[05:11] hello, i'm here again to say something awful about u1! xD [05:11] its eating all my little upload internet band! >.< [05:24] Igoru-san: is that bad? [05:24] Igoru-san: if it's bad, i.e. you have ADSL and it eating your upload kills your download, try wondershaper [05:25] Igoru-san: if it's bad just in the sense that you want to use your upload for something else, Real Soon Now we'll have throttling [05:56] omg i didn't see Chpaca_ talking to me. what wondershaper do? that's EXACTLY my problem. [05:56] with this program i can say 'ubuntuone will just use 10kbps'? [12:59] buenos días #ubuntuone! [13:13] hola thisfred [13:35] thisfred, aquarius, https://code.edge.launchpad.net/~cmiller/desktopcouch/replication-daemon/+merge/10732 [13:36] I'm AFK for ~40 minutes to take care of the kid. [13:36] "This daemon shoud probably be merged into the desktopcouch-start program somewhere." ? :-) [13:37] Yeah, something. We don't really need a separate program, I think. [13:37] Rather, when couchdb fires up, it should also bring the replicator up with it. [13:37] yeah. I think that it being so merged is a prerequisite for being accepted, though? Or do you not thing so? [13:37] AFK! [13:38] sorry, yes, go away, talk when you come back :) [13:38] * thisfred reviews as well, will give wildly biased and unfiltered thoughts in a bit [13:46] hi aquarius, did CardinalFang tell you about the desktopcouch listening problem (opportunity!)? currently it binds to 127.0.0.1, so you can't connect to a desktopcouch running outside your machine. this is secure [13:46] but for machine-to-machine replication to work, we would need to bind to something like 0.0.0.0 [13:46] before we have OAuth turned on, this is very insecure === rmcbride_ is now known as rmcbride [14:27] statik, he did [14:27] statik, this is what I said yesterday about the problem with not enabling auth :( [14:28] statik, I think the approach is to not enable LAN sharing until dc0.4 [14:29] agreed [14:30] it's easy to change the config for testing/integrating [14:30] but by default we should be secure [14:30] *nod* [14:30] I agree entirely; that's why I wanted to get the oauth stuff in for 0.3 :( [14:30] it's only a problem for lan sharing; cloud sharing doesn't need it because it's driven by localhost [15:03] It's that special time of the morning. [15:03] (or mid-afternoon or whatever) [15:03] when folks in the Desktop+ clan get together to shout their slogan in unison: DONE! TODO! BLOCK! ARGH! [15:03] if you're a member of the clan, present your tartan, and shout out ME [15:04] in other words, MEETING BEGINS ;-P [15:04] * urbanape shouts ME [15:04] me (I must write a bot to do this :)) [15:04] me [15:04] me [15:04] what? who? me [15:04] me [15:05] me [15:06] DONE: Pushed a few simple branches for ubunet. Stared sadly at the ZipStream stuff. Made a separate project for it. [15:06] TODO: Bindwood tweaking. [15:06] BLOCK: None [15:06] ARGH! [15:06] aquarius: you're up! [15:06] ⚀ DONE: helped chad, thisfred work on existing DC0.3 branches; reviews of DC stuff; tidied up my bug list a bit [15:06] ⚁ TODO: piston oauth in snowy; help release of DC0.3; write DC talk for Ubuntu Developer Week [15:06] ⚂ BLOCKED: couchdb patch which lets OAuth read users from the ini file doesn't seem to work [15:06] rock me big teknico [15:06] DONE: reviews [15:06] TODO: more reviews, working on the contacts CRUD web ui [15:06] BLOCKED: many twisted.internet.defer.AlreadyCalledError appear while testing trunk [15:06] next: CardinalFang [15:06] DONE: submitted buggy replicator so we can get desktopcouch out the door. [15:06] TODO: fix some bugs in replicator, if time permits. [15:06] BLOCKED: None. [15:06] statik! [15:06] aquarius, was it "rock me, big teknico" or "rock me big, teknico"? :-) [15:07] DONE: Worked on Erlang MIR and embedded zlib/pcre problem. Reviews. Branch to buffer desktopcouch trunk from pqm. Fell asleep. [15:07] TODO: Write email about tracking bugs in the 18 packages that we have a vested interest in, not just our projects. help aquarius and cardinalfang and thisfred with desktopcouh 0.3 release. [15:07] BLOCKED: only by my own ambition. [15:07] teknico, I'm happy to be guided by you ;) [15:07] :) [15:07] rodrigo_, welcome to the big time [15:07] • DONE: Debugged with Ken an evo-couchdb problem he's having. Fixed uuid dependency checking in couchdb-glib. Finished URL fields support in evo-couchdb/couchdb-glib. Released new versions of couchdb-glib and evo-couchdb. Started looking at becoming a MOTU [15:07] • TODO: Add more tests in couchdb-glib test suite. Add social services accounts config to about-me. Talk to Ara about writing mago tests for evo-couchdb. Propose couchdb-glib/evo-couchdb for GNOME 2.29. Store UUIDs for postal addresses. Conflict resolver tool in pair tool. oAuth authentication and signing of all couchdb-glib requests. Add changes notifications to couchdb-glib. Look at becoming a MOTU (https://wiki.ubuntu.com/UbuntuDeve [15:07] • BLOCKED: none [15:07] dobey: su turno señor :) [15:07] ☭ DONE: Finished prefs dialog [15:07] ☭ TODO: 0.93.0 feature freeze release, OAuth [15:07] ☭ BLCK: None. [15:07] fin [15:07] jblount out today? [15:09] statik, that email sounds interesting [15:09] statik: ambition is a bitch, isn't it? [15:09] sure is [15:16] pfibiger, you'll be happy to hear (!) that Lenovo is sending someone to replace my noisy laptop fan. [15:16] wow. sending someone, that's pretty cool. [15:25] Hope your fireplace is in working order. They travel by floo powder. [15:26] aquarius, CardinalFang: hows it looking for cutting a desktopcouch 0.3? [15:27] what's the best way to check on the pqm status? [15:27] urbanape, I live in the deep South. We have no fireplace, but I have a barbeque stack outside. [15:28] CardinalFang, I'm sure they adapted something. [15:28] statik, the reviewers are revolting! [15:29] CardinalFang, yeah, sorry about that [15:29] aren't they just? [15:29] * CardinalFang sets up the jokes. That's his job. [15:29] statik, I believe that once CardinalFang's branch is done and merged, that can be 0.3 [15:29] statik, I think it's close. [15:29] aquarius, once we build the package for that everything goes onto the CD. let me know when it's ready to merge :) [15:30] thisfred, aquarius, I think I addressed your complaints in my resubmission. [15:30] CardinalFang, what's the best way to test it? [15:30] rodrigo_, you have such big TODO lists! it inspires me [15:31] statik: heh, I just append there everything :D [15:31] ps axu |grep couch [15:31] oops. [15:32] heh [15:32] kenvandine, thanks a million for your patience and help with all our packages [15:32] statik, i'll do whatever i can :) [15:32] CardinalFang: will rereview in a bit [15:32] kenvandine, so when I build the 0.3 package, should I branch from your packaging branch? or how does that work? [15:33] statik: make him do more! [15:33] jcastro, i need your help [15:33] ok [15:33] CardinalFang, we should use xdg.BaseDirectory to find the cache folder rather than hardcoding $HOME/.cache [15:33] i count at least 18 different packages that I care deeply about in ubuntu because we have put our dirty code in them [15:33] statik, i can do it if you like [15:33] aquarius, Eh, it's not easy, and it's probably broken anyway. Bind couchdb to 0.0.0.0, turn off authentication (because replication should require admin acct, and that's not available yet). Then, seed the "management" database with the local ID and a paired-host ID. Oh, and do this all on two machines (or accounts). [15:33] is the tarball published? [15:33] kenvandine, just waiting on the last branch and i'll put up the tarball [15:34] ok [15:34] aquarius, I have a program for seeding couchdb. [15:34] statik, ping me and i will do the packaging [15:34] hrm, I can't test on two machines in a hotel room :) [15:34] jcastro, we have 8 projects under the launchpad.net/ubuntuone suite, but i want to look out for bugs in these other 18 packages too (erlang, couchdb, ubuntuone-client, etc. etc) [15:35] jcastro, any tips or tricks for managing bugs on both packages and projects? all my experience is with working with launchpad from the project perspective [15:35] kenvandine, thanks! [15:36] CardinalFang, approved based on code inspection but not testing that it actually works [15:37] test it or i kill you [15:37] Ignore statik. [15:37] oops, did i say that out loud? :) [15:37] kenvandine, ^_^ [15:37] and rodrigo_ ^^ [15:37] need me to test anything? [15:37] pygi, what? [15:37] just saying hi xD [15:38] statik, I'm happy to test it if you're happy for me to go out and buy a second laptop on expenses ;) I only have the one machine here :( [15:38] hey pygi [15:38] pygi, you are in all the channels i am in [15:38] kenvandine, that is so not true! [15:38] rodrigo_, how are you doing?:) [15:38] pygi, close :) [15:38] pygi is a spy :) [15:39] pygi: doing great, and you? [15:39] kenvandine, nah, I left a few... PK is not my regular channel anymore, foresight isn't either [15:39] a bit tired, otherwise good, thank you [15:39] CardinalFang, can I have your program to seed the db? [15:39] just discussing KDE client with one folk [15:39] pfibiger, the gwibber folks have a problem with images in their web-browser-like widget. Is it possible to notice in JS/DOM if an attempt to load an "file:///" IMG fails? [15:39] statik: how about an lp tag? that would cover across projects and packages. [15:39] pygi: kde client of what? [15:39] rodrigo_, U1 :P [15:39] duh [15:40] ah :) [15:40] he's going to write it? [15:40] CardinalFang, pfibiger: trap img.onerror [15:40] rodrigo_, no, I am thinking of it xD [15:40] I've been talking to you about that earlier [15:40] statik: hmm but then you'd have to have someone tagging them. [15:40] yeah, I remember [15:40] you're getting old=) [15:41] statik: we should complete dobey's lp-tools, to show branches, bugs, etc, of all the projects you tell it to watch [15:41] pfibiger, My idea is to load a default image, and then do { try: fill-in-real-image; break; catch error: sleep(3 sec); } while true; [15:41] statik: you could just subscribe to all the bugmail for each project, but then you'd have to have ninja-mail filtering to make it sane [15:41] statik: I had planned to add that, but got no time in the last few weeks [15:41] pygi: heh [15:42] rodrigo_, that was a joke :p [15:42] aquarius, Hmm, that may be it. [15:42] pygi: I know, I feel like a 10 years boy :D [15:42] rodrigo_, we should really talk about that tho [15:42] I should have some time next month [15:42] yeha, whenever you want [15:43] jcastro, yeah i'm using mail right now but i think we'll end up writing something custom using the launchpad API. was curious if you had any other teams that heavily used launchpad both as an upstream and as a package maintainer [15:44] you'd be the first I know of [15:44] lemme ask around [15:44] though I get the feeling a mail to launchpad list would get you more advice from the lp experts [15:44] beuno knows I bet. :D [15:47] aquarius, have you done anything with virtual machines? i keep meaning to set up virtual machines on my laptop, but ran into problems with both virtualbox and vmware, and i'm not smart enough to use kvm [15:47] aquarius, seeder in email. [15:48] statik, last time I tried running virtualbox stuff I could use the VM but couldn't make it see the network, and I tried fiddling with all the tun/tap stuff and got horribly confused and borked actual ordinary networknig on my laptop :( [15:49] ouch. i had total system lockup when i tried it [15:49] *nod* virtualisation seems like a good idea, it just doesn't like me very much afaict :( [15:50] vmware works well for me on stable ubuntu releases, but it never works with the newer kernels in the dev version of ubuntu, which is what i'm always running [15:59] statik: I have hardy and jaunty kvm's, the only weird config bits is the bridge configuration, once you have the bridge working, kvm works like a charm [16:00] statik: let me know if you need to double-check any kvm config file ;) [16:02] does vmware still laugh at you when you try to install vmware inside vmware? [16:04] Chipaca: only you tried to do such thing :) [16:09] verterok, ah cool, thanks! [16:14] verterok: evidently not; the dialog says "You just had to try, didn't you?" or something like that [16:14] Chipaca: :) [16:18] * aquarius has virtualbox crash :( [16:22] I'm using Karmic in VMware Fusion right now [16:22] it's not been too bad. [16:23] Chipaca: it laughed as it ate my OS when I booted the host ubuntu system as the guest one [16:23] Have to rebuild the vmware tools each time the kernel gets revved, though. [16:23] CardinalFang, so that's two reviews based on "it looks ok on my machine" [16:23] aquarius, Thanks! [16:23] I've taken an action to do see if someone can do a "How to set up KVM easily and awesomely" workshop at next UDS. [16:23] does anyone have two machines, or working virtualisation, so they can test this last branch...then we can merge it and release 0.3 [16:24] I personally have the same problem and I see other people having it so I can't be the only one. [16:24] aquarius, i do [16:24] aquarius, what branch should i test? [16:24] ooo, cool. kenvandine, CardinalFang should be able to talk you through the branch, I think... [16:49] CardinalFang, what branch should i try? [16:50] kenvandine, https://code.edge.launchpad.net/~cmiller/desktopcouch/replication-daemon/+merge/10740 [16:51] ok [16:54] kenvandine, Problems: couchdb binds to addr 127.0.0.1 . You must turn off any admin authentication it's doing, too. Finally, add records on each machine to tell it to look for the other machine. Then, run the replicators on each. [16:54] CardinalFang, how? [16:54] * CardinalFang mails program to seed data into desktopcouch. [16:55] kenvandine, er, which part? [16:58] all of it :) [16:58] ok [17:18] CardinalFang, getting an error [17:18] running desktopcouch-pair [17:18] return server.desktopCouch.getPort() [17:18] AttributeError: _DeferredMethod instance has no attribute 'getPort' [17:18] http://pastebin.ubuntu.com/259931/ [17:19] kenvandine, Er, yeah, I didn't change that. [17:20] kenvandine, I can fix it, though. [17:20] thx [17:26] kenvandine, well, I thought so. Please, $ ps xww |grep desktopcouch-servic[e] [17:30] CardinalFang, it's running [17:30] this is a python error though [17:30] from desktopcouch-pair [17:36] kenvandine, Weird! aquarius, any idea why ken can't use the getPort dbus function? [17:36] kenvandine: #419324 and #419326 [17:37] CardinalFang, TypeError: find_port() takes no arguments (1 given) [17:38] why is it returning a deferred rather than a d-bus object? [17:39] aquarius, Ah, I think I got it. Never mind. [17:39] find_port has changed [17:40] Reads the portnumber from the logfile, now that it is written there by couchdb. [17:41] dobey, thx [17:44] ok, i'm off to get some lunch [17:47] CardinalFang, fixing it on your branch? [17:48] kenvandine, yes. Maybe discovering that it never should have worked. Mmm, schroedinbugs. [17:57] dobey: hi, would you be the best person on this team to speak to about python-oauth? [18:00] I'm off for dinner with my daughter. Will be back later if people need me [18:00] ttfn [18:01] dobey: ah, in fact it's use that has contributed some 1.0a support to them [18:10] james_w, dobey is out for lunch... but i think he would be the best person [18:11] cool, thanks [18:11] we need to get that unblocked soon [18:34] dobey, so if you set the applet to "Never" displayed... how do you get to preferences? [18:35] kenvandine: i'm not sure yet :) [18:35] james_w: hey [18:35] hehe [18:35] hey dobey [18:35] what's up? [18:35] * kenvandine can't get it back now :) [18:35] * kenvandine unsets it [18:36] ubuntuone needs python-oauth, ubuntuone needs to be in main, therefore python-oauth needs to be in main [18:36] kenvandine: you can change it in ~/.config/ubuntuone/ubuntuone-client.conf [18:36] python-oauth had a known security problem, which you fixed [18:36] so we need to get that fixed version in to karmic [18:36] but it changes API [18:36] james_w: well, my changes weren't committed [18:36] dobey, thx [18:37] so we need some co-ordination between the current users [18:37] http://code.google.com/p/oauth/source/detail?r=1092 [18:37] james_w: but yes, to fix 1.0a in the server side API of python-oauth, it required API changes [18:37] CardinalFang, any luck with that find_port issue? [18:37] james_w: and ubuntuone is still using it's own copy of oauth.py [18:37] urgh [18:37] kenvandine, Not yet. DBus is weird. [18:38] patched or unpatched? [18:38] CardinalFang, shouldn't it just be removing pid? [18:38] find_port() [18:38] unpatched (though i'm not sure what that question means exactly) [18:38] kenvandine, Yes, but there's a separate problem. [18:38] does it use the old API or the new one? [18:38] ok [18:38] CardinalFang, i'll wait :) [18:39] :( Sorry! [18:39] i.e. do I need to get changes to u1 to allow it to use the fixed oauth? [18:39] no worries [18:39] just getting a little stressed :) [18:39] james_w: well, the client doesn't use the server API, but the revision of oauth.py in ubuntuone is the pre-1.0a code [18:39] james_w: upstream is still broken [18:39] james_w: and is a horrible pain in the ass to deal with [18:39] what about the change I just pointed to? [18:40] james_w: yes, it's not complete. and it takes forever for upstream to reply to any e-mails/issues/etc... [18:40] but uploading a fixed one won't break u1, which is good [18:40] no, changing python-oauth won't break u1 [18:40] and why do you use a private copy of oauth.py? [18:40] because we've always used a private copy [18:41] and i haven't finished poking at the backporting [18:41] ok [18:41] and at this point i am set on the idea that it just needs to be forked [18:41] but i haven't had time to fork it with trying to get features in for ubuntuone-client, before the freeze [18:43] the "security issue" isn't entirely clear, and i don't think we are "vulnerable" either, given the way we're using oauth [18:43] i also don't think 1.0a actually fixes the problem, but eh [18:47] bug 419365 filed for tracking [18:47] Launchpad bug 419365 in ubuntuone-storage-protocol "Uses embedded copy of oauth.py" [High,Triaged] https://launchpad.net/bugs/419365 [18:48] if you fork let me know and we'll get that packaged as well, and move launchpadlib as well [18:48] I don't want more than one copy of this code in Ubuntu [18:48] james_w: do we have any time to get it in for karmic? [18:48] get what in? [18:49] james_w: a fork [18:49] maybe [18:49] in main? [18:49] it wouldn't make to much difference to karmic whether it was a fork or not [18:55] james_w: right. i was just asking about it in regards to feature freeze [18:56] james_w: entirely new software doesn't tend to get added after FF does it? [18:56] not without good reason [18:57] but, as I said, the karmic package is a defacto fork, so even if you want to fork upstream I don't think there's much benefit to getting that fork in to karmic [18:57] unless you plan to rewrite it [18:58] also the changes to upstream oauth.py don't require 1.0a usage. so it doesn't exactly "fix" the issue [18:58] 1.0 still works (though servers still need to update for the API change, even if they just ignore it) [18:58] well, 1.0a explicitly allows both sides to work in both versions, and detect what the other side is doing [18:59] so I think that is possibly reasonable [18:59] it at least allows 1.0a, which is a step in the right direction [18:59] the 1.0a spec doesn't support 1.0 working [19:01] but it doesn't clash with 1.0, intentionally [19:02] if you don't get oauth_callback in the first request then it is 1.0 [19:04] kenvandine, Okay, it doesn't crash now. I also stopped the pairing tool from allowing local pairing, because it doesn't work anyway. [19:04] assumably [19:04] or it could just be a broken client [19:04] CardinalFang, locally, like on the same box? [19:04] CardinalFang, same branch? [19:04] kenvandine, same branch. Locally on this network. [19:04] relying on any means of detecting a version other than the oauth_version parameter seems like a rather bad idea to me [19:04] humm... so no pairing? [19:05] i am confused [19:05] dobey: given that the didn't change oauth_version it makes it rather difficult to rely on it for this change [19:05] and if the problem is in the protocol, continuing to allow the broken protocol to work doesn't exactly seem like a fix to me [19:06] kenvandine, no pairing. couchdb has no authentication yet, so we can't safely replicate anyway. [19:06] humm.. so what am i testing then? [19:07] just the cloud? [19:07] no, but they wanted to leave that choice up to the people using it [19:08] james_w: and my fork will probably be considered more of a rewrite than a fork [19:08] e.g. you said u1 isn't vulnerable, so forcing upgrades of clients for something that isn't a problem wouldn't be great [19:08] kenvandine, well, I haven't changed the pairing program until just now, so I'm glad you caught that API change. I'm don't think there's anything new to test. [19:08] by all means, i generally like choice, but sometimes it's just asking for trouble [19:08] see "smooth upgrade path" on http://wiki.oauth.net/OAuth-Session-Fixation-Advisory [19:08] james_w: i said i don't think we are actually "vulnerable" [19:09] since all our oauth bits are done through callbacks and such [19:09] CardinalFang, ok... did you seeding script add U1? [19:09] s/you/your [19:10] kenvandine, No, but the pairing app does add U1. U1 isn't set up to accept replication yet, though. [19:10] ok [19:10] so pairing is getting punted for 0.3? [19:11] kenvandine, yep. All the features are there, but we can't safely use them for local, and can't use them at all for U1. [19:11] ok... [19:12] CardinalFang, let me know when it is ready for release [19:13] thisfred, aquarius, Do you mind reviewing rev [51 -- 55] ? https://code.edge.launchpad.net/~cmiller/desktopcouch/replication-daemon/+merge/10740 [19:14] None are earth-shattering. [19:15] damn, I was hoping for earth shattering ;) [19:15] Earth-shattering takes longer to review! [19:15] CardinalFang: it's the next thing on my queue, after reviewing current branch [19:15] true, but it's more fun :) [19:16] earth shattering code -> earth scorching reviews :) [19:32] CardinalFang: any particular areas I should concern myself with? I don't see anything wrong (other than long lines...) and the tests pass, so my approve stands [19:33] thisfred, Eyeball the diffs, to see if they make sense to you. Releasing is a big deal, or else I'd just say they're so simple that it's fine. [19:33] Actually, I have a bug I hope I can fix, too. [19:33] thisfred, do you know the couchdb py module well? [19:36] couchdb-python you mean? a little [19:38] CardinalFang: I've dug through the code before [19:38] so hit me with any questions [19:42] thisfred, I want to post a new document to database "_replicate". I just noticed that serverobj["_replicate"] gives a value error because "_replicate" isn't a valid database for user use. [19:42] I'm hacking out my use of serverobj.__getitem__ and its check, now. [19:43] It's better. erverError: (405, '') [19:43] It's better. ServerError: (405, '') [19:43] CardinalFang, i have a gripe... [19:43] +""""exec ${PYTHON:-python} -t $0 "$@";" """ [19:43] dpkg hates that! [19:43] please just use #!/usr/bin/python [19:43] kenvandine, Heh., well, you can replace that and the #!/bin/sh above it with "#!/usr/bin/python" [19:44] yeah... i did that with pair.py :) [19:44] * CardinalFang says meekly "Okay.". [19:44] dpkg wouldn't build a package with it :) [19:44] i don't see why dpkg gets so upset over it, but it does [20:06] kenvandine, Hey! I have a new change to the source. You knew that would happen, right? [20:06] yeah :-D [20:11] kenvandine, Okay, what's arriving now, I am happy with. [20:11] ok [20:11] merged into trunk yet? [20:16] Hrm, does PQM handle it too? [20:16] Is there any way to change the dir that ubuntone uses? [20:16] does pqm handle what? [20:18] desktopcouch. I just tried submitting a branch to it. [20:18] * CardinalFang waits for email. [20:19] no [20:21] CardinalFang, we merge it manually [20:22] with tarmac [20:22] not entirely 'manually' [20:22] yeah [20:22] CardinalFang, http://bazaar.launchpad.net/~cmiller/desktopcouch/replication-daemon/revision/56 reverses the meaning of the if statements at lines 96 and 104, doesn't it? [20:22] aquarius, it sure does. It was backwards. :( [20:22] ah, good, OK :) [20:23] cor, what's %r in a format string? repr(whatever) ? [20:24] * aquarius approves up to r56 [20:25] * dobey does some updates [20:25] vbabiy, there isn't at the moment, but work is going on to make that happen [20:25] and python-avahi [20:25] and then hopefully desktopcouch tests will all pass [20:25] aquarius: any idea on date? [20:26] vbabiy, I haven't, but someone else will have, I suspect. statik? :) [20:26] lies [20:27] cool, i can land desktopcouch branches with tarmac now [20:27] or rather, i can land them and have it run tests [20:28] * CardinalFang will learn about tarmac one day. [20:28] do i need to land things? [20:32] re [20:32] CardinalFang, aquarius: hows the last desktopcouch branch looking? [20:32] statik, It replicates! At least, between users on my box, using external addresses. [20:33] happiness and glory [20:33] has it been through review yet? [20:33] nobody answered me :-/ [20:33] * statik reads the backlog [20:34] dobey, what's the question? [20:34] CardinalFang, aquarius: are there dc branches that need to land [20:34] dobey, about landing, i'm not sure yet just looking at the status of the last branch now [20:34] yay, it looks approved [20:34] ah, well CardinalFang was asking if pqm dealt with it [20:34] dobey, probably. We're looking at https://code.edge.launchpad.net/~cmiller/desktopcouch/replication-daemon [20:34] and i just went through the trouble of fixing my machine to be able to run desktopcouch tests [20:35] i see stuart approved it 10 minutes ago, so i will land it with tarmac now [20:35] CardinalFang, can you set the commit message for that merge proposal? [20:35] heh, all that work for nothing [20:35] https://code.edge.launchpad.net/~cmiller/desktopcouch/replication-daemon/+merge/10740 [20:35] dobey, you are welcome to land it [20:36] * dobey wonders who on lp team to talk to about oauth [20:37] dobey, salgado and flacoste i think [20:40] statik, merge message set. [20:40] ok, landing now [20:42] it looks good to me, I think, yes [20:43] it is a big shame that local pairing isn't in it :( [20:45] CardinalFang, aquarius: I get this traceback when running tests, any ideas? Traceback (most recent call last): [20:45] File "/usr/lib/python2.6/dist-packages/twisted/trial/runner.py", line 555, in loadPackage [20:45] module = modinfo.load() [20:45] File "/usr/lib/python2.6/dist-packages/twisted/python/modules.py", line 380, in load [20:45] return self.pathEntry.pythonPath.moduleLoader(self.name) [20:45] File "/usr/lib/python2.6/dist-packages/twisted/python/reflect.py", line 456, in namedAny [20:45] topLevelPackage = _importAndCheckStack(trialname) [20:45] File "/home/emurphy/canonical/ubuntuone/shared/desktopcouch/trunk/desktopcouch/records/tests/test_couchwidget.py", line 25, in [20:45] [20:45] exceptions.ImportError: cannot import name URI [20:45] gah, shoulda pastebinned that [20:45] really? all the tests pass for me ;( [20:46] statik, is this specifically in CardinalFang's branch, or once that branch is merged to trunk? [20:46] aquarius, once it's merged to trunk [20:47] It should have been no merging, just appending to trunk, fwiw. [20:47] I'll merge trunk into the branch here and see if I get the same failure [20:47] I haven't touched any couchwidget stuff, but I'll look. [20:48] that _importAndCheckStack thing looks kinda fishy, like a false alarm. line 25 in test_couchwidget.py looks fine to me [20:48] cd .. [20:48] Hrm. "PASSED (successes=47)" [20:49] dobey: a user gets this when he starts the client from the command line: process 5009: The last reference on a connection was dropped without > closing the connection. This is a bug in an application. See > dbus_connection_unref() documentation for details. [20:49] erm? [20:49] I branch trunk, then bzr merge lp:~cmiller/desktopcouch/replication-daemon and it says "nothing to do" [20:49] confused [20:50] aquarius, it's already merged to trunk [20:50] oh. heh. [20:50] i'm just paranoid and run tests a lot [20:50] ok...all 47 tests pass in trunk too [20:50] joshuahoover: uhm. that's new to me [20:50] joshuahoover: and probably python, since we don't drop references... [20:51] dobey: ok, wanted to make sure it wasn't known or fixed already...thought you would know :) [20:51] statik, should you even have test_couchwidget? isn't it test_couchgrid now? [20:51] ooh, maybe it's a stale .pyc file [20:52] all better, it was a stale .pyc file [20:52] thanks! [20:52] Dang, I need a beer. [20:53] i need a quantum clone [20:55] dobey, so whenever you do one thing it does the opposite and one of you will definitely get it right? [20:55] CardinalFang, sing it, brother [20:55] * aquarius drinks some crappy lemon water [20:56] aquarius, CardinalFang: can you guys add some release notes here? https://code.edge.launchpad.net/desktopcouch/trunk/0.3 [20:57] omg release notes [20:57] aquarius: no. so that all of us can do different things in an infinite number of universes, simultaneously :) [20:57] kenvandine, here is the 0.3 tarball for desktopcouch: https://code.edge.launchpad.net/desktopcouch/trunk/0.3 [20:57] statik, aquarius, I'll try. [20:57] blimey, so, what's in 0.3, then? can I get a list of all branches that were in it to remind myself? :) [20:57] super! [20:58] have we...(hushed voice) released 0.3? [20:58] * aquarius watches the world tilt on its axis [21:01] CardinalFang, so, I think in this release there is: replication, oauth token creation (although compulsory oauth is not enabled), the contacts picker, couchgrid, specifying databases/design docs in the filesystem to be loaded on startup, no stupid port finding tricks [21:01] CardinalFang, anything else massive I'm missing? [21:01] CardinalFang, also, a beer is due you for getting the replication stuff done ;) [21:01] dobey, do you know how to set a tag on trunk? i can create a tag locally on rev48 which is 0.3, but I can't figure out how to get the tag up to launchpad [21:02] you have to do it before you push [21:02] i see, so we almost need a dummy revision going through tarmac to do that then [21:02] i don't know if there is a bug already, but it seems like a bug that bzr doesn't let you do a push when you only change tags [21:02] yeah [21:03] and tags don't get picked up by the 'missing' command [21:03] i forgot to tag stuff before pushing today when i did releases [21:07] https://code.edge.launchpad.net/desktopcouch/trunk/0.3 Will you please describe the "specifying databases" item? [21:07] aquarius, ^ [21:10] CardinalFang, done. I knew there was a reason I docuemnted all this in teh readme ;) [21:11] Great! [21:13] statik, question about the freeze. Presumably we can still merrily commit new features to desktopcouch, we just don't push the packages into karmic? dobey, doesn't that also apply to -protocol etc? [21:14] aquarius, yeah we just need to sort the series/branches before polluting trunk [21:14] aquarius, we'll have a karmic series that stays stable [21:14] and trunk will be crackful as ever [21:14] so I should avoid committing new stuff to DC trunk for now? [21:14] you can still write code and push branches up for review [21:14] aquarius: yeah, we need to figure out the branching/maintanence stuff first [21:15] right, gotcha, cheers [21:15] we should just get our story straight before merging branches [21:15] yeah [21:15] I was assuming that launchpad just had a button with "ahaha! karmic freeze at this point!" written on it or something :) [21:16] launchpad needs to let you propose one branch for merging into multiple others [21:16] although that probably wouldn't be too great [21:16] nevermind that [21:16] backporting fixes will just have to be painful i guess [21:18] theres a discussion on launchpad-dev about exactly that right now, proposing a bugfix branch for merge into multiple series [21:18] problem is that it won't work so well [21:18] yeah, especially once trunk starts diverging [21:18] because you probably want to propose a branch from trunk [21:19] in which case the merge to !trunk is going to probably end up bringing in all the changes from trunk as well [21:19] kenvandine, so now that the desktopcouch 0.3 release is out, do you still want to do the package update for it, or should I do that now? [21:24] good work, CardinalFang, statik, thisfred. We've done it. Thank gawd. :) [21:24] hi fives all around! [21:25] Hurrah! [21:26] aquarius, thisfred, CardinalFang: maybe we can have a call tomorrow to brainstorm the OAuth stuff and the FFE for enabling contact sync from the cloud [21:26] thank you guys for working so hard on this [21:26] statik, yep, good idea on the call [21:27] sounds like a good idea [21:27] Yes, please. [21:27] I have other oauth questions, which I was just about to harass aquarius about. [21:28] It shouldn't be on the call, prob. [21:30] go for it [21:34] aquarius, I don't yet understand what generates oauth data. I need to get my machine-a and machine-b paired. Presumably they exchange information with something or between themselves. Where does that information come from? [21:34] And, which parts are shared and which are secret? [21:34] it's done on ini file creation. [21:35] CardinalFang, desktopcouch/start_local_couchdb.py, create_ini_file [21:35] CardinalFang, that creates a random token, token secret, consumer key, and consumer secret for this server [21:36] aquarius, I guess I'm still surprised that everything shares information. Is this the way I authenticate others, or is this how others authenticate me? [21:36] (we could define a comsumer key and consumer secret for desktopcouch generally and use them; they don't all need to be random. I just didn't.) [21:36] if other people know that key/secret/token/secret then they can authenticate to me [21:37] so someone else can read and write my data if they have those key/secret/token/secret [21:37] So, I give some or all parts away to let others prove themselves to me later? [21:38] yes. [21:38] essentially, it's a password. it's just broken up into four bits. [21:38] the thing here is: we could issue a separate "password" to each consumer. We just don't, at the moment. [21:39] kenvandine, i'm creating the new package for desktopcouch per bug 416591 [21:39] Ah. That's what I expected. [21:39] So we're not particularly getting the benefits of oauth (other than unsniffability over the network) [21:39] Launchpad bug 416591 in desktopcouch "New version, 0.3" [Undecided,New] https://launchpad.net/bugs/416591 [21:40] CardinalFang, part of the reason for that is that couch implements two-legged oauth (you give people an access token) rather than the more standard 3-legged oauth (they ask for an access token and the user gets sent somewhere in a browser to say "yes they can have an access token") [21:40] hey. what's new in 0.93? anything special one should test? [21:43] What is the reason for ubuntuone being one of the more memory-demanding processes in standby? (comparable to compiz and Xorg..) [21:44] Or rather - in idle [21:45] arand, is it resident memory or cached of files on disk? [21:47] Hmm, not sure, would top tell me? [21:52] CardinalFang: category "resident memory" shows 36mb, which seems comparatively high. [21:56] statik, cool [21:57] statik, sorry i had to walk down the street to meet my daughter's school bus, which was late [21:58] kenvandine, no worries. i'm hitting an error in the pbuilder about POTFILES.in missing [21:58] humm [21:58] * kenvandine looks [21:59] statik, it needs to be added to MANIFEST.in [21:59] ah, i see that now. so that needs a new branch and a new tarball [21:59] so that means a new tarball [21:59] 0.3.1 ? or whats the best way to handle that [21:59] just delete the old one [21:59] ok [21:59] it has only been a few minutes :) [22:00] move the tag [22:01] if you're going to re-use 0.3 [22:01] yep, good point [22:01] yes... definately [22:01] freaking launchpad ajax, i just deleted the wrong tarball [22:01] i deleted the one i meant to delete, but the labels were wrong [22:03] kenvandine, so this looks right to you?=== modified file 'MANIFEST.in' [22:03] --- MANIFEST.in 2009-08-20 09:14:26 +0000 [22:03] +++ MANIFEST.in 2009-08-26 21:02:34 +0000 [22:03] @@ -2,6 +2,7 @@ [22:03] recursive-include data *.tmpl [22:03] include desktopcouch-pair.desktop.in [22:03] include po/desktopcouch.pot [22:03] +include po/POTFILES.in [22:03] yes [22:03] huh [22:03] I'm gonna hit the road, gang, unless anyone needs me for anything right now? [22:03] -include po/desktopcouch.pot [22:03] later aquarius! [22:04] dobey, don't need that anymore? [22:04] oh [22:04] kenvandine: what for? [22:04] ok [22:04] aquarius, that's all for me. Thank you! Good night. [22:04] kenvandine: it should get regenerated at every build [22:04] yeah... we need that [22:04] humm [22:04] * kenvandine didn't think it was [22:04] POTFILES.in is needed [22:04] it should be [22:04] yeah... your right [22:04] it being there isn't harmful [22:04] with that it gets generated :) [22:04] just wondering why it's there :) [22:05] cause we didn't have POTFILES.in :) [22:05] heh [22:05] i don't think it was being generated [22:05] Hrm. Is that for translations? [22:05] yes [22:05] statik, so yeah -include po/desktopcouch.pot [22:05] http://identi.ca/notice/8633544 <- everyone go favorite this :) [22:06] done :) [22:06] i hope nobody downloaded the "broken" 0.3 [22:06] since it's been twitted [22:09] yay, that built nicely and installed(upgraded) just fine [22:11] woot [22:11] * kenvandine wished we could pair! [22:11] but oh well [22:11] kenvandine, do i need to attach the .changes file too, or just the .dsc, orig, and diff? [22:12] just point the bug at the packaging branch [22:12] oh you can't... [22:12] give me a debdiff [22:12] actually no [22:13] you just changed the changelog right? [22:13] new entry in changelog, added a depends on python-setuptools in debian/rules (and mentioned it in the changelog [22:13] a depends or build depends? [22:13] build-depends [22:13] ok [22:14] can you push the branch somewhere? [22:14] i don't have it in a branch [22:14] ok [22:14] give me those 2 files then :) [22:14] or a debdiff [22:14] * kenvandine is easy [22:15] ok, but the debdiff is going to be big because of the changed .orig.tar.gz right? [22:15] just give me those files :) [22:15] or the changed lines and i can apply them [22:16] ok, so all the work i did to generate and test the package is going to be wasted, and i won't get credit for the upload :P [22:16] hehe [22:16] you will get credit [22:16] you are in the changelog [22:16] i'm just teasing, i don't mind [22:17] i have a pretty long changelog already prepared for 0.3 actually [22:17] just trying to learn the right way to do it. when i saw you process the other new versions, i thought you attached the .dsc and orig.tar.gz to the bug [22:17] oh nice [22:17] that will be better then [22:17] statik, there are 22 different ways of doing it [22:17] :) [22:17] so i'm learning [22:18] this one is in a packaging branch under ~ubuntu-desktop [22:20] kenvandine, i attached debian/control and debian/changelog, along with the correct orig.tar.gz to bug 416591 [22:20] Launchpad bug 416591 in desktopcouch "New version, 0.3" [Undecided,New] https://launchpad.net/bugs/416591 [22:20] thx [22:20] i've testbuilt in a pbuilder, and installed both binary packages to be sure they upgraded [22:21] i've moved the tag on trunk, and committed the change to manifest.in [22:21] i've deleted the old tarball from the project page, and uploaded a new tarball with new gpg signature [22:21] anything i'm missing? [22:21] kenvandine, i'm really sorry this one was so close to the wire, i really did try to schedule it with plenty of buffer but it didn't work out that way [22:22] heh [22:22] statik, understand [22:23] - python-gtk2, [22:23] - python-gnomekeyring, [22:23] - python-oauth [22:23] is that right? [22:23] welcome to schedules with freezes :) [22:23] oh [22:23] is desktopcouch using python-oauth? [22:23] yes [22:23] not for long! :) [22:23] yeah yeah... but for right now it is :) [22:23] statik, that doesn't look right [22:24] oh... nm... that is the diff from 0.2 [22:24] i had already fixed it up quite a bit in my 0.3 branch [22:24] kenvandine, what looks wrong? the couchgrid and contactspicker and pairing tool use gtk right? [22:24] and we're storing tokens in the keyring [22:24] and signing requests for oauth [22:25] using oauth [22:25] yes... your changes are removing the deps [22:25] later! [22:25] wow, you totally saved me on that one then [22:25] later dobey [22:25] statik, no worries [22:25] kenvandine to the rescue! [22:26] i really do want to learn how to use these packagebranches, it makes so much sense to be able to accumulate packaging fixes there [22:27] kenvandine, need anything else from me? my ride is leaving, but you have my phone number if you need it [22:28] statik, nope... i just pushed it [22:28] fabulous! i should buy you a deer at the next UDS [22:28] hehe [22:28] no deers for me [22:28] i prefer beer [22:29] yeah, but i'm sure everyone buys you beer. if i buy you a deer then it will be memorable [22:29] alright, back for more pain later [22:30] hehe