=== mattgriffin changed the topic of #ubuntuone to: | https://one.ubuntu.com | https://launchpad.net/ubuntuone | Please honk if you want a music store === mr_ian_ is now known as mr_ian === aquarius changed the topic of #ubuntuone to: Talk to aquarius if you have problems | https://one.ubuntu.com | https://launchpad.net/ubuntuone | Please honk if you want a music store [14:14] hello, cann anyone tell me why I'm getting a 500 when I try to view contact data in ubuntuone? [14:18] mandel, hm, looks fine for my contacts. Any additional messages? [14:18] nah, just a 500 [14:19] but I'm testing my application on desktopcouch and I'm using the new attachment code added my Chad, I was wondering if that makes any diff [14:20] is there anyone that can look at the logs looking at my user? I;m wondering if is the avatar added by my app as an attachment 'cause evo and my code have no problem [14:21] I forgot to mention that the source of the docs in couchdb is my app [14:21] well i'm sure there's an OOPS id :) [14:25] Morning, all [14:27] * rtgz looks at his watch - 16:27... [14:27] dobey, anyway to get it? [14:28] (modulo localtime) [14:29] mandel: it should have been printed on the error page. [14:29] but it's working here ok too... [14:29] * aquarius checks too [14:29] really? 'cause I see the nice robot like cartoon but nothing else... [14:30] I'm wondering if it has to do with the format of my doc... [14:30] mandel: yes. but maybe the robot image thing broke that.. jblount ^ [14:32] is there any kind of test error page, when you get there it just manually crashes giving you the overview how the error message will be displayed? something like /oh-noes/ ... [14:34] he he bug #498194 [14:34] Launchpad bug 498194 in ubuntuone-client "Syncronization marks are incorrect" [Undecided,New] https://launchpad.net/bugs/498194 [14:35] mandel, interesting. I wonder if your contact records have changed and the web UI can't cope with that? That would be bad, if so. (Contacts web UI works for me too, btw) [14:36] the contacts that I create through the web ui work perfectly [14:36] * rtgz has a virtual machine for u1 tests, need new virtual lp account for a virtual u1 server :) [14:36] mandel, is it only one account that causes this? [14:36] mandel, sorry, one contact entry [14:36] the ones generated by my client are listed, but If I try to access the data, they brake [14:37] mandel, i guess the server side devs might be interested in the record itself :) [14:37] rtgz, all the one generated from my side, Ill post one doc example, give me a sec [14:38] this is what I've got in my desktopcouch: http://pastebin.com/m4aab8848 [14:39] only diff I can thing is that my collections do not have an _order, could that be it? [14:39] and I did not added it because it was not in the spec [14:40] * rtgz is waiting for replication to break his contact list as well... [14:41] this is how contact sharing is implemented ATM :) [14:42] mandel: which url are you getting the errors on? [14:43] thisfred, this is one of them: https://one.ubuntu.com/contacts/1019e2b361cd4217a9bdb25cdd736878/details/ [14:43] but I can replicate the error with any contact that was made in my side [14:44] I can give you one that has no problems with my account, wanna it? [14:46] mandel: I can not see your contacts so, that's ok, I was just looking for the shape of the url [14:47] thisfred, is just test data, so no problem... [14:51] bah, make start keeps dying [14:51] which makes it hard to run up a test version of the contacts web UI to put mandel's contact in :) [14:51] yup [14:51] Something has gone wrong (500) [14:52] rtgz, you reproduced the error? [14:52] * aquarius updates the sourcedeps. Again. [14:52] mandel, yup [14:52] le sigh [14:52] so, what could it be, 'cause I think there is nothing diff in the doc [14:53] and edge server dies in the same way, no extra info to report.. :( [14:53] mandel, what is your user id, so that we can grep through the logs for that? [14:54] mandel@themacaque.com [14:54] or mandel, is the email or the username? [14:54] mandel: I mean the numerical id you'll find in the couchdb urls [14:55] oh, webm0nk3y found it I think [14:56] you mean the id of the contact? [14:56] that would be this: 1019e2b361cd4217a9bdb25cdd736878 [14:57] mandel, i guess this is where replication goes to should be part of the URL, there is some kind of ID as well... [14:58] mandel: rtgz yeah, the replication url is something like u/foo/bar/xxx/contacts, xxx would be your user id. But as I said, we've found it. No errors logged [14:59] rtgz, thisfred, ok [14:59] so the error is not in couchdb, or in the python layer around it, it seems. The web ui itself maybe doesn't know how to deal with the attachment, although I find it strange that it would even notice it;s therte [14:59] there [15:00] desktop+ meeting begins. aquarius, chad, dobey, jblount, mandel, rodrigo, rtgz, teknico, urbanape, vds: you know how it works: say "me" to get a turn, then in your turn say your done/todo/blocked status. [15:00] thisfred, the error happens with a simple doc with no attachment [15:00] me [15:00] aquarius: I'll try locally [15:00] me [15:00] me [15:01] * Chipaca notes rodrigo and teknico are on holiday and thus won't me [15:01] me [15:02] aquarius: dobey: urbanape: vds: PING [15:03] me [15:04] me [15:04] aquarius aquarius aquarius [15:04] urbanape urbanape urbanape [15:05] CardinalFang: go [15:05] DONE: Sorted distro d-c bugs. Some progress on releasing desktopcouch stable update. Re-tagged d-c on releases, as the source-package bzr tools require tags. [15:05] TODO: file a bug on tarmac about discarding tags. Finish releasing d-c. Pester james_w a lot. [15:05] BLOCKED: Understanding. [15:05] rtgz, give me the low-down! [15:05] DONE: Marshallers and file emblems fixed (waiting for merge), diagnosed nautilus crash with Share on Ubuntu One dialog. [15:05] TODO: Update ubuntuone-client-diagnose with the latest bug info. Check what can be done with folder emblems (got some ideas) - bug #440848. Dig into nautilus to find out why our routine is called twice. Ping dobey regarding bug #498136. [15:05] BLOCK: none [15:05] Launchpad bug 440848 in ubuntuone-client "UbuntuOne sync status emblems should apply to folders as well as files" [Wishlist,Triaged] https://launchpad.net/bugs/440848 [15:05] Launchpad bug 498136 in ubuntuone-client "All "Shared With Me" folder items have emblems broken" [Undecided,New] https://launchpad.net/bugs/498136 [15:05] jblount, *whatever I should say* [15:06] DONE: Took last random holiday off [15:06] TODO: Make forum icon overlay thing, Work on layout issues for /plans and /support/account-assistance [15:06] BLCOKED: Nope [15:06] mandel: you! [15:07] DONE: Played with tags and group implementation in my app [15:07] TODO: Update UI in macaco-contact to refect changes [15:07] BLOCK: NONE [15:07] vds, go [15:07] DONE: more testing sync with mobile phones, finally got a new sim card for the testing handset, started a branch to fix problem noticed testing with handset, code review [15:07] TODO: more testing with mobile devices, setup funambol to send sms [15:07] BLOCKED: no [15:07] dobey: vai! [15:07] ☺ DONE: More initial new client work [15:07] ☹ TODO: Reviews, New Client Code, E-mail motu-council [15:07] ☹ BLCK: None. [15:08] aquarius, urbanape: fight! [15:08] me [15:08] woops [15:08] mandel: when did you first notice the 500 error? [15:08] DONE: Made some progress on manifest branch, ubuntuone-servers branch finally passed PQM, review of next branch should be easier [15:08] urbanape: go [15:08] (our logs may be up to an hour out of date) [15:09] TODO: Get other branch reviewed/submitted, keep moving on manifest branch [15:09] BLOCK: None [15:09] thisfred, yesterday, during the night, but that is because I've never looked before [15:09] aquarius: you're up, I guess [15:09] mandel: ah ok, so we should have some in the logs then [15:10] thisfred, yes, and it is quite easy to reproduce, any contact I do in my side will break [15:10] mandel: I'm going to try to reproduce this on my machine, if I can set the web ui and server all up correctly [15:10] desktop+ EOM. CardinalFang, dobey, jblount, mandel, rtgz, urbanape, vds: thanks! [15:11] CardinalFang: are you really blocked by (presumably your) understanding? is there anything we can do to unblock you? [15:12] thisfred, do you need any testing data? I can send you a bunch of different docs [15:13] mandel: the one you pastebinned above fails yes? I'll try that one first [15:14] thanks for helping us track this down! [15:14] thisfred, yes that one fails also rtgz got the same error. [15:15] taking an early lunch to run some errands. BBL. [15:15] Chipaca, I may not be blocked. I'm just bellayaching about how releasing is supposed to be easier and it isn't so far. [15:15] thisfred, I hope I could help more, but from the client side there i no much I can do :P [15:15] CardinalFang: apt-get install digestive-salts? [15:16] It Depends on ubuntu-patience, which isn't available. [15:17] grr, lp apparently does funky stuff with diffs now [15:20] wrote "invalidate extension date". Yeah, extension does not want to date again :/ [15:31] verterok: is get_metadata going to start sending full paths? [15:31] dobey: fullpaths? [15:32] verterok: it looks like get_metadata currently has only "foo/blah" as the "path" in the dict. i think that will be problematic with the user defined folders, no? [15:32] dobey: get_metadata returns the metadata of a full path [15:33] dobey: no, because the info also contains the share_id (probably will be renamed to volume_id) [15:34] verterok: but that's not the local path of the folder is it? [15:34] dobey: yes, it is :) [15:34] dobey: but we can add a more fine grained methods for whatever metadata you guys need ;) [15:34] oh, it's not a hash id? [15:34] dobey: hash_id? [15:35] verterok: share_id [15:35] verterok: it's a path, and not a hash? [15:35] dobey: it's a uuid [15:35] or a uuid [15:35] i need the path [15:36] dobey: but you already have the path... it's the argument passed to get_metadata..right? [15:36] verterok: yes and no. we have the path when we make the call, but not when we finish it [15:36] verterok: async APIs are hell :) [15:37] dobey: don't know about that, I use twisted ;) [15:37] twisted doesn't do dbus :) [15:37] dobey: it does ;) [15:37] but yes, it's easier to deal with in a dynamic language like python [15:38] dobey: so, you need the fullpath in get_metadata? for what are going to use get_metadata output? [15:38] dobey: maybe we can add a new and optimized method...get_metadata is quite generic [15:38] verterok: we're calling get_metadata to determine what files need updating [15:39] dobey: so, we'r doing a get_metadata call for each path? [15:40] verterok: for every file under Ubuntu One, that gets displayed in an open nautilus window, yes [15:40] ouch? :) [15:40] sounds like that should be a single call [15:40] Chipaca: I was thinking on the same thing :) [15:41] dobey, Chipaca: I was also thinking in a get_me_the_sync_status_plzzz() method [15:41] there is no way to do a single call for it in nautilus [15:41] dobey: no? [15:41] nope [15:42] dobey: get_metadata_for_all_files_in(directory) ? [15:42] we get a update_fileinfo call for every file individually, in the extension [15:42] Chipaca: we don't ever get a list of files [15:42] we'd have to do a lot of extra work to do that, and it would be very painful [15:43] though... one of rtgz's branches might make that a little easier, but not a lot [15:44] darn [15:44] dobey, don't see how any of the branches might help that :( [15:45] rtgz: the one that stores the weak references to NautilusFileInfos [15:45] dobey, yep, but there references like to go away after some time... [15:46] and we can't get a fs snapshot... this will mean more caching inside the plugin which already contains a lot of them. [15:47] well, the big problem is we will never know when all the files are ready to pass on to the get_all_metadata call [15:48] so yeah, can't do that [15:48] we might ... hm... there are 2 calls in nautilus - first for get_file_items and second get_background_items. The background one is called once per folder... we might try to collect all files, issue some combined request and later on wait for it to complete... [15:48] need to check this [15:50] that's not what you think it is [15:50] get_background_items() is the call to get the menu items to add when you right click on the "background" [15:51] verterok: ok, so we can't bunch up calls without more work than makes sense [15:51] dobey, true... debugged the sharing stuff for too long, now I try to attach these calls to whatever seems distantly connected. [15:51] verterok: can you make any async call response include everything that was passed in (that makes sense)? [15:51] hi sorry back now [15:52] if the path were just the full path, we'd be fine [15:52] i thought it was anyway [15:52] dobey: we can add the fullpath [15:52] ⚀ DONE: desktopcouch HTML API docs branch; bugfix branch for DC; add more DC docs to freedesktop.org [15:52] ⚁ TODO: test music store!; continue work on desktopcouch developer docs; publish DC HTML API docs somewhere (where?); look at Tomboy xml/html translator; work with rodrigo on Music Store; write up things learned at UDS; step-by-step guide to what happens during contact sync; hand off "pipe" to transfer data between two HTTP connections with twisted 9.0 to lucio's team; make tomboy first-sync experience nicer [15:52] ⚂ BLOCKED: not being able to think of where to put DC API docs because I am clearly stupid [15:52] Chipaca: /me don't understand, please elaborate? [15:53] dobey: please file a bug about this, so it don't get lost in the holidays ;) [15:53] mandel: one thing I see wrong with your document: you don't use uuids for the keys in the mergeable lists [15:53] verterok: do you have the full dbus api somewhere? (I know you put it on pastebin yesterday) [15:53] Chipaca: https://chinstrap.canonical.com/~guillo/syncdaemon_dbus_api.txt [15:53] I'm trying the web ui locally now [15:54] Chipaca: oops, not public url [15:54] thisfred, really? let me check [15:54] verterok: how about we just fix it right now :) [15:54] Chipaca: http://pastebin.ubuntu.com/343768/ [15:54] (although a bug needs to be filed too) [15:54] dobey, yes, otherwise desync will happen once more :) [15:55] thisfred, from my code => (self.id = uuid.uuid4().hex) [15:55] dobey: I'm doing something completely different right now, and want to get it done asap :) [15:55] verterok: i'll make the branch, it's trivial :) [15:55] dobey: ok :) [15:56] thisfred, they are Mergeable sets, for example phone_numbers is a dict of keys and data, just like the spec [15:56] dobey: thanks btw :) [15:56] verterok: schedule_next is blocking, right? [15:56] please please please merge my branches first, otherwise this would be the 3rd set of 3 branches :) [15:56] verterok: by blocking I mean not part of an async conversation [15:56] mandel: yes, but the keys need to be uuid4 uuids [15:57] if the spec isn't clear enough about that, we need to fix it [15:57] rtgz: no, your one branch will just need to be updated [15:57] Chipaca: I think it's a yes: self.action_queue.content_queue.schedule_next(share_id, node_id) [15:57] dobey, I can commit while it is in proposed state O_O? [15:58] yes [15:58] Chipaca: fwiw, all dbus calls can be async for the client it's just a matter of how the client make the call [15:58] Chipaca: yes, it's "blocking" [15:58] verterok: yes. So does the schedule_next throw a ContentQueueChanged? [15:58] mandel: I don't think the web ui should break over this though ;) [15:58] thisfred, you mean that my ids are not uuid4? as I said, I use uuid.uuid4().hex [15:59] honk, btw [15:59] thisfred, isn't that correct? [15:59] and yes, I agree, it should not break 'cause of that [15:59] mandel: hmm, we use a different notation [16:00] Chipaca: I can't find where ContentQueueChanged is fired [16:00] mandel: str(uuid.uuid4()) [16:00] right [16:00] thisfred, ahh let me know which one, so I can change the code (is a one liner) [16:00] ok, I'll change it [16:00] I'm pretty sure this is the issue. I'll file a bug against the web ui [16:01] Chipaca: found it [16:01] Chipaca: no, it's poking with the deque dirtectly [16:02] thisfred, ok I made the changes on my code, lets try it with a new doc [16:03] Chipaca: basically calling: self.waiting.remove(cmd) + self.waiting.appendleft(cmd) [16:03] verterok: me suena, me suena :) [16:03] mandel: let me know how that goes or if you want me to test it here [16:04] Chipaca: :) [16:04] Chipaca: should it fire a signal? [16:04] thisfred, will do, I'm working on it right now [16:05] verterok: I don't know, I just started going down the api aking myself that kind of quesiton :) [16:06] Chipaca: ok, please file a bug if you think it should ;) [16:06] it's not used outside of u1sdtool, and won't be used outside of that during lucid [16:06] so, probably not. But as soon as dobey needs to integrate it into the desktop, hell yes :) [16:08] thisfred, is going to take me longer than I though (Can't set '1f1526e8-df92-465a-9eb5-9c0ef304395e'. uuid-like keys are not allowed.) [16:09] I'll try to be quick to fix the issue [16:09] mandel: no hurry, let me know what happens, or chad/aquarius next week (I'll be on holiday) [16:10] thisfred, oki [16:12] dobey, sorry if I ping you too much, but what should I set in lp if i have committed a fix for "Needs Fixing" review, "Work in progress"? [16:14] rtgz: no. you just push your fix, and reply to the comment saying you fixed it [16:15] dobey, thanks :) [16:19] oh [16:19] launchpad is smarter now [16:20] so you don't even need to comment [16:20] it adds the commit info in the middle of the comments :) [16:21] verterok: https://code.edge.launchpad.net/~dobey/ubuntuone-client/getmetadata-fullpath/+merge/16340 :) [16:23] thisfred, made the changes, waiting for replication [16:24] cool [16:40] thisfred, confirmed, it was the format of the id [16:40] thisfred, interesting bug :P [16:41] mandel: yeah, python-desktopcouch enforces that format, but nothing else should take it for granted [16:42] mandel: note that not using that format will break stuff, in that your telephone numbers might not show up or be deleted, but it should not cause errors ;) [16:43] ?? couchdb _id has to have a specific format? [16:43] aquarius: not the id [16:43] aquarius: the keys in the mergeable lists [16:43] ah, ok [16:44] thisfred, from my point of view by problems was related with the fact that I was not relying on desktopcouch to convert my set to a dict, something very probably for people using other languages [16:44] I meant very probable, pardon my english [16:44] :P [16:44] mandel: yeah absolutely, so we need to make this a lot clearer [16:45] what's the problem with using different ids? [16:45] thisfred, adding it in the spec should be good enough [16:45] aquarius: that it will break [16:45] thisfred, what will break? sorry, I know I'm catching up late here [16:45] aquarius: we detect the uuids by the format with the dashes [16:45] oh... [16:45] so that's how we know if something's a mergeablelist? [16:45] oof [16:45] which is a hack, but there's really no other way [16:45] that's a bit magic. :) [16:46] that needs documenting, then :) [16:46] indeed [16:46] and: I haven't thought of another way, rather than there isn't one [16:47] so nobody stop looking for one! ;) [16:47] can't you just say that if it has _order it is a mergeable list? [16:48] for the list should be true, for a mergeable set, that is another story [16:48] mandel: no, because we have mergeable sets as well, or at least we don't always have _order I think [16:48] mandel, mergeablesets don't have _order, though [16:48] so if we have to have the hack for one we might as well be consistent about it :) [16:50] documenting should be enough for know, at least that way someone will be able to find the problem [16:51] is there a doc validator? [16:53] * aquarius wags the waggy finger of disapproval about validators. [16:53] :) [16:56] at least for the "bloody" id.... [16:57] or for debugging, not to use them but for this things, they do help [16:58] I agree [16:58] the ID's a weird one, though. I don't think we'd have thought to put that in a validator anyway :) [17:00] well, is good to know why happens, I'll get back to my code, I wanna get ahead during the xmas [17:07] laters [17:09] hey. new One user, general question .. i have a file on two different machines. have been updating between them manually. i upload through One, how will it know which version to sync from? is there a timestamp process .. or ?? [17:10] burzki, the easiest way to deal with that, at first, is to put that file in your Ubuntu One folder on only *one* of the machines, and let Ubuntu One copy it to the other machine [17:11] burzki, so, put the version that you consider "most recent" in the Ubuntu One folder, and wait until it's been synced to the other machine [17:13] aquarius, got it. thx. second part of my question relates to tomboy updates .. my notes on the two different machines are different sets of notes. is there a way i can use One to stnc my notes so that the two collections become one and are then complete on both machines? i'm concerned about one set wanting to 'write-over' the other .. ? [17:20] burzki, that should Just Work [17:20] burzki, one set should not overwrite the other [17:21] burzki, however! I haven't actually, strictly, tried that [17:21] so you might want to copy your .local/share/tomboy folder somewhere else just in case I'm wrong ;) [17:21] great. i'll try it. [17:21] aquarius: that sounds like the kind of thing that could go in a screencast [17:21] thanks for clarification. [17:22] Chipaca, syncing notes? [17:22] burzki, when it works, do please come back and tell us that it worked so we don't have to add the caveat for the next person who asks :) [17:22] aquarius: "I'm Joe User and already have stuff on two notebooks. This is how I start using Ubuntu One to keep things in sync automagically." [17:23] aquarius, will do [17:23] Chipaca, mm, good thought. the key thing is, for files, don't start off with two copies of the same file on the different machines, I think [17:23] Chipaca, do we plan to change that? or will we always get conflicts if a file is alredy in the Ubuntu One folder when we start syncing, even if the file being synced down and the file already in the folder are the same? [17:24] Chipaca, i agree. would be nice to have a page off the One homepage saying how all this works, or a little tutorial for this scene [17:25] aquarius: No Can Do; there is no application-independent way to tell that case apart from a conflict [17:25] aquarius: if they happen to be *exactly* the same, you shouldn't get a conflict, however [17:26] Chipaca, oh, really? I thought we conflicted even if the files were identical [17:26] aquarius: at one point we did. I believe we no longer conflict on file creation erroring out with ALREADY_EXISTS [17:27] __lucio__: can you confirm or deny ^? [17:27] ah, fantastic, that cheers me up no end, then :) [17:28] urbanape, I'm trying to switch to GOOG Chrome browser. How's the bindwood plugin coming? [17:29] <__lucio__> Chipaca, we dont conflict on file creation. but we had a bug where a user copied the same stuff to two machines and had conflicts, so we have a bug there [17:29] <__lucio__> aquarius, ^^ [17:30] __lucio__, cool, cheers [17:30] __lucio__, since if that's a bug, it will get fixed :) [17:31] thx folks [17:32] <__lucio__> aquarius, yeah, but i cant promise when, and i wouldnt bet critical stuff on that being fixed soon [17:32] <__lucio__> aquarius, someone asked for udfs, we are doing those, but we had to not do something else, you know :) [17:33] __lucio__, ah, I'm not asking for this conflict stuff to be fixed because I'm depending on it, it's just bitten me in the past :-) pure personal desire, which takes a back seat to stuff we actually need to get done! [17:33] aquarius: do not mention fixing bugs to __lucio__! You'll hear all about UDFs [17:33] aquarius: and you want UDFs for the music store, so shaddup :-D [17:33] * aquarius grins. I sho' do [17:36] CardinalFang: it's a bit blocked on getting the env into the plugin. Looks like the only way that's gonna happen is with an NPAPI plugin. [17:36] urbanape, yeah, I can't find a way around that, either :( [17:42] yeah, sad [18:09] btw, is tomboy note formatting fixed? [18:17] jblount: if I submit the new-folder-inline-edit branch, will we be able to get the CSS fixed up in time for the next deployment? [18:20] push it to zed :) [18:20] then you are guaranteed that the related UI branches get done before it's merged to trunk [18:22] is zed edge? [18:23] zed is like an edge for edge [18:23] edgy. [18:23] z.one.ubuntu.com [18:23] and how does one push to zed? [18:24] zed is not yet working, pushing to zed sounds like a bad idea. [18:25] urbanape: I can only give you a maybe, is there a bug that tracks the specific problems? [18:26] test driven development is like driving an SUV in the mountains [18:26] jblount: nope, but I'll make some. It's about the positioning of the actions icons [18:27] dobey: Overrated and not as safe as you think? [18:27] jblount: pretty much, yes [18:27] urbanape: That'd be super great. If you assign them to me with a tag "next" they are really likely to get done soon(ish). [18:28] jblount: not built for handling rough terrain [18:28] Also, is anyone else creeped out when they see Jonathan Coulton after meeting urbanape ? Cause I am. [18:29] jblount: will do [18:29] i haven't met jonathan coulton [18:29] urbanape: Also, z.one.ubuntu.com is for disruptive web ui changes, it will roll automatically every hour from lp:ubuntuone-servers/zed (or something) and will be using the production database. [18:41] verterok: i've been running contrib/test forever, and those tests aren't failing for me :( [18:42] dobey: the first error is from test_tools.py it should fail, as you changed get_metadata [18:42] dobey: there is a bug for the second failure [18:45] dobey: tests/syncdaemon/test_tools.py, line 141 is doing an assertion on the relative path [18:49] verterok: i changed the tests too [18:50] dobey: when? after proposing the branch? [18:50] oh test_tools i didn't realize was testing the same thing again [18:50] but that doesn't explain why it wasn't failing for me [18:51] dobey: right, it's weird that isn't failing [18:53] verterok: anyway, i pushed the fix [18:56] dobey: do we really need a different branch for the contrib/test change? [18:57] dobey: it's just an new option I forgot to mention in the merge proposal comment [19:02] verterok: maybe less so in this case, but one of the branches i worked on backporting, had unrelated changes to contrib/test, and was more work to backport. i don't know if this needs backporting also, but when i went to review the branch the comment said it was to fix the tests... but the contrib/test changes aren't to fix the tests, they were to help you encounter the failures [19:03] dobey: ok, sorry, I missed to add that to the comment [19:03] dobey: there is no need to backport this, as is justa fix to the a test [19:03] verterok: it also makes it easier to reference specific changes later, if they're separate branches [19:05] dobey: yes, sure. also if the "extra" changes modifies other parts of the system, this is just a tool [19:05] and I included that changed because helped me to catch the bug [19:05] *change [19:11] dobey: my point is that his branch is a unit of work, splitting it is just overhead [19:12] verterok: i understand your point [19:13] verterok: i am thinking now :) [19:13] dobey: :) [19:23] verterok: hrmm, why did you have to add the test to POTFILES.skip? [19:24] and what the heck is creating tmp/ and not deleting it [19:27] dobey: because I'm using __import__, and intltool-update is... [19:27] ugh, tests [19:28] dobey: no idea about the tmp/ probably some old tests [19:28] verterok: if that's the line complaining, please file a bug against intltool also, so we can fix that :) [19:28] dobey: sure, in which package/project? [19:28] verterok: intltool [19:29] heh [19:29] thanks [19:33] dobey: Bug #498292 [19:33] Launchpad bug 498292 in intltool "intltool-update thinks __import__('') is a translation" [Undecided,New] https://launchpad.net/bugs/498292 [19:35] verterok: thanks! [19:44] * CardinalFang boggles at key modifiers getting stuck on. First Shift, then Ctrl. [19:44] CardinalFang: syrup enhanced sticky keys? [19:49] Could be. Seems to be synergy related, though I'm not ruling out keyboard layout wonkiness either. [19:50] * rtgz is in user-mode-linux instance [19:51] is anything proxy-related needs to be tested for ubuntuone? [20:02] Chipaca: ping [20:02] rtgz: I think there is no proxy support :) [20:03] verterok, yep, i want to experience this :) [20:03] hehe [20:03] which makes me wonder "why?"... [20:05] verterok: pong [20:06] rtgz: time? :) [20:06] Chipaca: hi, I'm doing some configglue stuff, and wondering if you thinked about upgrading config files :) [20:07] verterok: what is this "upgrading"? [20:07] i think core python doesn't work well with http(s) proxy yet [20:07] i'd love to get the patches for that out of the python tracker and into python stdlib that ships in ubuntu [20:08] Chipaca: e.g: I have a config file with an option in [__main__], lets call this option: log_level ;) [20:08] Chipaca: and now I'm changing it to be in [logging] [20:10] statik, but I've seen some bits of proxy support in syncdaemon sources o_O [20:10] verterok: you create a parser that sets the right value in [logging] while also prints a deprecation warning [20:11] rtgz: do you have a proxy handy where you can test? [20:11] Chipaca: a whole new parser right? extending TypedConfigParser [20:11] verterok: yes. Patches welcome! :) [20:11] Chipaca: jajaja [20:11] Chipaca: ok, I'll see how it goes, thanks! [20:11] Chipaca, yup, right here, squid [20:12] rtgz: what happens if you run syncdaemon with the right http_proxy environment variables? [20:13] not sure if it's http_proxy, HTTP_PROXY, HTTPS_PROXY or https_proxy [20:13] one of those ;) [20:14] Chipaca, yep, it is something lowercase [20:14] * rtgz needs PPA, NM is not working here... [20:15] rtgz: ppa:ubuntuone/stable [20:16] rtgz: or test the version in karmic-proposed instead [20:16] rtgz: then you can confirm the bugs are fixed and we can get the SRU out :) [20:22] erm... do I have to do anything extra apart of just adding karmic-proposed deb line?... Candidate: 1.0.2-0ubuntu2 [20:26] guys, are you sure that it went to karmic-proposed? the Packages file does not list anything ubuntuone related [20:27] dobey, ^ [20:28] let me check [20:28] rtgz: you might be on a slower to update mirror [20:28] dobey, main server, archive.ubuntu.com, just updated :-/ [20:28] rtgz, btw here is the relevant bug about proxy support in python stdlib, looks like there has been progress https://bugs.edge.launchpad.net/python/+bug/94130 [20:28] Ubuntu bug 94130 in ubuntuone-client "HTTPS over proxy fails" [Undecided,Confirmed] [20:29] statik, yep... now i just need the client to test :) [20:31] rtgz: ok i guess it's not in -proposed yet [20:31] hrmm [20:34] * rtgz uses ppa:ubuntuone/stable [20:36] 443 port blocked, proxy started. launching ubuntuone-client applet :) [20:37] Authorization Error: [Errno socket error] [Errno 110] Connection timed out. Now trying magic variables... [20:45] statik, have you seen "bzr builddeb" fail at the debsign stage? I've worked around this once with someone, but I can't get it to succeed now. [20:46] gpg: skipped "Chad MILLER ": secret key not available [20:46] CardinalFang: bzr builddeb -- -k'chad.miller@canonical.com' [20:46] $ gpg --list-keys "Chad MILLER " # yields exactly what I expect. [20:47] CardinalFang: debsign apparently apparently doesn't get a proper match for some reason, when using "First Last " but 'email' works [20:48] CardinalFang: I set DEBSIGN_KEYID= in ~/.devscripts [20:48] dobey, gpg: skipped "chad.miller@canonical.com": secret key not available [20:48] I've tried DEBSIGN_KEYID and DEBSIGN_MAINT, too [20:48] or pass the short hex keyid to -k [20:49] weird [20:49] yeah... it goes to the proxy [20:49] someone should fix debsign [20:49] POST https://ubuntuone.com/oauth/request/ HTTP/1.0 [20:49] someone not me [20:50] rtgz: getting an oauth token definitely doesn't work with a proxy [20:50] meeting begins [20:50] oops [20:50] wrong ... thingie [20:50] wtf? [20:50] heh [20:50] oh, good, OK :) [20:50] aquarius: there is no meeting [20:50] I thought, god, I've been here 24 hours :) [20:50] Well, he got our attention. [20:51] CardinalFang, I have DEBSIGN_KEYID=608C0112 and it works ok [20:52] dobey, yes, but it is... just several lines of code away... of course, if it is not buried down somewhere completely [20:54] statik, Got it. Thanks. I was setting that as environment variable, which doesn't work. [20:56] 2008!!!??? [20:56] 2008? [20:56] the year of the awesome! [20:57] rtgz: but it is what? [20:58] Chipaca, dobey, urllib bug is 1 year, 3 months old.. [21:02] and the bug report in ubuntu is from march 2007 [21:03] what urllib bug? [21:04] rtgz: your statements are so incoherent right now :) [21:04] i have no idea what you're talking about [21:05] RST [21:06] ok, will try to not to output debug messages to the channel [21:06] rtgz: i am confused. i have no idea what bug you're talking about :) [21:08] dobey, i have read the lp bug report about python urllib not supporting HTTPS proxies and thought that it is silly to have a year w/o working urllib-based apps. Then I opened the original bug report and it goes back to 2006 :) [21:08] oh [21:11] statik, ok i will update my urllib2, hope that it won't break too much :) === john__ is now known as webm0nk3y1 === webm0nk3y1 is now known as jdobrien1 [21:24] statik, erm... it is already there for urllib2.py and httplib.py... but not for urllib (hm, no idea what's the difference, not python guy :) ) [21:25] rtgz: what are you trying to do anyway? [21:26] dobey, find out why ubuntuone client does not connect via HTTPS proxy [21:26] rtgz: the things in syncdaemon or storageprotocol with "Proxy" in them, probably aren't the kind of proxy you might think they are :) [21:27] rtgz: doing http[s] proxy with urllib2/httplib requires some extra work that we aren't doing yet, and haven't yet had time to do, as i understand it [21:27] rtgz: do do ssl verification, we also had to do some special magic with httplib [21:28] dobey, storageprotocol/proxy_tunnel.py ... [21:28] rtgz: yes [21:28] rtgz: not exactly what you might think it is :) [21:29] rtgz: i think to support https proxies in syncdaemon, it needs to do some stuff with twisted, similar to how the bw throttling works [21:52] ok. proxy support is really broken [21:52] yes :) [21:58] alright, i'm outta here for now [21:58] later