[02:56] <leonel> Hello .. How can I tell ubuntuone to NOT get connected when I logon to my desktop ?
[10:40] <duanedesign> good morning U1inites
[10:43] <mkarnicki> morning duanedesign
[11:04] <kermiac> can someone please take a look at bug 589818 - the OP is unsuccessfully trying to sync approx 4500 files (paid account)
[11:04] <ubot4> Launchpad bug 589818 in ubuntuone-client (Ubuntu) (and 1 other project) "High processor load for ubuntuone-syncdaemon (affects: 1) (heat: 10)" [Undecided,Incomplete] https://launchpad.net/bugs/589818
[11:09] <rye> kermiac, hm, syncdaemon will be extremely busy on startup reading a) metadata b) checking whether directories changed and querying the server about that, as far as i can tell
[11:10] <kermiac> hello rye :)  - yeah, I couldn't see anything that was obvious in the logs :(
[11:10] <rye> 2010-06-13 13:06:30,048 - ubuntuone.SyncDaemon.Main - NOTE - ---- MARK (state: <State: 'QUEUE_MANAGER'  (queues WORKING_ON_BOTH  connection 'With User With Network')>; queues: metadata: 197; content: 4022; hash: 0, fsm-cache: hit=295826 miss=15769) ----
[11:11] <rye> suggests that he has 197 metadata entries to go and 4022 content items were not yet uploaded
[11:12] <kermiac> yup, so it's still processing. I know the service has sped up a *lot* but they are trying to sync a *lot* of files
[11:13] <kermiac> and if I understand them correctly, they keep stopping the service, so it will begin doing the server rescan stuff again when they next reboot or re-enable the syndaemon
[11:15] <kermiac> rye: would you suggest that I ask them to leave the PC on & explain that the process of syncing 4500 files will take some time & point them to your u1sdtool wiki page if they wish to check on the sync progress?
[11:16] <rye> kermiac, i will create 4500 files now and see how that works. It definitely is not good with 30k files but with 4k it needs a better look into
[11:16] <kermiac> ok, thanks rye :)
[11:19] <rye> + 3900 files so i now going to sync 5000 files. fire extinguisher is ready
[11:20] <kermiac> haha :)
[13:06] <thecondor> Hi! Just a little question I couldn't find an answer for: does anybody know, whether there is a broken download problem with WinXP IE, when trying to download big files published via UbuntuOne? A friend of mine, who is using internet explorer I guess, is just getting corrupt files. Same files downloaded here with ubuntu/firefox are fine.
[13:06] <thecondor> honk
[13:08] <beuno> thecondor, that's super odd
[13:08] <beuno> I wouldn't expect files to be currupt on IE
[13:08] <thecondor> Hi beuno. Yes it's oddset.
[13:08] <beuno> does he by any chanve have a non-sensitive file he can download and send us to inspect?
[13:09] <thecondor> This would be the next test. Sadly i cannot reach him atm.
[13:10] <thecondor> Just wanted to know whether other users might have experienced similar issues.. or recommending firefox e.g.
[13:12] <thecondor> I think he told me
[13:13] <thecondor> ..something like 'all files downloaded have same checksum and cannot be unzipped'
[13:13] <thecondor> (Should be different sized working zip files i uploaded. Strange.)
[13:14] <rye> thecondor, hmmmm
[13:15]  * rye fires his expired trial copy of win7
[13:16] <thecondor> I guess it could be helpful to get more info bout the files he downloaded.. but i cannot reach him.
[13:17] <thecondor> Shall i make a test upload with non sensitive data for you to download?
[13:19] <jdobrien> g-morning ubuntu oners
[13:21] <thecondor> G-morning jdobrien
[13:22] <thecondor> @rye: test upload for you in progress
[13:31] <rye> looks like my libvirt/apport is somehow broken now, brb
[13:48] <nisshh> CardinalFang: hey, id like to code a function into my app that can tell when desktopcouch is being synced by U1 and when its not, are you able to help me out?
[13:53] <CardinalFang> nisshh, I have planned some DBus signals that your app could listen for.  This isn't on the 10.10 roadmap, though, so I won't be working on it, but I can review and merge code you supply.
[13:56] <rye> beuno, hm, i picked up random link - http://ubuntuone.com/p/2Di and it does not open in IE but it opens fine in Firefox in Windows. Not really a corruption but something is really broken there
[13:57] <CardinalFang> nisshh, so, these signals I have planned are not *exactly* what you're looking for.  You need a state variable that the signals toggle, and at startup your app doesn't know the current state, but it's the cleanest thing I can think of, at present.
[13:58] <beuno> rye, that's intersting. Can you file the bug?
[13:58] <thecondor> Wrong http headers?
[13:59] <thecondor> Another test: http://ubuntuone.com/p/7ED/
[13:59] <thecondor> Should be 40.4 MB
[14:02] <thecondor> Thx, rye and beuno, i will mail my friend he shall try firefox for the moment.
[14:03] <rye> thecondor, hmmmm that file is downloadable via IE
[14:03] <thecondor> Oh
[14:03] <rye> thecondor, the question is now whether checksum matches
[14:03] <thecondor> Now i'm baffled.
[14:04] <thecondor> A sec..
[14:04] <rye> thecondor, looks like not - 7zip could not open that file
[14:06] <thecondor> Should be: f0f9e7685c5a00f12a01a38c46d6c486 for beryll.avi.zip
[14:07] <thecondor> Filesize 42,331,421 bytes
[14:07] <thecondor> Standard zirpel created from nautilus
[14:07] <thecondor> -zirpel +zip
[14:12] <thecondor> Rye, what did u get with firefox/ie, speaking of filesize/md5?
[14:12] <rye> thecondor, still trying to get that file out of that win instance
[14:12] <thecondor> Ok
[14:16] <rye> thecondor, fda01bd702b248dc06263e4e5e67d500
[14:16] <rye> thecondor, d/ling w/ ff now
[14:18] <thecondor> Wow, maybe zip files get re-compressed when uploading to U1... really dont hope so.
[14:18] <rye> thecondor, hm, ie's version is actually bigger - 42,342,089 bytes
[14:19] <thecondor> Rye, what a messe
[14:19] <thecondor> Additional data? Where did it can fromm?
[14:19] <thecondor> Carmen
[14:20] <rye> thecondor, the only possible explanation I have is that there is gzip compression
[14:20] <thecondor> Sry, typos
[14:20] <rye> Content-Encoding: deflate
[14:20] <rye> Content-Length: 42342089
[14:20] <rye> yes
[14:20] <rye> cool
[14:20] <thecondor> Indeed.. thatcher could be. So it might be deflate
[14:21] <rye> IE does not support deflating?
[14:21] <rye> beuno, ping, it looks like IE does not decompress the deflated stream
[14:22] <thecondor> I guess deflate was always trouble making with ie.. because of some weired apache settings on the other end or so.
[14:23] <thecondor> I remember i had to remove deflate from apache often, when i was a lamp admin and php dev.
[14:30] <thecondor> ..or its some kind of missing x-handler directive for gz or so.
[14:34] <nisshh> CardinalFang: sorry i was afk for a while, yes so dbus sounds good, but i suppose my app could find out the current state easily enough, id be happy to contribute code, although i havent done anything with dbus at all and im a pretty noobish programmer still, be aware that this is just an idea i had, i dont even know if ill implement it or not.
[14:35] <CardinalFang> nisshh, DBus isn't very tough, especially signals.  I'll be here if you have questions.
[14:35] <nisshh> CardinalFang: ok, thanks
[14:36] <rye> i nearly know why it could not display that web page then... because jpeg stream was damaged according to IE...
[14:37] <thecondor> IE sucks so badly..
[14:40] <thecondor> Rye, do you file a bug report for this deflate IE thingy or should i do so?
[14:40] <rye> thecondor, i am filing one right now
[14:41] <thecondor> Thx very much! I'm outta here then for now..
[14:45] <rye> hm
[14:49] <rye> if I pretend to be IE then i get 42331421 bytes content, so it is ok
[14:50] <rye> need tcpdumping
[14:51] <rye> beuno, it is really some kind of broken browser detection
[14:52] <rye> beuno, http://paste.ubuntu.com/450597/
[14:52] <beuno> rye, I'd be surprised if we're doing any kind of browser detection
[14:53] <rye> beuno, we do, but for attachment names, not for deflate/not deflate thing. This seems to be dealt with by the frontend servers
[14:55] <beuno> rye, hrmf.
[15:00] <rye> beuno, frankly speaking - i don't get it :-/
[15:16] <rye> Accept-Encoding: gzip, deflate
[15:17] <rye> hm, IE says that it wants deflate
[15:36] <gnomefreak> when using funambol with thunderbird-3 what would i use for location?
[15:36] <beuno> gnomefreak, http://syncml.one.ubuntu.com
[15:36] <gnomefreak> beuno: thanks
[15:36] <beuno> gnomefreak, have you created the mobile user/pass?
[15:37] <gnomefreak> beuno: no im not syncing with a mobile device just to ubuntuone site
[15:38] <rye> beuno, our IE support is definitely broken so this is definitely a bug report
[15:38] <gnomefreak> i was assuming the user/password would be the one i use to log into ubuntuone site
[15:39] <beuno> gnomefreak, no no, you need the mobile user/pass to use funambol
[15:39] <beuno> we can't get it to use openid, so we need to generate a special user/pass
[15:39] <gnomefreak> oh, how do i create one than
[15:39] <beuno> rye, thanks
[15:40] <gnomefreak> i dont see anyway to do that in Funambol
[15:40] <beuno> gnomefreak, in the web ui
[15:40] <beuno> go to contacts
[15:40] <beuno> you should have a "set up phone sync" option
[15:41] <beuno> I know, this needs to be a bit less confusing  :)
[15:41] <gnomefreak> beuno: i dont see it. i have "sync all"
[15:41] <beuno> gnomefreak, on the website
[15:41] <gnomefreak> in the menu there isnt anything like that either
[15:41] <beuno> one.ubuntu.com
[15:41] <gnomefreak> on
[15:44] <gnomefreak> beuno: this is also for the free accounts?
[15:45] <beuno> gnomefreak, well, funambol is for paid accounts, you get a 30 day trial
[15:45] <beuno> that said
[15:45] <beuno> we're going to waive that until Maverick
[15:46] <gnomefreak> oh
[15:46] <beuno> now that everything is out there and working, we're planning a juicier mobile package
[21:06] <mkarnicki> verterok: Just got up from a nap & doing UML project for tomorrow =_= .. Any news on that Android bug maybe ;_; ?
[21:12] <verterok> mkarnicki: hi
[21:12] <mkarnicki> verterok: hello
[21:13] <verterok> mkarnicki: nope, I think one option is to stop using netty (and probably NIO altogether), in this case of the android client...and use plain old socket + ssl
[21:13] <verterok> mkarnicki: that means rewrite part of the protocol client
[21:14] <mkarnicki> verterok: ...
[21:14] <verterok> but is better than no client at all
[21:14] <verterok> :)
[21:14] <mkarnicki> verterok: PipeLine factory will suffice?
[21:14] <mkarnicki> *pipelinefactory
[21:14] <mkarnicki> I mean, that one file where I changed SSL to TLS ?
[21:14] <mkarnicki> or there's more to change?
[21:14] <verterok> mkarnicki: no, a bit more :)
[21:15] <mkarnicki> verterok: not good..
[21:15] <verterok> mkarnicki: that's the pipeline for the RequestHandler, which is a subclass of netty SimpleChannelHandler (or something)
[21:15] <mkarnicki> verterok: oh. so it's still netty..
[21:15] <verterok> mkarnicki: the way the client connects to the server needs to be changed
[21:15] <mkarnicki> verterok: sorry, I'm still unconsious
[21:16] <verterok> mkarnicki: yes, we need to ditch the parent class of the RequestHandler, and implement all the socket handling, protobuf en/decoing, etc  using plain-ld sockets
[21:16] <mkarnicki> verterok: you can imagine I feel devastaded
[21:16] <verterok> *old
[21:16] <verterok> mkarnicki: isn't that bad
[21:17] <mkarnicki> verterok: I have used sockets before, but 1. not secure ones, 2. not on such level of detail as we probably need
[21:17] <verterok> mkarnicki: the client will need to be executed in a different thread, as the old java IO is blocking...
[21:17] <mkarnicki> verterok: that's not a problem, I'll make it a service
[21:17] <mkarnicki> verterok: with it's own thread
[21:18] <verterok> mkarnicki: no need to do that just yet ;)
[21:18] <verterok> mkarnicki: let's get it running in a thread, then move it to whatever a service is :)
[21:18] <mkarnicki> verterok: brb 1 sec
[21:19] <mkarnicki> verterok: I'm back. ok, I'll run it in another thread for now
[21:20] <mkarnicki> verterok: if I can help somehow, tell me what I can do (after I finish my stupid college assigmnent)
[21:21] <verterok> mkarnicki: simple SSL wiht sockets is quite "simple", it will look like this: http://pastebin.ubuntu.com/450760/
[21:22] <mkarnicki> verterok: aha..
[21:22] <mkarnicki> verterok: if you would just start that, like with handling one request, I could build up on that
[21:22] <verterok> mkarnicki: we need to translate some of the stuff in RequestPipelineFactory to blocking IO
[21:23] <verterok> mkarnicki: my idea is to keep the rest of the code, that's the mainloop and should handle all requests
[21:23] <mkarnicki> verterok: ok then. if you could leave me few pointers on priv, I'll start building on that. actually you already gave me a good start I think.
[21:23] <mkarnicki> verterok: I see, good idea
[21:24] <verterok> mkarnicki: the wire protocol is: header[4bytes]+message[nbytes]
[21:24] <verterok> mkarnicki: the 4 bytes header is the length of the message
[21:24] <verterok> mkarnicki: you should take a look to the python implementation, the file is request.py
[21:25] <verterok> mkarnicki: I'll try to play a bit tonight, but not sure if I'll be able to get enough time to get somethign working
[21:25] <mkarnicki> verterok: got yet another phone call, brb!
[21:25] <verterok> mkarnicki: np
[21:30] <mkarnicki> verterok: ok, gotcha! I'll read up on that python code. you're already tons of help, so maybe I'll be able to input more on that this time.
[21:30] <mkarnicki> verterok: thanks again. and now I have hope again that the project may go on
[21:31] <mkarnicki> verterok: I was really upset it could have had really bad impact. I mean it already had, but hell, we've done much job!
[21:31] <verterok> mkarnicki: I think I have a partial (no-working) implementation of the code somewhere, before I started with netty...I'll try to find it
[21:32] <mkarnicki> aquarius: ^ good news up here. we'll rewrite part of the protocol and not use netty, that will hopefully workaround any Android bugs.
[21:32] <mkarnicki> verterok: that would be lovely, if you'll have it around, hit me up with a paste and I'll read up on that, too. thanks!!!
[21:33] <verterok> mkarnicki: sure
[21:33] <verterok> mkarnicki: we first need to be sure that plain sockets+ssl works on android ;)
[21:39] <mkarnicki> verterok: heheh. there was a reply from netty author Tristan, that one guy should use plain (old?) OIO transport - do you know what he meant?
[21:39] <verterok> mkarnicki: a reply where?
[21:40] <mkarnicki> verterok: I'll dig it up
[21:40] <verterok> mkarnicki: yes, plain OIO is the blocking IO
[21:40] <verterok> mkarnicki: NIO is the new non-blocking IO
[21:41] <mkarnicki> verterok: right, so that e-mail can prove useful. it was on some mailing list
[21:41]  * mkarnicki digs it up
[21:42] <mkarnicki> verterok: 2nd mail - I think it's what you suggested, isn't it? http://www.jboss.org/netty/community.html#nabble-td5096427
[21:42]  * verterok reads
[21:44] <verterok> mkarnicki: no, different issue
[21:44] <mkarnicki> verterok: but we also got java.net.SocketException: Bad address family - still different issue?
[21:45] <verterok> mkarnicki: we got that?
[21:45] <mkarnicki> (by we got - I mean running that on Froyo anyway :/  )
[21:45] <mkarnicki> verterok: umm.. yeah, but.. did you get my late msg's yesterday?
[21:45] <mkarnicki> verterok: the bug has been fixed in Android 2.2 (sic!)
[21:45] <mkarnicki> verterok: but then there was that Bad address family anyway :/
[21:46] <verterok> mkarnicki: we can try to switch the current client to netty's OIO transport
[21:46] <mkarnicki> verterok: so no good either
[21:46] <mkarnicki> verterok: that would be less changes in the code?
[21:46] <verterok> mkarnicki: the issue was with ikvm or something, that compiles java to .net stuff
[21:46] <verterok> mkarnicki: it should!
[21:46] <mkarnicki> verterok: :)
[21:46] <verterok> mkarnicki: http://docs.jboss.org/netty/3.1/guide/html/architecture.html
[21:46] <verterok> mkarnicki: search for OIO
[21:46]  * mkarnicki reads up
[21:46] <verterok> mkarnicki: actually, read 2.2. Universal Asynchronous I/O API
[21:48] <mkarnicki> verterok: yup, right there
[21:48]  * mkarnicki reads
[21:48] <mkarnicki> verterok: you know what.. I have this dang assignment waiting for me for tomorrow. I'll do it now and go back to all that OIO stuff, if it's ok with you?
[21:49] <verterok> mkarnicki: sure
[21:49] <mkarnicki> verterok: ok thanks. I'll let you know, till then.