> apol: so, the /credentials path is not the same as /com/ubuntu/sso/credentials
[15:30] <nessita> apol: /credentials path is deprecated, you should use /com/ubuntu/sso/credentials instead
[15:30] <apol> ah
[15:30]  * apol tries...
[15:30] <nessita> apol: the API changes a bit, so the same calls as before will not work
[15:31] <nessita> apol: please check the API in that wiki page :-)
[15:35] <apol> wonderful
[15:36] <nessita> dobey: would you please let me know if I need to do something specific to add the new dep python-openssl to ussoc?
[15:36] <nessita> (re: packaging)
[15:38] <briancurtin> ralsina: should we change SHOW_CMD (conf.py) back to False now or just right before the official release?
[15:38] <apol> this new API has a method called register which is a C reserved keyword, so it's not possible to generate the C interface of your dbus
[15:38] <apol> nessita: where you aware of that?
[15:38] <apol> It's quite a common way to deal with dbus in Qt
[15:38] <ralsina> briancurtin: I keep it True for QA builds and turn it off for release
[15:39] <nessita> apol: not sure what you mean... would you please explain a bit?
[15:39] <apol> sure
[15:39] <briancurtin> ralsina: good plan
[15:39] <apol> in Qt, to use a dbus interface, what you do is to generate the xml file of the interface using a line such as:
[15:39] <apol> qdbus com.ubuntu.sso /com/ubuntu/sso/credentials org.freedesktop.DBus.Introspectable.Introspect > ubuntu_sso_dbus_interface.xml
[15:39] <apol> then it generates a .h file
[15:40] <apol> like this http://paste.ubuntu.com/893773
[15:41] <nessita> apol: ack. So, the problem is...?
[15:41] <apol> nessita: the problem is that there's a method called "register" and you can't have anything called "register"
[15:42] <apol> I probably can find ways to workaround this, but well, it's still workarounding
[15:42] <nessita> apol: can't you generate the .h so the method names use the namespace they are in? in dbus you have hierarchy (service name, object path, interface) so you can group your method logically
[15:43] <apol> well, the method is ComUbuntuSsoCredentialsManagementInterface::register
[15:43] <apol> it's a method inside ComUbuntuSsoCredentialsManagementInterface class
[15:43] <nessita> apol: and, given that the method is under a specific path/interface, I don't see a conceptual reason to rename it... we can't rename methods considering all the languages reserved words...
[15:43] <apol> ok then
[15:43] <apol> never mind
[15:44] <nessita> apol: can you propose another name for it, that would make more sense in that context?
[15:44] <mandel> apol, what is the complies telling you?
[15:44] <nessita> such as sso_register (first name I came up with)
[15:45] <mandel> nessita, the problem that apol has is that if he wants to auto generate the api via the qdbus xml he finds a register method, since register is a key word in c++ the compiler is probably going bananas
[15:45] <nessita> dobey: for u1client, be sure to require ussoc >= 2.99.91, since the new use of webclient requires it
[15:46] <nessita> mandel: I understand that
[15:46] <apol> mandel: http://paste.ubuntu.com/893779
[15:46] <nessita> mandel: but what I say is that we can't rename our methods considering every language reserved keyword...
[15:46] <mandel> nessita, yeah.. I know, is a PITA coming from the c++ compiler
[15:47] <nessita> mandel: perhaps the generation can be tweaked and a given prefix can be added to every method name?
[15:47] <nessita> such as sso_
[15:47] <apol> register_credentials?
[15:47] <mandel> nessita, just what I was going to say!
[15:47] <mandel> nessita, stop reading my mind!
[15:47] <nessita> mandel: lol
[15:47] <mandel> apol, yes, or even just register_ to be closer to the dbus definition
[15:47] <nessita> apol: just FYI, the method goal is not registering credentials but registring a new sso account
[15:48] <nessita> oops, so many typos, sorry
[15:48] <nessita> apol: can you add a fixed prefix to every method?
[15:48] <apol> i don't really know
[15:48] <apol> I'll investigate
[15:48] <mandel> apol, qdbus does not have many params, but what about adding an extra step in the middle, grab the xml and append _ to any cpp keyword?
[15:49] <briancurtin> -sys.maxsize on prefixing methods something. i would need to have lazer eye surgery
[15:49] <briancurtin> s/something/with something/
[15:50] <apol> mandel: well, I'm seeing if qdbusxml2cpp has something like that
[15:50] <mandel> apol, are you cheking the generated files in a source control? or are you doing this as a make/cmake step?
[15:50] <apol> cmake yes
[15:52] <mandel> apol, afaik you should not be running that as a make step, but just do it the first time and then adapt the generated code
[15:52] <mandel> apol, since there is no guarantee that the machine compiling the code will have the sso dbus interface
[15:52] <apol> i know
[15:52] <apol> I have the xml versioned
[15:53] <apol> what I do compile-time is to run qdbusxml2cpp
[15:55] <alecu> nessita, ralsina, mandel: good news: nightlies u1cp, sso and sd are crossing an authenticated proxy just fine.
[15:55] <nessita> alecu: AWESOME NEWS
[15:55] <nessita> alecu: really, great great news
[15:55] <ralsina> alecu: awesome!
[15:55] <mandel> alecu, I'm updating the clean vm and will try the software center behind a proxy
[15:55] <ralsina> apol: just out of curiosity, what are you doing with it? :-)
[15:56] <apol> ralsina: I'm working on kubuntu's software center, I'm using it to login for the review and ratings service
[15:56] <ralsina> apol: great!
[15:57] <ralsina> apol: my next sunday project is better integration with KDE (maybe in a month of two)
[15:57] <ralsina> apol: but I need to refresh my C++ *a lot*
[15:58] <apol> ralsina: if I/we can help you, don't hesitate to get in touch :)
[15:58] <ralsina> apol: will do!
[15:59] <ralsina> apol: you may want to try this for a taste of things to come: https://launchpad.net/u1-toys
[15:59] <nessita> gatox: something happened with the reset password page
[15:59] <nessita> gatox: the error is no longer shown :-/
[15:59] <apol> :)
[16:00] <gatox> nessita, mmmmmm is there any bug for that?
[16:00] <nessita> gatox: opening
[16:00] <gatox> nessita, thanks
[16:01] <nessita> gatox: https://bugs.launchpad.net/ubuntu-sso-client/+bug/961315
[16:01] <gatox> nessita, thanks
[16:05] <alecu> dobey, I've tried dist-upgrading the whole Precise VM, and I still get those GI errors when notifications are to be shown.
[16:05] <alecu> dobey, the only issue is notifications not being shown; SD keeps working fine otherwise.
[16:07] <alecu> dobey, I'm creating a new bug for this... shall I assign it to you?
[16:08] <alecu> dobey, or should I assign it to the notifications GI project?
[16:09] <ralsina> I have to switch to linux to test some styling bugs, should be back in a few minutes. Keep talking to me, I have logs ;-)
[16:10] <nessita> alecu, dobey: I packaged ussoc and tested it on a clean VM (without proxy involved). Will upload as soon as dobey gives me some ack about the python-openssl package
[16:13] <apol> nessita: another question, why do you use "a{ss}" maps instead of "a{sv}"? It would make sense to me if it always were strings
[16:13] <apol> but considering that the winid has to go there, a{sv} would make more sense
[16:13] <nessita> apol: they are always strings so we can use the dbus bindings from python
[16:13] <nessita> apol: using variants from python is more complicated than what we need
[16:13] <apol> ugh xD ok
[16:14] <nessita> apol: so, the windid is expected as string and cast to int
[16:14] <nessita> casted*
[16:16] <briancurtin> u1cp/gui people: i added a call to UbuntuOneWizard.done to add SD to the Windows autostart. i need the same functionality to occur for users who don't run through the wizard (they already have creds) - i'm not sure where this should go
[16:16] <nessita> briancurtin: you added what where? :-)
[16:16] <nessita> briancurtin: not sure I understood
[16:16] <briancurtin> nessita: https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/add_to_autostart
[16:17] <briancurtin> i took add_to_autostart from u1-win-installer and added it to the current u1cp Wizard, so SD will startup automatically
[16:17] <briancurtin> but it only does so in the case where it's a first time install. if you are upgrading and already had creds, we don't currently add to the autostart, but need to find a place to
[16:18] <nessita> briancurtin: why not add that code to the installer itself?
[16:18] <mandel> nessita, I've not installed the ubuntu one control panel, that is, I'm in a clean P vm, and ubuntu-sso-client-qt is installed, is that correct?
[16:19] <nessita> mandel: unless something is pulling it in, which I doubt, you should not have -qt
[16:19] <nessita> mandel: you sure?
[16:19] <briancurtin> nessita: im not sure if we can do that - looking
[16:20] <mandel> nessita, 100% sure
[16:20] <mandel> nessita, I just added nightlies via add-apt-repository ppa:ubuntuone/nightlies
[16:20] <nessita> briancurtin: is extremely difficult to know, when there are crendentials in the system, if the user just installed
[16:20] <mandel> nessita, and did a sudo update-manager -d to make sure I have the latests of the latests
[16:21] <nessita> mandel: something in our nightlies is pulling that in... but definitely is not the case in ubuntu
[16:21] <nessita> 110% sure
[16:21] <mandel> nessita, I'll remove it then
[16:22] <nessita> otherwise the platform team would be screamiung at our faces
[16:24] <alecu> dobey, bug #961342
[16:25] <dobey> alecu: you can assign to me and i'll look into it
[16:25] <alecu> done.
[16:25] <nessita> ralsina: do you know what I need to add to the debian pacakging to deal with the python-openssl license isue?
[16:25] <dobey> nessita: just add the python-openssl dep
[16:26] <nessita> oh, a dobey is here! :)
[16:26] <mandel> nessita, nevertheless, requesting sso creds over auth proxy in software center fails with an error stating 'Authentication failed'
[16:26] <nessita> dobey: didn't you mention we needed to "grant" the license or something like that?
[16:26] <mandel> nessita, which is the 407 error from the proxy
[16:26] <nessita> mandel: aaaaarghhhh so confusing :-/
[16:26] <nessita> but I guess we'll have to settle with that
[16:27] <mandel> nessita, it does not brake in a crazy manner, but I guess that we can add a bug to return an exception with a more decent error
[16:27] <alecu> ralsina, bug #961348
[16:27] <dobey> nessita: we need to tweak the license text a bit to add an exception yes; but we'll do it after this release. anything that needs it now, just add it.
[16:27] <nessita> mandel: yes, like "proxy in the way, can't do anything else"
[16:27] <nessita> dobey: ack then
[16:27] <nessita> uploading before leaving for lunch
[16:28] <mandel> nessita, I'll add the bug and tag it with the u1-tag, should be a simple one to fix
[16:28] <briancurtin> nessita: we can add to the autostart via installer, but we should really only do it after the user has logged in. is there a good place in u1cp that would only execute right after login?
[16:28] <mandel> nessita, one questions, if this messages are shown in the ui, shouldn't they be translated?
[16:28] <mandel> nessita, not now, not another UIFe
[16:29] <alecu> hey all: I need to run some car related errands, will be back later.
[16:30] <nessita> briancurtin: yes, ubuntuone/controlpanel/gui/qt/controlpanel.py -> on_wizard_finished
[16:30] <nessita> briancurtin: but I guess is the same you had before?
[16:31] <nessita> briancurtin: there is no place to detect when there was credentials in the system *already*, but the controlpanel just installed
[16:31] <briancurtin> I added it to wizard.py. looking at controlpanel.py now
[16:32] <nessita> briancurtin: is basically the same, though I like it a bit better on the controlpanel. Also, please make sure to add a test confirming the call to the add_to_autostart is made
[16:33] <briancurtin> nessita: thanks, i'm looking into adding it here
[16:33] <briancurtin> (here = on_wizard_finished)
[16:33] <mandel> nessita, FYI bug 961357
[16:34] <mandel> ralsina, FYI moving back to windows
[16:44]  * gatox lunch
[16:48] <nessita> briancurtin: I just realized there is a much better option
[16:48] <briancurtin> i'm interested
[16:48] <nessita> briancurtin: I think you need to call to autostart inside the controlpanel's main when the --installer option was passed...
[16:48] <nessita> briancurtin: though now that I'm writing that, we still have the "new credentials" issue, no?
[16:49] <apol> nessita, mandel: I have it working now, thanks for the help
[16:49] <nessita> apol: nice!!!
[16:50] <apol> it would be cool if you looked into the register renaming though, the code looks really ugly :(
[16:50] <mandel> apol, superb!
[16:50] <briancurtin> nessita: in main via --installer does seem better, i'll look into it. i'll see what it would have to do for new creds - maybe it just stays inside the wizard's done function as well (it's harmless to call it multiple times)
[16:50] <mandel> apol, problem is, we not only have cpp clients.. and since we mostly we use python we did not see that comming
[16:50] <mandel> apol, good things is, you wont get one called if/while/try/except etc :P
[16:51] <apol> mandel: well, take it as positive criticism
[16:51] <apol> if you want people to use it, at least
[16:52] <mandel> apol, I know, I know, era broma :)
[16:52] <ralsina> mandel: thanks!
[16:52] <mandel> apol, problem is, we are very late in this cycle, but.. why don't you add a bug?
[16:53] <apol> where would I report such a bug?
[16:54] <mandel> apol, here https://bugs.launchpad.net/ubuntu-sso-client/
[16:55] <mandel> apol, go to report a bug and explain all the details, we will later deal with it (not at the moment due to the deadline)
[16:56] <apol> i'll see what I can do
[16:58] <mandel> ralsina, turns out that automatic detection on windows is 'heavier' on windows than manual, which meant that altough I had manually set up the proxy it was not used..
[16:58] <mandel> ralsina, puto windows!
[16:59] <mandel> apol, sorry for the burden, but trust me we will do our best to get it working
[16:59] <ralsina> mandel: no sure what that means. When you set to "automatic" the manual is disabled
[16:59] <mandel> ralsina, yes
[16:59] <mandel> ralsina, and is a tiny little check box..
[17:00] <ralsina> mandel: he, ok, so now you know :-)
[17:09] <nessita> dobey: ussoc 2.99.91 uploaded, if you want to move forward with u1client
[17:09] <nessita> (which must depend on that)
[17:10] <dobey> nessita: ok, thanks
[17:15] <mandel> ralsina, I found the bug on windows, applicationProxy on windows returns a hostName that is '' therefore the credentials manager complains because you cannot have '' as the key of a pwassword
[17:15] <mandel> password*
[17:22] <ralsina> mandel: WTF?
[17:22] <ralsina> oh, got it
[17:22] <mandel> ralsina, and to make it more annoying the proxy type is 2, which means that qt is not able to detect the proxy..
[17:22] <ralsina> I read that as the hostname beig "therefore the credentials manager..."
[17:23] <mandel> ralsina, that would be an awesome proxy name
[17:23] <nessita> lunchtime on this end
[17:23] <mandel> ralsina, but, yeah, proxy detection, the way we have it, is broken, but there is a fix
[17:27] <mandel> alecu, it seems that the proxy setting on windows return without a proxy, which is a proxy with hostname '', port 0 and type 2
[17:27] <mandel> alecu, it means that we might have to query the system to set the proxy in the setup
[17:28] <ralsina> gotta run for a few minutes, brb
[17:55] <mandel> ralsina, FYI windows bug fixed, setting tests and will propose
[17:55] <ralsina> mandel: awesome!
[17:56] <mandel> ralsina, I hate that qtreactor, makes debugging really hard with pdb
[17:56] <ralsina> mandel: you mean the constant "event loop already running"? That's Qt actually
[17:56] <ralsina> mandel: I just 2> /dev/null
[17:57] <mandel> ralsina, yeah, is what I do too.. on linux, but windows..
[17:58] <ralsina> mandel: yes, on windows, you just can't
[17:58] <ralsina> mandel: maybe with bash for windows you can
[17:59] <mandel> ralsina, well, is sorted so lets get this merged hehe
[17:59] <ralsina> mandel: ack
[18:02] <urbanape> oof, hello, all.
[18:03] <urbanape> briancurtin: ping
[18:03] <briancurtin> urbanape: pong
[18:03] <urbanape> heya, you up for any pairing today?
[18:04] <briancurtin> urbanape: trying to add the last half of a branch for release, then yes
[18:04] <urbanape> awesome, thanks.
[18:04] <urbanape> I'll be here. Amber's watching Lex for the rest of the afternoon and tomorrow morning.
[18:08] <briancurtin> sweet, got it working. commence testing
[18:09] <nessita> gatox: so, I'm doing the review for https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/backend-problem/+merge/98513
[18:10] <gatox> nessita, aja
[18:10] <mandel> ralsina, do you mind if I EOD and propose tom?
[18:10] <nessita> gatox: any chance you do not remove and add the main.py, but bzr move it? the way is now, the diff (and the bzr history) will show that file as  removed and a new file added
[18:10] <nessita> gatox: so, the ideal way to handle that is (in a clean branch):
[18:10] <nessita> bzr mkdir ubuntu_sso/qt/main
[18:11] <nessita> bzr move ubuntu_sso/qt/main.py ubuntu_sso/qt/main/__init__.py
[18:11] <ralsina> mandel: no, it's ok
[18:11] <nessita> and then apply your changes
[18:11] <nessita> gatox: same for the test file
[18:11] <nessita> bzr mkdir ubuntu_sso/qt/main/tests
[18:11] <gatox> nessita, in a clean branch or this one?
[18:12] <nessita> bzr move ubuntu_sso/qt/tests/test_main.py ubuntu_sso/qt/main/tests/test__main.py
[18:12] <nessita> gatox: I don't think you can do this in this one
[18:12] <nessita> since you already remove it and re add it
[18:13] <mandel> ralsina, ok, thx! tom you will have the working version, I'll ask you and alecu for reviews :)
[18:13] <mandel> all., EOD here!
[18:13] <mandel> see you tom!
[18:14] <gatox> nessita, but maybe i could revert that..... ok.... i'll try in a new branch
[18:14] <nessita> gatox: if you can revert, you're welcome to
[18:14] <nessita> I have no idea how to do revert that and preserve the changes
[18:22] <gatox> nessita, so, here it is: https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/main-moved/+merge/98703
[18:22] <gatox> nessita, i think you should accept that before, so i can merge my backend branch with trunk
[18:22] <gatox> nessita, or should i mark this one as prerequisite?
[18:23] <gatox> what do you prefer?
[18:23] <nessita> gatox: a single branch, of course :-)
[18:23] <nessita> gatox: merge your changes in that branch ;-)
[18:23] <gatox> nessita, and then drop the backend branch?
[18:23] <nessita> gatox: exactly
[18:24] <nessita> gatox: the branch originally should not have removed and re-added files
[18:24] <gatox> nessita, ok, i'll do that
[18:24] <ralsina> gatox, nessita, alecu: two very simple branches for review, just qss changes, with pictures: https://code.launchpad.net/~ralsina/ubuntuone-control-panel/fix-zion-inverted/+merge/98701 https://code.launchpad.net/~ralsina/ubuntu-sso-client/fix-zion-inverted/+merge/98696
[18:24] <nessita> gatox: we should bzr move files when we're moving then
[18:24] <gatox> ralsina, on it
[18:24] <gatox> them
[18:30] <briancurtin> ah great. not even going to try to explain it, but this add_to_autostart will be a fun merge.
[18:31] <nessita> briancurtin: enjoy! and welcome
[18:31] <nessita> dobey: would you please let me know when u1client is uploaded? I'm trying to avoid having it in source dep wait
[18:32] <dobey> nessita: what needs the new version?
[18:33] <nessita> dobey: nothing specific, but you know how much I love to depends on the latest version for consistency sake (controlpanel does depend on 2.99.91 ussoc)
[18:34] <dobey> nessita: stuff should only depends on the newest version necessary.
[18:36] <nessita> dobey: yes, I usually agree. But since we're gonna say "version 2.99.91 of U1 supports proxy" I would like the controlpanel 2.99.91 to have the same proxy support as u1client, but if u1client is not 2.99.90, that will not be true...
[18:37] <nessita> (since controlpanel web request will have proxy support, but syncdaemon will not work)
[18:37] <ralsina> Also, https://code.launchpad.net/~ralsina/ubuntuone-control-panel/u1cp-windows-styling/+merge/98704 should cause no visible changes in linux, fix bug #961229 on windows
[18:37] <nessita> ralsina: any reason why the underline was removed from the tabs when focused?
[18:37] <nessita> ralsina: visually, not in code I mean
[18:37] <ralsina> nessita: on what branch?
[18:38] <nessita> ralsina: I noticed that since a couple of days ago
[18:38] <ralsina> nessita: hmmm I have the underline here
[18:38] <nessita> ralsina: no particular branch, just IRL
[18:38] <nessita> ralsina: when tabbing into a tab?
[18:38] <nessita> ralsina: not when hovering it
[18:38] <nessita> gatox: you let me know when the branch is ready?
[18:39] <gatox> nessita, yes.... i'm testing it IRL.... because something went wrong and it wasn't working the new branch now
[18:39] <nessita> ack, thanks
[18:39] <ralsina> nessita: yes, by tabbing. But then again, I have not been in ubuntu much lately, so maybe there is something broken there. I do have them in KDE right now
[18:40] <nessita> ralsina: hum...
[18:40] <nessita> I don't have them in ubuntu...
[18:40] <ralsina> Let me switch to unity since I have to test some stuff there
[18:40] <nessita> (just double checked)
[18:40] <nessita> ralsina: nice
[18:41] <gatox> nessita, done: https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/main-moved/+merge/98703
[18:41] <nessita> ack
[18:42] <nessita> gatox: also, can you please explain in the commit message how you fix the backend getting stuck? :-) when that goes to the changelog, does not provide that much info to someone reading the release notes
[18:42] <gatox> nessita, ok
[18:47]  * alecu is back
[18:48] <ralsina> nessita: I have underlines usig 3.1+r292-49~precise1
[18:48] <ralsina> nessita: I have not, however, updated the system in a few days, I only updated u1cp and dependencies today
[18:49] <nessita> ralsina: is your system up to date?
[18:49] <nessita> ah, je
[18:49] <nessita> that ;-)
[18:49] <gatox> ralsina, is this necessary? https://code.launchpad.net/~ralsina/ubuntuone-control-panel/fix-zion-inverted/+merge/98701
[18:49] <ralsina> gatox: oh, yes
[18:49] <gatox> ralsina, in the other branch you are setting background-color: white to QWidget
[18:49] <ralsina> gatox: look at the screenshot
[18:49] <gatox> spinbox should inherited
[18:49] <ralsina> gatox: that is on sso
[18:49] <dobey> nessita: the new binary package for u1client will hold it up a bit btw
[18:49] <gatox> ralsina, ahhhhhhhh my bad
[18:50] <gatox> ralsina, only look at the code.... thought that was the saame project
[18:50] <ralsina> gatox: if I set qwidget to white here, I break all the app
[18:50] <nessita> dobey: ah, right...
[18:50] <gatox> ralsina, roger that.... i thought it was the same qss.....
[18:50] <dobey> but it's uploaded now
[18:53] <nessita> dobey: thanks
[18:54] <nessita> gatox: could you please commit with --fixes and the bug number for the branch https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/main-moved/+merge/98703 ? also, in the commit message it should specify that the fix applies to windows only
[18:54] <gatox> ok
[18:55] <nessita> (otherwise people from linux, suffering from this - like perrito - will think we fixed it for them as well)
[18:55] <nessita> which so far we did not :-/
[18:58] <gatox> nessita, i realize there wasn't a bug for this.... the bug was only for linux..... creating now
[18:58] <nessita> gatox: thanks
[19:01] <ralsina> nessita: I am not sure I understand what you mean by "I wonder why you added a new method start_from_license instead of redefining setStartId in the QWizard"
[19:02] <nessita> ralsina: this question may be a non-sense, I might be getting something wrong from the qwizard documentation... but why adding a custom method when the api provides one?
[19:02] <ralsina> nessita: if you change the startID outside __init__ you need to restart the wizard
[19:02] <ralsina> which is exactly what the new method does
[19:03] <nessita> ralsina: right.. but in my head, by doing that (providing a method in our custom wizard), you're leaking info to the mainwindow about the controlpanel widget structure
[19:03] <nessita> ralsina: so it would make more sense (to me, please correct me) to have the start_from_lincese in the controlpanel widget
[19:03] <gatox> nessita, done
[19:03] <ralsina> nessita: I could add that method to contrlpanel instead
[19:03] <ralsina> right
[19:03] <nessita> and let le controlpanel widget do whatever it considers necessary, like doing startID on the wizard and restarting it
[19:04] <nessita> ralsina: so no need to add the method in the wizard itself, given that its API already provides what is needed
[19:04] <nessita> no?
[19:04] <nessita> gatox: ack
[19:04] <ralsina> makes it awkward to test, and means I have to add a installer parameter to control_panel too
[19:04] <ralsina> or no, forget that
[19:04] <nessita> ralsina: why?
[19:04] <nessita> ok, forgotten
[19:04] <ralsina> :-)
[19:04] <nessita> I'm just saying to have: self.ui.control_panel.start_from_license()
[19:04] <ralsina> right
[19:04] <nessita> instead of accessing the "internal" (not necessarily hidden) wizard
[19:05] <dobey> bah, am hungry again
[19:06] <ralsina> nessita: usually, I don't consider UI components private (or I would call it _ui instead of ui) but sure, I will do that and keep it in mind for the future.
[19:13] <briancurtin> great, crashing test.
[19:13] <briancurtin> eh, maybe its not me. the baseclass crashes as well
[19:17] <briancurtin> is u1cp ubuntuone.controlpanel.gui.qt.tests.test_gui.MainWindowTestCase currenty passing on Linux?
[19:18] <dobey> nessita, alecu: https://code.launchpad.net/~mandel/ubuntu-sso-client/libsoup-ssl-dialog/+merge/97057 should be 'rejected' right?
[19:19] <nessita> dobey: not sure... it also has several branches depending on it
[19:19] <gatox> briancurtin, yes, they are working on liunx
[19:19] <nessita> briancurtin: yes
[19:20] <dobey> nessita: but we aren't doing ssl proxies, because nothing supports ssl to the proxy, right?
[19:20] <briancurtin> i know this works IRL but the tests crash python :/
[19:20] <nessita> dobey: right
[19:21] <alecu> dobey, right, but that branch checks ssl against the server, not against the proxy.
[19:21] <dobey> oh
[19:21] <dobey> hmm
[19:22] <alecu> dobey, and for a library that might be useful at some point if we want to allow the user to accept invalid certificates for some servers.
[19:22] <alecu> dobey, but certainly we do not want users to accept invalid certificates for *our servers*.
[19:23] <dobey> alecu: unless of course, it's connecting to a different server that isn't *our* server
[19:23] <alecu> so the branch is not needed right now, but may make sense at some point if we want our desktop apps to download say, pictures or videos or something from user defined servers. Or something like that.
[19:24] <dobey> well, anyway; it's probably good to note, that practically nobody gets ssl right, anywhere.
[19:25] <alecu> but it's gotten more right than proxies usually.
[19:25] <dobey> also; fml, dandelions grow way too fast :(
[19:25] <dobey> only because proxies are always wrong
[19:29] <briancurtin> if anyone has a minute for lp/bzr question, can you let me know how to un-screw this up: https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/add_to_autostart -- rev293 is ralsina's branch that i had applied and didn't know it would get pushed (294 used to be called something different, but i remerged onto a clean checkout)
[19:29] <briancurtin> i might just export it as a diff and just start over...
[19:30] <ralsina> briancurtin: that's what I would do
[19:34] <briancurtin> ralsina: since it depends on your --install branch, once i get a clean u1cp branch, how do i apply your branch but not have it be a part of any work i checkin/push to my own branch (so yours is dependent)
[19:35] <ralsina> briancurtin: you merge my branch and in the proposal declare it as prerequisite
[19:36] <briancurtin> ralsina: ah, so it's lp that figures out that it's not my work?
[19:36] <ralsina> briancurtin: right, the proposal does it
[19:36] <briancurtin> ralsina: cool, thanks
[19:39] <alecu> briancurtin, usually when I merge from a side branch I put the branch name in the commit, so it's easier to track. Like "merged from lp:~ralsina/project/branch"
[19:48] <ralsina> nessita: I get some strange test failures in installer-option and can't find why. Have a minute to give me a hand?
[19:49] <nessita> ralsina: sure
[19:50] <ralsina> nessita: I am getting failures in UbuntuOneWizardCloudToComputerTestCase which I have not touched
[19:50] <nessita> ralsina: have a trace?
[19:51] <ralsina> nessita: https://pastebin.canonical.com/62768/
[19:52] <nessita> ralsina: wanna push the branch that is giving you that?
[19:52] <nessita> they look like legit "test" errors... but if you have not changed anything...
[19:52] <ralsina> nessita: pushed, it's lp:~ralsina/ubuntuone-control-panel/installer-option
[19:52] <nessita> branching
[19:52] <ralsina> nessita: the tests there are a bit weird. For example, the "button behaviour" tests depend on absolute page IDs
[19:53] <ralsina> nessita: which is why I had to shift them all when I added a page at the beginning
[19:53] <nessita> ralsina: yes, they do. We can improve them by putting a name instead of the id and in the test doing the getattr in the self.ui
[19:53] <ralsina> I can do that after I get rid of these :-)
[19:54] <nessita> ralsina: at the time, I consider that a good intermediate solution between time-results, considering that I thought the page order  will not change (I don't like adding the license page if we're never showing it...)
[19:55] <ralsina> nessita: I *could* add the page in start_from_license...
[19:55] <ralsina> nessita: but that's even hackier
[19:55] <nessita> ralsina: right
[19:55] <nessita> so, no
[20:00] <nessita> ralsina: if you test your branch IRL, the test failure are reflecting IRL bugs. For example, in the cloud-to-computer page, the back button is gone
[20:00] <ralsina> nessita: hmmmm ok, I'll dig then
[20:00] <nessita> ralsina: you forgot to remove something:
[20:00] <nessita>     self.setButtonText(self.NextButton, self.next_button_text)
[20:00] <nessita> AttributeError: 'UbuntuOneWizard' object has no attribute 'next_button_text'
[20:01] <ralsina> nessita: where's that?
[20:01] <nessita> ralsina: grep for it, i got that in the test run
[20:01] <ralsina> nessita: ok
[20:01] <nessita> from the trace:
[20:01] <nessita>   File "/home/nessita/canonical/controlpanel/review_installer-option/ubuntuone/controlpanel/gui/qt/wizard.py", line 252, in initializePage
[20:01] <nessita>     self.setButtonText(self.NextButton, self.next_button_text)
[20:01] <nessita> AttributeError: 'UbuntuOneWizard' object has no attribute 'next_button_text'
[20:02] <ralsina> nessita: got that one, and that explains it. Thanks!
[20:02] <nessita> ralsina: still 2 tests failures, they may be legit
[20:02] <nessita> ralsina: my guess is that a CommitButton does not make currentIdChanged be emitted
[20:03] <nessita> but some other signal
[20:03] <ralsina> nessita: well, that's totally undocumented anyway
[20:03] <ralsina> it's an implementation detail of QWizard, really
[20:07] <gatox> ralsina, nessita eod for me..... and i really need to go to the university to an import errand..... i'll try to apply nessita's comments to the backend branch tonight
[20:07] <ralsina> gatox: ok, thanks!
[20:10] <nessita> ralsina: why changing the next button to be a commit button?
[20:12] <ralsina> nessita: so the user can't go back
[20:12] <ralsina> you can't "disagree" on a license
[20:13] <nessita> ralsina: but the way it was before, you couldn't go back either. Though I understand is more correct using CommitButton
[20:13] <ralsina> nessita: before there was no way to see the page, so I am not sure what you mean :-)
[20:15] <ralsina> nessita: you mean having a next and removing the back in the next page?
[20:16] <ralsina> I am just trying to use the features provided instead of doing things manually as much as possible
[20:16] <briancurtin> ralsina: https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/windows-autostart/+merge/98720 -- i'm confident in the branch via IRL testing, but due to some GUI tests crashing on Windows (higher up the chain that my test) i'm not exactly sure what to do
[20:17] <ralsina> briancurtin: you can run your specific tests
[20:17] <briancurtin> ralsina: even running my specific test fails due to a class higher up the inheritance chain
[20:17] <ralsina> briancurtin: using u1trial, although I am fuzzy in the details beyond running all the tests in a single file by u1trial --reactor=qt  --gui foo/bar/test_stuff.py
[20:17] <ralsina> briancurtin: oh, that sucks
[20:18] <ralsina> briancurtin: you get the same failure in trunk?
[20:19] <nessita> ralsina: yes, since we already are customizing every button layout on every page
[20:19] <nessita> ralsina: you could see the license page by manually changing the show_license to True and running it from source
[20:19] <nessita> was working great :-)
[20:20] <ralsina> nessita: hahaha, ok, I just tried to make it nicer, and use the features we have
[20:20] <ralsina> nessita: but I can revert if you really feel it's the right thing to do
[20:21] <nessita> ralsina: not really, I just did not know about commitpage
[20:21] <nessita> commitbutton
[20:21] <ralsina> commit*
[20:21] <ralsina> yeah, ok, I have the tests fixed now, need to add one for start_from_license and I'm done with it
[20:22] <briancurtin> ralsina: yep, with trunk it fails
[20:23] <ralsina> briancurtin: then if with good IRL tests it passes, the tests should not block you, I guess
[20:23] <ralsina> nessita: are you ok with landing briancurtin's page with preexistant failing tests in windows?
[20:23] <ralsina> nessita: I am starting on test-fixing on friday
[20:23] <briancurtin> ralsina: yep, i've tested on 7 and XP, locally and via a bundled installer
[20:24] <nessita> ralsina: can I see the trace?
[20:24] <ralsina> briancurtin: ^
[20:24] <briancurtin> nessita: there's no trace per se...it crashes the Python interpreter
[20:24] <nessita> briancurtin: ah, that's very likely a qt seg fault
[20:24] <nessita> briancurtin: are you patching "everything" that you should?
[20:24] <nessita> :-)
[20:24] <nessita> briancurtin: I can look at the branch to confirm that
[20:25] <nessita> what branch is it?
[20:25] <briancurtin> nessita: https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/windows-autostart/+merge/98720
[20:26] <nessita> branching
[20:26] <nessita> briancurtin: which one is the last test being reported in the terminl?
[20:27] <briancurtin> nessita: when running a clean u1cp trunk it fails - i'm not aware of any current branches that need to be applied for that, but i could have missed it
[20:27] <ralsina> briancurtin: no branch pending is supposed to fix it
[20:28] <briancurtin> nessita: it gets to test_animation_is_active (AccountPanelTestCase), but the issue is further down in some base class. the test that i added fails, then i ran the tests of the base class and they fail, and on up the chain
[20:28] <briancurtin> but that's just very early on in the tests to be run
[20:28] <nessita> briancurtin: how are you running the suit?
[20:28] <nessita> suite
[20:29] <briancurtin> nessita: i started by just doing run-tests to run everything. even when i run just the individual test i added (by doing u1trial and passing -t with the exact class to run) it crashes
[20:29] <nessita> briancurtin: are you passing -r qt4 --gui?
[20:29] <nessita> if you do not use --gui, UI tests will explode
[20:30] <nessita> booting my windows VM now
[20:30] <briancurtin> nessita: i'm 99% sure i was doing that but it's past my command history (or i closed the window). i will try again and be 100% sure to use that
[20:30] <ralsina> nessita: finished fixing installer-option. Calling dibs on your morning review slot tomorrow
[20:31] <nessita> ralsina: dibs granted
[20:31] <nessita> ralsina, briancurtin: overall, I think we can land a branch if all is green in linux, but I think we need to fix this (sooner or later)
[20:32] <nessita> so will try to debug a little
[20:34] <briancurtin> nessita: yep, i'm running with --gui
[20:34] <briancurtin> nessita: python ..\..\bin\u1trial --reactor=qt4 --gui -p"ubuntuone\controlpanel\dbustests" -i "test_linux.py, test_libsoup.py" ubuntuone
[20:36] <briancurtin> also crashes when i give it a -t to filter down to specific tests (like the one i added)
[20:36] <ralsina> nessita: yes, I filed a bug for these test failures yesterday, I think
[20:37] <nessita> briancurtin: reproduced, will give it 5 minutes of debugging
[20:39] <nessita> ja, found out what;'s the proble,m
[20:39] <nessita> but have no solution to propose
[20:40] <nessita> briancurtin, ralsina: this avoid the crashes http://pastebin.ubuntu.com/894239/, but is not sematically correct to remove that line...
[20:40] <nessita> ralsina: any idea why showing a widget makes qt seg fault?
[20:40] <ralsina> nessita: not the slightest
[20:40] <ralsina> nessita: unless we are not having a QApplication for some reason
[20:40] <nessita> will add a print before the .show() call
[20:41] <ralsina> nessita: god news are, a core dump on the crash should give useful info
[20:41] <nessita> ralsina:  this is windows we're talking about, no?
[20:41] <nessita> :_P
[20:41] <ralsina> nessita: oh, right, windows
[20:41] <ralsina> nessita: me engolosiné despues de 4 horas de ubuntu
[20:42] <nessita> ralsina: when adding these prints:
[20:42] <nessita>     271         print '\n\n\n[20:42] <nessita>     272         print '\n\n\n[20:42] <nessita> I got:
[20:42] <nessita> [20:42] <nessita> [20:43] <ralsina> nessita: ok, so. We could bisect the failure, I guess
[20:43] <nessita> ralsina: so the seg fault is a pain in the butt
[20:43] <nessita> ralsina: bisect how?
[20:43] <briancurtin> ralsina: we could get a nice dump file if i ever have time to finish https://bitbucket.org/briancurtin/minidumper
[20:43] <nessita> ralsina: every single ui tests fail with that
[20:43] <ralsina> nessita: try revision 100, see if it fails, and so on
[20:43] <ralsina> nessita: we used to have working windows tests in the past, we could find the revno that broke them
[20:43] <nessita> ralsina: I added the self.ui.show() in the past 4 weeks
[20:44] <nessita> so clearly that broken it
[20:44] <nessita> but why self.ui.show() will break it?
[20:44] <briancurtin> they weren't broken too long ago. i can run the manual bisect if you want
[20:44] <ralsina> nessita: no idea whatsoever, but why did you have to add it?
[20:44] <nessita> briancurtin: you'we welcomed to
[20:45] <ralsina> briancurtin: you can run bzr blame to see the change for that line, so we can confirm that broke it
[20:45] <nessita> ralsina: 2 reasons: several tests were doing the self.ui.show() in them and the adding a cleanup for hiding... and also, I consider that semantically is more correct to test how a widget *looks*, which requires the widget is shown, no?
[20:45] <nessita> ralsina, briancurtin: 100USD bzr blame says nessita ;-)
[20:45] <nessita> actually, nataliabidart
[20:46] <ralsina> nessita: the problem may be that we are faking something and then we are showing() something that's not a widget somewhere
[20:46]  * nessita knows she added it
[20:46] <nessita> ralsina: not in this tests
[20:46] <nessita> these*
[20:46] <ralsina> nessita: the only things that need showing are checking for geometry and isVisible()
[20:46] <nessita> we're faking backends
[20:47] <nessita> ralsina: leaving the discussion of  to show or not show appart for a while, qt should not seg fault when showing a widget, no?
[20:47] <ralsina> nessita: it usually doesn't :-)
[20:48] <nessita> ralsina: this is the trace http://pastebin.ubuntu.com/894254/
[20:48] <nessita> (what I could get from windows, of course)
[20:48] <ralsina> hehe
[20:49] <ralsina> Oh, right, old 0a9e! (not really)
[20:49] <ralsina> nessita: BTW, you probably should update your Qt/PyQt
[20:50] <nessita> ralsina: to which version?
[20:50] <ralsina> nessita: Qt 4.8.0 PyQt 4.9.2 IIRC
[20:50] <ralsina> 4.9.1
[20:51] <briancurtin> nessita: rev 264 http://bazaar.launchpad.net/~ubuntuone-control-tower/ubuntuone-control-panel/trunk/revision/264.2.11
[20:51] <ralsina> not that it will help here, of course
[20:51] <ralsina> briancurtin: so yes, adding the show breaks it
[20:51] <nessita> I won! I won!
[20:52] <nessita> ralsina: found the problem
[20:52] <ralsina> nessita: what's self.ui.destroy?
[20:52] <nessita> briancurtin: instead of removing the show, change this:
[20:52] <nessita> ralsina: YES :-D
[20:52] <ralsina> use deleteLater!
[20:52] <nessita>     300             self.addCleanup(self.ui.hide)
[20:52] <nessita> ralsina: what's deleteLater?
[20:53] <nessita> ralsina: can't we delete it ASAP?
[20:53] <ralsina> nessita: deleteLater deletes it after it gets to the event loop
[20:53] <nessita> ralsina: is there some method to deletes it sonner? (when calling it)
[20:54] <ralsina> the problem is, if you remove the object before reaching the event loop, then maybe a signal is triggered with a stale pointer
[20:54] <ralsina> nessita: it will delete it soon enough :-)
[20:54] <nessita> ralsina: not really... is not deleted until the whole suite finishes
[20:54] <nessita> I just tested it IRL
[20:54] <briancurtin> nessita: i think our line numbers aren't matching up. do you want that as the last line of the setUp right there?
[20:54] <ralsina> add a processEvents as cleanup
[20:54] <nessita> briancurtin: sorry, was doing tons more of tests, will give a proper diff
[20:55] <ralsina> a cleanup that does thing.deleteLater() ; QtCore.QCoreApplication.instance().processEvents() should delete immediately
[20:55] <nessita> ralsina: ack
[20:55] <ralsina> nessita: if even with that it's not deleted, then we may be keeping a reference somewhere
[20:56] <nessita> that worked
[20:56] <nessita> briancurtin: diff is:
[20:56] <ralsina> nessita: http://stackoverflow.com/questions/4888189/how-delete-and-deletelater-works-wrt-to-signals-and-slots-in-qt
[20:56] <nessita> http://pastebin.ubuntu.com/894265/
[20:59] <nessita> briancurtin: with that diff ^, I got: FAILED (errors=16, successes=988)
[20:59] <nessita> which is much better than a crash :-D
[21:00] <briancurtin> nessita: ah, there we go. working
[21:00] <nessita> briancurtin: and I will leave that fix to you, since symlinks are involved ;-)
[21:00] <briancurtin> nessita: yes, my favorite part of python on windows!
[21:00] <ralsina> nessita: are cleanups stacked? So it's hide->deleteLater->processEvents?
[21:01] <nessita> ralsina: yes, guaranteed
[21:01] <ralsina> nessita: cool then
[21:01] <nessita> briancurtin: merry christmas!!!
[21:01]  * ralsina goes play Another One Bites the Dust
[21:02] <nessita> ralsina: let's do 1-1- tomorrow?
[21:02] <nessita> before the weekly call
[21:02] <ralsina> nessita: yeah, I am drained
[21:02] <nessita> me too, and I need to clean my habitat a bit
[21:02] <nessita> dust and pelusas are eating me
[21:02] <ralsina> nessita: and still have to fix a bunch of crap on windows, which I really don't want to
[21:03] <ralsina> but it's taking shape :-)
[21:04] <nessita> yes
[21:04] <nessita> ok, I'm off crowd
[21:04] <nessita> see ya all tomorrow
[21:04] <ralsina> bye nessita!
[21:04] <nessita> bey all!
[21:07] <dobey> later all
[21:09] <ralsina> bye dobey
[21:10] <ralsina> briancurtin: if you get to propose that branch today, I can add it to tomorrow's installer for QA
[21:10] <briancurtin> i'm proposing the symlink branch now, will be proposing the branch with nessita's fix in a minute
[21:10] <ralsina> specially since the test failures don't seem to be IRL fails
[21:10] <ralsina> beuno: cool
[21:10] <ralsina> oops, that was for briancurtin
[21:19] <briancurtin> ralsina: https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/nessitas-crash-fix/+merge/98735 and https://code.launchpad.net/~brian.curtin/ubuntuone-control-panel/no-symlinks-on-windows/+merge/98736 are ready
[21:20] <briancurtin> now that i have this nessita fix i'm going to try out the autostart tests that i was ignoring before when it was broken
[21:20] <ralsina> globally approved the crashfix
[21:22] <ralsina> briancurtin: on the symlink test, it's better if you move that part to a separate test and make it skipIfOs("win32", "No symlinks on windows" )
[21:23] <briancurtin> ralsina: noted - i'll fix that after the other branches
[21:23] <ralsina> Or maybe doing that is a pain in the ass
[21:23] <ralsina> wait, it is a pain in the ass
[21:23] <ralsina> since that's not a test, and is rather part of the setup
[21:24] <ralsina> approving what you have
[21:36] <ralsina> and EOD for me
[21:37] <ralsina> byr briancurtin and whoever else is still around :-)
[21:37] <briancurtin> ralsina: i'll mail you with status and branches once im done
[21:38] <briancurtin> and the test i added now works with all of the other branches, so i'll repropose the autostart branch