=== pfibiger` is now known as pfibiger === Kkkhbbnj is now known as Cibort === Cibort is now known as Martez16 === Martez16 is now known as Cibort [08:12] morning all! [10:17] Happy St Patrick's Day, everyone! :D [10:29] morning all === christoph_ is now known as natschil [11:16] So, looking at some work that gets synced to ubuntu one, I noticed that some of it was a horribly old version where I am certain beyond any doubt that it is not the newest copy. I suspect ubuntu one, as it frequently refuses to save a new version of work and instead creating a file ending in .u1conflict with the contents of the work...Is there some sort of changelog for ubuntu one? [11:17] natschil: what version of ubuntu are you on? [11:18] (no, there isn't a changelog, but you shouldn't be seeing those auto-conflicts any more) [11:24] Chipaca: I'm not on the current version, but on one behind.... seriously, it has no changelog? What kind of backup system has no changelog? [11:25] s/behind/before/ [11:25] natschil: a backup system with no changelog would suck and be terrible and unusable [11:25] natschil: but, we are not a backup system [11:26] natschil: if you use the file sync service as part of a backup system, the onus is on you to add the missing bits [11:26] aquarius: I always forget, what's the name of the backup system that's building on us? [11:33] Chipaca: something that uploads files to the internet and even replaces files on my computer (i.e. moving more recent files to be called .u1conflict) should have some sort of changelog at least. I mean seriously, I'm sorry but if u1 has no changelog it's a serious failure, changing some 'synced' file to delte somewhere would cascade across all the synced computers. So now way to get the files back? [11:34] natschil: deleting a synced file deletes it all over the internet, yes [11:35] natschil: and a changelog is not the same as an undo or an undelete [11:35] natschil: we're working on having an eventlog, and we already have a rudimentary undelete feature [11:35] natschil: but we are not a versioning system, and we are not a backup system [11:36] natschil: the auto-conflict issue is a bug that has been fixed, however [11:36] natschil: (auto-conflcit being the case where a single client will produce .u1conflict files because of inadequate tracking of changes) [11:37] Chipaca: that's all very nice, but how does it help me get my data back? [11:37] natschil: which data have you lost? [11:38] a file which I am quite sure I saved, and is now nowhere to be found. [11:38] natschil: and there isn't a .u1conflict file? [11:38] natschil: (if your problem is that there is an older version in its place) [11:39] Chipaca, deja-dup, but not this cycle [11:39] Chipaca: there is no .u1conflict file as far as I can see. I found one in my trash, but it also seemed to be of an older version. Sorry I need to go, be back later. [12:03] clarita: mandel. mandel: clarita. [12:03] you guys need to talk :) [12:26] good morning! [12:37] Stecchino, ping [12:44] aquarius: hi [12:46] Stecchino, heya; I wanted to talk to you about the music store and amarok; ralsina mentioned he'd been chatting to you [12:47] yes he has [12:50] aquarius: I'm going to make a time estimate for this, based on what I consider to be the minimal feature set [12:50] Stecchino, so...I'm probably the person to chat to about this :) [12:50] good to know :) [12:51] Stecchino, can you talk about what you're planning to do and I can lay out how you might want to do it, and possible alternatives [12:51] aquarius: 2 major parts: a collection to access music stored on u1 and an integrated store [12:52] amarok currently have 2 stores integrated already in the internet services view [12:52] Stecchino, OK; you may have to talk me through a few amarok things, like what a "collection" is, here [12:52] but want to make it a bit more interactive by including the web view [12:53] aquarius: a collection is a source of music. We have the local collection (local files, sql database), media device collections (iPod, USB mass storage and MTP), UPnP, DAAP, etc [12:54] Stecchino, ah, right; so this collection would talk to the music streaming API we provide, gotcha [12:54] anything that can support queries by artist, genre, year, etc and has it tracks accessible can be a collection [12:55] aquarius: was a bit confused about the streaming package actually. In the regular package, would the tracks have to be downlaoded to disk before they can be played? [12:55] the music streaming API currently allows you to browse by artist, then by album, then to a song; I don't think you can query it for, say, one particular year. Does a collection need to support *all* those things? [12:56] aquarius: yes, in order to be displayed according to the users wishes. But for smallish collections we can cache all that info in memory and query that way [12:56] I say smallish, but we even do that for DAAP and UPnP collections, which can be huge [12:57] aquarius: the QueryMaker class also allows stuff like similar artist lookup and integration in amarok's dynamic playlists [12:58] OK, bit of a summary. The music streaming package gives you access to the streaming API; the streaming API is an HTTP API which lets you browse your music by artist and so on and get a streaming URL for any of your songs. So, you request http://streaming.../getIndexes and that returns an XML list of all artists and artist IDs; you can then request http://streaming.../getMusicDirectory?id=artistid to get a list of [12:58] albums or songs for that artist. Make sense? [12:58] yes [12:58] the streaming API is basically http://www.subsonic.org/pages/api.jsp -- subsonic is a run-your-own-music-server Java thing, and we implemented their API so that subsonic clients will work with our streaming system. [12:59] so borwsing but no querying? [12:59] All music that's stored in your personal cloud is available via the streaming API. [12:59] you can search, but the search is "pass a string and we return all matches". You can't search for, say, a year specifically. [13:00] so if you search for "Johnny" we'll return both songs by Johnny Cash *and* the song "Johnny We're Sorry" by the Fine Young Cannibals. [13:00] I don't think it's a problem to cache the missing metadata maps in memory in order to comply completely with the querymaker concept [13:01] I was also concidering making QM less strict so we can have more "full" collections in amarok [13:01] so when you drill down to, or search for, a song, you get an HTTP URL to stream that song, and you just point your playback engine at that URL directly (or if you want to you could download from that URL to a local cache and point the playback engine at the locally cached file, if that's what you prefer) [13:02] Alternatively, Ubuntu One file sync will synchronise those files to your computer directly, and then as far as Amarok is concerned they're just local files, but of course you need the U1 syncdaemon running to do that (that's what does the downloading). [13:02] Does that make the distinction clearer? [13:03] aquarius: but this streaming API is only available with the streraming package? [13:03] Stecchino, correct, yes. [13:03] aquarius: is there a standard folder u1 filesynced files end up? [13:04] Stecchino, u1 can sync any folder. So, if on your first machine you mark your ~/Music folder as synced, then it'll be synced to ~/Music on your second machine if you ask to sync it. [13:04] which folders are synced is up to the user [13:04] aquarius: what about tracks purshased on u1 music store? [13:05] Stecchino, right. Tracks purchased in the U1 store are added to the cloud first, in the folder "~/.ubuntuone/Purchased from Ubuntu One" and then that folder is synced down to your machine (if you're running the syncdaemon). [13:05] * Stecchino should try it. Just has not found music he likes one the u1 store [13:10] aquarius: so it probably makes sense to have a specific collection for the contents of that folder [13:10] amarok can copy those tracks to the main collection [13:11] and I think it's also possible to still keep a link to the U1 collection so it's always known that it was purchased there [13:11] not sure if that would be that useful though [13:11] Stecchino, yeah; you could handle this two different ways [13:12] in Rhythmbox, there are two separate libraries; one for the "main" music library, and one for the U1 store [13:12] in Banshee, they're all integrated into one library [13:12] aquarius: in amarok it would have to be like rythembox. because the location of the local (sql) collection is user configurable [13:13] *nod* [13:13] besides, there is branding for a separate collection which I'm sure canonical will like :) [13:14] that makes sense; that's not really a technical decision (from our perspective) -- it's basically up to you as the media player guy :) [13:14] hmm [13:24] Stecchino, so I am more than happy to explain stuff as and when you need it; these APIs will be documented soon, but not yet (I'm still working on that!) [13:25] aquarius: they seem to make sense so rather self documenting I hope [13:33] Stecchino: are you using libubuntuone as the store widget, or rewriting it in qt to use khtml/qtwebkit? [13:34] dobey: not using anything at the moment since I have not started this yet, all depends on how much time I'll have for it. But I'll probably use a qwebkitview [13:34] or html plasmoid [13:37] hello! [13:39] dobey: hi [13:39] Stecchino: ah. i wish there was some way to easily make libu1 work for both gtk+ and qt [13:39] pavolzetor: hi [13:39] hola alecu [13:39] dobey: https://bugs.launchpad.net/couchdb-glib/+bug/736308 [13:39] Launchpad bug 736308 in couchdb-glib "cannot compile simple Database file (affects: 1) (heat: 6)" [Undecided,Confirmed] [13:39] hola dobey! [13:40] dobey: I have reported it, but I need know, how long takes it to fix [13:40] pavolzetor: greater than 0, but less than infinity [13:41] dobey: hmm, month, week or year? because I have coded some objects,but I need database and sqlite is not well suited for this [13:43] pavolzetor: i don't know. [13:44] dobey: who is in charge of this? why python works and vala not? [13:44] uhm, python is the native language that desktopcouch is developed in [13:45] but couchdb is in erlang [13:45] sure, but python-couchdb isn't [13:46] couchdb itself is just a server, not a library for accessing it [13:46] hmm, I don't like python and I don;t knwo, why everything is coded in this lang (gwibber, ubuntuone) and therefore it is so slow [13:46] JSON is used for accessing over CURL? [13:47] pavolzetor: dobey is setting up a club of python haters [13:47] chipaca: :-D [13:47] The Roppongi Club [13:47] but it only has one member [13:47] me? === m_conley_away is now known as m_conley [13:48] no [13:48] so who? [13:49] there is only one person who has enough hate [13:49] okey, keep your secret ;-) [13:49] but who is in charge for vala and couchdb? [13:50] pavolzetor: what is "in charge"? [13:50] chipaca: someone who works on binding for C and vala [13:51] pavolzetor: I'm responsible for it, but there is nobody working on it at this precise instant [13:52] chipaca: so you know when it will be fixed? [13:52] some kind of schedule [13:52] well i'm working on it, but in another dimension, due to quantum mechanics, so it won't appear in this reality for some time [13:53] pavolzetor: it will be probably fixed by next month [13:53] dobey: why due to quantum mechanics? (is it so small or?) [13:54] chipaca: long time [13:54] pavolzetor, what kind of application are you using this for? [13:55] happy st patricks day :) [13:55] fagan, beware of guiness and records! [13:55] alecu: my rss reader [13:55] alecu: I don't like liferea (I made patch for it, but I don;t like it) [13:56] pavolzetor, nice! Do you have the sourcecode published? [13:56] alecu: not, it is in preprepre alpha [13:56] alecu: I have palns to publish it, but it is for my own usage (primary goal) [13:56] alecu: Guinness is horrible man and the only record ill be getting today is the most sleep ever [13:57] pavolzetor: are you using webkit in vala? [13:57] fagan, I always thought irish people had only Guiness for breakfast! [13:57] dobey: no, why? [13:57] pavolzetor: how are you planning to render the feeds? [13:57] dobey: no [13:58] alecu: well Guinness actually originated in England interestingly enough [13:58] dobey: just server and client, articles will be opened in browser (I think almost nobody reads them in readers) [13:59] what are you using desktopcouch for then? just to cache the title and url? [13:59] eh, why have a reader then... :) [13:59] thisfred: yeah, srsly. use google reader. [14:00] dobey: yes :-D, and for settings to have them synced [14:00] ugh [14:00] thisfred: because of UI and notification and so on [14:00] * ralsina has written 3 or 4 RSS readers already, and rendering the actual page in the reader makes users very very happy. [14:00] chromify-osd + google reader [14:01] rendering the summary... not so much. [14:01] ralsina: i hate seeing the "... [more]" stuff on planet sites. super annoying [14:01] well, it's your time, I guess. [14:01] dobey: my planets use the full feeds if they are available [14:02] stand up people, sorry about the missing ping! [14:02] me [14:02] also, calling them planets annoys me [14:02] dobey: I don't even use planet to create them :-) [14:02] I use liferea to auto download enclosures and mark those items as read. I do my reading in google reader, and it's pretty hard to beat [14:02] me [14:02] they should be called aggros [14:02] me [14:03] meh [14:03] me [14:03] thisfred: reader sucks for partial feeds [14:03] thisfred: okey, maybe reader is better, I have a look at it, I spend about 1 day at vala, so it is not so much time [14:05] alecu, go [14:05] DONE: proposed two branches. Together they fix most known zeitgeist bugs. Bug #708145, bug #723832, bug #735613 [14:05] alecu, please? [14:05] TODO: weekly meeting, peer review, get tickets, sort bugs [14:05] BLOCKED: still no input on bug #729259, still need a deep chicharra review for the branch that fixes bug #692730 and bug #693545 [14:05] NOTE: bought a car this week. Will need to do some paperwork tomorrow morning. Still need to learn to drive :P [14:05] Launchpad bug 708145 in ubuntuone-client "ZeitgeistListener tracebacks on AQ_CREATE_SHARE_OK (affects: 1) (heat: 19)" [Undecided,New] https://launchpad.net/bugs/708145 [14:05] Launchpad bug 723832 in ubuntuone-client "Zeitgeist tests skipped because of intermittent failure (affects: 1) (heat: 42)" [High,Confirmed] https://launchpad.net/bugs/723832 [14:05] Launchpad bug 735613 in ubuntuone-client "Zeitgeist should save a user friendly file name (affects: 1) (heat: 820)" [Medium,Triaged] https://launchpad.net/bugs/735613 [14:05] Launchpad bug 729259 in ubuntuone-control-panel (Ubuntu Natty) (and 2 other projects) "Scrollable list of devices looks broken (affects: 1) (heat: 300)" [Medium,Triaged] https://launchpad.net/bugs/729259 [14:05] Launchpad bug 692730 in ubuntuone-client "ZeitgeistListener tracebacks on SV_FILE_DELETED (affects: 1) (heat: 7)" [High,Confirmed] https://launchpad.net/bugs/692730 [14:05] Launchpad bug 693545 in ubuntuone-client "ZeitgeistListener tracebacks on AQ_UNLINK_OK (affects: 1) (heat: 10)" [High,Confirmed] https://launchpad.net/bugs/693545 [14:05] NEXT: mandel [14:06] DONE: more sso ui. Realized that I ALWAYS forget to update setup.py and that it is not tested. Wrote a bad patch to test setup.py, yet I've got a better idea of how to do in using u1-dev-tools. Reviewed alecus branch. [14:06] TODO: doctors appointment, do that setup.py testing thing. [14:06] BLOCKED: no [14:06] thisfred, please [14:06] * DONE peer reviews [4/4] [14:06] * DONE talked to vds about a strategy for couchdb view performance optimization [14:06] * INPROGRESS get ubuntuone-couch into natty [1/2] [14:06] - [X] 0.2.0 https://code.launchpad.net/~thisfred/ubuntuone-couch/release-0.2.0/+merge/53317 [14:06] - [ ] ubuntuone-couch FFE http://pad.lv/729117 [14:06] * INPROGRESS dbus activation of control panel http://pad.lv/728722 https://code.launchpad.net/~thisfred/ubuntuone-control-panel/dbusify [14:06] * INPROGRESS notifications/alerts of quota events http://pad.lv/702172 https://code.launchpad.net/~thisfred/ubuntuone-client/quota-notifications [14:06] * TODO assist in couchdb migration [14:06] NEXT: dobeyh [14:06] λ DONE: more libu1/musicstore frustration [14:06] λ TODO: new webkit api in libu1, bug #727558, bug #733327 [14:06] λ BLCK: My brain hurts. [14:06] Launchpad bug 727558 in libubuntuone (Ubuntu Natty) (and 3 other projects) "Need to notify user when Purchased Music folder is not subscribed (affects: 2) (dups: 1) (heat: 295)" [Medium,Confirmed] https://launchpad.net/bugs/727558 [14:06] Launchpad bug 733327 in libubuntuone (Ubuntu) (and 1 other project) "Notify user of missing MP3 support (affects: 1) (heat: 6)" [Medium,Confirmed] https://launchpad.net/bugs/733327 [14:07] ralsina: arriba [14:07] DONE: reviews, doctor's appointment, canonicaladmin, trying to get into allhands.canonical.com, almost fixed rhythmbox [14:07] TODO: make sure we have a reasonable release soonish, send mail to invite testing, several calls, evaluations [14:07] BLOCKED: no [14:07] comments? [14:07] i hate JS/HTML/CSS [14:07] (aka, the web) [14:08] dobey, it probably hates you back [14:08] and it's everywwhere [14:08] alecu: i thought for #729259 we agreed to just make the scrolledwindow be SHADOW_IN for the shadow type [14:08] we should put you on rotation in beuno's team to break your spirit [14:09] Weird, I had never noticed quassel has rich text support. I wonder how that looks for others [14:09] beuno: i think it hates me, and i'm doing the returning of hate [14:09] ok, eom? [14:09] eom [14:11] no really, i have spent 3 days trying to figure out how to make this all fit together, but every single page in the music store is completely different, and i can't find the css for the thing i want to use anywhere, and my brain just goes into wtf-mode when i look at this code now :( [14:13] maybe i'll just hack it up and pop up a dialog or something [14:15] ralsina: I have a doctors appointment in 5 mins, is just around the block, I should be back on time for the weekly, I might be like 5 min late max [14:16] mandel: ok, no problem [14:17] forget this making it pretty junk [14:18] thisfred, hey... your latest branch for u1-couch didn't include any packaging... was that an oversight? [14:19] hmm, I was assuming there was a web version of the u1 music store. There isn't one right? [14:19] dobey: I say make it work, and pretty we can ask for help [14:19] Stecchino: it is web based, but is not accessible via the web yet [14:20] kenvandine: uhm, I'm not sure I understand, perhaps I pushed the wrong thing then? [14:20] dobey: but the same htmls are used by banshee and rthythmbox? [14:20] ralsina: making it work by putting stuff in-page is much more complicated than i thought it waas going to be, it seems :( [14:20] kenvandine: yeah looks like it [14:20] :) [14:20] Stecchino: they both use the same widget we provide in libubuntuone for the music store, yes [14:22] ralsina: so i'll do something similar to the "you need to install plug-ins" thing that drops down at the top of the page in firefox [14:22] dobey: go ahead [14:24] kenvandine: ah, just forgot to bzr add debian/ and apparently the strict bzr push doesn't complain when pushing to +junk or something like that [14:24] kenvandine: fixed now [14:24] Stecchino: that's why i asked if you were going to use it, or reimplement with qt :) [14:24] I see now [14:25] I think I'll rather make a qt version of thatr [14:25] thisfred, cool, i'll look later today [14:25] thx! [14:25] thisfred, i will get it sponsored this week... sorry i keep getting busy [14:25] :/ [14:25] or at first use the html directly with some url rewriting to integrate in amarok well [14:26] kenvandine: I know how that is ;) [14:26] dobey, for bug #729259 nessita specifically asked for design input. When she went on vacation I was tasked to ping that up, but I haven't gotten any reply yet. [14:26] Launchpad bug 729259 in ubuntuone-control-panel (Ubuntu Natty) (and 2 other projects) "Scrollable list of devices looks broken (affects: 1) (heat: 300)" [Medium,Triaged] https://launchpad.net/bugs/729259 [14:26] ralsina, perhaps I should listen to dobey and "just make the scrolledwindow be SHADOW_IN" [14:29] alecu: so, I have this branch which I *think* does the minimal work necessary for dbus activation, but I'm not sure how to test this. I guess the only way is to also do the packaging branch that picks up the new dbus service, and then build the package and install it? [14:29] thisfred, can you point me at it? [14:29] Stecchino: the thing is that oauth is required, so it's not so much rewriting URLs, as dealing with the authentication [14:29] alecu: sure: https://code.launchpad.net/~thisfred/ubuntuone-control-panel/dbusify [14:30] alecu: you were trying to change the background color instead? [14:30] dobey: isn't the easiest way to do oath redirect to a webpage to log in? [14:30] ok, got the thing about input design now [14:30] thisfred: you can run the bits by hand, and then just run what rquires them [14:30] the rewriting of urls is more for playing the previews and such [14:30] dobey: awesome, now I just have to find out what bits and how [14:30] ;) [14:31] ralsina, I was doing nothing about it until getting input from design. [14:31] Stecchino: not quite, we use ubuntu-sso-client [14:31] alecu: do the border, then tell to design what you did, show a screenshot and ask for input on that [14:31] ralsina, keeping it like that is an option too (ugly for me, but who knows) [14:31] ralsina, cool. [14:31] alecu: I'd use blink tags, and make everything pink 24pt comic sans. That usually gets you feedback the quickest ;) [14:31] alecu: no, as it is now is just too ugly [14:32] thisfred: if you run ubuntuone-syncdaemon from the tree for example, and then run u1sdtool, it will talk to the one you ran from the tree (assuming system one wasn't already running, in which case it would be obvious it failed anyway) [14:32] thisfred, nice idea! [14:32] make the border raised in 3d and 5-pixels wide [14:32] Stecchino: i don't think we've quite set things up such that we support doing "web apps" yet, for talking to u1, which is basically what you seem to propose doing [14:32] dobey: right, but if the service does not exist at all yet, does that still work? [14:33] aquarius: ^^ is that right? [14:33] thisfred: yes, unless you're wanting to test that activation works, instead of talking to the service [14:33] dobey: that's exactly what I want to test [14:33] thisfred: if you run a service, it's what registers itself on dbus. the .service file doesn't have anything to do with registration [14:33] since the service does nothing other than that for now [14:34] dobey: no, I mean that signing on to other oath services such as twitter is usually handled by letting the user log in via a website [14:34] thisfred: you can also start a private dbus session bus daemon, with a custom config that points at the directory where your .service file is [14:35] dobey: but I don't have a .service file. I guess I just rename the service.IN and hardcode the path in there? [14:35] dobey, that's what I was about to propose as well. [14:35] Stecchino: like i said, not quite :) [14:35] thisfred: right [14:35] thisfred: although i don't much see the point of testing activation [14:36] dobey, you are correct. Stecchino, we use 2-legged oauth, not 3-legged oauth. So you retrieve an authenticated access token from ubuntu-sso via D-Bus, and then use that token to sign requests. You don't need to request a new token for the amarok music store. [14:36] ok, then I just need to know how to make a custom config :) [14:36] no problem, there is QOath: http://qt-apps.org/content/show.php/QOAuth?content=107420 [14:36] thisfred: if the resulting .service file has the right path and is installed in the right place, it will work [14:36] dobey: well, the bin file uses dbus to not start twice, which is the part I want to test [14:36] thisfred, also: try to move the REQUEST_NAME_REPLY_EXISTS bits to some file that gets tested [14:36] thisfred: dbus activation is well tested [14:37] thisfred, and log something before you do the "sys.exit" [14:37] thisfred: oh, well did you implement it in the same way that ubuntuone-preferences did before? [14:37] dobey: I have no idea [14:37] I copied bits from ubuntu-sso, but I don't know whether I did too much or too little [14:38] thisfred: http://bazaar.launchpad.net/~ubuntuone-control-tower/ubuntuone-client/stable-1-4/view/head:/bin/ubuntuone-preferences#L1093 [14:38] dobey thanks, that looks subtly different again [14:40] alecu: so, I still don't know how to test those parts. I could mock dbus, I guess [14:42] thisfred, I was going to say... "ubuntuone/controlpanel/integrationtests/test_dbus_service.py" [14:43] thx :) [14:43] thisfred, but I just saw that the only test skipped is the one that tests that bit :P [14:43] i was going to say "don't" [14:43] I was going to say "oh" [14:44] I could just propose the branch, and take the day off and when it breaks nightlies hope that someone will fix it? [14:45] :D [14:45] it makes to sense to me that people are often so insistent on testing that underlying libraries work in apps [14:45] no sense [14:45] dobey: I don't I want to know that MY code works [14:45] which I'm less than 50% sure of [14:46] unit tests aren't the way to check that you're calling APIs correctly though [14:46] error handling is [14:47] Integration tests are, or alternatively, mock tests, but for those, you have to know the correct API calls first [14:47] thisfred: if you want to test that your code works, then run one instance, and try to run a second instance [14:47] if your code works, you won't get two instances [14:47] et voila, done [14:47] Right [14:47] you don't need to test dbus activation [14:47] dobey, we want that to be repeatable. [14:47] dobey, if that code gets changed we won't know till it breaks. [14:48] dobey, that's why we have tests in the first place. [14:48] So my original question was mostly: how do I do that (manual test) [14:48] Then I can add mock tests that prevent regressions [14:48] PYTHONPATH=. bin/foo [14:48] then do the same thing in another terminal in the same tree [14:49] again that has nothing to do with activation [14:49] and tie us nice and tight to the current implementation ;) [14:50] the current implementation is deprecated anyway [14:51] dobey: well, since the stuff uses dbus to check whether the service is running, I have to have the service working. So create a fake service, somehow point dbus at that (which is the part I don't know how to do) run the code, profit [14:51] no [14:51] loss? [14:51] you keep conflating service file with actual service [14:51] they are not the same [14:51] fake service file [14:52] the service file is completely irrelevant [14:52] oh ok [14:53] it is only needed for activation. if you just run the script, the service will be registered [14:53] testing [14:53] after I upgrade since sso is still broken here [14:57] yay got it working [15:00] ah, but now we need to still make the switch to panel work [15:01] so I guess I do have to implement methods on the service. bah [15:02] alecu: ^ There is no other (acceptable) way to get the currently running instance of the ControlPanelWindow if it's already running, right? [15:04] if you want to talk to an already running service, dbus is the way to do it [15:05] thisfred, "implement methods on the service" <- yes, that's the way to get called in the process of the current ControlPanelWindow. [15:06] right, so this branch is not, as I'd hoped, gonna be small and simple :) [15:06] ok, diving in [15:06] well you can make another branch [15:06] that's the great thing about branches [15:06] dobey: yeah, but this branch as is will break the command line arguments [15:06] you can have so many of them [15:06] all full of leaves [15:07] thisfred, dobey, ralsina, mandel, joshuahoover, Chipaca, cparrino: meeting in 10'? [15:07] if the panel is open already [15:07] ralsina, mumble or irc? [15:07] mumble [15:08] thisfred: well, i don't think that's a big problem [15:08] ok, well then I'll propose it, and move on to the next one [15:09] dobey: if we do the packaging branch which adds the service right after that, we'll have activation [15:09] anybody around mumble, that can help me test it? [15:09] alecu: i just used skype call testing service to test my mic [15:10] alecu: am now [15:10] dobey, I know my mic works: the gnome control panel shows it moving [15:10] woah [15:10] why did i just get the mumble audio wizard [15:10] thisfred, thanks! [15:11] dobey: if the headphone/mic it's configured to use is not plugged in, mumble breaks horribly. Is that it? [15:11] * thisfred makes coffee [15:13] mumble? [15:14] thisfred: no, my mic is always plugged in, and i use the speakers [15:14] oh well [15:17] ralsina: mubmle! [15:17] my mumble is having trouble starting, I'm getting there === nathanael is now known as natschil [15:37] Chipaca: hello, I am back.... so you mentioned something about an eventlog or an 'undo' feature for ubuntu one... are these features unusable still? === christoph_ is now known as natschil [15:38] Chipaca: sorry, wrong nick. What christoph_ said was actually from me. [15:39] natschil: I said we were working on having an eventlog, and we have a rudimentary undelete feature [15:41] Chipaca: ah I see. bummer. So ubuntu one does not store old versions of files at all? I would assume when syncing it only syncs a diff anyways, so aren't those kept? [15:41] natschil: that's a good assumption, but it's counterfactual [15:42] it would be awesome if we could rsync, but cloud storage doesn't work that way, usually [15:42] thisfred: it could be made to. [15:42] natschil: depends on the storage [15:42] yep, and the plan is to do that at some point [15:42] natschil: you need CPU on the storage side [15:42] natschil: amazon S3 doesn't have that for instance [15:43] so you could build an intermediate layer [15:45] Chipaca: well, to be frank, not having a chengelog really sucks. I'm lucky that I have a hard copy of the file, so I'll type the 1,5k words all up again, but right now it seems the culprit is ubuntu one, which I stupidly assumed to be a backup system, something I have heard mentioned several times. Maybe the website should mention that it is not useful for backups, when, for example, it mentions storage. And to be honest, if Ubuntu One wants to gain [15:45] any substantial, non-fed up userbase, a changelog feature would probably be useful, i mean, even other 'cloud' apps, such as google docs have changelogs. Anywho, I don't want to be inflammatory, just slightly angry that I lost this work, and probably won't use ubuntu one to backup any more. [15:47] long time no see ;-) [15:47] I always feel stupid saying bye on mumble :-D [15:48] thisfred: true, you need a cpu there... but at least for ubuntu one, you do have a cpu. [15:49] dobey: I need to leave for 15 min, but can I pick your brains later about the setup.py? [15:49] mandel: yes of course [15:49] dobey: I'd like to do it using ubuntuone-dev in a smart way, rather that a crap check [15:49] mandel: i need to get lunch myself, so will be gone for a bit [15:49] natschil: we do, but not on the "machine" the storage is on, which makes it harder, not impossible, which is why Chipaca said we're thinking about it ;) [15:49] dobey: cool, I'll ping you later then [15:50] it's software, nothing is impossible. it is only inconsequential :) [15:50] nothing is impossible, everything is terrible [15:50] http://www.everythingisterrible.com/ [15:50] thisfred: what do you mean? ubuntu one runs (ran) on my machine all the time, and did all kinds of things, it could easily have simply sent the diffs up. In cases where diffs weren't used, I don't think it would use that much server cpu to compute a diff, though I'm not so sure. [15:50] for one user it's easy, for 50 users it's slightly harder, for millions of users it is very hard [15:52] natschil: the server doesn't need the diffs between what you have on your machine and what you had before. It needs the diff between what the server has and what your computer has. That's why you need CPU on the server side. [15:52] natschil: he means the data is not necessarily stored on the same machine that handles the uploads/downloads service. the server architecture is a farm, not a barn [15:52] natschil: so you send a diff, then something on the server has to apply that to the original file, which is non trivial with most cloud storage, since the files there are write once, so you'd have to load the file into memory or onto some other storage, apply the diff, and save it to cloud storage again [15:53] dobey: ralsina: thisfred: I see, that makes sense. [15:53] anyway, the answer is "we have been and are continuing to, look into improvements in this regard" [15:53] * mandel away for 15 min [15:53] also, lunch time [15:53] bbiab [15:53] also COFFEE [15:53] natschil: or we could save the diffs, and then make your browser apply the diffs to the original patch when you click download [15:53] I'm not feeling that itch yet [15:54]