[13:09] Till: I have a question about your desktop-couch-resource [13:12] till: As for now this is only for 'contact' PIM data (mime type -> text/contact ?) [13:12] till: ? === joshuahoover changed the topic of #ubuntuone to: Have a question? Ask joshuahoover | https://one.ubuntu.com | https://launchpad.net/ubuntuone | Please honk if you want a music store [13:46] Hi all. Here's my list of what items might be good and attractive in U1 but they are not yet there === kenvandine_ is now known as kenvandine [13:47] Web UI: Search for filenames and/or metadata. In case tons of photos get into Ubuntu One, you might add a Photo view page that will allow search for photo tags (e.g. added by F-Spot or other software). Text documents and Office docs may also be good. [13:48] This raises a privacy question. Yet, if this is done via per-user couch db so that the user could see how his metadata is stored then this may alleviate many questions [13:49] rtgz, is this for me to search my photos, or me to search everyone's photos? [13:49] aquarius, for me to search my photos [13:49] aquarius, otherwise that would be the new social network [13:49] how is there a privacy question, then? [13:50] aquarius, somebody is not comfortable with Google reading their mail and providing relevant ads. [13:51] aquarius, by setting up per-user metadata storage you might show that this is all the info you've got on them. [13:52] rtgz, ah, but this isn't us processing their data and knowing about it; this is them searching their own files [13:53] aquarius, data is being processed on the server, and it is stored on the server. By granting access to the stored data people may be assured that you are not hiding anything else. Of course if there is nothing else hidden :) [13:53] Ok, next: File preview. [13:54] at the moment we're not processing the data, either; we just store it. We don't separately store (or even calculate) file metadata -- this would be a good reason to do so, though [13:55] aquarius, yep, in this case I might browse my music collection from U1 interface and don't think about my records/old/a/backup/file.ogg hierarchy [13:55] How can I put this? You are not the first person to think of that idea :) [13:55] I really, really want to be able to do that :) [13:56] File preview is really needed for media files, since it is hard to find out what is inside 123874328734.ogg file. And picture.jpg is a perfect name but it does not say anything. [13:57] By providing file previews you might gain more people that will start using 50Gb plans. Hey, all your music collection is there and you can click 'Play!' and it will stream to your browser while you are browsing the web even on non-ubuntu computers. [13:58] This will require much more bandwidth but it may be worth doing so [13:58] yes. People want this. I want this :) [13:59] there is a distinction between "preview a file" and "stream a whole file", though [14:00] Image previews may be created right when file gets to U1 server, i.e. take a picture, squeeze it to some meaningful size (20+ megapixel photo of a flash in the mirror?). [14:03] rtgz, yep. that's already on the list of things we want to do, we just haven't quite got to it yet [14:05] aquarius, yep, the preview of a media file is a strange thing. A screenshot of a single non-black frame of the movie file might be useful, but one can't do a single frame of ogg file [14:05] * rtgz got an idea, the waveform [14:05] * rtgz dumps the idea [14:06] so the audio file might be clipped ... [14:06] but [14:08] the license might prohibit that :) [14:08] Next: Look here: https://www.asuswebstorage.com/member/login [14:09] A regular Member login except of a tiny note on top telling us about the planned outage. Two days ago I received the notification from ASUS telling about this planned outage. [14:09] hello [14:10] I'll be AFK at 10, so here's my status update: [14:10] DONE: Got https://code.edge.launchpad.net/~jblount/ubuntuone-servers/public-files-webui/+merge/17424 up for review, worked on layout bugs, helped mattgriffin with some bazaar stuff [14:10] TODO: REVIEW DAY, get layout stuff up for review [14:10] BLOCKED: Nope [14:11] I can't seem to upload files. The browser is sending a request to files.one.ubuntu.com. Is this a bug on my side or a server issue ? [14:12] compareTo("UbuntuOne Jan13-14 Outage"). When it turned out that the contacts are no longer accessible right when I needed one :), at 02:00 AM local time. (SSH X forwarding and Evolution rules in this case, but...). You have ubuntuone-users mailing list. Outage info might be well accepted there. [14:12] calliscope, checking... [14:13] calliscope, looks fine from here, what issue do you get? [14:14] The request is endless, although the file is only ~25Mo. Could it be prohibited to store encrypted .cbs ? [14:14] calliscope, Hm... What is your upload bandwidth? [14:14] calliscope, the files are not parsed for contents so it is possible to store virtually everything [14:16] rtgz, How do I see the upload bdwth ? I use Gnome. [14:16] calliscope, no, I mean the ISP bandwidth you are granted. (building the 25Mb file ATM...) [14:18] rtgz, ~100Mb/s [14:23] calliscope, ok, just tested with 21 Mb file, everything went through. For how long have you been uploading your file? [14:24] About 8mn. The applet also is showing a disconnected status and the contacts do not sync. Have I missed a step (is there a port to leave open or something ?) [14:25] The last time I could upload something was 2 days ago. [14:30] thunderbird 3.1 is making me stabby. [14:31] calliscope, ok, I advice you to try uploading the file again as I was not able to reproduce it even on my 1Mbit uplink. [14:32] rtgz, I restarted the thing, and actually the server is putting me into a loop where I could confirm acces to my computer / opens a new webpage, asks again to confirm/opens a webpage etc. [14:33] joshuahoover, calliscope: Bug #437165 [14:33] Launchpad bug 437165 in ubuntuone-servers "UbuntuOne stuck in an endless authentication loop (always changes to «Authentication failed»)" [Medium,Confirmed] https://launchpad.net/bugs/437165 [14:33] ok thanks, what can/should I do ? I am available for tests. [14:37] calliscope: can you run the following from a terminal session: sudo apt-cache policy ubuntuone-client [14:39] rtgz, I got the latest 1.0.2-0ubuntu2 installed. System fully updated, no extra reps apart from the multiverse [14:39] * rtgz is killing his OAuth record... [14:40] calliscope: ok [14:41] I use FF, adblock is on and I have no trouble connecting to the net (obviously, I'm here) [14:42] calliscope: can you paste the contents of ~/.cache/ubuntuone/log/oauth-login.log to http://paste.ubuntu.com/ and post the url here so i can take a look? [14:42] I should add that I have changed my password a few days ago [14:44] do I put my real name or just this nick ? [14:45] thunderbird 3.0 is far less stabby-inducing. [14:46] joshuahoover> Nevermind, here the URL : http://paste.ubuntu.com/357082/ [14:47] dobey: is the fix for #437165 in the beta ppa? [14:52] is that the networkmanager bug? or? [14:52] oh no [14:52] endless auth loop [14:53] dobey: right...i see it says the fix is in the beta ppa for https://bugs.edge.launchpad.net/ubuntuone-client/+bug/450633 [14:53] Ubuntu bug 450633 in ubuntuone-storage-protocol "Problem with computer verification in Ubuntu One." [Critical,In progress] [14:53] dobey: but i don't think it's in karmic, at least not according to lp [14:54] https://bugs.edge.launchpad.net/ubuntu/+source/ubuntuone-client/+bug/437165/comments/21 [14:54] Ubuntu bug 437165 in ubuntuone-servers "UbuntuOne stuck in an endless authentication loop (always changes to «Authentication failed»)" [Medium,Confirmed] [14:54] joshuahoover: it's in the 1.0.3 update i believe, so it's in karmic-proposed [14:55] dobey: ok, so if calliscope wants to get around this, is the best bet to install the beta ppa right now? [14:56] joshuahoover: preferably, the version in karmic-proposed, so that he can verify it fixes the bug, so we can have more documentation on the bug stating it's fixed, and we can actually get the SRU done :) [14:57] I'm opening aptitude and checking this branch of updates. I'll get back at you after it. Would hate to get back to apt-pinning here in Ubuntuland :P [14:57] dobey: ok, how does one go about installing that though? it doesn't show up yet for most users or am i mistaken? [14:58] no you have to enable proposed updates in the Software Sources [14:59] System->Administration->Software Sources, Updates tab, check "Proposed updates (karmic-proposed)" and close the dialog, and refresh when it asks you to [15:00] then you can install the updates (there will be other updates too, though) [15:00] Desktop+ MEETING BEGINS. Say 'me' to claim a slice of the stand-up meeting, then take your turn by saying DONE/TODO/BLOCKED. [15:00] but you can select them individually, and pin down the rest (sorry to meddle) [15:01] Or is it actually better to update the system to these patches ? [15:01] me [15:01] me [15:02] me [15:02] me [15:02] fix doesn't work, or do I have to log out/in ? [15:02] me [15:03] calliscope: you need to restart all the ubuntuone stuff, yeah. log out/in is easy way to do that [15:04] dobey> yes but quitting the client through the applet and restarting it through the menu does it also, right ? Or is there a deamon to restart as well ? [15:04] calliscope: i just install all the updates myself, and i haven't had any issues with them. they are updates destined for karmic, that haven't gotten enough testing before doing so, is all. if you could help test them, we and the greater ubuntu community would be greatful for it :) [15:04] calliscope: quitting the applet tells the daemon to quit [15:05] calliscope: whether it actually does or not, is up to the daemon though :) [15:05] it is supposed to quit though [15:05] urbanape, jblount: poke [15:06] dobey> :) well then the fix doesn't work. I'll try the hard way. Guess I'll quit (I'm not on irssi but on xchat right now :/ ) [15:07] me [15:07] ok, shall I start? [15:07] sure [15:07] Por favor. [15:07] • DONE: Fixed some issues in new XML<->HTML branch. CouchdbCredentials API. Music store widget work to support non-logged-in urls [15:07] • TODO: Conflict resolver tool in pair tool. Look at becoming a MOTU (https://wiki.ubuntu.com/UbuntuDevelopers). Make sandy's snowy test suite work with our server (http://git.gnome.org/cgit/snowy/tree/api/tests.py). Discuss with jdo and aquarius about oauth token per app, not per machine? Move JS code to its own file and add jslint tests to check: [15:07] • BLOCKED: no [15:08] next: aquarius [15:08] ⚀ DONE: redo music store URLs; adapt music store plugin prototype to use new URLs [15:08] ⚁ TODO: have music library page send message to downloader daemon and row to database; make workitems of outstanding todo items; make tomboy first-sync experience nicer [15:08] ⚂ BLOCKED: [15:08] CardinalFang, rock the house [15:08] DONE: not much, too much going on outside work. Worked on strange failure with notes-tomboy fix, which may be wrong. :| Talked with M about jobs. [15:08] TODO: bugs and blueprints. More bugfixing. [15:08] BLOCKED: None [15:08] I'm on vacation the next few days. See you guys next Thursday. [15:08] teknico, your turn. [15:08] DONE: more fighting with bzr (#506974); more admin work to keep connected to the net; more work implementing the mobile sync REST API for client app (#504689) [15:08] TODO: win the fight with bzr (#506974); fix a user's problems accessing the contacts web IU (#506584); more work implementing the mobile sync REST API for client app (#504689) [15:08] BLOCK: bzr misbehaving (#506974) [15:08] next: dobey [15:08] ☺ DONE: More testing code for login script [15:08] ☹ TODO: More new UI code and tests [15:08] ☹ BLCK: None. [15:08] urbanape: your roll [15:08] DONE: Finished the public-urls-webui branch, with the exception now of suppressing it within the Shared With Me folder. Noodled out the behavior of the Bindwood migration code path. [15:08] TODO: Finish the public-urls-webui branch and Bindwood migration. Get on with subsequent client behavior. [15:08] BLOCK: None [15:09] hello, the fix does not fix the issue. I'm still in the auth loop [15:09] jblount, if you please (and if you're around) [15:11] gabrielsimon: hrmm. i think you might be having a similar, but different issue then... can you please right click on the applet and choose "Report a Problem..." and follow the steps there? :) [15:11] sorry, I am calliscope, registered under another nick. === gabrielsimon is now known as calliscope [15:12] But should I do what you just said ? [15:15] yes :) [15:16] calliscope: and you're definitely using 1.0.3 now? [15:16] and python-ubuntuone-storageprotocol 1.0.1? [15:16] Absolutely [15:16] dobey> that I'll check [15:17] daboey> got version 1.0.1-0ubuntu1 [15:18] yes, please file a new bug with the method i described then :) [15:18] ping Till [15:18] ping till [15:18] it sounds like something else is getting masked as AUTH_FAILED [15:19] you mean a NEW bug ? because it seems pretty much the same as https://bugs.launchpad.net/bugs/437165 [15:19] Ubuntu bug 437165 in ubuntuone-servers "UbuntuOne stuck in an endless authentication loop (always changes to «Authentication failed»)" [Medium,Confirmed] [15:24] dobey> actually it is more like https://bugs.launchpad.net/ubuntuone-client/+bug/450633 [15:24] Ubuntu bug 450633 in ubuntuone-storage-protocol "Problem with computer verification in Ubuntu One." [Critical,In progress] [15:27] calliscope: those two are the same bug [15:27] calliscope: i think some other error is getting masked as AUTH_FAILED for you perhaps [15:28] dobey> ok, I'll file one. should I try the stuff a few times for the logs to be clearly about it ? [15:32] calliscope: the logs should be clear enough i think. if not, maybe we'll have to enable debug logging to get more info, but we'll see [15:42] brb [15:48] ok, i filed a report. hope it'll be better soon :) have a nice day. [16:46] bbiab, lunch. [18:05] CardinalFang, hello, got time to talk about a bulk update method? [18:05] mandel, sure [18:06] Maybe I should read up on couchdb to know its capabilities first. [18:06] CardinalFang, couchdb has no understanding of transactions [18:06] CardinalFang, so the only advantage in terms of performance is the fact that you are using a single request [18:07] Right. So, even in one request, some may fail. [18:07] the result is a collection of tuples with true if a particular doc succeeded or false otherwhise [18:07] plus the doc id of course [18:08] [(docid, TrueFalse), (...), ...] [18:09] On failure, what is the doc_id? [18:09] Suppose it's a new document, with some error, like invalid _rev. [18:10] the result id of the form (success, docid, rev_or_exc) where success is a bool and doc id is the id of the doc, the third is either the revision or the exception raised [18:11] I believe that currently you have to pass docs with an id [18:11] Ah. [18:12] I was thinking of passing a patch that allows to perform update like it is currently execute right now, but taking the single request advantage and changing the return result [18:13] let me send you a pastebin (5 mins) [18:23] CardinalFang, ok there it is http://pastebin.ca/1752631 [18:24] CardinalFang, something similar with a batch delete [18:26] only things is with the put_attachment (copy pasted it) which takes the updated record due to its version number, right? I think we would just need to pass the id [18:28] Only one thing:You can pull the "records_hash[record.record_id] = record" out of the condition results, since it's common to both. [18:28] indeed, I was just typing out of my head [18:29] Are there any errors possible with putting attachments? [18:30] the docs from python couchdb does not say anything about it [18:35] the actual like of the put attachment is better like this: http://pastebin.ca/1752643 [18:36] the problem with the attachments is that if there was an error you will have to roll back and since transactions do not exist it could mean that someone is using the new values beofre you can do that [18:38] Yeah. Grr. [18:41] CardinalFang, although the same would happen in a single doc operation so it would be consistent :P [19:02] rodrigo_: ping [19:24] calliscope: interesting. if the log i'm looking at is for you... it looks like the syncdaemon isn't actually starting for some reason.... [19:24] calliscope: care to do a little debugging? [20:30] aquarius, ping [20:42] urbanape, pong [20:44] heya. Got a couple questions about dc-related stuff. [20:44] so, in the course of robustificating Bindwood, I've introduced a few more record types. I've tried to make them generic enough that they could be standard (not introducing any Firefox-specific nomenclature) [20:44] and I wanted to talk to you about the idea of schema versioning [20:44] hey. Just about to show my daughter some things and then she's going to bed, but ask, and I'll be thinking while doing that :) [20:45] k [20:45] aquarius: hey [20:45] aha, dobarama. How can I help you today, too? [20:46] bookmark, folder, feed, and separator are the four record types in Bindwood now. And I was wondering if you thought the schema version should be a field on the records or in some meta-ish document? [20:46] aquarius: was wondering if you could reply to the other piece of my reply to you, about the possible keyring issues if we switch some stuff to python-keyring [20:46] we can't really make our C code use python-keyring for example, which presents some interesting issues when running rbox under kde to buy music, for example [20:47] * dobey wishes the cross-gui keyring thingy was ready (and why the hell we have gui-specific keyrings anyway) [21:00] dobey, yeah, that's interesting...there's no c-keyring [21:01] well, there's libgnome-keyring [21:01] until the whole shared keyring thing gets some levity [21:01] yeah, but there's no easy to use cross-keyring C library [21:02] urbanape, record version is on the document itself [21:02] dobey> I am available for debugging. [21:03] urbanape, unless I'm misunderstanding your question? 'cos I need to be able to tell from a record (a) that it's a separator, and (b) which version of the separator record type it is [21:04] calliscope: can you open a terminal and try to run /usr/lib/ubuntuone-client/ubuntuone-syncdaemon and let me know if it fails and outputs an error? [21:06] dobey> it says "Another instance is running" [21:06] and if I typ "sudo" it tells me not to launch it as root, "please" [21:07] yeah, don't run it as root [21:07] hrmm [21:07] weird [21:08] calliscope: can you quit the applet, and in the terminal then run u1sdtool --quit [21:09] if I type "killall /path/to/syncdaemon" it says "no process found". should i add i disabled ipv6 in the /etc/default/grub ? [21:09] killall doesn't work for taking paths, you just specify the program name itself [21:09] ok the comand you gave me got the daemon to stop. [21:10] ok, i didn't know that. [21:10] no, localhost having an ipv6 address would present a different problem [21:12] aquarius, that's fine, just wanting to match the consensus. [21:15] dobey, so...the existing oauthdesktop stuff will stay in Python, as will its talk-to-REST replacement? so that will be using python-keyring? [21:15] aquarius: it will remain in python, yes [21:16] aquarius: i'm not so sure we should switch it to python-keyring after my sudden discovery of these potential issues, though [21:16] right, so that's OK, then. But...applications that need to get your oauth token from the keyring, either to talk to U1 or to talk to desktopcouch, need to know which keyring it's in [21:16] well, the alternative is: stick with gnome-keyring and just don't work off gnome. Which is untenable in a world with Windows and N900 ports in... [21:16] aquarius: another option of course, is to do per-app authorization, so they all have separate tokens [21:17] doesn't help [21:17] aquarius: then they can store them with whatever API they use [21:17] also, nightmare when you have to authorise each app separately [21:17] i think Maemo uses gnome-keyring [21:17] not according to thomasvs, unless I misunderstood him. He used python-keyring in the desktopcouch port because gnome-keyring isn't there [21:17] or it's massively different or something [21:18] aquarius: actually, no nightmare. i'd really like to move to doing bipedal oauth [21:18] aquarius: to get rid of the "click this crap" [21:18] it's a nightmare for people who aren't you. :0 [21:18] aquarius: well, *python*-gnomekeyring might not be there? [21:18] aquarius: but i was pretty sure that gnome-keyring proper was [21:18] ah that might be it, indeed. [21:18] the python bit might not be, because it's packaged pretty crappily [21:18] I remember there was some problem, but not what it was [21:19] iow, with all the other gnome python crap [21:19] aquarius: the only thing that should still use tripedal oauth for lucid, should be tomboy versions using the Snowy API [21:20] so...how do you propose authing a new app? [21:21] well, it might be annoying if you don't save your password in the keyring, when doing the initial login or something, but that is no more annoying than having to click a button that says "let this computer which i obviously am trying to connect to my account, connect to my account" [21:21] probably less, since it's not rhetorical :) [21:23] nah, hang on, what you're proposing is that the first time you run any app that talks to U1, you have to auth it, even if you've authed other apps before? [21:23] yes [21:23] so the first time you run the music store, you have to auth it even though your machine is already connected to U1 and downloading files? [21:23] and you can remove individual apps from accessing stuff [21:23] music store probably not [21:23] that's what I think's a nightmare, especially when there are lots of apps that talk to U1, which is the plan. [21:23] music store is a weird special case thing [21:24] since it's using file sharing [21:24] which is already authed [21:24] nah, it needs to do more than that. It uses the website, too, it's not just file sharing. [21:24] aquarius: why do you think it would be a nightmare? [21:25] because I want to connect my *computer* to U1, not all my apps separately. [21:25] then I have to do it once, not an unbounded number of times [21:25] dobey> sorry to interrupt ; the only package listed in aptitude that i do not have on my system is the ubuntuone-client-tools. do you think this could help bypass the pb ? [21:25] calliscope: no, that's a separate client [21:25] a cli client ? [21:26] calliscope: not exactly, but yes [21:26] calliscope: a not really support, experimental, cli client thing (which still requires X for some pieces) [21:27] i see, it's just a trigger of some kind. [21:29] calliscope: ok. can you verify that none of the ubuntuone pieces are running? [21:30] dobey> well how do i make sure of it ? [21:30] the applet is disabled, and the syncdaemon is stopped. is there more ? [21:31] calliscope: ps efx|grep ubuntuone [21:31] calliscope: that should hopefully show nothing [21:32] calliscope: it might show the grep [21:32] i got an enormous text [21:32] hmm [21:33] should i kill the corresponding pid ? [21:33] calliscope: can you paste it on http://paste.ubuntu.com and paste the link to that page here? [21:33] of course [21:34] dobey> http://paste.ubuntu.com/357258/ [21:36] calliscope: ah ok, those are fine. it's firefox and grep. but you can probably close firefox, so we can try fresh [21:36] calliscope: but if you can now try to run the ubuntuone-syncdaemon again as I asked before and let me know if it does anything, please [21:37] i just closed ff and evolution and retyped the command : i still get the pid 3594 running [21:38] dobey> to restart the syncdaemon is /path/toit/syncdaemon on ? [21:38] yes [21:38] then i get an error [21:38] try running that in the terminal please [21:39] should i copy/paste it ? [21:40] yes, with paste.ubuntu.com please [21:40] dobey> http://paste.ubuntu.com/357263/ [21:42] calliscope: oh, don't run it with "on" after it. just run it with no arguments :) [21:45] dobey> it's not doing anything, the shell hangs [21:45] calliscope: no, it's running, which is good [21:45] calliscope: now try to run the applet from the menu again please [21:47] dobey> there i am in the loop again :) [21:49] calliscope: ok, weird. [21:49] calliscope: do you have any syncdaemon-exceptions.log in ~/.cache/ubuntuone/log/ ? [21:50] i have lots of it ! [21:52] dobey> but the file syncdaemon-exceptions.log is empty. only the others (with numbers after) print something === gnomefreak76 is now known as gnomefreak [22:05] calliscope: ok, can you paste what the older copies of that file have in them? [22:12] dobey> http://paste.ubuntu.com/357274/ [22:13] calliscope: can you log into the one.ubuntu.com site, and remove all the machines from the list there? [22:15] dobey> allright, done. [22:17] calliscope: ok, now can you go to Accessories->Passwords and Encryption Keys, on your computer, and remove all the UbuntuOne keys from your "login" keyring (or any other keyring they might be in)? [22:18] dobey> allright, done. [22:19] calliscope: ok, now close the firefox window, and right click on the applet and choose "Connect" please [22:19] calliscope: and let me know if it still sticks you in a loop [22:22] dobey> sorry, the loop is still on [22:22] well boo :( [22:24] think it is related to my config ? i got nothing exotic though... [22:28] calliscope: do you have ~/.config/ubuntuone/syncdaemon.conf ? [22:29] i do. want me to paste it ? [22:29] i got no limit and on is "false" [22:29] yes please [22:30] http://paste.ubuntu.com/357281/ [22:32] dobey> might i add i am still running the karmic-proposed version on the client (sorry if this is irrelevant) [22:34] can you quit all the ubuntuone pieces, and remove that file, and try again please? [22:42] i erased the entries on the web, cleaned ff cache/cookies/etc., shut the syncdaemon, closed the applet, removed the .conf file and restarted the service and the applet. i still get the loop, sorry :( [22:43] ok, i'm really confused as to why, and it's late in the day now [22:48] calliscope: maybe we can continue debugging later? it really annoys me that this isn't working for you :) [22:50] of course :) thanks you [22:54] calliscope: thank you for letting us help, and helping with debugging :) [22:55] no way man, thank those who make it fun to do that :) [23:01] well, i gotta go [23:01] later everyone!