[00:36] ok, basic up/download notifications in place, EODing. good night guys! [02:30] in natty, left or right click on tomboy indicator-applet, and sync notes is hidden ammong all other notes. Same for everyone? [04:30] espen77: hmm. not for me [04:31] espen77: I will see If I can find a bug report on this [04:59] duanedesign: for you it is allways one of the three first choices? [05:02] might be just me, sins i am behind a big bad fw and cashing proxies...even impossible to get a fresh iso from here, [05:02] never sure what state anything is in any more [05:05] espen77: it is at the bottom with Preferences, Help, About Tomboy. [05:18] .names [05:32] purge and reinstall helped.... [05:36] I like the u1 notifications on steriods, atleast it shows something is happening with sync every once in a while [06:22] duanedesign: you know of any plans to do anything with pictures in u1 (kinda like picassa)? Not looking for a full clone of the picassa web, but would love a sync of my pictures to u1 based on sd card id or camera id or something. [06:25] espen77: yes there are plans to integgrate Ubuntu One with Shotwell [06:26] espen77: you can keep track of the proggress on the blueprint https://blueprints.launchpad.net/ubuntu/+spec/multimedia-ubuntuone-n-shotwell-integration [06:26] duanedesign: that's cool...will wait to see what that will look like before start whining. [06:35] heh, cool. Does not look like it will ship with 11.04 but likely be in 11.10 [06:43] hm....almost half way from lucid to next lts [06:48] Yes, there were problems doing the shotwell integration. It was decided it's better to let shotwell's plugin architecture settle a bit. [06:58] ralsina: was reading up on shotwell plugins, seems plugin support just made it in the trunk, [06:58] espen77: precisely. We are very close to feature freeze, so there was no time to do things. [07:00] ralsina: figured that, just hoping for it in or before next lts [07:00] espen77: I expect it will be there for next cycle, so yes. [07:00] * ralsina checks if next cycle is the LTS, it's 4AM here :-) [07:01] ralsina: when i have moved my photos from my camera to u1, then what? [07:01] ralsina: still a year to go [07:02] basically, you will have them on the web without having to publish them, and probably can sync back/forth with your mobile later [07:02] ok, gotta play with my insomniac kid [07:03] ralsina: what i was thinking, kinda like the google setup. [07:04] espen77: what part? [07:05] the pictures being on the web, and awailable on the cellphone....that way i can give a link to friends and family, and they can see the pictures, and they will be in the cell phone in the pocket so allways with you. [07:23] espen77: that's the general plan, yes :-) [08:35] hi folks [08:36] I'd like to file a bug but I wanted to ask for your opinion first [08:36] u1 is quite unusable in my system by default by a collection of issues: [08:36] (all issues taken together make me disable u1 daemon as soon as I log in) [08:36] the issues are: [08:38] 1) default limits for outgoing bandwidth are poorly selected, the default is 20Mbit upload, nobody I know has that much upstream speed, it might work fine in a data center but in "at home" environment it always hits the network cap created by the ISP. [08:39] 2) The way most ADSL connections work as soon as you start sending too much you cannot receive traffic normally (lots of packet loss, very large latency). From users' point of view the "internet" does not work anymore. Websites don't load before timing out,DNS often fails to resolve names and so on. [08:41] 3) Users that experience this will (hopefully) want to reconfigure upload speed in u1 control panel, here's another set of issues: the panel requires communication with the server (via the daemon most likely), when issue 2) hits this is largely impossible and on my system times out every time I try to access the panel while uploading hits my network limits [08:42] 4) Then the nail to the coffin, the upload speed control is broken, at low speeds (10-20KB on my machine) it never manages to connect and send anything, on slightly higher speeds (~50KB) it behaves erratically - sometimes it saturates the network (causing issue 2 again) and sometimes it stops doing anything (almost no upload for a brief period of time) [08:44] from my point of view the root cause are bad defaults, I have no such issues with dropbox because they seem to do some smart control of upload speed by default (they have three options: no upload control, "smart" upload control which is selected by default and manual upload control) [08:44] so that's it [08:44] I'd love to know what you think, I'm going to report this later today [08:45] Hmmmm.... the default is actually unlimited upload [08:46] Also, it doesn't require a network connection to set the cap, justa connection to syncdaemon that's on your own box [08:46] Other than that, ouch [08:47] ralsina, so the defaults are wrong either way, the no cap never works on ADSL and the default limit is sky-high for ADSL as well [08:48] zyga: I have ADSL and never capped it [08:48] ralsina, but the UI that sets the cap times out and fails to operate when you have dropping outgoing packets all the time [08:48] zyga: but I understand the problem [08:48] ralsina, interesting, what network speed do you normally have? [08:48] 3Mbps download, 256Kbps upload [08:48] ralsina, I have 5Mbit down and 1Mbit up (but 0.7 really) [08:49] ralsina, I suspect a lot depends on the modem that ISP provides [08:49] or maybe 512Kbps upload, I don't quite recall [08:49] zyga: yeah [08:49] ralsina, but I suspect that still might be a common case and we should look at fixing this - it gives very bad impression IMHO [08:50] zyga: really, the cap should not even require an internet connection [08:50] It's all done through D-Bbus [08:50] ralsina, that's right but the way UI is constructed it's not really working this way today [08:50] ralsina, also, when you ask the daemon to disconnect it seems to _finish_ last transfers, that can take a lot of time (minutes, hours?) depending on the size of the uploaded file [08:51] ralsina, I can record a video of what is happening on my system if you want [08:51] zyga: no need really, I believe you [08:51] ralsina, I think it will explain how this behaves to a lot of people that do not experience this on their systems [08:51] ralsina, right, I don't want to convince you I just think it could help to illustrate the problem :) [08:51] zyga: ok, record it then and post it here [08:52] I'll log in with english locale so the video is easiler to follow [08:58] morning all [08:58] good morning duanedesign [09:02] got it [09:03] let me upload that to p.c.c [09:03] duanedesign, hi, could you please check the backlog for my messages, I'd like to know what you think [09:11] morning rye [09:12] duanedesign, morning! [09:21] duanedesign, are you using a laptop? [09:21] rye, yes [09:22] duanedesign, is plugging out the AC working for you fine? [09:23] duanedesign, what do you think about the upload cap issue? [09:23] ah, bug #539467, seems like they have disabled sata alpm completely again. [09:23] Launchpad bug 539467 in pm-utils-powersave-policy (Ubuntu Natty) (and 12 other projects) "SATA link power management causes disk errors and corruption (affects: 25) (dups: 1) (heat: 160)" [High,Invalid] https://launchpad.net/bugs/539467 [09:24] zyga, upload cap? bug # ? I've been poking bandwidth throttling recently so would like to know what's not working [09:24] zyga: i could not add much that ralsina did not already touch on. rye might be able to add something. He was testing upload/download recently [09:24] http://paste.ubuntu.com/568089/ <---rye [09:24] rye, did you see my earlier messages? [09:24] because it was working for me no matter what i tried to do [09:25] zyga: I pastebin'd it :) [09:25] rye, I'm uploading the video of the session to people.c.c, it should be there soon [09:25] duanedesign, it's hard to pastebin that ;-) [09:26] zyga, so you set 10-20KiB and it behaves horribly, right? [09:26] testing [09:26] rye, it either respects the limit and fails to work (no upload, stuck trying) or does not respect it and saturates the link [09:27] rye, the video I created is pretty interesting - I think you should see it, pictures speak more than words I believe [09:27] 18 minutes left [09:28] zyga: I am worried about not being able to set the cap once the internet connection is saturated [09:28] zyga: so I am anxiously waiting for the video [09:28] ralsina, I think there are a few bugs to file here, just wait please :) [09:29] 2011-02-17 11:28:59,168 - twisted - INFO - pause writes for: %s 18.1292480469 [09:29] ralsina, you know what, i think i know what's happening [09:29] so, syncdaemon works with bandwidth throttling the following way: [09:31] it measures the amount of bytes it should write/read to/from the socket for the given speed, pushes that amount to the socket then waits for some time in order to maintain the average speed specified by user [09:31] but [09:31] since it pushes the whole block at once it is basically using 100% of bandwdith on low bandwidth links for a small amount of time [09:32] something tells me it has to do delays for sending between every sent packet [09:33] ralsina, does that make sense ^ [09:33] and we need to fix that pause writes for: %s format [09:33] rye: well, I have never looked at that code [09:34] zyga, do you happen to know about any quick ways to plot the network link usage? [09:34] So I don't want to say anything that can be misunderstood [09:36] rye, gnome-system-monitor and sudo iftop -B -i ethXX [09:37] rye, 8 minutes left to upload [09:37] rye, define "low bandwidth links" please [09:37] rye, the behavior to upload whole block without capping is IMHO broken, this should not work like that [09:38] rye, how large is the block size? [09:40] erm, i don't understand, i can't find the code [09:41] 5 minutes left... [09:41] Aloha! [09:45] because i am searching in the client, not protocol, phew. [09:47] rye, ralsina: http://people.canonical.com/~zyga/Maverick%20Movie.avi [09:47] 139MB [09:48] urgh [09:52] zyga: I'm in the middle of an epic code review, but I will look at it today [09:52] ralsina, thanks [09:52] ralsina, I [09:52] ralsina, I'll be here if you want to chat about it [09:53] zyga: cool [10:03] hm, looks like my system does not want to play that [10:09] well, great, on 20KiB it does not want to start uploading :-/ [10:09] rye, mplayer works for me [10:09] rye, ah, so you reproduced that! great [10:09] rye, in my video you can also see that the network cap is never respected, as if the setting did not exist [10:10] rye, it's more interesting when you consider 3G connections - that speed is all you might get (and even then you pay through the nose) [10:10] rye, we might want to ask n-m about connection type and offer separate cap for 3G [10:11] rye, does it also try to connect and fails for you? it's using bandwidth without getting any progress [10:11] zyga, i agree, i think we may need to poke facundobatista when he returns online about the implementation of the bandwidth throttling. [10:12] rye, I don't get one thing, I keep coming back with the throttling implementation bugs, is it really working for everyone else? [10:13] zyga, like i said before i've been testing it with various values changing during the transfer, but, since my network could accomodate those values it all went fine, i got the average readings around the requested values and it looked like that was fine. [10:14] rye, what values did you try? [10:14] zyga, from 10 to 2000 [10:14] rye, wow, you can get 2000 of upload? [10:15] zyga, yes, max is 3750 for me [10:15] you have almost 40Mbit of uplink? [10:16] wow :-) [10:16] rye, you could try with iptables rules that would emulate different connection [10:16] rye, I used a dedicated box a few years ago, it had controls for latency, bandwitdth, packet loss and a few other quirks [10:16] rye, worked like a hub otherwise [10:17] rye, it was really good for testing IPTV software we were writing at the time [10:17] zyga, i think i'll need to find out how tc works and cap myself [10:18] rye: check the Linnux Advnced Routing and Traffic Control Howto [10:18] the LARTC [10:18] rye, one more question, do you want me to file a bug about this? I think there are a few more than one based on the video I posted [10:18] rye: short version, it'sa pain [10:18] zyga: yes. Filing bugs is good. [10:19] ralsina, yes, last time i did that it was pretty quick but i did not understand what i have been doing [10:19] ralsina, is the video working fine for you? my variant is a slideshow [10:19] rye: that's the usual patter with the lartc. I have even charged clients while not understanding it. [10:19] rye, try without compiz perhaps [10:19] zyga, is md5sum of that file 2e6d91408e3ca2fa317aa22cc5713855 [10:20] zyga, i am running nouveau, no compiz for me [10:20] and no bouncing icons in unity too [10:20] rye, well the video is rather large but I can play it on my desktop [10:21] zyga, it starts fine but after 5 second it starts being a turn-based strategy game [10:21] rye, cpu load/ [10:21] rye, I can transcode it to some other code/resolution [10:22] rye, the md5 sum is correct [10:23] zyga, ok, will try to transcode here then [10:23] rye, that recording was done by vmware btw [10:25] hm, no, that does not work for me too [10:25] rye, let me do it [10:26] rye, what is your screen resolutin? [10:26] 1280x800 [10:27] zyga, vlc, mplayer, gstreamer-things all show slideshow, should it be a smooth video? [10:27] zyga, wait, looks like it is ok further in the file [10:27] rye, not smooth, a few fps [10:28] rye, it's the code, it's optimized for capturing screen sessions [10:28] ok [10:28] I just played this on atom netbook [10:28] cpu usage is barely noticeable [10:29] zyga, ok, at 02:24 syncdaemon is in STANDOF state, but network monitor still shows the activity, that's something i have not seen before, i must say [10:29] STANDOFF [10:30] zyga, re: indicator - i need to update it to support correct event canceling... 70% hmmm [10:30] rye, STANDOFF is "sending last file"? what does it do in that state? [10:32] rye, is it possible that the control panel is stuck waiting on a dbus call to the server that in turn requires network connection (and takes a long time) ? [10:32] rye, this might explain why I cannot disconnect when the network is saturated [10:33] zyga, control panel talks directly to the server-side rest api for account, devices, it talks to syncdaemon to get connection status, bandwidth throttling readings and the list of the volumes and their subscription status [10:33] rye, what about speed control? [10:34] rye: there is no way to remove a package from the "affects" in launchpad? [10:34] bug #487875 is really about either the screensaver or compiz or something like that, not ubuntuone [10:34] Launchpad bug 487875 in ubuntuone-client (Ubuntu) "Problemas com os efeitos especiais (dup-of: 465230)" [Undecided,Incomplete] https://launchpad.net/bugs/487875 [10:34] Launchpad bug 465230 in ubuntuone-client (Ubuntu) "Need better error handling for timestamp failures (affects: 93) (dups: 13) (heat: 370)" [High,Triaged] https://launchpad.net/bugs/465230 [10:34] zyga: that should only require local communication between the control panel and syncdaemon. [10:35] zyga: it should even work if you unplug your adsl [10:35] ralsina, okay [10:35] rye, I recorded another shorter session that shows the UI is not responding in that case, I'm transcoding for upload [10:36] rye, and also observe that the speed limit was never effective, I could not see any limit being enforced [10:38] zyga: about that, the best persons to ping are probably lucio and facundobatista [10:38] weird, i pushed zyga's movie to ubuntuone folder and current transfer shows nothing, and nothing in waiting content [10:38] zyga: they should be here in a cuple of hours [10:38] ralsina, I'll keep that in mind, thanks [10:39] ralsina, i think we need to push to get this looked into since something has been happening with bandwidth throttling for a long time [10:39] rye: sure [10:39] ralsina, what link is facundobatista on? [10:39] rye: no idea [10:39] uh, the second movie is just 5MB, vmware codec is crappy compared to h264 [10:40] rye: but he can't have more than 5Mbps since he lives nearby and noone sells much more than that here [10:40] well, selecting 20KiB in controlpanel enough for the things to go weird [10:40] is [10:41] rye: that's good, if it's repeatable. [10:41] because then we can really bug foundations about it :-) [10:41] please have a look at http://people.canonical.com/~zyga/Maverick%20Movie002-1.m4v it's very short and shows the same thing [10:41] checking [10:44] rye, perhaps the amazon server is dropping connections if the speed is too low? [10:45] zyga, hm, could you please try u1sdtool --quit the client then run it in debug mode - /usr/lib/ubuntuone-client/ubuntuone-syncdaemon --debug from the terminal? [10:48] o_O [10:48] Maverick Movie uploaded 12 minutes ago [10:49] rye, sure just a second [10:50] rye, done, running in a script session with full transcript [10:50] rye, do you want me to interact with the preferences in any way while that happens> [10:52] rye, as before it's saturating my link (the cap is set at 10K [10:54] rye, another thing, I think that the client is not able to resume previous uploads, is that true? I still see it trying to push one .mp3 file I have [10:54] zyga, could you please check me, 139MB on 20KiB throttling should take about 2 hours to upload, is that correct? [10:55] 118 minutes, yes [10:56] * zyga has crazy idea [10:56] nope, not this -- for a second I thought the upload/download caps were swapped [10:58] rye, why is the client uploading multiple files at the same time? [10:59] zyga, because it is a new feature we have, concurrent uploads and no more separate meta and content queue which means that files will be uploaded faster [10:59] I see [10:59] it definitely wan not 2 hours since i started uploading [10:59] well I have lots of data already, do you want me to stop this session? [10:59] rye, I have a movie + script again [11:00] rye, I was doing the same thing, playing with throttling speed, looking at network usage etc [11:00] rye, pausing / resuming [11:01] 2011-02-17 12:30:46,298 - hashing starts; 2011-02-17 12:35:55 - file is uploaded [11:01] 5 minutes [11:01] that's not 20KiB [11:01] rye, that's not 20 :- [11:01] right [11:01] rye, so shall I stop the --debug session now? [11:02] zyga, yes, you may stop debugging [11:02] zyga: I see the UI problem now. We should try to show the local device without pulling the whole list first (that is what requires network) [11:02] zyga: if you had no internet, then syncdaemon would return immediately because it would know it can't pull the list from the server, and it would be quick. [11:03] zyga: if your net was working well, it would be quick, too. [11:03] zyga: but with a very slow net, it sucks badly. [11:03] re [11:03] sorry, vm crashed [11:03] ralsina, that's great news then [11:04] zyga: well, no, I think it sucks :-D [11:04] * zyga feels QA work for u1 is becoming useful :-) [11:04] ralsina, It's good you found it :-) [11:04] zyga: the bad news is, we don't have time slots to handle it, I think. [11:05] zyga, if we assume that syncdaemon writes whatever it can during 1 second (lets say i have 1MiB/s connection to amazon) then it would have taken 2.5 minutes to upload the file but since we set up some waiting time it could have made this last 5 minutes. [11:05] But that part at least deserves its own bug assignerd to ubuntu one desktop+ [11:05] i need to talk to facundobatista, it looks weied [11:05] weird [11:07] I know you are all busy working on things (bugs and features) but I hope someone can still have a look a this [11:07] quality > features [11:07] zyga: yes, I agree. [11:07] zyga: but there is no quality freeze [11:07] hehe right [11:19] rye, http://people.canonical.com/~zyga/ , I've just uploaded the --debug session + corresponding video [11:27] rye, is the --debug log what you wanted? [11:27] zyga, yes, please [11:27] rye, I mean is the log I uploaded what you wanted ;-) [11:28] http://people.canonical.com/~zyga/u1-debug-session.txt [11:29] zyga, ok, so at 2011-02-17 11:52:11,396 it was trying to upload the mp3 file [11:30] yes [11:30] zyga, ok, i suppose we will talk to facundobatista about the implementation and then try to find out what goes wrong [11:31] ok, I'll get back to work, ping me when you want any feedback from me [11:59] nessita: we should really add a link from somewhere in Ubuntu One to the music store UDF if the machine has it subscribed. Don' t you agree? [11:59] Otherwise the music is hidden unless you know where to look [11:59] I do agree, but I'm pretty sure that is in place already. If the link is not there, we're having a bug [12:00] nessita: it' s there in the machine that buys the song [12:00] it' s not there in your other devices, AFAICS [12:00] hum, so is not that trivial [12:00] since we can't add that symlink always [12:01] people that doesn't purchase music will not want the symlink [12:01] nessita: yes, trivial things I don' t ask around ;-) [12:01] ralsina: do you know who is creatying the symlink today (in the box you buy music)? [12:02] I was thinking, that if you click on subscribing the folder via control panel, then it could be added but separation of concerns bit mr in the ass :-) [12:02] nessita: I am guessing the store plugin [12:02] nessita: but that' s just a wild-ass guess [12:02] I think that too (that is the plugin) [12:02] ralsina: so, I'm thinking this: [12:03] * we can't move the burden to create the symlink to syncdaemon since syncdaemon is music-agnostic [12:03] * the one in charge to create the link in other computer should be the same music store plugin [12:03] I was thinking control panel because the user has to subscribe it manually. [12:04] we may not even have the plugin installe [12:04] d [12:04] * but, what happen if the user has purchased music but the plugin is not installed? [12:04] right [12:04] ralsina: I see that as a hack, you know. But doable, of course. [12:04] yep, a hack [12:04] but I see noone else that we know is installed except syncdaemon [12:05] ralsina: 'is installed'? [12:05] right, if the store plugin is in charge of symlinking it, it may not be installed, or may never run at all. [12:05] So the symlink won' t get done. [12:05] right [12:06] If the user is subscribing to thepurchased music folder, he has control panel [12:06] (or knows how to use u1sdtool and we don' t care about him ;-) [12:06] So we can assume control panel is there and is being used [12:06] well, I guess we should do both: ensure that the store plugin creates the symlink in any box, and also do the check in the control panel [12:07] nessita: yeah, sounds good. [12:07] I will chew on it a bit [12:07] ralsina: file me a bug please :-) [12:07] ah, ok [12:07] maybe inspiration will strike [12:10] confirmed, rhythmbox-ubuntuone-music-store creates the symlink. I' ll check if the banshee store does too. [12:12] rye, zyga, hola? [12:13] hi [12:13] facundobatista, good day :) [12:14] zyga, good day! [12:15] awesome, the banshee music store crashes banshee [12:15] ralsina, crashee [12:15] the banshee ubuntu one music store plugin, that is [12:15] I am going to hack the servers and ship xmms [12:15] facundobatista, hola! We have a confirmed issue with bandwidth throttling. But first of all, how does it work? [12:16] rye, it counts how many bytes are sent/received, and stops for the second when you get the maximum for that second [12:17] facundobatista, how does it sends the data? as blocks of some size or as a stream? [12:19] zyga, it does not change that.... [12:19] zyga, think it as a "waiting tcp proxy": [12:19] zyga, let's suppose you have 10KBps max [12:19] facundobatista, so basically it can write 2Mb of data per second then halt then write 2Mb in one second again? [12:19] zyga, so, some second starts, and it receives 2KB, it just sends it [12:20] it receives 5kB, insid the same second, it just sends it [12:20] facundobatista, something is not working then, we reproduced that on two different systems [12:20] it receives 5KB more, in the same second.... oops, max exceeded, let's wait for the next second... ... (wait) ... send the 5KB [12:20] facundobatista, I uploaded a --debug traece and a video that shows what I was doing [12:21] facundobatista, http://people.canonical.com/~zyga/ check out the third movie and the .txt file [12:22] nessita: I subscribed you to bug #525210 which is kinda the root of the music folder thing we discussed. [12:22] Launchpad bug 525210 in rhythmbox-ubuntuone-music-store (Ubuntu) (and 2 other projects) "UbuntuOne musicstore data isn't saved according to the XDG Base Directory specifications (affects: 3) (heat: 22)" [Wishlist,Confirmed] https://launchpad.net/bugs/525210 [12:22] facundobatista, what about upload speed setting, will it send maximal amount of data for the given second and then notice that it has sent it? [12:23] rye, don't understand your question [12:23] rye, zyga, btw, if you want to check the code, it's ThrottlingStorageClientFactory in client.py in the storageprotocol [12:24] facundobatista, how soon are the changes in throttling speed "noticed", let's say you upload 100MB and play with the throttling setting, will it affect the speed in real itme? [12:24] facundobatista: the granularity may be a bit coarse, by the way it affected zyga [12:24] zyga, nop, you need to restart the client [12:24] facundobatista, does it measure the sent and received data at second boundary only or it applies the delay for every sent/received chunk of data (say, 1024 bytes) [12:24] O_O [12:24] facundobatista: ok, that explains it :-) [12:25] facundobatista, why do we have this as a realtime setting? [12:25] facundobatista, uh, that's a bug then [12:25] facundobatista, on the second hand, it's not respected - from what I've seen [12:25] rye, "realtime setting"? [12:25] facundobatista, it's still sending the fastest way possible, the cap was set to various low settings and was never respected [12:25] nessita, we have bandwidth throttling in ubuntuone-preferences and control panel and it requires syncdaemon restart to become effective [12:26] zyga, that may be a bug, yes [12:26] nessita, as we have just found out [12:26] nessita, we need to restart the client to get new configs, right? [12:27] (maybe we could send kill and it will reload? [12:30] zyga, you can not use "system monitor" to see the bandwidth usage for syncdaemon [12:30] zyga, we had a lot of "almost bug reports" of "SD uses a lot of my network", and it wasn't syncdaemon at all [12:31] zyga, you need to use a tool like nethogs [12:31] facundobatista, why not? [12:31] zyga, because something else may be using the network [12:31] facundobatista, so you claim that something else was using the link at that time? [12:31] facundobatista, I also used iftop - believe me it was u1 [12:31] facundobatista, I know how to check things like that [12:31] zyga, I don't know, but it's no use for a bug report [12:31] facundobatista, I was annoyed by this bug long enough to check this over and over [12:32] facundobatista, I can do another session if you really want but it _was_ u1 this time [12:32] zyga, which one? using more than the bandwith it had, or not changing the behaviour after changing the setting while running [12:32] facundobatista, you can also see this in the video when the daemon was stopped and started multiple times [12:32] ? [12:32] facundobatista, let me scroll back and copy my original bug [12:33] hi folks [12:33] I'd like to file a bug but I wanted to ask for your opinion first [12:33] u1 is quite unusable in my system by default by a collection of issues: [12:33] (all issues taken together make me disable u1 daemon as soon as I log in) [12:33] the issues are: [12:33] 1) default limits for outgoing bandwidth are poorly selected, the default is 20Mbit upload, nobody I know has that much upstream speed, it might work fine in a data center but in "at home" environment it always hits the network cap created by the ISP. [12:33] 2) The way most ADSL connections work as soon as you start sending too much you cannot receive traffic normally (lots of packet loss, very large latency). From users' point of view the "internet" does not work anymore. Websites don't load before timing out,DNS often fails to resolve names and so on. [12:33] 3) Users that experience this will (hopefully) want to reconfigure upload speed in u1 control panel, here's another set of issues: the panel requires communication with the server (via the daemon most likely), when issue 2) hits this is largely impossible and on my system times out every time I try to access the panel while uploading hits my network limits [12:33] 4) Then the nail to the coffin, the upload speed control is broken, at low speeds (10-20KB on my machine) it never manages to connect and send anything, on slightly higher speeds (~50KB) it behaves erratically - sometimes it saturates the network (causing issue 2 again) and sometimes it stops doing anything (almost no upload for a brief period of time) [12:33] from my point of view the root cause are bad defaults, I have no such issues with dropbox because they seem to do some smart control of upload speed by default (they have three options: no upload control, "smart" upload control which is selected by default and manual upload control) [12:33] so that's it [12:33] I'd love to know what you think, I'm going to report this later today [12:34] facundobatista, from my point of view the throttling is not working, no matter what you do with the setting and restarting the daemon [12:34] zyga, ok, let me tackle each point [12:35] 1) why you say the default limit is 20Mbit? in /etc/xdg/ubuntuone/syncdaemon.conf, I see 2097152 [12:35] (bytes/sec) [12:35] so 2MBit [12:35] Chipaca, yeap [12:35] 2M bytes/s is roughly 20Mbit [12:35] facundobatista: 2MB == 16Mbit? [12:35] hmm? [12:36] ah, ok [12:36] facundobatista, nope, not 2 [12:36] sorry [12:36] ah, bytes per second? [12:36] I know it's not 20 but it's easier to speak like that [12:36] zyga, no, I was confused with the bit/bytes, ok [12:36] zyga: with TCP overhead is more than 20Mbps as your ISP measures it [12:36] rye: I honestly would not know, I'm just exposing the DBus iface, If it requires restart, is a pain :-( we can force it but from my POV is too violent [12:37] zyga, anyway, the default is high, but the default is "on = False", so the feature is not activated [12:37] facundobatista, that's another bug IMHO, see my rationale [12:37] facundobatista: I don't know, I would expect you to know that. I would say no, but just because I would think is common sense [12:37] zyga, so, by default, syncdaemon will try to use all your network [12:38] nessita, maybe verterok knows about the restarting for the config [12:38] maybe. verterok you around? [12:38] nessita: whasup? [12:38] zyga, if that is a bug for you, open a bug in launchpad but do not mix that with the other points [12:38] facundobatista, did you check what dropbox does by default? they monitor the network and try to see how much they can send and slow down if some other traffic shows up - while that's tricky to implement it's infinitely better from users point of view - and it's the default [12:38] zyga, we will fix it, or not (try to find if something similar was filed before) [12:38] verterok: when setting new throttling limits, are those values taken into account without restarting syncdaemon? [12:39] nessita: setting the limits how? [12:39] verterok: using the DBis iface [12:39] nessita: if you change the config file manually, syncdaemon needs to be restarted [12:39] zyga, awesome, didn't know what dropbox do regarding network, do you have some code to show that? [12:39] facundobatista, my points were not bug reports per se, they were points that lead to conclusion "u1 does not work out of the box" [12:39] verterok: DBus config iface [12:39] nessita: no, if you use the Dbus iface the config is applied [12:40] facundobatista, rye, zyga: ^ [12:40] facundobatista, nope but it's easy to reproduce this - I did not try to automate that if that's what you are asking [12:40] verterok: thanks! I thought so [12:40] verterok, thanks [12:40] nessita: at least that was the bahaviour at the time it was implemented :) [12:41] verterok: facundobatista: the bandwidth throttling is not working for a lot of people, I can confirm that much. [12:41] Chipaca: I'ld say that it's working, but not as expectd :) [12:41] verterok, how so? [12:42] verterok: crashing network hardware isn't "working" [12:42] verterok, it's not respecting any limit I set [12:42] Chipaca: crashign hardware? [12:42] verterok: (yes, that is happening also) [12:42] zyga: don't know about your specific issue [12:42] verterok: yep :) [12:42] zyga, so, let's continue with your points? [12:42] facundobatista, sure [12:43] verterok: probably flaky hardware to start with, but still [12:43] zyga, Chipaca: the main bug with throttling is that it's too deep in twisted/netwrok stuff, and doesn't know anything what is actually thtrttling [12:43] *about what is throttling [12:43] 2) you're right! [12:44] so, SD limit the network usage no matter what...and maybe it's stopping sending bytes in the middle of a protocol message [12:44] facundobatista, do you agree with 1 too? that was not conclusive [12:44] Chipaca, zyga: ^ [12:44] verterok, SD? [12:44] zyga: syncdaemon, sorry :) [12:44] ah [12:45] verterok, but how would that break the throttling? [12:45] facundobatista: what dropbox do is monitor the packet queue (/proc/net/tcp's tx_queue and rx_queue), apparently [12:45] Chipaca: if an aplication can crash network hardware...it sounds like a serious problem in the stack (IMHO) [12:45] Chipaca, if it does we can check that with strace easily [12:45] zyga, regarding 1), I don't know why it's configured with that, you have a better suggestion for the default? [12:45] zyga: I'm just pointing the known bugs :) maybe some change completetly broke it [12:46] facundobatista, yes, enable the limit by default and change the upload speed to a much much lower value, perhaps show this UI when setting up u1 on a new computer each time [12:46] verterok, it's not "completely" broken, remember my problem when I had the limits in 200 and 500 bytes [12:46] facundobatista, I have no hard data to back my claims but 20Mbit uplink is probably not something 10% of users even dream of [12:46] verterok: is there a way to just limit the content queue and not the meta queue? [12:47] Chipaca: no! that's "the bug" [12:47] :) [12:47] zyga, the default is off, maybe users will change that number when activating the feature? [12:48] facundobatista, the feature _should_ be active by default - that's my claim in 1) [12:48] facundobatista, if you enable it by default and keep existing speed limit then it's as good as disabled [12:48] Chipaca: throttling currently works in a lower level, and know nothing about queues or meta or content or syncdaemon :) [12:48] zyga, ok, it should be active and in a lower limit... say 1Mbit? I like that default, it worths a bug report [12:48] 1Mbit for download, 100kbit for upload [12:48] those are nice limits [12:49] facundobatista, the actual limit is still not as simple, consider three cases: modem, broadband and 3g, but I agree in general [12:49] facundobatista, those sound sane to mee to [12:49] zyga, yeap, it's not as simple [12:49] me too even [12:49] zyga, we don't want to give the image of "syncdaemon is slow" because of too low limits [12:49] zyga, it's a business decision, not technical, IMHO [12:50] facundobatista, I understand but "sync is slow" vs "this breaks internet whenever I use it" is also a business thing [12:50] facundobatista, we should not be broken out of the box, whatever we do [12:50] facundobatista: nobody on adsl in .ar has an uplink greater than 512kbps, fwiw [12:50] Chipaca, yeap [12:50] ok [12:50] onwards to claims 3 and 4 [12:51] 3 is about recovery when the user is smart enough to link slow internet to u1 (and the new notification might help with that) [12:51] zyga, 3) there it may be a bug also, maybe the control panel should be usable with the network saturated, or even with the network disconnected? [12:51] 3 is the failure of the control panel UI to work under saturated link [12:51] facundobatista, right AFAIR ralsina already found a bug there, correct? [12:51] zyga, I don't know enough of the control panel to help you, nessita or ralsina will be of more help [12:51] zyga: does it also not work without a network link at all? [12:51] Chipaca, AFAIR not [12:52] Chipaca: it works without a link [12:52] nessita: ^? [12:52] Chipaca: it works with a fast link [12:52] zyga, so, point 4, the one that seems more like a problem... which client are you using? [12:52] Chipaca: it doesn' t work with a slow link, because it asks for the list of devices before you can throttle your own [12:52] ralsina: does it work when network manager thinks you have a link but the link isn't connected to the internets? [12:52] (I think) [12:52] Chipaca, if you disconnect and n-m notices this it tells you that you need to be online, if you drop the link in a way that n-m is not aware (like disconnect your network modem while still being connected to a switch to keep the link up) the U1 will not work [12:53] facundobatista, the latest nightly [12:53] Chipaca, (and by U1 I mean the control panel here) [12:54] zyga, so, let's see how we can reproduce that, in your machine and mine [12:55] facundobatista, which point do you want to focus on reproducing now? [12:55] zyga, 4, that the limit works [12:55] ok [12:56] zyga, give me a couple of minutes so see if we could log a couple of things better [12:56] sure [12:56] zyga: hey there. What is broken with the control panel regarding net limits? [12:56] nessita, you cannot get to them on a saturated link [12:57] nessita, and once you do they don't work (But that's another issue not related to UI) [12:57] zyga: what does it mean 'you can not get them'? [12:57] nessita, I posted a few videos on http://people.ubuntu.com/~zyga/ [12:57] nessita, they show what happens, please try the smaller ones first [12:57] nessita, various things time out [12:57] nessita, and the whole section is grayed out [12:57] nessita, this is all on latest nightly [12:58] er [12:58] http://people.canonical.com/~zyga/ [12:58] zyga: I know what is happening, and, I'm not sure what we can do about it. The network limits are asked to syncdaemon via dbus, but the device list is asked to our servers [12:58] so, we're using http traffic to get the device list [12:58] which can not be retrieved since you have a saturated link [12:58] nessita, we should have a separate part for the local device that works regardless of the rest [12:58] hi folks, I see we've got some long discussion :) [12:58] nessita, IMHO the setting of speed limits should not be there :-) [12:58] zyga, why you have a "debug" item in your [bandwith_throttling] section? [12:59] facundobatista, where? [12:59] zyga: maybe, but that's a whole completely separated issue [12:59] facundobatista, I ran the third video with --debug [12:59] nessita, agreed [12:59] zyga: the UI design puts that info there [12:59] zyga, you say that file was changed because you run the client with --debug? [12:59] facundobatista, which file, I don't follow you, sorry [13:00] zyga: can you please file me a bug for that? description should be 'limits can be set because device list can't be retrieved due to slow connection' [13:00] ralsina: not sure what we can do about it ^ [13:00] nessita, perhaps you can construct the list with the "local" machine and give control to the throttling part and asynchronosuly populate it with additional devices [13:00] /home/zyga/.config/ubuntuone/syncdaemon.conf [13:00] nessita, I'll queue it, there are a lot of bugs here [13:00] nessita: I don' t know the internals enough either [13:01] zyga, you do a cat of it around 1:15s [13:01] If we could show the local device entry without asking syncdaemon, that may help [13:01] facundobatista, ah - no that was present from a previous debugging session [13:01] ralsina: the device list is fetched doing a REST api call, we don't have one device at a time [13:01] zyga, can you remove that item, please? if all is fixed, *that* was a problem [13:01] ralsina: and we don't have the information of the local device either, in order to show it separately [13:01] facundobatista, sure [13:01] facundobatista, that would be a fail on my part too ;-) [13:01] nessita: so ajo & agua [13:01] nessita: so ajo & agua [13:02] zyga, in that case, it would be a real bug in the config management code [13:02] zyga: sadly we can' t fix the control panel [13:02] ralsina: in the same scenario occurs if you have files sync disabled, devie tab is useless [13:02] nessita: why don't we have the info of the local device? [13:02] Chipaca: because that info is in the servers, not in syncdaemon [13:02] facundobatista, it did not help [13:02] facundobatista, I captured the session as video again [13:03] Chipaca: syncdaemon knows nothing about the local device (name, type, date added, etc) [13:03] nessita: that's a regression from u1-prefs [13:03] facundobatista, the connection saturated the link instantly [13:03] zyga, wait, wait [13:03] facundobatista, the whole traffic was going to amazon/u1 servers [13:03] facundobatista, ok [13:04] Chipaca: you sure? [13:04] nessita: positive :) [13:04] Chipaca: what does preferences show when no device list is in place? [13:04] ralsina, why not? design constraint? [13:04] nessita: u1-prefs had placeholder values for the local machine until it got the server list [13:05] Chipaca: right, the UNKNOWN placeholders, which people complains about... not sure what is best/worse [13:05] zyga: we don' t have the right info to show without connecting to ask for it, but maybe Chipaca and nessita are getting somewhere [13:06] ralsina, could we not use the hostname before we fetch the true values? [13:06] zyga: the host name may not be the device name [13:06] facundobatista, I uploaded the session as movie 004 [13:06] nessita, right but it's better than nothing, another idea - since we cannot edit the device name today perhaps you could cache it in the UI? [13:07] Chipaca: we could guess the name using the token name. Not sure how to redraw later with the real info without making an overwrite in the UI, maybe surprising the user [13:07] zyga: we could, yes. We're not doing for this cycle due to lack of time [13:07] nessita: the local machine should always be first IMHO, so the rewriting would be more of an appending, which wouldn't be so surprising [13:07] zyga, rye, Chipaca, verterok, found it! [13:08] zyga, rye, I was bad with how I told you this worked [13:08] zyga, rye, it works by getting what twisted indicates *what already is written or read" [13:08] Chipaca: I mean, rewritting the local machine with server info, that may differ or be more complete [13:08] facundobatista, so did you find the bug or did you confirm that it must work (while it does not for me at least)? [13:09] wait, written _or_ read? [13:09] anyway, my question stands [13:09] zyga, rye, so, if you say that the limit is 10kbps, and twisted just wrote 50kb in the same chunk, it will wait *5* seconds before writing more [13:09] facundobatista, so it's not working in practice, now let's figure out why [13:09] zyga: it is! [13:10] Chipaca, explain pease [13:10] please [13:10] zyga, the efect is a lot of small pikes that saturates your network, with pauses [13:10] zyga: twisted is probably writing too much at a time [13:10] facundobatista, how long will it wait? proportional to the overcommitment of bandwidth limit? [13:10] zyga, on the average, the limitwrks [13:10] zyga: say twisted writes a chunk of ... what's the default kernel-side buffer size? Probably something big [13:10] chunks of 5MB, say [13:10] facundobatista, nope, I can show you that it does not work [13:11] hello all! [13:11] facundobatista, it never goes below my cap, I don't know how much theory you put behind this it is never slower than without the limit in the first place [13:11] then we notice that and wait an hour before doing something else [13:11] hi alecu ! [13:11] hey there karni, how is it going? [13:11] but it's already too late! the kernel is already writing this info out, too much data at a time [13:11] Chipaca, 5MB? what? [13:11] Chipaca, i had wait times of 18 seconds during my tests [13:11] alecu: good! made a downloading/uploading notification yesterdays evening! :) [13:12] Chipaca, how did you come up with 5MB value? [13:12] alecu: x file(s) uploading, y file(s) downloading (in case there's both up and down tranfsers at the same time) [13:12] rye, exactly [13:12] alecu: have you poked around oauth yet :)? [13:12] what's twisted chunk before it calls us back? [13:12] zyga, you can see how SD waits in your log, see the pauses [13:12] zyga: made it up [13:12] karni, I started with it yesterday. still wrapping my head around it. [13:13] facundobatista, checking now [13:13] zyga: I'm looking around to see what the actual value is :) [13:13] alecu: I'll be happy to help anytime [13:13] great, thank! :-) [13:13] :) [13:13] facundobatista, I see values like 5-18 seconds [13:14] zyga, pauses... in that time, SD does not write anything [13:14] facundobatista, what does twisted use at the very bottom to write? send? [13:14] zyga: with what rate limit? [13:14] Chipaca, yes [13:14] zyga: with a rate limit of how much? [13:14] alecu: oh, by the way. at the moment that logic uses the browser as a source for the oauth tokens. we may switch to the new API some time soon (1-2 months? that one is not scheduled yet) [13:14] Chipaca, during the whole session the link was never idle, always saturated, I don't think you can do that this way, the kernel will block if the socket is full, the socket has a limited buffer, something is writing fast enough [13:15] Chipaca, between 19 and 25Kb/s I was changing it during the test [13:15] Chipaca, actual traffic was around 80Kb/s [13:16] Chipaca, I just noticed it wanted to pause for 2+ minutes [13:16] zyga, don't change the limit during the test, it does not work, and confuses everything here [13:16] Chipaca, IMHO it should decide how much to write not how much to sleep afterwards [13:16] zyga: I think we all agree, there [13:16] facundobatista, when does it measure how much bandwidth was used? after each block or after each file? [13:16] * karni is afraid to think about throttling in U1F seeing all this discussion ;) [13:16] Chipaca, facundobatista, zyga: ^ that was what I tried to say...but poorly explained :p [13:17] U1F? [13:17] Chipaca: the Android client for U1 files sync [13:17] ah :) [13:17] zyga, *after* [13:17] Chipaca: ubuntu one files, to to speak [13:17] facundobatista, after what? [13:18] zyga, oh, sorry, after each block of bytes twisted wrote, whatever that is [13:18] facundobatista, do we set the size of that block today? is it by any chance using sendall to send the data? [13:18] facundobatista, doe we reconnect for each new file we transfer? [13:18] verterok, facundobatista: " don't change the limit during the test, it does not work," - so does it in real time or not? [13:18] i am confused [13:19] * zyga opened a can of worms here, :/ [13:19] zyga, we don't reconnect on each new file [13:19] rye: the config change via DBus is in real time [13:19] verterok, do we log when that happens? [13:19] facundobatista: when what happens? :) [13:19] verterok, we had bugs in the past with the GUI setting the syncdaemon in strange limit values [13:19] transfer limit change [13:19] facundobatista, the daemon notices, there's a log for that [13:20] 2011-02-17 11:56:10,260 - ubuntuone.SyncDaemon.DBus - DEBUG - called set_throttling_limits [13:20] zyga, but it does not say with which values? [13:20] facundobatista, not in the log, no [13:20] zyga, could you please open a bug for *that* [13:20] ? [13:20] facundobatista, let me queue the bugs and let's recap before I start loosing context [13:20] zyga, the log should be explicit for it... [13:21] facundobatista, I added this to the bug queue [13:22] nessita, I also got your bug in the queue ('limits can be set because device list can't be retrieved due to slow connection') [13:22] zyga, and then please open the real bug we have for throttling: "it does not throttles inside the second, but sends a bunch of data and waits then some second to be below the limit" [13:22] facundobatista, ok [13:22] facundobatista, and one for the defaults, okay? [13:22] *"...waits then some secondS..." [13:22] default network limits are too high [13:23] ok [13:23] and "default network speed limit should be enabled" [13:23] zyga, please, give me the number of the throttling one, so I can comment that with some code [13:23] ok, let me start reporting them [13:23] facundobatista: verterok: zyga: what happens if we set SEND_LIMIT on the connection? [13:23] zyga: 'can't be set' (I made a mistake before) [13:23] nessita, fixed, thanks [13:23] that way only up to SEND_LIMIT will be sent at a time [13:24] Chipaca, what is SEND_LIMIT? [13:24] facundobatista, let's see if we missed anything from my original 4 points [13:24] (so we'd get to check how much we send more often) [13:24] ah [13:24] right [13:24] one more thing [13:24] "sync daemon does not manage do upload any files when speed limit is low (10KB)" [13:25] facundobatista, I don't know if that description is good but in general with low speed limit it kept disconnecting AFAIR [13:25] facundobatista, I can try to see how this worked [13:26] facundobatista, but from my point of view it was stuck uploading the same MP3 file over and over [13:26] rye: remember this? https://bugs.launchpad.net/ubuntuone-client/+bug/517847 :) [13:26] Launchpad bug 517847 in ubuntuone-client "Bandwidth Throttling is Applied to Content and Control Connection. (affects: 2) (heat: 3)" [Undecided,Confirmed] [13:26] all bugs against ubuntuone, ok? I will let you reassign to proper places [13:27] zyga, if the content transfer saturates the network, it complicates the other commands, yes [13:27] zyga, that is already known, see the bug verterok just pointed [13:27] verterok, sure, i posted that :) [13:27] rye: :) [13:28] facundobatista, I don't think that's my bug, the speed limit _was_ observed earlier (but on on this nightly installation) [13:28] facundobatista, but I'll skip that until I get some hard facts for it [13:28] "it does not throttles inside the second, but sends a bunch of data and waits then some second to be below the limit" - YES! I am not that crazy after all! Wooo-hooo! [13:28] facundobatista, does the client resume previous uploads btw? [13:28] * rye goes back to tomboying [13:28] * zyga starts reporting bugs now [13:29] zyga: not yet....but will start doing it in ~1 week :) [13:29] i hope this word does not translates to something weird [13:29] verterok, great :) [13:29] zyga, yes [13:30] facundobatista, verterok: response conflict [13:30] rye: I think it does (and that is good :) ) [13:30] zyga, ah, "resume" from which point? [13:30] facundobatista: jaja, it doesn't resume uploads (yet) [13:30] facundobatista: it restarts the uploads ;) [13:30] yes [13:30] facundobatista: or resume from 0 :-) [13:31] verterok, so that's what I observed, if it's not reported yet I can add that to the queue [13:31] facundobatista, bug 720702 [13:31] Launchpad bug 720702 in ubuntuone-client "sync daemon does not log the value of set_throttling_limits (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720702 [13:31] zyga: report what? [13:31] zyga: the control panel bug should go to ubuntuone control panel project [13:31] verterok, missing feature / bug [13:31] nessita, ok [13:32] zyga: the feature is already in nightlies, but we need to rollout the server to make the client actually resumae :) [13:32] *resume [13:32] verterok, I see [13:32] nessita, bug 720704 [13:32] Launchpad bug 720704 in ubuntuone-control-panel "limits can't be set because device list can't be retrieved due to slow connectionnes (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720704 [13:33] zyga: thanks! [13:34] dobey: tarmac is not landing u1client branches failing with 'OSError: [Errno 2] No such file or directory: '/var/cache/tarmac/ubuntuone-client/trunk/_trial_temp/dbus-session.conf' make: *** [test] Error 1' [13:35] facundobatista, bug 720707 [13:35] Launchpad bug 720707 in ubuntuone-client "it does not throttles inside the second, but sends a bunch of data and waits then some seconds to be below the limit (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720707 [13:36] zyga, gracias! [13:39] facundobatista, bug 720711 [13:39] Launchpad bug 720711 in ubuntuone-client "default network limits are too high (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720711 [13:41] facundobatista, and finally 720712 [13:41] that's all folks, thanks for working on this! :-) [13:41] (bug 720712) [13:41] Launchpad bug 720712 in ubuntuone-client "default network speed limit should be enabled (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720712 [13:43] zyga, thank you [13:46] zyga, thank you very much! [13:46] rye, it was my pleasure [13:47] * zyga still wants to do LAN sync for u1 but as usual does not have enough time :-( [13:48] facundobatista, re: autoconflict bug, i should start marking all other as a duplicate of that one? [13:49] facundobatista, bug #718924, and maybe remove logs from that first and then mark it as non-private? [13:49] rye: Bug 718924 on http://launchpad.net/bugs/718924 is private [13:51] rye, the one I commented on? I found a case! [13:52] rye, if you write continuosly enough in a file, you will hit a point when a command that send EOF to the server is cancelled, and that is "conflict" [13:53] facundobatista, so the file gets sent to remote storage (in any state it is at that time, be it empty or anything else) but the client does not know that; Upon subsequent saving it checks the remote hash and it has changed so it marks local as the conflict, right? [13:55] alecu dobey nessita mandel thisfred standup in 6' [13:55] ack [13:55] ayeaye [13:55] ok [13:57] ayea [13:58] mandel: one of your branches landed out of order and needs extra reviews [13:59] ralsina: what do you mean? [13:59] and which ? [14:00] mandel: the complex dbus landed before the simple one [14:00] so it failed because the prerequisite was not merged [14:00] and got set to needs review again [14:00] oh, ok [14:00] me [14:00] rye, the process is this [14:00] me [14:00] rye, the file in the server and client is "A" [14:01] me [14:01] me [14:01] rye, there's an upload of B, it sends to the server "the previous was A" [14:01] alecu, thisfred, dobey? [14:01] ha! [14:01] :-D [14:01] rye, the upload gets to the server, but before getting the answer, it is cancelled in the client [14:01] me [14:01] rye, next upload sends C, and says previous is "A", but in the server it's currently B -> conflict [14:02] facundobatista, awesome! [14:02] alecu: say me! :-) [14:02] facundobatista, so, can I start marking all other bug reports as a duplicates? [14:02] me [14:02] DONE: got reviews and tried to land branch for Bug #702616, Bug #709200, Bug #719376. Talked with aquarius re: sync/async credentials API. Started project for files storage API. Control panel bug triage. [14:02] TODO: start working on sync files API. Weekly meeting. IRC discussion re throttling (syncdaemon and control panel). [14:02] BLOCKED: nopes [14:02] NEXT: ralsina [14:02] DONE: lots of bug triaging, bunch of reviews [14:02] TODO: reviews, management stuff, sprint planning, triage another 5 tons of bugs, some coding (my days are all alike) [14:02] BLOCKED: no [14:02] mandel por favor! [14:02] Launchpad bug 702616 in ubuntuone-client "Passing tests with odd DBus errors (affects: 1) (heat: 20)" [Medium,In progress] https://launchpad.net/bugs/702616 [14:02] Launchpad bug 709200 in ubuntu-sso-client (Ubuntu Natty) (and 3 other projects) "Service quits even if ref count is not zero (affects: 1) (heat: 6)" [High,Fix released] https://launchpad.net/bugs/709200 [14:02] Launchpad bug 719376 in ubuntuone-client "ubuntuone.credentials should allow passing a window xid (affects: 1) (heat: 6)" [Medium,In progress] https://launchpad.net/bugs/719376 [14:03] DONE: UX talks. A lot of IPC. Tests, tests... [14:03] TODO: More test [14:03] BLOCKED: No [14:03] thisfred: go [14:03] DONE: worked on tests for u1 couchdb API [14:03] TODO: finish and package API, some u1client UI wrap up [14:03] BLOCKED: still no unity [14:03] NEXT: dobey [14:03] λ DONE: bug #717829, started maverick SRU release [14:03] λ TODO: banshee mustic store fixes, finish maverick SRU [14:03] λ BLCK: on-call for jury duty [14:03] dobey: Bug 717829 on http://launchpad.net/bugs/717829 is private [14:03] uh, sprint planning, I should book tickets [14:03] alecu: go [14:04] DONE: branch to disable notifications from control panel. Worked on OAuth for DroidCouch [14:04] TODO: OAuth for DroidCouch [14:04] BLOCKED: no [14:04] LOVE: eclipse [14:04] HATE: java [14:04] COMMENTS: ? [14:04] rye, if they are about saving lot of times in a row, most probably [14:04] ralsina: shall I buy tickets to buenos aires? [14:04] nessita: print is a go then? [14:04] sprint? [14:04] facundobatista, like gedit, openoffice [14:04] thisfred: not sure. ralsina? [14:04] don' t buy tickets yet but it' s close [14:04] ack [14:04] ah ok :) [14:05] I need to save 20 minutes to finish writing the official proposal [14:05] thought I'd missed a mail [14:05] then it enters the canonical machinery and out comes a sprint (I guess) [14:06] we have the weekly call in a little while [14:07] eom? [14:07] eom! [14:07] dobey: verterok filled bug #720696. Is the control panel supposed to work on lucid? [14:07] Launchpad bug 720696 in ubuntuone-control-panel "control panel nightlies doesn't work on Lucid. (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720696 [14:08] it should i guess, shouldn't it? [14:08] why do i have like 6 russian spam messages in my canonical folder [14:11] ralsina, Chipaca: does this look right to you? https://wiki.ubuntu.com/UbuntuOne/Tutorials/LucidCouchDB [14:11] ralsina, Chipaca: anyway we can streamline that for lucid users? [14:12] i don't think there is [14:12] dobey: is that error fixable on lucid? [14:12] joshuahoover: that looks good, or at least I don' t know how to make it shorter [14:13] nessita: probably. [14:13] although, using GLib via gi seems a bit overkill [14:13] ralsina: yeah...there's 2 things that concern me: 1) the number of steps & 2) that we have lts users enabling "unsupported" backports (at least that's the wording next to the check box) [14:14] joshuahoover: well, we could have them enable "unsupported" PPA instead, and add a couple steps [14:14] or use a terminal and make the steps uglier [14:15] But it' s pretty much the same thing. [14:15] dobey: i'd prefer a ppa over this actually [14:16] dobey: then we could potentially get them other fixes more easily (maybe) [14:19] well we already have one, but it might break the world [14:19] and we really haven't devoted any time at all to supporting all this stuff :( [14:20] hi. i'm on natty, the "Share this folder" link in the web ui doessn't seem to do anything [14:20] neither does the 'More' link [14:20] smoser, firefox 4? [14:20] yeah [14:20] a FAQ? [14:20] https://wiki.ubuntu.com/UbuntuOne/FAQ/WhyDoesntTheFilesWebUIWorkWithFirefox4 [14:20] i see. [14:20] right :) [14:21] any other way for me to share a file ? [14:21] smoser, chrome? or on the desktop? [14:21] i could use chrome, but would actually prefer desktop if thats possible [14:22] sure, just right-click on the folder, Ubuntu One > Share [14:22] in nautiuls ? [14:22] yeap [14:22] i dont see an option. [14:23] Package nautilus-ubuntuone is not available, but is referred to by another package. [14:23] ^ i get that after 'apt-get install nautilus-ubuntuone' [14:23] smoser: ubuntuone-client-gnome is tha package [14:23] joshuahoover: I'd add a /usr/lib/desktopcouch/desktopcouch-stop; /usr/lib/desktopcouch/desktopcouch-get-port as 5b [14:24] joshuahoover: (or "reboot") [14:24] Chipaca: ahhhhhh!!!!! [14:24] joshuahoover: please don't run screaming from the room [14:24] Chipaca: when i tested it, it started syncing without rebooting or restarting desktopcouch...that was on 10.04 with latest updates...i was surprised [14:25] Chipaca: but maybe i just got lucky? [14:25] joshuahoover: you probably didn't have desktopcouch running before the update, and then did something that triggered it to start running [14:26] joshuahoover: it only replicates if it's running [14:26] Chipaca: right..hmmm [14:26] so, i right click -> properties -> Share [14:26] then [14:26] no [14:26] wait. yeah, right click -> Ubuntu one [14:26] not the sharing tab in file properties [14:26] then i type someting [14:26] rightclick->ubuntuone->share [14:27] an email [14:27] dobey: maybe ubuntuone-client-gnome should 'provide' nautilus-ubuntuone just for that to work? [14:28] ok. sorry for being quite dense. [14:28] now, i think i did that. [14:30] Chipaca: i don't think it works that way [14:54] nessita: do you have any pending branches for u1cp? [14:55] ralsina: ping [14:56] mandel: pong [14:56] ralsina: ming poping in mumble 3 min before our meeting [14:56] as in now hehe :P [14:57] ok. popping [14:57] dobey: kinda, why? [14:58] dobey: do you need me to do a release? [14:58] nessita: no, because need new revno for new nightlies to build :) [14:59] dobey: heh. I can give you a needed branch in a sec [15:03] ralsina: here you go http://twistedmatrix.com/trac/wiki/QTReactor [15:04] gah [15:04] i wonder if they break the qt main loop there too [15:06] mandel: https://github.com/ghtdak/qtreactor [15:08] alecu dobey mandel nessita thisfred Chipaca mumble in 8' [15:09] oh right. blah [15:09] yessir [15:09] dobey: https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/align-devices/+merge/50164 [15:10] dobey: I can give it a trivial approve for quicker landing [15:10] i'm not in that big of a hurry [15:10] well, it is trivial :-) [15:10] i was just wondering because my u1cp merges folder was empty :) [15:11] heh [15:11] and i pushed a change to the nightlies packaging which should fix verterok's issue [15:11] dobey: should I test? :) [15:13] verterok: do you have gir1.0-glib-2.0 installed? [15:13] * verterok checks [15:13] dobey chipaca, mumble? [15:13] no [15:13] :) [15:13] dobey: yes, Installed: 0.6.8-1 [15:13] ha [15:14] verterok: and python -c "from gi.repository import GLib" fails? [15:14] dobey: yes, ImportError: No module named gi.repository [15:15] verterok: and "from gi import repository" also fails? [15:16] dobey: yeap: ImportError: No module named gi [15:16] ok [15:20] ugh, glade === yofel_ is now known as yofel [15:23] ugh, and splitting the label up into separate labels with the entry in the middle :( [15:30] i can't find the mail/wiki/whatever that says "don't do that" [15:35] dobey: I do think glade is a great idea. over half a year ago there was a problem that it couldn't attach signals automagically. then something else broke. I ended up coding the whole GUI myself, to be on the safe side. hyhyh [15:35] and the gui was complex ;[ [15:38] karni: rading a .diff of a glade file isn't very informative, though [15:38] dobey: I can imagine :/ [15:42] ok, so, we have a broken tomboy sync at the moment in case the note was edited via the web ui, bug #709378 [15:42] Launchpad bug 709378 in tomboy (Ubuntu) (and 1 other project) "Tomboy fails to sync after editing note via web: "String was not recognized as a valid DateTime." (affects: 17) (dups: 2) (heat: 276)" [Undecided,Invalid] https://launchpad.net/bugs/709378 [15:44] __lucio__: anything up with the servers? I think there's a slowdown [15:46] beuno: sometimes (twice since a moment ago), I get a handshake (conn+set_caps+auth) 30 second timeout. I'm thinking whether I should pop a dialog saying 'This is taking longer then usual' ;)? [15:46] oops *than [15:47] connected this time :) [15:48] beuno: I'll just show 'Trying to connect' Toast and bail after n-th attempt and inform the user. [15:49] nessita: does u1cp run ok on maverick? [15:49] dobey: yes, using it right now [15:49] ok [15:52] karni, sounds good [15:52] beuno: ack :) [15:56] brb, lunch groseries shopping [15:58] hrmm, volumes are not subscribed by default now, right? [16:05] facundobatista, verterok: ^^ do you know if new volumes are still subscribed by default? === beuno is now known as beuno-lunch [16:11] ok, I' ll be gone for a few hours. Have a nice day everyone! [16:12] ralsina: you too! === JanC_ is now known as JanC [16:21] so here's the script that will tell what notes are broken http://people.canonical.com/~roman.yepishev/us/tomboy-sync-validator.py [16:26] nessita: do you know the answer to my question re: volumes autosubscribe? [16:26] dobey: I do, sorry, I missed your question. Yes, volumes are not subscribed by default (neither are shares) [16:27] dobey: starting on natty and nigthlies. Plain maverick should be autosubscribed by default [16:27] nessita: including music? [16:27] dobey: yes [16:27] yeah, i'm wondeirng about narwhal [16:27] dobey: we do no special treatment for music (we might, though) [16:27] ok. that makes for an interesting scenario i have to fix then [16:31] alright i'm off to lunch, bbiab [16:39] is there something special about classmethods/factory functions, when using mocker? I'm not having much luck mocking one. [16:50] is there anybody on maverick w/o nightlies packages installed to test http://people.canonical.com/~roman.yepishev/us/tomboy-sync-validator.py ? [16:56] anybody running natty without ubuntuone nightlies? === beuno-lunch is now known as beuno [17:16] nessita, is there any way for maverick standard ubuntu_sso to be contacted via dbus? [17:17] nessita, i am trying to get sso info from both maverick and natty stable and nightlies and it already works on natty, but in maverick there is no ubuntuone.credentials [17:23] facundobatista: I have a question.. states_manager.svg (under the is_connected line) says, that SERVER_RESCAN, QUEUE_MANAGER and STANDOFF have a selfloop of SYS_HANDSHAKE_TIMEOUT event [17:24] facundobatista: how would that event occur in QM, if SYS_HANDSHAKE_TIMEOUT is generated in check_Version, set_caps, and authenticate [17:26] karni, the handshake timeout is turned off when entered in SERVER_RESCAN [17:26] ah! that's what it means [17:27] karni, as it may take a lot of time, and of course it has no sense in QM [17:27] rye: hey there. what do you need in particular? u1 creds in maverick? [17:27] facundobatista: got it! [17:27] facundobatista: thanks [17:27] karni, the detail is what happens if the event that changes state is just generated a little before the timeout triggers? [17:27] ok [17:27] facundobatista: right :) thank you [17:31] nessita, yes, u1 creds in maverick (preferably w/o twisted magic, just dbus calls) [17:31] rye: sure, open d-feet, access the com.ubuntu.sso service [17:32] nessita, hmmm [17:32] rye: from there, /credentials path and in the interface (can't remember the name), find_credentials [17:32] parameter is the app_name, 'Ubuntu One' [17:32] why do i need to import from ubuntuone.credentials then... i am so smart [17:33] rye: you don't, in maverick it does not exist [17:44] thisfred, ralsina: [Branch ~unity-team/libunity/trunk] Rev 40: Add a Python example for the launcher API [17:45] nice! [17:45] wheeeeee! [17:54] what the heck. where did all of these sudden g-s-d crashes come from [18:04] can I have a couple of reviews for https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/align-devices/+merge/50164 ? [18:08] i must abstain [18:12] nessita, sure, give me a little while [18:14] dobey: why? [18:16] because splitting strings into multiple labels like that is horrible for localization; and i don't want to spend time arguing about it [18:24] ok [18:40] dobey: I miss the answer before, will you be able to fix bug #720696? or shall I do something special? [18:40] Launchpad bug 720696 in ubuntuone-control-panel "control panel nightlies doesn't work on Lucid. (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720696 [18:41] i am fixing it; i think [18:48] nessita, can I see a quick screenshot for your branch? [18:48] beuno: yes, sorry, i should do that already [18:49] it's ok, I should of set up a good workspace to test the desktop client, I'm just lazy [18:52] beuno: http://ubuntuone.com/p/dnA/ [18:52] beuno: don't frek out becasue the lack of styling [18:52] beuno: Daniel Fore is working on coding a RC file (kinda like a CSS file) [18:53] beuno: changes are: labels left aligned, and units to the ight [18:53] nessita, +1 [18:53] very subtle buy nice change [18:53] oh ugh, and the labels were moved :( [18:54] * beuno hugs dobey [18:54] awesomeness can be painful [18:54] thanks! [18:54] another approval, anyone? thisfred, maybe you? [18:54] and they aren't disabled when "[] Limit bandwidth" is unchecked [18:54] this is not awesomeness [18:55] it's breaking a perfectly fine layout for no good reason [18:55] nessita: sure [18:55] thisfred: https://code.launchpad.net/~nataliabidart/ubuntuone-control-panel/align-devices/+merge/50164 [18:56] gotcha [18:56] and if i review the branch and say so and disapprove it, it will devolve into a stupid argument about how "well design said it should be like this and you're just trying to block the branch and blah blah blah" [18:56] I think the UI looks better [18:56] not design said [18:57] nessita: can I trade you a review? (if you don't have time that's fine) https://code.launchpad.net/~thisfred/ubuntu-one-couch/add-tests/+merge/50206 [18:57] better than what? [18:57] than what there is now [18:57] to me it looks like i can change the bandwidth limit values and have them work, without actually clicking "limit bandwidth" [18:57] people don't read right-aligned text very well [18:57] the text wasn't right-aligned was it? [18:58] yes, now it's left aligned [18:58] where is the before screenshot then? [18:58] I looked at my current control panel [18:58] because it wasn't right-aligned in ubuntuone-preferences [18:58] didn't want to make nessita do even more work just to get work through [18:59] it is in control-panel, and this improves it, so I +1ed [19:02] alecu also maybe, since you've looked at it already? :) https://code.launchpad.net/~thisfred/ubuntu-one-couch/add-tests/+merge/50206 [19:02] and breaks localization [19:02] anyway, bah. i need chocolate [19:06] thisfred, ack [19:08] thisfred: sure! [19:09] dobey: about disabling the spin buttons when limit bandwidth is not clicked, is a pending taks, but we're dpoing it [19:10] actually I can add it to that same branch, but I thought you needed a quick branch [19:10] thisfred: before moving further with this project, you may wanna ask for a project rename [19:11] thisfred: all u1 related projects are ubuntuone-something (instead of ubuntu-one-something) [19:12] thisfred, will the project be named: "ubuntu-one-couch" or "ubuntuone-couch" ? [19:12] oh [19:12] didn't read nessita saying the same. [19:16] alecu: :-) [19:17] beuno: by the way, if the user would make a folder syncable, it should start downloading it's contents as soon as it confirms there's enough space, right? any objections/suggestions to such scenario? [19:18] thisfred, what's the right way to run the tests in your branch? [19:18] alecu: ah: PYTHONPATH=. u1trial tests/ [19:19] for now [19:19] thisfred: did you see the project name concern? [19:19] nessita: sure, I'll ask a losa [19:19] nessita: ping [19:20] thisfred: also, the python package we provide should be, for consistency, ubuntuone.couch [19:20] joshuahoover: hey there [19:20] sure [19:20] thisfred: thanks :-) [19:20] nessita: any ideas on what would cause u1-prefs to have contacts, bookmarks, etc. grayed out (only files selectable)? [19:20] nessita: on maverick [19:21] beuno: the only user-story I can come up against this feature is "I want to sync this folder, but without particular file/folder inside it" - but we can tackle that later. [19:21] joshuahoover: yes, desktopcouch is not available [19:21] joshuahoover: that indicates that the module import failed [19:22] nessita: so not started up? user appears to have it installed [19:22] joshuahoover: as per my code knowledge, the module could not be imported [19:23] dobey: is that correct? ^ [19:24] nessita: another question: do you know how I can set the default reviewers to team ubuntuone for my project? [19:25] thisfred: yes, you have to make the project owner and driver 'ubuntuone-hackers' team [19:25] thx [19:25] thisfred, I don't like that the request method returns very different things... strings for some errors, nothing for other errors, and parsed json for success. [19:25] alecu: I agree [19:25] thisfred, perhaps it should only return parsed json, and exceptions otherwise. [19:26] alecu: this will be fixed. The first passthrough was just picking apart the big ball of twine of the original, without functionality changes [19:26] well, I made it work, but other than that ;) [19:26] thisfred, yeah, it looks like a lot of untwingling :-) [19:27] thisfred, and it looks great that there many tests now :-) [19:27] thisfred, so, I'll approve, but I'll set up a bug for that change [19:27] awesome, thank you! [19:30] nessita: that sounds right [19:30] thisfred: which project? [19:30] dobey ubuntu-one-couch, soon to be ubuntuone-couch [19:31] dobey, I want to add that to tarmac, also, but for that it needs automatic testing obviously, which I have not yet set up [19:32] thisfred: it also needs to be owned by the right team, and not by you [19:32] dobey: that's done already I think [19:32] I just changed that [19:32] "just changed it"? [19:32] I will copy all the testing infrastructure from d-c [19:32] as in just now [19:33] thisfred, added bug #720917 and approved. [19:33] changed it how? [19:33] Launchpad bug 720917 in ubuntu-one-couch "the request method returns very different things (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720917 [19:33] dobey: I changed the driver and maintainer. Or is there yet another thing? [19:33] thisfred: you own the branch [19:34] thisfred: thus, nobody else can write to it (ie, tarmac) [19:35] Oh right. I have no idea how to fix that [19:35] or how launchpad works in general ;) [19:36] right. Again, no idea how to fix that [19:37] well you have a branch proposed to merge into it right now, so ii'll wait and fix it after you land it, i guess [19:37] awesome, thx [19:37] I'll ask for a project rename as well [19:39] i think it's about time to pkill -9 firefox here [19:40] yup. 632M RES is not fun, no matter how much RAM you have [19:48] thisfred: I have a question. WHy do you use ubuntu_sso instead of ubuntone.credentials? [19:48] thisfred: I landed a branch yesterday where you can just use: [19:49] from ubuntuone.platform.linux.credentials import CredentialsManagementTool [19:49] cmt = CredentialsManagementTool() [19:49] creds = yield cmt.find_credentials() [19:49] nessita: awesome, did not know this. Again, I mostly just started to refactor what was there, there is *lots* of room for improvement [19:50] the .request method is about 10 times longer than I'd like it to be for instance :) [19:50] mmm, those were some very necessary key lime white chocolate cookies [19:50] now look, you made me drool [19:50] thisfred, I'm running examples/launcher.py from libunity trunk, and it's working as advertised. It shows the progressbar on the launcher icon, sets the progress and shows an emblem. [19:50] awesome! [19:50] thisfred, it's not packaged yet, so I had to make install it :-) [19:51] thisfred, it looks as simple as this: http://bazaar.launchpad.net/~unity-team/libunity/trunk/view/head:/examples/launcher.py [19:51] unity doesn't have super awesome built-when-landed packages? [19:51] dobey, perhaps they have some nightlies ppa, don't know. [19:52] alecu: that looks very usable [19:54] thisfred: so, shall I file a bug or you change the creds stuff in this branch? [19:54] nessita: file a bug please, this branch is long already, and was really about adding tests [19:54] ack [19:54] thanks! [19:56] thisfred: bug #720928 [19:56] Launchpad bug 720928 in ubuntu-one-couch "Use ubuntuone.platform.linux.credentials module instead of ubuntu_sso (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/720928 [19:56] great! [19:57] nessita: should that not be: from ubuntuone.platform.credentials import *? [19:58] the whole point is that we don't want to have the code depend on the platform right? [19:59] thisfred: maybe, I'm not sure how this window magic is supposed to work [19:59] thisfred: you'll have to verify but i don't think mandel did anything beyond just moving it into the platform dir [19:59] ah right [20:00] so we still need to make the __init__ there do the detection magic [20:00] thisfred: approved [20:00] thanks! [20:00] what do you need me for??? [20:01] well you probably need to add a platform/credentials.py that does the right imports from platform.foo.credentials [20:01] mandel: nothing, just some confusion about the ubuntuone.platform stuff [20:01] thisfred: tell me, I might be able to help :) [20:01] well, what dobey said ^^ [20:04] thisfred: all platform code that is going to be present on all platforms should be imported as from ubuntuone.platform import x if the code is specific for a platfrom (such as credentials AFAIK) you have to import it as ubuntuone.platfrom.linux import * [20:05] thisfred: or at least that is the logic I have been following so far, I do welcome other approaches :) [20:05] mandel: so we're not gonna have credentials on windows? [20:05] well, i doubt credentials is specific if we're going to have sso on windows [20:08] thisfred, dobey: by looking at the code in trunk, atm you can import the CredentialManagement from platform [20:08] ok, awesome :) [20:08] verterok: ping [20:08] I could have done that myself, but I only have two hands :P [20:08] dobey: pong [20:09] verterok: are you on 32 bit or 64? [20:09] thisfred: nessita mentioned that such a class was just used on linux tools and that is why I considered the removal of it, is you code multiplatfrom? [20:09] dobey: lucid-64bits, maverick-32bits [20:09] mandel: yes it is supposed to be [20:09] thisfred: cool :) then we have a reason to leave it there :) [20:10] that is a double smile, oh yeah! [20:10] thisfred: if there is anything missing just let me know and i'll fix it asap [20:10] verterok: ok, the amd64 build of pygobject is publishing now, so you should be able to apt-get upgrade and have a working control panel on lucid. and hopefully not have everything else break [20:10] mandel thx, should be fine [20:11] thisfred: ok, got to go now, laters! [20:13] dobey: cool, will test it later [20:14] mandel: my ocde is not multiplatform, it uses DBus and twisted [20:15] nessita: yeah, no your code doesn't need to be. Mine will be by (optionally) passing in the u1 couch tokens on the command line (for now) [20:15] ah [20:16] nessita: yes, but I know, but the 'need' is multiplatform, so your code is an implementation of that need on linux, it just means I have to write a CredentialsMnagement class for other platforms [20:16] we need to get some other dumb ass to help me with this multiplatform crazyness, any volunteer in the room? [20:17] * mandel is going to teach cparrino to code…. [20:22] dobey: ping [20:22] dobey: rename of project, and merge of branch into trunk done [20:22] ok [20:23] dobey: I have u1client installed at Version: 1.5.4+r867~maverick1, which is the same as trunk: [20:23] nessita@dali:~/canonical/u1/client/trunk$ bzr revno [20:23] 876 [20:23] But the code in trunk/ubuntuone/platform/linux/credentials.py does not match /usr/share/pyshared/ubuntuone-client/ubuntuone/platform/linux/credentials.py [20:23] ah no [20:23] I'm dislexic, again [20:23] right [20:23] dobey: any idea why u1client is still at 867? [20:24] packages? not specifically. launchpad might hate us for requesting so many recipe builds though [20:24] dobey: isn't the3re any failed build? [20:25] no [20:26] ok [20:29] ooooh! It's kindergarten time! [20:29] * alecu bbls [20:34] wtf [20:35] why is ubuntuone-control-panel failing to build as a package on narwhal, but works fine on maverick and lucid [20:35] dobey: what error? [20:36] dh_install: python-ubuntuone-control-panel missing files (debian/tmp/usr/lib/python2.*/*-packages/*/ubuntuone/__init__.py), aborting [20:36] which of all the possible things that could be a valid error, this makes the least sense [20:40] dobey: how long is been failing? [20:41] since i bumped the version and added the missing indicator file, i guess [20:47] ah [20:48] ? [20:48] 'ah' as in 'ok, not sure how can I help there' [20:49] yeah, it makes no sense. the package is the same for all 3 versions [21:16] JamesTait: beuno: one day we'll have to adjust the "machines/computers" terminology on the website. during oauth it's "Add this computer", in webUI it's "View machines connected to this account", at the bottom it's "Remove selected computers". [21:16] JamesTait: beuno: and none of that will suit the forthcoming Android app [21:17] the phone/device, I mean. actually, it'll be a bit misleading.. Android phone will be under connected machines section, and not "phones" [21:18] meh, dealing with multi-bug SRUs is annoying [21:18] I'm mentioning this only because I headed to remove some of the tens of tokens I have generated during testing. [21:21] JamesTait: I love the Ubuntu one SSO page. I know the changes have been a while around already, but I really like it! [21:22] ok, I'm off [21:22] bye all! [21:22] bye nessita [21:32] hmm, no synchronous way to get the credentials. That's unfortunate. [21:41] worst log message I came up with evar 'discarding connect, connector is connected' [21:57] verterok: did you try to upgrade from nightlies to see if your issue is fixed? [21:57] dobey: not yet, let me check [22:00] dobey: a new issue :/ [22:00] dobey: from gi.repository import Soup, SoupGNOME [22:00] ImportError: cannot import name Soup [22:00] dobey: mayeb it's better to just not package it for lucid? :( [22:00] verterok: and you have gir1.0-soup-2.4 installed? [22:00] ah, no idea. let me check [22:01] you should, since it should be in the Depends [22:01] yes, Installed: 0.6.5-5ubuntu2 [22:01] hrmm [22:01] i wonder why it won't import then [22:02] dobey: in case it helps: RepositoryError: Failed to load typelib file '/usr/lib/girepository-1.0/Soup-2.4.typelib' for namespace 'Soup': Typelib version mismatch; expected 3, found 2 [22:02] brb [22:02] ok, so it's because libgirepository is incompatible :( [22:03] so i guess we will have to stop building for lucid [22:04] verterok: you should downgrade your python-gobject and gobject-introspection related packages back to what came in lucid [22:09] alright, well, i'm out for the night. cheers all [22:19] dobey: ok, will do. thanks1