[10:26] <alecu> morning...
[10:26] <alecu> jamesh, ping
[10:51] <jamesh> alecu: hi.  I've replied to your email now.  Had it open earlier but hadn't finished it off.
[11:06] <alecu> hi jamesh, thanks!
[11:33] <duanedesign> hello rtgz
[11:34] <rtgz> duanedesign, hi!
[11:34] <duanedesign> hope you had a good weekend
[11:34] <rtgz> duanedesign, pretty nice, + 2 bugs :)
[11:35] <rtgz> erm, +1 bug, another one is duplicate. Nope, not that great :)
[11:42] <rtgz> duanedesign, and I got the following nice dialog from ubuntuone ppa: http://picasaweb.google.com/lh/photo/aN3amTKoEd7OfPiYWFO0kw?feat=directlink
[11:42] <rtgz> and I believe I know why appled dies and apport fires when applet is clicked
[11:44] <rtgz> it is ... tada! metadata loading that takes a lot of time for a big number of files
[11:45] <duanedesign> o.0
[11:46] <duanedesign> rtgz: isnt there a bug open for loading the meta data more effeiciently
[11:46] <rtgz> duanedesign, there is
[11:46] <duanedesign> i have never seen that dialog box, lol
[11:47] <duanedesign> rtgz: this weekend i worked on a bug graphing tool.
[11:47] <rtgz> duanedesign, i have never seen it either and i could not find the actual reason behind it and its intended purpose... Some text there would be good
[11:47] <rtgz> duanedesign, from lp bugs?
[11:47] <duanedesign> lol
[11:47] <duanedesign> rtgz: yes, my first try was using data from the API
[11:48] <rtgz> PPA is broken
[11:49]  * rtgz is installing pastebinit in vm to show how much it is broken
[11:50] <rtgz> duanedesign, btw, have you found a way to get bug id from searchTasks call?
[11:51] <rtgz> http://paste.ubuntu.com/371618/
[11:51] <rtgz> syncdaemon-exceptions.log
[11:52] <rtgz> http://paste.ubuntu.com/371620/
[11:52] <rtgz> and syncdaemon stops syncing.
[11:53] <rtgz> but if there are no shares, then it is probably working fine, until clicked on applet icon while metadata is loading for a very long time. Dbus exception is not handled and it BOOM, apports :)
[12:08]  * rtgz needs to change irc nick once again :-/
[13:33]  * rtgz needs pokable developers
[13:34] <verterok> rtgz: http://paste.ubuntu.com/371618/ it's fixed in trunk
[13:34] <rtgz> verterok, good, 'cause it is broken for everyone running ppa
[13:35] <verterok> rtgz: yeap :(
[13:35] <rtgz> verterok, i believe ppa needs to be rebuilt, since it does not sync all files if shared folders are present
[13:35] <verterok> rtgz: hmm, probably, but we can't do that until we get a new version of the server rolledout
[13:36] <rtgz> verterok, still, it has to be rebuilt because of broken dependency on ubuntuone-storage-protocol, and everyone upgrading to ppa will have syncdaemon broken
[13:36] <verterok> rtgz: ooh
[13:36] <rtgz> verterok, ah, protocol changed?
[13:36] <rtgz> verterok, it is easy to fix, but it should be mentioned somewhere
[13:36] <verterok> rtgz: good point :)
[13:37] <verterok> rtgz: yeap, actually not protocol but a new capability
[13:37] <rtgz> bug 518598
[13:38] <rtgz> not ubuntuone-storage-protocol, python-ubuntuone-storage-protocol
[13:38] <rtgz> verterok, http://picasaweb.google.com/lh/photo/aN3amTKoEd7OfPiYWFO0kw?feat=directlink
[13:39] <rtgz> verterok, any ideas? :)
[13:39]  * verterok looks
[13:39] <verterok> rtgz: oh, crazy applet
[13:39] <rtgz> :)
[13:39] <verterok> rtgz: any errors in the logs?
[13:41] <rtgz> verterok, hm... I've got an idea on how to reproduce it
[14:00] <duanedesign> put some bug stats on one html page to make it a little easier to view.  http://people.ubuntu.com/~duanedesign/index.ubuntuone-client.html
[14:01] <rtgz> duanedesign, wow
[14:40] <CardinalFang> aquarius, thisfred, https://code.edge.launchpad.net/~cmiller/desktopcouch/get-port-without-fd-race/+merge/18840
[14:40] <thisfred> CardinalFang: awesome!
[14:40] <aquarius> CardinalFang, nice one. I saw your email. Great catch!
[14:42] <CardinalFang> I'm glad kenvandine codes late at night in bars, or else I wouldn't have looked there.
[14:42] <kenvandine> :)
[14:43] <aquarius> kenvandine codes everywhere. It is a most useful talent :)
[14:43] <kenvandine> CardinalFang, awesome... so that should fix this bug for real right?
[14:44] <CardinalFang> kenvandine, Yes.  This and my previous patch a week ago, I'm sure will be the last for this problem.
[14:45] <kenvandine> woot!
[14:45] <kenvandine> CardinalFang, you rock
[14:45] <cherva> can I use ubuntu one with root ? there is no Ubuntu one dir in /root
[14:47] <statik> CardinalFang, so the desktopcouch SRU for karmic should probably be reworked to include both patches then? congrats on the fix
[14:53] <CardinalFang> statik, yes, backporting to 0.5 line now.
[14:56] <urbanape> morning, all
[14:58] <mhall119|work> so, I saw the new MeMenu screenshots
[14:58] <mhall119|work> is UbuntuOne going to be integrated with the new social network features?
[15:00] <mhall119|work> it would be cool to right-click a file or photo, and says "Share with friends", and have it uploaded to UbuntuOne, made public, and a link to it sent to Twitter/Identi.ca/Facebook
[15:03] <teknico> urbanape, hi
[15:03] <urbanape> heya, teknico
[15:04] <CardinalFang> mhall119|work, it wouldn't me in Me menu, but I think it's likely.  Announcing it would be an optional check-box at share time, I suspect.  I hope we have time for that.
[15:06] <aquarius> welcome back urbanape
[15:07] <aquarius> urbanape, we looked at getting bindwood packaged last week during the distro sprint, but it didn't seem to work for me or CardinalFang, so we decided to wait until you got back
[15:09] <CardinalFang> I tagged trunk 1.0.0 before we decided there was more to do.  Please don't whip me, urbanape.
[15:09] <urbanape> aquarius, thanks, I just managed to get a little wifi time in the jungle, so I saw the approval messages.
[15:09] <urbanape> CardinalFang, no whipping, you guys went above and beyond, so thanks.
[15:09] <rye> mhall119|work, MeMenu screenshots? I can find only UI design drawings
[15:11] <xtephan> i have a question, anybody can help me?
[15:11] <xtephan> hi
[15:13] <CardinalFang> xtephan, just ask.  Interested or knowledgeable people will try.
[15:14] <xtephan> I see a warning who says "The app "desktopcouch service" want to acces the password for "ubuntu One token for https://ubuntuone.com" in the predeterminated keyring
[15:14] <xtephan> i transalate the text of the warning because my computer is in spanish
[15:15] <CardinalFang> xtephan, okay.  What is your concern?  That looks normal to me.
[15:15] <xtephan> will i allow it always?
[15:16] <CardinalFang> If you like.  I think you should.
[15:16] <xtephan> what is the "desktopcouch service" application?
[15:16] <dobey> oh good, i didn't miss the me stuff yet
[15:16] <rodrigo_> xtephan, yes, you should allow always, unless you want to have that dialog every time
[15:17] <xtephan> but, what is the "desktopcouch service" application?
[15:17] <CardinalFang> desktopcouch is a component of ubuntu one.  It is a distributed database, for data you care about.  Contacts, bookmarks, et c.  These could be synchronized to other computers you own.
[15:18] <dobey> mhall119|work: no there won't be any any integration with the gwibber stuff for sharing files. we don't have time to do it
[15:18] <xtephan> @CardinalFrang thanks
[15:18] <dobey> mhall119|work: maybe in +1 it could do that, feel free to file a feature request bug
[15:23] <mhall119|work> dobey: I'll think it through
[15:23] <mhall119|work> will the new public option in UbuntuOne be something that can be configured from the client-side?
[15:24] <rye> CardinalFang, question - karmic updated, on session login desktopcouch service is running. When i stop desktopcouch and start it again keyring daemon asks the question about desktopcouch access. How does it start then on session login?
[15:24] <dobey> i don't think there will be any configuration. it will just be "Publish on Ubuntu One" or something
[15:24] <CardinalFang> rye, it starts only when it is needed by some app.
[15:25] <mhall119|work> dobey: I ask because if it's something that can be done all client-side, I may just try hacking at it
[15:25] <CardinalFang> rye, oh, I see what you're asking.  Hrm.
[15:25] <CardinalFang> rye,  I don't know.
[15:26] <dobey> mhall119|work: I'm not sure of the specifics of how it will work in the client
[15:26] <mhall119|work> I guess I have to get a Lucid alpha in vbox
[15:27] <mhall119|work> is Xubuntu using U1?  I didn't see it
[15:27] <dobey> CardinalFang: is that the bug we fixed in PDX?
[15:27] <dobey> mhall119|work: we don't have integration with Thunar. we only have a Nautilus extension at the moment
[15:27] <mhall119|work> ok
[15:30] <CardinalFang> dobey, I don't think so.  Which fix do you think?
[15:31] <dobey> CardinalFang: I just remember on Tues or Wed, you were getting the password pop-up, or was that for a different password in futon vs. keyring?
[15:31] <urbanape> did we change how/when we do standups?
[15:31]  * urbanape is still catching up on mail
[15:32] <dobey> urbanape: you don't get to post one since it is just "DONE: Belize, TODO: Un-tan"
[15:32] <jblount> heh
[15:32] <urbanape> bah. I wore SPF 70. Didn't even turn a little pink
[15:32] <CardinalFang> dobey, yes, I fixed a password bug, but that was for the bookmark file, not seahorse.
[15:33] <rye> CardinalFang, i believe syncdaemon needs couchdb for account info, so it should invoke desktopcouch... and desktopcouch invoking results in keyring access, which is not happening on session login
[15:34] <dobey> CardinalFang: ok, right.
[15:34] <aquarius> rye, not quite. the applet uses desktopcouch -- it doesn't need the data, but when the applet signs in, it sets up a pairing between local desktopcouch and Ubuntu One couch, so that your couchdb data is replicated as well as your files.
[15:35] <rye> aquarius, yes, it just checks for desktopcouch db, i believe, but this is enough for dc to trigger keyring access, i believe
[15:35] <dobey> no, it tells dc to pair with u1, which involves getting the u1 token from keyring
[15:35] <aquarius> rye, it checks for the management DB to exist, and if it doesn't, creates it, and then adds a record to it. This should prompt keyring access, yes
[15:36] <aquarius> because the applet gets the U1 token from the keyring
[15:36] <aquarius> in order to add the pairing record, which needs to contain the keyring token
[15:36] <CardinalFang> aquarius, so, why not ask first, and then ask later?  I have no idea.
[15:36] <rye> aquarius, erm... is desktopcouch running as a parent process to couchdb?
[15:37] <aquarius> rye, possibly. And possibly not. :)
[15:37] <aquarius> CardinalFang, don't understand
[15:37] <rye> aquarius, i mean someone triggers replication, right? it is not done internally by couchdb so there's an externally running couchdb
[15:37] <rye> ^W desktopcouch
[15:38] <aquarius> rye, yes. d-bus starts desktopcouch-service; desktopcouch-service (a) runs couchdb, and (b) triggers replication
[15:38] <CardinalFang> aquarius, he is asking why an app starts desktopcouch daemon (seemingly), and does not need keyring access.  Then, later, on second run it does need access.
[15:38] <aquarius> CardinalFang, yeah, but I don't understand why that would happen :)
[15:38] <dobey> rye: what specifically is listed as needing access, in the keyring dialog that pops up? it should show the application name there
[15:39] <rye> CardinalFang, aquarius my couchdb runs w/o desktopcouch companion after login...
[15:39] <rye> so, obviously, it does not replicate stuff...
[15:40] <CardinalFang> rye, er, as which user does it run?
[15:40] <rye> CardinalFang, rtg
[15:40] <rye> CardinalFang, i.e. me
[15:40]  * CardinalFang boggles.
[15:40] <rye> and neither my contacts nor notes are replicated
[15:40] <rye> but couchdb.html file IS created
[15:41] <CardinalFang> Sometimes I envy Windows developers who can distract people by saying "you need to reboot 4 times."
[15:41] <rye> and the only things to update are: evolution-couchdb libcouchdb-glib-1.0-1 python-pyinotify
[15:42] <rye> CardinalFang, vm was rebooted 8 times so far to test this :)
[15:43] <rye> CardinalFang, dc installed is 0.5-0ubuntu1
[15:43] <rye> what if...
[15:43] <rye> there is no seahorse daemon running when couchdb is started... no, it's weird
[15:43] <rye> need to trace the execution path or otherwise this is useless
[15:46] <rodrigo_> statik, have you done the packaging for the new couchdb-glib/evo-couchdb?
[15:46] <rye> beta ppa is enabled
[15:46] <statik> rodrigo_, i have not finished it, no. am planning to work on it again today
[15:47] <rodrigo_> statik, ah, ok, let me know if you need any help
[15:49] <mandel> hello, does anyone know if there is anyone to tell the CouchGrid to update according to changes in the db, I filed a bug due to my ignorance :(
[15:53] <dobey> rodrigo_: is there an accompanying libubuntuone branch for your move of contacts picker?
[15:54] <rodrigo_> dobey, not until it's merged
[15:55] <aquarius> mandel, I don't know -- rickspencer3 is the chap you want to talk to about that
[15:55] <dobey> rodrigo_: ok
[15:55] <mandel> aquarius, ok, is he at quickly irc(if any) or other irc?
[15:55] <aquarius> mandel, he will be, but he's not online atm -- he's in the US
[15:56] <rye> okay, trying here. is there any debug info given by desktopcouch on startup? I have a perfect setup - a vm which can reproduce the bug with broken replication with 100% reliability. Just need a hint on whether it is built in or i need to stuff something into it...
[15:56] <mandel> aquarius, ok, so it would be around 9 pm here, right?
[15:57] <aquarius> mandel, I'd expect him in the next hour or so, if he's working today
[15:58] <rye> ok, will be back in an hour or 1h30m
[15:58] <mandel> aquarius, superb, another question, what happened with the python-keyring idea?
[15:59] <dobey> keyrings are hard, mmkay
[16:01] <aquarius> mandel, the big problem with it is how couchdb-glib and other non-python things know where to get the data from. We could provide a desktopcouch API to get that data (which would then use python-keyring), but you can't do d-bus APIs on Windows...
[16:01] <mandel> dobey, I know.. I'm just asking 'casuse is one of the few things that block me from porting destopcouch to windows.. I already have the IPC, a c# library to talk with contacts and I'm close to do outlook integration.
[16:02] <rodrigo_> aquarius, a d-bus call would work for couchdb-glib
[16:02] <dobey> mandel: we can't really use it on ubuntu though, because it doesn't make sense in the end
[16:02] <mandel> aquarius, that is not a problem at all on windows, I can take care of that
[16:03] <dobey> rodrigo_: i don't want us wasting time writing an abstraction for an abstraction, so we can lose useful API because pykeyring is a lowest-common denominator
[16:03] <mandel> dobey, what do you mean, porting to windows or using python keyring?? Windows I understand (although trust me I'm sure a lot people will love the idea) but python keyring...
[16:04] <rodrigo_> dobey, writing a dbus method is a 10 min "waste of time", if that eases the port to windows, I'm all for losing those 10 mins
[16:04] <dobey> mandel: the benefits to python-keyring on GNOME/KDE are minimal and diminishing
[16:05] <aquarius> dobey, yes, especially since the secret-storage stuff is coming, but they're much larger benefits in the context of a Windows port.
[16:05] <dobey> rodrigo_: i'd rather just have those 10 minutes spent doing if WINDOWS: import keyring; else: import gnomekeyring;
[16:06] <mandel> dobey, for windows that would do the trick, but I got people complaining in FOSDEM for not using KWallet...
[16:06] <dobey> aquarius: i'm fine with using python-keyring ONLY on windows/osx, but I would rather avoid the headaches on ubuntu
[16:06] <dobey> mandel: kwallet is going away
[16:06] <aquarius> mandel, I need to hear how your fosdem talk went, btw! Was it videoed?
[16:06] <mandel> dobey, I know, but people complain, what can I do hehe
[16:06] <dobey> mandel: in lucid+1 we should hopefully finally be able to use the unified keyring
[16:07] <dobey> mandel: yes, we know they complain. we hear it enough at UDS :)
[16:07] <aquarius> dobey, as per http://www.freedesktop.org/wiki/Specifications/secret-storage-spec you mean?
[16:07] <dobey> aquarius: yes
[16:07] <mandel> aquarius, no video AFAIK, it went well but not enough time and got some people very interested but couchgrid failed me in the middle of it hehe
[16:07] <dobey> aquarius: the nice thing is that the API is almost exactly the same as gnome-keyring
[16:08] <mandel> dobey, I don't mind using keyring just for MacOS and windows, but I dont like lots of if statements in the code for that, looks ugly...
[16:09] <mandel> aquarius, that is why I want to patch couchgrid and add for functionality for reporting changes etc..
[16:09] <dobey> mandel: the problem is that any abstraction is going to mean we can't use useful API from gnome-keyring
[16:10] <dobey> and python-keyring is only python
[16:11] <mandel> dobey, I know... that is the pain of being multi-platform.. well if it is a problem I can think a way to go around it, implement something similar on windows that uses the same calls but that uses something more native to the platform
[16:11] <mandel> I meant same signatures as the gnome-keyring
[16:15] <mandel> dobey, I can follow the gnome-kerying signature and use DPAPI instead, that would do the trick and will not make the code to dirty
[16:16] <dobey> mandel: well you're going to have to do some tricks, because gnome-keyring does more stuff afaik
[16:16] <dobey> like ACLs
[16:17] <mandel> dobye, if that is truly used I might just use WIF which should provide all the work, it will be extra code but I'll be able to provide everything that is needed
[16:18] <mandel> ups, misstyped dobey^
[16:19] <dobey> i don't know if desktopcouch uses the ACLs and other stuff
[16:21] <rodrigo_> it doesn't right now afaik
[16:22] <mandel> well, right now I'm doing some work on linux later I'll give it a go to see what happens
[16:27] <dobey> u1-client definitely uses ACls though
[16:28] <dobey> anyway, need to get some lunch and all that
[16:28] <dobey> bbiab
[16:35] <mandel> aquarius, I need to go to get food, if you see rickspencer3 let him know I'm try to get him :)
[16:35] <mandel> laters
[18:06] <rye> heh
[18:06] <rye> question
[18:06] <rye> who can start couchdb but not via dbus ?
[18:07] <verterok> rye: /usr/lib/desktopcouch/desktopcouch-service ?
[18:08] <rye> verterok, it looks like something is broken but I can't say what
[18:08] <rye> i disabled dbus startup and couchdb is started... after applet startup
[18:09] <rye> which means that it is starting it directly
[18:09] <rye> so DBus does not play here
[18:10] <verterok> rye: how do you disabled dbus startup?
[18:10] <rye> verterok, /bin/true as Exec handler :)
[18:10] <rye> i mean i disabled desktopcouch startup via dbus
[18:10] <rye> verterok, ^
[18:11] <verterok> rye: and you killed it properly before? :)
[18:12] <rye> verterok, reboot. Kills everything (tm)
[18:12] <verterok> hehe
[18:13] <verterok> rye: ps aux | grep desktop-couch ?
[18:14] <verterok> rye: maybe it's the system wide couchdb?
[18:14] <rye> ps aux | grep desktop-couch
[18:14] <rye> yes, right...
[18:14] <rye> verterok, http://paste.ubuntu.com/371914/
[18:15] <verterok> rye: nice! :/
[18:15] <verterok> rye: I don't see any code in the applet that could start desktopcouch :(
[18:16] <rye> it may not be applet... there is ubuntuone-login as well (r321, ppa)
[18:16] <verterok> rye: that's the oauthdesktop "service"
[18:16] <rye> hmmm
[18:17] <rye> verterok, and Oauthdesktop needs couchdb...
[18:17] <verterok> rye: no, it shouldn't need couchdb
[18:17] <rye> verterok, ah. grr. keyring only... but who will then establish couchdb pairing?
[18:18] <verterok> rye: sorry, I'm a bit lost there...I think that's desktopcouch itself
[18:18] <dobey> rye: uhm
[18:18] <dobey> rye: it's evolution-data-server loading the couchdb plug-in
[18:19] <rye> verterok, but it will need to find info about oauth first
[18:19] <dobey> rye: which talks to couchdb directly
[18:19] <rye> dobey, hm...
[18:19] <verterok> dobey: isn't using dbus?
[18:20] <rye> dobey, but how can does it start desktopcouch dbus or plain /usr/lib/desktop-couch/*-service ?..
[18:20] <dobey> verterok: it's not using desktopcouch, and couchdb doesn't provide a dbus api, so i presume it just starts it if it isn't running
[18:20] <dobey> rodrigo_: ^^ is that true with couchdb-glib?
[18:20] <verterok> dobey: ok, thanks
[18:20] <dobey> rodrigo_: or does it call the desktopcouch dbus api?
[18:20] <rye> dobey, no, evolution-data-server is not running here :-/
[18:21] <rye> or it died
[18:21] <dobey> i would suspect the latter
[18:21] <dobey> the clock applet on the panel uses e-d-s :)
[18:21] <rye> /usr/lib/evolution/2.28/evolution-alarm-notify is running, though...
[18:22] <dobey> weird
[18:24] <rye> dobey, and i have seen such behavior in 3 or 4 cases so far here.
[18:24]  * rye is disabling ubuntuone
[18:25] <dobey> i don't guess e-d-s needs to be running all the time for the alarm notifier to work
[18:27] <rye> nope, couchdb process is not started when ubuntuone-applet is not in startup notification thing
[18:27] <rye> *WALL*
[18:29] <rye> started ubuntuone-client-applet and got couchdb processes WITHOUT desktopcouch
[18:29] <rye> moreover
[18:29] <rye> on my host machine: 2010-02-05 00:29:44,704 ERROR    replication of services aborted
[18:30] <rye> this is the last time replication tried to be performed
[18:30] <rye> and now i don't have desktop-couch wrapper running too
[18:30] <rye> verterok, dobey, anybody, with running replication, could you please paste the output of  ps aux | grep couch  to pastebin?
[18:31] <verterok> rye: I don't have replication configured :/
[18:31] <rye> verterok, not syncing evolution contacts?
[18:32] <verterok> rye: not in the laptop (I'm not in my desktop ATM)
[18:33] <rye> ah
[18:33] <rye> true
[18:33] <rye> i have additional pc
[18:34] <dobey> i don't have anything in couch being synced
[18:34] <dobey> hrmm
[18:34] <rye> dobey, but do you have additional python service for desktopcouch ?
[18:35] <dobey> rye: i don't think so. i tend to run stuff from the source tree
[18:37] <rye> O_O
[18:37] <rye> netbook - replication last time occured on 2010-01-11
[18:44] <rye> hm.
[18:45]  * CardinalFang tries to land 5 branches of desktopcouch at once.
[18:45] <rye> no desktopcouch on jaunty, so no use to test :-/
[18:58] <dobey> huh
[19:30] <rye> this is definitely applet
[19:31] <rye> because i started ubuntuone-login, then ubuntuone-syncdaemon and couchdb service appeared only after ubuntuone-client-applet.
[19:31] <rye> dbus could not have started it, because the handler is replaced by /bin/true
[19:32] <rye> and applet has started w/o any output (e.g. starting..., e.g. here's your couchdb .html)
[19:47] <manselton> Hi, Nice to meet. I'm new to ubuntuone and I've got u1conflicts. What does this mean, please?
[19:49] <dobey> hi. it generally means that the file was modified in two different places before it could be synced to your computer.
[19:49] <dobey> it was modified on your computer and another location, and the other location got synced to the web, and the sync to your current computer generated the conflict
[19:50] <dobey> it may be that you're also hitting a bug
[19:51] <manselton> Thanks. Can I try to do anything myself? Single files upload OK but can I just rename the .u1conflict files?
[19:54] <dobey> yes, just look at the original and the .u1conflict, and either replace the original or remove the u1conflict, depending on which is supposed to be there. or manually merge any differences if necessary
[19:55] <manselton> OK thanks, I'll give it a go.
[20:10] <rye> when backup isp is present the bottleneck is in the router...
[20:27] <CardinalFang> It's taking forever to update to Lucid on my main machine.  Aigh!
[20:33] <rye> pdb
[20:33] <rye> icon.main() and couchdb is created
[20:33] <rye> G r R Rr r
[20:33] <dobey> guess i should upgrade my workstation to lucid too
[20:37] <dobey> rye: yeah, it pokes desktopcouch whenever NewCredentials signal is called
[20:38] <dobey> rye: so desktopcouch should be running... if it's not, it sounds like there's something weird in desktopcouch causing it to exit, or one of the desktopcouch.records methods is perhaps causing it to start
[20:38] <dobey> rye: what is the exact program trying to access the keyring?
[20:39] <rye> dobey, desktopcouch-service
[20:39] <rye> dobey, when I start it manually
[20:39] <rye> dobey, "desktopcouch service" (/usr/bin/python2.6) to be precise
[20:41] <dobey> rye: i mean in the instance for the problem you're trying to debug
[20:41] <rye> dobey, applet startup does not trigger any message
[20:41] <rye> dobey, but couchdb processes are created
[20:41] <rye> dobey, it looks like it is half-started
[20:43] <dobey> rye: trace the code from setup_desktopcouch_pairing() then :)
[20:45] <rye> dobey, do you mind if I ask how does applet relate to couchdb?  I could not find anything couchdb related in ubuntuone-client-applet (r321)
[20:46] <dobey> rye: it doesn't. it calls desktopcouch stuff only, to create the pairing to the u1 cloud couchdb
[20:47] <rye> dobey, but... grep couch /usr/bin/ubuntuone-client-applet is silent...
[20:47] <dobey> rye: see /usr/lib/ubuntuone-client/ubuntuone-login then
[20:48]  * rye slaps his face
[20:48] <rye> I was assuming that if ubuntuone-login is started and no couchdb is brought up then it is not ubuntuone-login. I forgot that procedures tend to be called at a later time :-/
[20:52] <rye> done
[20:52] <rye> debug finished
[20:52] <rye> BUG
[20:53] <rye> dobey, CardinalFang, verterok, aquarius , this is a bug in desktopcouch implementation
[20:54] <rye> when CouchDatabase object is created in /usr/lib/ubuntuone/ubuntuone-login then couchdb is started, right
[20:54] <rye> BUT
[20:54] <rye> neither replication nor dbus service to advertise the port is started
[20:55] <rye> this leads to broken replication
[20:55] <rye> the replication and dbus service to advertise the port is created in /usr/lib/desktopcouch/desktopcouch-service, which is NOT called via d-bus
[20:57] <rye> It appears that couchdb startup is performed directly. Not yet found how but I am on my way
[20:57] <rye> What is the worst part?
[20:58] <rye> This can last unnoticed until the user starts to use couchdb for his contacts. There will not be any.
[20:58] <rye> I think this is a major flaw
[20:59] <dobey> or bookmarks, or gwibber, or etc...
[21:00] <aquarius> rye, you are correct. I think desktopcouch.records.CouchDatabase should call D-Bus to get the port, and it doesn't.
[21:00] <rye> dobey, or life (if the user actually backups his life via UbuntuOne as marketing copy says :) )
[21:00] <aquarius> it does it the old-fashioned way of calling desktopcouch directly.
[21:00] <aquarius> CardinalFang, would you agree?
[21:01] <CardinalFang> That's a good idea, probably.  Let's think.
[21:02] <rye> CardinalFang, i would agree, otherwise replication never works
[21:02] <rye> erm
[21:02] <CardinalFang> Synchronous call, so no need for event loop and thread-safe.
[21:02] <rye> aquarius, i would agree, otherwise replication never works and it is bad for users
[21:03] <CardinalFang> Requires dbus, though.  Support is good, right?  And speed?
[21:03] <rye> ok, if you don't mind I'll go file a bug
[21:03] <CardinalFang> rye, that's perfect.  Please do.
[21:03] <aquarius> CardinalFang, you have to require d-bus, though; d-bus is how you're meant to find the port
[21:04] <aquarius> CardinalFang, desktopcouch.records sneakily avoids doing that by knowing how to call DC directly, but it isn't supposed to.
[21:04] <aquarius> Also, this may explain why some people don't have replication happen. :)
[21:05] <rye> aquarius, those who have don't have applet set to automatic login
[21:05] <aquarius> *nod*
[21:05] <rye> aquarius, the replication is started via e.g. evolution calling dbus method to find the port, etc.
[21:05]  * rye hopes that evolution does this via d-bus
[21:05] <rye> :)
[21:05] <jcastro> erm
[21:07] <rye> aquarius, bug #489963
[21:07] <CardinalFang> aquarius, so, change find_port to be dbus client, and make a new, private function to be the dbus server function?
[21:08] <aquarius> CardinalFang, I think so, yes.
[21:08] <dobey> isn't there already some dbus thing to do that?
[21:08] <CardinalFang> Okay.  branch coming up in ~10.
[21:09] <aquarius> CardinalFang, hang on
[21:09] <CardinalFang> wut?
[21:09] <rye> aquarius, no, it is something different, probably related but in my case it does not work completely
[21:09] <aquarius> CardinalFang, what do you mean by "to be the d-bus server function"?
[21:09] <CardinalFang> You'll see.  A few minutes....
[21:10] <aquarius> CardinalFang, what we need is just to replace the find_pid/find_port calls with a call to teh d-bus API to get the port, and that's it
[21:10] <aquarius> unless I'm missing something?
[21:14] <CardinalFang> Er, this is harder than I thought.  One can't bind execution contexts to any number of DBUS functions.  Tests break.
[21:15] <rye> yes, evolution does getPort via dbus
[21:16] <CardinalFang> aquarius, what do you think -- org.desktopcouch.CouchDB{,2,3,4,5,6,7...} ?
[21:17] <CardinalFang> Okay, that's more than a ten-minute patch.
[21:18] <dobey> CardinalFang: what is the problem? i think you're making it harder than it needs to be?
[21:19] <rye> CardinalFang, aquarius: bug #519028
[21:22] <CardinalFang> dobey, short answer:  dbus function names are session-specific, but tests are independent of session.
[21:23] <dobey> no. if calling a dbus method was session specific, it would never work. how would any external app know what to call?
[21:23] <CardinalFang> Not calling the function.  Hosting the function.
[21:23] <CardinalFang> The server end.
[21:24] <CardinalFang> Running tests should fire up a new server, not pollute your personal db.
[21:26] <dobey> the tests need fixed to not hit the real keyring, also
[21:26] <CardinalFang> Hrm, that's a good idea.
[21:27] <dobey> and for the dbus stuff, it might be better to just replace the dbus stuff and mock some more necessary bits
[21:27] <aquarius> easy way to do it: mock the d-bus method and have the mock look up the port with desktopcouch.find_port.
[21:27] <aquarius> hard way to do it: run your own d-bus bus, which is what the ubuntuone-client tests do.
[21:33] <dobey> do the tests actually need to stick anything in a couchdb though?
[21:38] <aquarius> dobey, the tests do need to test that we can successfully talk to couch; one of the things they protect against is regressions in either couch itself or python-couch
[21:44] <dobey> hmm
[21:46] <dobey> ok
[21:46] <dobey> brb
[21:47] <rye> i believe this will need to go into SRU, since it gradually decreases u1 functionality for evolution
[21:47] <aquarius> we're creating an SRU atm for desktopcouch, with a couple of other fixes, so it should be able to go into that.
[21:47] <rye> aquarius, thanks
[21:56] <rye> good night, everyone!
[22:10] <dobey> hrmm
[22:26] <rodrigo_> dobey, couchdb-glib calls desktopcouch's getPort method, so that starts it if it's not running
[22:27] <dobey> ok
[23:03] <mesula> When will symlinking support come?
[23:03] <mesula> I'd love to symlink my ~/Documents and ~/Pictures to sync online and be an automatic file backup service.
[23:15] <statik> mesula, in 10.04 you will be able to right-click on ~/Documents and select "sync with ubuntu one"
[23:15] <mesula> statik: Great. :D
[23:16] <mesula> statik: That'd sure beat the hell out of my current solution of placing the folders in the 'Ubuntu One' folder and symlinking from my home directory.
[23:22] <duanedesign> put some ubuntuone-client bug graphs on an html page.  http://people.ubuntu.com/~duanedesign/index.ubuntuone-client.html