[00:59] <tefal> hi, there !
[01:00] <tefal> does someone know how to have the ubuntuone applet in ubuntu 10.04 ?
[01:02] <duanedesign> tefal: hello
[01:02] <duanedesign> tefal: theapplet was replaced by the Ubuntu One Preferences Panel accessed by going to Me menu --> Ubuntu One
[01:03] <tefal> duanedesign: ok, so is there a way to have notifications, like the applet used to do ?
[01:17] <duanedesign> tefal: Ubuntu One does not use the Notify USD Notifications anymore
[01:18] <tefal> duanedesign: ok, don't know if it's a good thing, yet... :)
[01:18] <tefal> duanedesign: thanks !
[01:19] <duanedesign> tefal: no probem, anytime :)
[01:19] <duanedesign> problem*
[02:25] <Chipaca> mwhudson: ping
[02:25] <mwhudson> Chipaca: hi
[02:25] <Chipaca> mwhudson: lucid?
[02:25] <mwhudson> Chipaca: yes
[02:25] <mwhudson> i just read that this might be much better in maverick...
[02:26] <Chipaca> more or less (a.k.a. "oh **** yes"
[02:26] <Chipaca> )
[02:26] <Chipaca> mwhudson: lots of small files was the worst case we had. It still won't be brilliant, but the difference on reconnect makes it usable
[02:27] <mwhudson> Chipaca: is there a ppa i can try?
[02:27] <Chipaca> mwhudson: I don't know if it'll work, but you can try
[02:28] <Chipaca> mwhudson: some other bits have changed, and I'm not sure if they'll all work in lucid *today*
[02:29] <mwhudson> well
[02:29] <mwhudson> willing to try
[02:29] <Chipaca> mwhudson: also, it's a one way road
[02:29] <Chipaca> I mean, it upgrades the metadata
[02:29] <Chipaca> so the old client won't read it afterwards, you'd have to nuke it and start over
[02:29] <mwhudson> and indeed, it seems to be the rescan on reconnect that means this hasn't got anywhere yet
[02:30] <mwhudson> Chipaca: will an old client on another box be able to read it?
[02:30] <Chipaca> if you're experimenting, you should be ok
[02:30] <Chipaca> yep
[02:30] <mwhudson> then fine
[02:31] <Chipaca> ok. Let's see what's the least cutting edge ppa I can get you on :)
[02:32] <Chipaca> apparently, the nightlies :(
[02:32]  * ajmitch guesses that crack-of-the-day should work fine
[02:32] <Chipaca> mwhudson: ppa:ubuntuone/nightlies
[02:33] <mwhudson> Chipaca: should i u1sdtool -q before upgrading?
[02:33] <Chipaca> mwhudson: can't hurt
[02:34] <mwhudson> The following packages have been kept back:
[02:34] <mwhudson>   python-ubuntuone-client ubuntuone-client ubuntuone-client-gnome
[02:34] <ajmitch> mwhudson: you're trying apt-get upgrade, aren't you?
[02:34] <Chipaca> ugh
[02:34] <mwhudson> ajmitch: yes
[02:34] <Chipaca> ah, try dist-upgrade
[02:34] <ajmitch> that won't work if there are new packages
[02:34] <ajmitch> dist-upgrade ought to :)
[02:35] <mwhudson> ah ok
[02:35]  * mwhudson installs other updates first
[02:35] <Chipaca> if it says "I'll remove X", be suspicious
[02:35] <ajmitch> either ubuntuone gets no bugs, or subscribing to bugmail  just isn't working for me
[02:41] <Chipaca> sorry, got disconnected. Did I miss anything?
[02:42] <ajmitch> just me complaining about not getting bug mail
[02:42] <Chipaca>  ajmitch: what're you subscribing to?
[02:42] <Chipaca> (copied from my history :) )
[02:43] <ajmitch> ubuntuone, ubuntuone-client, ubuntu-sso-client
[02:43] <ajmitch> I seem to be getting bugmail for only the latter
[02:44] <ajmitch> it's probably something I need to take up with a LP person (hi mwhudson!)
[02:44]  * mwhudson puts his linaro hat on, sharpish
[02:45] <ajmitch> hah
[02:45]  * mwhudson escalates his efforts to get ubuntuone-syncd to exit
[02:45] <ajmitch> kill -9 tends to work in most situations
[02:46] <mwhudson> yeah, it might come to that
[02:46] <ajmitch> I'd hate to think what it may do to files on disk
[02:49] <Chipaca> just kill should work
[02:49] <Chipaca> especially if it's stuck in local rescan or something :)
[02:49] <Chipaca> killall ubuntuone-syncdaemon
[02:49] <Chipaca> tadaa :)
[03:03] <Chipaca> mwhudson: how's it going?
[03:04] <mwhudson> Chipaca: ah, i hadn't restarted the daemon :)
[03:11] <mwhudson> hm
[03:11] <mwhudson> the log now just says
[03:11] <mwhudson> 2010-08-19 14:04:13,970 - ubuntuone.SyncDaemon.fsm - INFO - loading metadata from old version '4'
[03:11] <mwhudson> i guess this will take a while?
[03:11] <mwhudson> lots of io is happening
[03:13] <Chipaca> yep
[03:13] <Chipaca> non-solid state drive?
[03:13] <mwhudson> yes
[03:13] <Chipaca> how many files?
[03:13] <mwhudson> ~70k
[03:13] <mwhudson> i like to make life hard for people :)
[03:14] <Chipaca> yep, that'll take a while, because we don't persist the metadata index
[03:14] <Chipaca> that'll happen during N
[03:14] <mwhudson> 'while' == 'an hour' ?
[03:15] <Chipaca> not sure
[03:15]  * Chipaca checks
[03:15] <Chipaca> mwhudson: i have ~30k files and it takes a minute on my ssd
[03:16] <Chipaca> so, x10 for ssd->non-ssd, and x2.5 for num of files
[03:16] <mwhudson> it's taken 10 so far
[03:16] <Chipaca> roughly, 15 minutes
[03:16] <Chipaca> :(
[03:16] <Chipaca> that must be so much *fun*!
[03:16] <mwhudson> so nothing unexpected yet i guess
[03:16] <Chipaca> you *like* suffering, dontcha
[03:17] <ajmitch> given what he works on..
[03:17] <Chipaca> mwhudson: I know who we'll go to to test perf improvements
[03:17] <lifeless> mwhudson: so; bzr ? :)
[03:17] <Chipaca> mwhudson: there is an implementation of FileShelf that used sqlite, but we scrapped it. That was a long time ago, before people storing 10k files was reasonable. We might have to revisit that.
[03:17] <mwhudson> lifeless: i'm helping make u1 better!
[03:17] <lifeless> flamingspork was putting his notmuch stuff into rev control
[03:18] <lifeless> mwhudson: you could put your repo in u1
[03:18]  * ajmitch is glad that he's not sticking all his mail in u1
[03:18] <lifeless> Chipaca: what does fileshelf do ?
[03:18] <Chipaca> one of our tests was checking out bzr inside Ubuntu One, and seeing how long it took to work on the other side
[03:19] <lifeless> Chipaca: generations?
[03:19] <lifeless> Chipaca: I meant to say
[03:19] <Chipaca> lifeless: persists a big-ass dictionary to disk
[03:19] <lifeless> Chipaca: you should add a busy commit-edit-commit-edit loop to that
[03:19] <lifeless> because what you
[03:19] <lifeless> what I saw your test do was unlikely to trigger bugs unless your code was -really-broken- :)
[03:19] <Chipaca> lifeless: I've lost track of the syncdaemon development in detail, as I now manage desktop+
[03:21] <mwhudson> ah hah
[03:21] <lifeless> congratsa
[03:21] <mwhudson> different stuff in the log now
[03:21] <Chipaca> mwhudson: "OM_NOM_NOM_FILES"?
[03:21] <mwhudson> so 15 minutes was a pretty good guess :)
[03:22] <Chipaca> I levelled guessing after three years of physics :)
[03:23] <Chipaca> mwhudson: u1sdtool -s should now say "READY"
[03:23] <mwhudson> last log entry is
[03:23] <mwhudson> 2010-08-19 14:20:17,881 - ubuntuone.SyncDaemon.local_rescan - DEBUG - comparing directory '/home/mwh/Ubuntu One/Maildirs/INBOX/cur'
[03:24] <Chipaca> so, first it reads in the metadata, then it does local rescan, then "ready" state (i.e. "let me connect already")
[03:24] <Chipaca> you can say u1sdtool -c to ask it to connect when ready
[03:24] <mwhudson> u1sdtool says "Failure: dbus.exceptions.DBusException: org.freedesktop.DBus.Error.NoReply: Did not receive a reply. Possible causes include: the remote application did not send a reply, the message bus security policy blocked the reply, the reply timeout expired, or the network connection was broken."
[03:25] <mwhudson> this is the big directory it's on i guess
[03:25] <Chipaca> not sure why, but local rescan seems to block a lot
[03:25] <Chipaca> I'd call it a bug, this being twisted
[03:25] <mwhudson> is guess even stuff like os.listdir maybe takes a while
[03:25] <Chipaca> what's your memory looking like? (32bits, or 64?)
[03:26] <mwhudson> 260 meg RSS, 64 bit
[03:26] <mwhudson> growing slowly
[03:26] <Chipaca> ok, not too bad
[03:26] <Chipaca> you're probably the biggest controlled-but-in-the-wild case of lots-of-small-files I've seen so far :)
[03:27] <Chipaca> os.listdir is pretty bad, by the way
[03:28] <Chipaca> (I guess you know that already)
[03:28] <Chipaca> the bad thing is that os.walk uses os.listdir, so memory use is stupid big
[03:28] <lifeless> not -that- bad, bit its slow enough
[03:29] <lifeless> so bzr has an optimised one
[03:29] <lifeless> (which you should be using :P)
[03:29] <Chipaca> lifeless: how is it memory-wise? we're not too worried about speed, but memory can kill us
[03:29]  * mwhudson wonders what fraction of the io is writing 500 lines to the log file every second
[03:29] <lifeless> Chipaca: tight
[03:30] <lifeless> Chipaca: bzr worries about memory too
[03:30] <Chipaca> lifeless: and have you done an optimized walk too?
[03:30] <lifeless> Chipaca: yeah
[03:30] <lifeless> different interface, stats optimially, that sort of thing
[03:32] <Chipaca> lifeless: where?
[03:33] <Chipaca> mwhudson: if you're going to run it with debug on, maybe make ~/.cache/ubuntuone/log a symlink to /dev/shm/u1log ?
[03:34] <Chipaca> (or just mount a tmpfs there)
[03:35] <mwhudson> actually i don't think it's so bad
[04:21] <Chipaca> mwhudson: still local rescan?
[04:22] <mwhudson> Chipaca: how do i tell?
[04:22] <mwhudson> u1sdtool is still hanging
[04:23] <Chipaca> mwhudson: that's a "yes", then :(
[04:24] <Chipaca> mwhudson: also, every MARK_INTERVAL, it prints the current state to the log
[04:25] <Chipaca> mwhudson: grep NOTE.*MARK | tail should work
[04:25] <Chipaca> (on the log)
[04:25] <mwhudson> ok
[04:33] <mwhudson> what's MARK_INTERVAL usually?
[04:33] <mwhudson> i haven't seen one in the last 6-7 minutes or so
[04:34]  * Chipaca looks in /etc/xdg/ubuntuone
[04:34] <Chipaca> 120
[04:34] <Chipaca> 2 minutes
[04:54] <duanedesign> Chipaca, I am liking the Nautilus improvements
[08:39] <mandel> morning!
[11:13] <leeb9972> anyone know how i can add machine on ubuntuone, i do not get the option?
[11:18] <rye> leeb9972, hi, what distribution are you running?
[11:19] <leeb9972> 10.4
[11:21] <rye> leeb9972, ok, have you had your computer authorized in the past or this is the first time?
[11:21] <leeb9972> i had it authorised in past, but i delted the machine
[11:22] <rye> leeb9972, ok, just a moment
[11:23] <leeb9972> nps
[11:23] <rye> leeb9972, # open seahorse utility (Karmic, Lucid: Applications/Accessories/Passwords and Encryption Keys; Maverick: System/Preferences/Passwords and Encryption Keys)
[11:23] <rye> #
[11:23] <rye> remove the entry called "Ubuntuone token for https://ubuntuone.com".
[11:23] <rye> #
[11:23] <rye> run ubuntuone-preferences from MeMenu again and it should prompt you to re-add the machine
[11:50] <mandel> vds, ping
[11:53] <vds> mandel pong
[11:53] <mandel> vds, did you get my last msg?
[11:54] <vds> mandel: yes but I see strange thing in xchat :)
[11:54] <mandel> vds, such as?
[11:54] <vds> mandel: that's the same error I get running the script from nant
[11:55] <vds> mandel: have you tried to run the script from cmd?
[11:55] <vds> it works
[11:55] <vds> mandel: btw did you get my msgs?
[11:55] <mandel> vds, so, it looks like nant makes the script to behave strange when dealing with sockets
[11:55] <vds> about multiple items, s3 and static file system?
[11:55] <vds> yes
[11:55] <vds> mandel: pretty strange
[11:55] <mandel> vds, bulloks, I'll try to get the debugger and see what is going on
[11:56] <vds> I have no idea of how fork works on win
[11:56] <vds> mandel: before you do that
[11:56] <vds> can we talk a bit about the idea of updating the rss instead of generating a new one>
[11:56] <vds> ?
[11:57] <mandel> sure, give me a sec
[11:57] <vds> as I said the name of the installer is the same so we need different url
[11:57] <vds> URLs
[11:57] <vds> I can easily do it with s3 now
[11:58] <vds> but we've been told we'll store the installers on a static folder
[11:58] <vds> well static folders
[11:58] <vds> so we'll have to change the script
[11:58] <vds> ok
[11:59] <vds> mandel: I do it with multiple items, later we'll figure it out
[11:59] <mandel> ok
[11:59] <vds> mandel: byw I need some help with the uids
[11:59] <mandel> vds, I'll be back in 2 min, something is going on with the network
[11:59] <vds> ah ok
[12:05] <mandel> vds, ok, I'm back I should be able tow ork now
[12:05] <mandel> vds, you were saying??/
[12:05] <vds> mandel: I was saying
[12:06] <vds> that we can list multiple items in the rss
[12:06] <vds> that means having more than one installer available at the time
[12:06] <vds> that means store the installers in different folders
[12:06] <mandel> vds, yes, that is the idea of the rss feed
[12:07] <mandel> vds, ok, is that a problem in s3?
[12:07] <vds> not at all
[12:07] <vds> but s3 is temporary
[12:07] <vds> so we'll have to change when we go in production
[12:07] <vds> not a big issue
[12:07] <vds> I'm working on it
[12:08] <mandel> vds, ok, so the plan is, you work on that and I debug the stupid nant issue?
[12:11] <vds> mandel: yep
[12:12] <mandel> vds, cool, I'll let you know how it goes
[12:33] <mandel> vds, ok, found the issue, when executing with nant we ran out of retries to make the connection, I'm looking at the exact reason right now
[12:33] <vds> wow...
[12:33] <vds> mandel: that's interesting
[12:35] <mandel> vds, we get several times the existing connection was forcibly closed by the remote host, this is comming from the TCP stack from windows
[12:35] <mandel> vds, I'll pocking the boto code to take a look
[12:52] <mandel> vds, looks like the issue is when we try to upload the package, I wonder if the S3ReposneError leaves the connection unusable and we have issues with that
[13:12] <mkarnicki> good afternoon!
[13:15] <mandel> vds, fixed the bug, you script does not like relative paths, why, no fucking clue but the boto connection does not like it, I fixed it by making the build copy the msi to the utils dir, upload and clean
[14:25] <dnielsen> there appears to be a problem with the last version of libubuntuone in maverick with regards to the CIL bindings
[14:25] <dnielsen> https://bugzilla.gnome.org/show_bug.cgi?id=626962
[14:25] <ubot4> Gnome bug 626962 in Community Extensions "Ubuntu One Music Store extension fails to build" [Normal,Resolved: notgnome]
[14:26]  * ajmitch takes a quick look
[14:27] <rye> dnielsen, bug #618945
[14:27] <ubot4> Launchpad bug 618945 in rhythmbox-ubuntuone-music-store (Ubuntu Maverick) (and 1 other project) "Maverick: with the ubuntuone store plugin enabled Rhythmbox can not import files or folders (affects: 2) (dups: 1) (heat: 1745)" [High,Confirmed] https://launchpad.net/bugs/618945
[14:29] <rye> o_O
[14:29] <rye> dnielsen, please ignore my reply
[14:30] <ajmitch> looks to be a slightly different issue
[14:34] <ajmitch> I see the bindings did change a bit in 0.3.2
[14:42] <dnielsen> ajmitch, okay so they aren't as feared borked?
[14:42] <ajmitch> I'm not sure, I'm rebuilding it now to check them :)
[14:42] <dnielsen> excellent
[14:42] <ajmitch> or trying to
[16:35] <rye> duanedesign, you might be interested in http://people.canonical.com/~roman.yepishev/ubuntuone-publish-service.py - automatic publishing of the files based on their path and nice pynotify notification about file publishing state (for all files)
[16:35] <rye> duanedesign, lives in bzr:~rye/+junk/ubuntuone-scripts
[16:36] <mkarnicki> rye: you mean, lp:~rye/+junk/ubuntuone-scripts ?
[16:37] <rye> mkarnicki, ah, yes, you are right
[16:37] <mkarnicki> np
[16:38] <mkarnicki> rye: I'll gladly branch it to learn some python on that :)
[16:38] <rye> because i find the need to publish files pretty often, and this script can be put to startup applications...
[16:39] <rye> mkarnicki, please note that i am learning python at this time as well, so the code can be baaaaad
[16:39] <mkarnicki> rye: no probs, I'm sure I'll only gain from it :)
[16:40] <mkarnicki> rye: btw I haven't started learning yet, so that branch is for pleasure reading :)
[17:17] <duanedesign> rye: that is neat! thank you
[17:27] <rye> http://blog.rtg.in.ua/2010/08/ubuntu-one-automatic-publishing.html
[17:37] <duffydack> nice idea
[17:38] <duffydack> how do I get the notification back like you have ?
[17:39] <rye> duffydack, that's not Ubuntu One notifications, thats generated by the script
[17:39] <duffydack> I know they are for the script, but  Ijus want it back for general
[17:49] <rye> duffydack, me too
[17:59] <lahwran> is it just me, or is updown going really slow?
[17:59] <dnielsen> updown?
[18:00] <Chipaca> lahwran: it seems to be tripping up more than usual, yes
[18:00] <Chipaca> not sure what's up with that
[18:00] <mbarnett> hmm, it does look like updown is having a bit of a hard time.  Beating it with a stick now.
[18:00] <lahwran> :/
[18:00] <Chipaca> dnielsen: updown servers are the ones that put things in and out of ubuntu one through the web
[18:01] <dnielsen> ah
[18:01] <Chipaca> dnielsen: for the file sync part, that is
[18:01] <lahwran> might it have anything to do with my file not having synced in a day or two?
[18:01] <lahwran> I thought updown was the web only
[18:01] <Chipaca> lahwran: nope, updown is only through the web
[18:01] <dnielsen> that will teach me to irc with a fever
[18:01] <Chipaca> dnielsen: yeah, only irc with people. ircing with fevers is always trouble.
[18:02] <lahwran> XD
[18:02]  * lahwran downloads file that didn't sync with this computer and adds it manually
[18:04]  * lahwran is a little annoyed at ubuone now ... looks like it doesn't have the latest version of this file
[18:04] <lahwran> it's not on my cloud part
[18:06] <Chipaca> lahwran: is your syncdaemon connected?
[18:07] <lahwran> not sure
[18:07] <Chipaca> lahwran: u1sdtool -s ?
[18:07] <lahwran> is on this machine at the moment
[18:07] <lahwran> I think it only just connected a minute or two ago
[18:07] <Chipaca> lahwran: what odes u1sdtool -s output?
[18:07] <Chipaca> *does
[18:08] <lahwran> connected
[18:08] <lahwran> and idle
[18:08] <Chipaca> and which is the file that didn't sync?
[18:19] <lahwran> it's a script i'm writing
[18:45] <rye> rodrigo_, ping, how do I use libsyncdaemon from python?
[19:33] <rye> anybody here tried using libsyncdaemon with python?
[20:42] <popey> hullo!
[20:42] <dobey> alo govna
[20:44] <popey> having fun playing with new music store features :)
[20:54] <dobey> :)
[22:28] <dnielsen> ajmitch, any update on the bindings issue?
[22:31]  * ajmitch didn't have much of a chance to look at it - it was nearly 2AM & things weren't cooperating :)
[22:33] <dnielsen> ah
[22:34] <ajmitch> now I recall, I was trying to build stuff & one of the NZ mirrors was dead
[23:01] <rodrigo_> rye, you need pygi, the new python bindings that use the introspection stuff
[23:01] <rodrigo_> rye, not sure how it's used, look for info on live.gnome.org
[23:01] <rye> rodrigo_, here's the problem
[23:02] <rye> rodrigo_, syncdaemon has connect() method and it overrides connect() method of GObject
[23:02] <rodrigo_> hmm, oh
[23:02] <rye> rodrigo_, so it is impossible to get any signal assigned
[23:02] <rodrigo_> rye, so, gobject's connect is g_signal_connect binding?
[23:03] <rye> rodrigo_, and if i rename connect to e.g. do_connect / do_disconnect it does the following: http://paste.ubuntu.com/480629/
[23:04] <rye> rodrigo_, so first bug report will be about inability to tie the signals and the second one is to convert gpointers to gslist or something python-list-compatible
[23:06] <rodrigo_> rye, well, we need probably to make the Python bindings do all that, or do you mean renaming the methods in libsyncdaemon?
[23:08] <rye> rodrigo_, regarding connect/disconnect - there is not much we can do w/o renaming connect/disconnect methods within syncdaemon, unfortunately
[23:08] <rodrigo_> hmm, ok
[23:12] <rodrigo_> rye, so, what do they do in the GLib python bindings to have GObject.connect match g_signal_connect
[23:12] <rodrigo_> that doesn't seem to be automatic at all
[23:13] <rodrigo_> it looks to me there's custom code to do that mapping, so can't we do that?
[23:13] <rodrigo_> mainly, because a API change now, not sure how it would get seen for Maverick
[23:13] <rodrigo_> but I'm ok with renaming the methods and provide #define's to provide the old API
[23:22] <rye> rodrigo_, bug #620735
[23:22] <ubot4> Launchpad bug 620735 in ubuntuone-client "libsyncdaemon gobject introspection cannot be used from python (affects: 1) (heat: 6)" [High,Confirmed] https://launchpad.net/bugs/620735
[23:26] <mwhudson> 2010-08-20 10:25:53,082 - ubuntuone.SyncDaemon.Main - NOTE - ---- MARK (state: <State: 'QUEUE_MANAGER'  (queues WORKING_ON_BOTH  connection 'With User With Network')>; queues: metadata: 64712; content: 75226; hash: 0, fsm-cache: hit=5109776 miss=504593) ----
[23:26] <mwhudson> this is going to take a while?
[23:26] <mwhudson> the queues don't seem to be going down very fast...