[08:22]  * rtagger advised his wife to use ubuntuone some weeks ago. Initial reaction - "don't need it, does not have KDE applet (now it does, but we haven't compiled it yet)", etc. Now all study info is stored in Ubuntu One, moreover most edits take place in Ubuntu One directory directly.
[08:37] <rtagger> Bug!
[08:38] <rtagger> In case oauth server replied with some server error, not auth_failed, or something like this, then oauth token gets removed from the keyring even if it is valid
[08:39] <rtagger> This is the reason behind Bug#452589 and the reason why this person - http://irclogs.ubuntu.com/2009/11/20/%23ubuntuone.html#t00:22 (slide) has missing auth token
[08:40] <rtagger> ubottu: Bug 452589
[08:40] <rtagger> am I alone at home?
[09:31] <rtagger> Because for elif message.type == protocol_pb2.Message.ERROR: self.error(request.StorageRequestError(self, message)) there is a check in ubuntuone/syncdaemon/action_queue.py, which checks message.type against protocol_pb2.Error.AUTHENTICATION_FAILED.
[09:33] <rtagger> And if we look at MessageType ERROR - descriptor.EnumValueDescriptor(name='ERROR', index=1, number=1, options=None, type=None), It is the same index as for descriptor.EnumValueDescriptor( name='AUTHENTICATION_FAILED', index=1, number=1, options=None, type=None).
[09:35] <rtagger> Later on, self.event_queue.push('SYS_OAUTH_ERROR', error=str(e)) is called, via syncdaemon/states.py SYS_OAUTH_ERROR is translated to AUTH_FAILED, and then...             elif state == "AUTH_FAILED": [...]  iface.clear_token(OAUTH_REALM, OAUTH_CONSUMER, reply_handler=token_cleared, error_handler=reauthorize_error)
[09:36] <rtagger> That's how we loose Auth Tokens when OAuth server is not cooperative
[10:01] <teknico> rtagger, interesting
[10:02] <rtagger> teknico: yep, the other issue is that for some reason it does not get re-authed. Will reproduce this in vm in an hour.
[10:03] <teknico> rtagger, great, thanks!
[10:10] <rodrigo_> hmm, that would explain why the u1 client tokens get lost, but what about the tomboy ones? some people seem to be losing them
[10:11] <rodrigo_> although, well, removing the dc ones from the keyring and trying again solves it for them, so might be relateds
[10:13] <rtagger> rodrigo_: erm, Do you mean that removing desktopcouch auth info allows tomboy to reconnect with web-based service?
[10:16] <rodrigo_> rtagger: for some people that were getting auth problems, yes
[10:18] <rtagger> rodrigo_: hm, yes, tomboy should receive tokens somehow. Okay, will check Tomboy as well
[10:25] <rodrigo_> rtagger: tomboy doesn't remove them, it just uses the same token he got when doing the initial oauth dance
[10:25] <rodrigo_> that's why it might be get lost on the server, not sure
[11:10] <rtagger> something is wrong with my assumption here. The client should use existing tokens at all times and clear them only if the server fails to authenticate with it... Hmm...
[11:49] <rtagger> Looks like this is pretty hard to reproduce w/o proper server support. I assume that there was some server-side problem that made clients forget their token after they were denied access by fs-1.one.ubuntu.com for some reason.
[12:19] <thomastp> why do ubuntuone-storage-protocal and ubuntuone-client both install in the ubuntuone python namespace ?
[12:20] <aquarius> cos they're both to do with Ubuntu One, no?
[12:39] <thomastp> aquarius: sure, but so they both try to install the same files
[12:39] <thomastp> aquarius: why isn't it simply included in the same package ?
[12:39] <thomastp> the practical reason is that currently I need to mark storage as noarch, but the ubuntuoneclient as arch-specific, so things break
[12:39] <aquarius> which files do they both install? that seems odd
[12:39] <thomastp> ubuntuone/__init__.py
[12:40] <thomastp> arguably that one will be empty, but still silly for packagers
[12:40] <aquarius> that's weird
[12:40] <aquarius> kenvandine, ping?
[12:40] <aquarius> kenvandine knows Much about packaging ;)
[12:40] <thomastp> I imagine that current packages hardcode exclusion of those __init__.py*
[13:35] <kenvandine> aquarius, pong
[13:36] <aquarius> kenvandine, heya, pal. thomastp asked why ubuntuone/__init__.py is in both  ubuntuone-storage-protocal and ubuntuone-client packages
[13:37] <kenvandine> shouldn't be
[13:37]  * kenvandine checks
[13:37] <thomastp> not in the packages
[13:37] <thomastp> in the install of each
[13:38] <thomastp> obviously they can't be in both packages, and so a packager I assume will currently have to exclude them by hand from the ubuntuone-client package
[13:38] <kenvandine> ah
[13:42] <kenvandine> so i suspect those files are in the source tarballs because the u1 developers like to run them from the checkout
[13:42] <kenvandine> so it needs to be importable
[13:42] <kenvandine> the packaging doesn't really work around it... it just doesn't include them in the .install files
[13:43] <kenvandine> so if are packaging for something other than ubuntu, i would suggest including it in the python-ubuntuone-storageprotocol package but not in the client package
[13:43] <kenvandine> just for consistency
[13:44] <kenvandine> thomastp, also feel free to ping me anytime if you have questions
[13:44] <kenvandine> i would love to see more packages for it :)
[13:45] <thomastp> kenvandine: what I know is that I install both with setup.py in my specs
[13:45] <thomastp> kenvandine: so the same file is installed by both packages
[13:45] <thomastp> kenvandine: sorry, by both install steps of the source tarballs
[13:45] <thomastp> I'm working around it for now by deleting them in the ubuntuone-client package
[13:45] <kenvandine> yeah, so exclude it from the client package
[13:46] <kenvandine> thomastp, was it your blog that i saw about packaging u1 for fedora?
[13:46] <thomastp> kenvandine: yep
[13:46] <kenvandine> awesome!
[13:47] <thomastp> kenvandine: I still recommend fixing the install though, it's still wrong
[13:47] <kenvandine> that would be up to aquarius, it would make it harder for them to run it from their source checkout
[13:47] <kenvandine> thomastp, but i agree it is a bit of a pain for packaging
[13:52] <thomastp> kenvandine: source checkout and install are two different ways of running no ?
[13:52] <thomastp> we do the same in flumotion
[13:52] <kenvandine> thomastp, yes
[13:53] <kenvandine> the developers are going to want to import ubuntuone from their checkout, but that can be worked around
[13:53] <kenvandine> i guess we can just exclude it from the tarball
[13:53]  * kenvandine will prepare a branch with that :)
[13:56] <kenvandine> good morning mattgriffin
[13:56] <mattgriffin> kenvandine: morning ken!
[14:00] <dobey> hmm
[14:00] <dobey> hi thomastp
[14:03] <dobey> thomastp: ubuntuone-client doesn't install ubuntuone/__init__.py
[14:03] <dobey> or it shouldn't anyway
[14:05] <dobey> kenvandine: we can't exclude it from the tarball either
[14:05] <kenvandine> why?
[14:05] <dobey> kenvandine: tests need it to be able to import things?
[14:05] <kenvandine> thomastp, i know it isn't installed in our package
[14:05] <dobey> "make test" should work from the tarball
[14:05] <kenvandine> dobey, but do tests run from the tarball?
[14:05] <kenvandine> true
[14:06] <kenvandine> it just means packagers need to work around it
[14:06] <dobey> or "make check" i guess is the right thing
[14:06] <kenvandine> which isn't terrible
[14:06] <dobey> well it shouldn't be installed to the system
[14:06] <kenvandine> right, so thomastp has to work around it and other packagers will too
[14:06] <kenvandine> which isn't that uncommon
[14:06] <kenvandine> actually
[14:06] <kenvandine> it should be in the tarball
[14:07] <kenvandine> but make install shouldn't install it :)
[14:07] <kenvandine> so we should just remove ubuntuone from pypackages?
[14:07] <kenvandine> or would that prevent the clientdefs.py from getting installed?
[14:08] <dobey> kenvandine: right we need clientdefs
[14:09] <kenvandine> yeah, that would prevent it from installing clientdefs
[14:10] <kenvandine> thomastp, basically it isn't trivial to fix that case and it is a pretty common issue packagers run into
[14:12] <dobey> it is trivial
[14:12] <dobey> i'll fix it today
[14:12] <kenvandine> ok
[14:12] <kenvandine> dobey, btw autotools is evil, mkay'
[14:14] <dobey> nah
[14:14] <dobey> i am evil
[14:14] <dobey> autotools is just a minion
[14:15] <kenvandine> hehe
[14:15]  * kenvandine agrees with that
[14:15] <kenvandine> dobey, btw at UDS aquarius asked me why our desktopcouch package can't be maintained with quickly commands...
[14:15] <kenvandine> i blamed you :)
[14:16] <kenvandine> although i do understand we want to build for < karmic
[14:16] <dobey> why can't it be?
[14:16] <dobey> i don't know what quickly wants or why :)
[14:17] <dobey> if quickly can't handle proper distutils/setuptools without the -extras stuff, i think that's a quickly bug
[14:37] <kenvandine> dobey, quickly was the driver for -extras, the point was make distutils smarter
[14:38] <dobey> well it was the driver for the auto() thing
[14:39] <dobey> using it by default doesn't mean it should depend on it though if you write a more explicit setup.py
[14:54]  * rtagger finally implemented simple local file status w/o weird hacks, perl-like code and strange assumptions: http://paste.ubuntu.com/326901/... And I love pylint :)
[15:02] <CardinalFang> Desktop+ MEETING BEGINS.  Say 'me' to claim a slice of the stand-up meeting, then take your turn by saying DONE/TODO/BLOCKED.
[15:02] <teknico> me
[15:02] <CardinalFang> me
[15:02] <aquarius> me
[15:03] <rodrigo_> me
[15:03] <jblount> me
[15:04] <dobey> me
[15:04] <urbanape> me
[15:05] <aquarius> vds, ?
[15:05] <vds> me
[15:06] <teknico> shall I? shall I?
[15:06] <urbanape> and where's our fearless leader?
[15:07] <jblount> teknico: Yes, please get started.
[15:07] <teknico> THEN WHO WAS PHONE
[15:07] <CardinalFang> I don't think it's mutinous to start without him.
[15:07] <teknico> ops
[15:07] <teknico> DONE: continued updating the Funambol code to v. 8.0 in our code (#403435)
[15:07] <teknico> TODO: land the branch to fix Funambol config and tests for v. 8.0 (#403435), finish updating the Funambol code to v. 8.0 in our code (#403435)
[15:07] <teknico> BLOCK: none
[15:07] <teknico> next: CardinalFang
[15:07] <CardinalFang> DONE: Triaged desktopcouch bugs.  Fixed a few minor bugs.  Worked a tiny bit on contacts views api.
[15:07] <CardinalFang> TODO: On the road, working.  Contacts views.
[15:07] <CardinalFang> BLOCKED: None
[15:07] <CardinalFang> aquarius, please...
[15:07] <aquarius> ⚀ DONE: start work on desktopcouch developer docs; talk to thomasvs about desktopcouch for Fedora
[15:07] <aquarius> ⚁ TODO: look at oauth-enabling twisted; make tomboy first-sync experience nicer; work on desktopcouch developer docs; write up things learned at UDS/sprint; work with rodrigo on Music Store
[15:07] <aquarius> ⚂ BLOCKED:
[15:07] <aquarius> go go rodri go
[15:08] <rodrigo_> • DONE: Auto* setup of libubuntuone. Looked at webkit-gtk and started music store widget that uses it. Talked with Thomas about couchdb on n900 and RPM packages
[15:08] <rodrigo_> • TODO: Conflict resolver tool in pair tool. Look at becoming a MOTU (https://wiki.ubuntu.com/UbuntuDevelopers). Make sandy's snowy test suite work with our server (http://git.gnome.org/cgit/snowy/tree/api/tests.py). Discuss with jdo and aquarius about oauth token per app, not per machine?
[15:08] <rodrigo_> • BLOCKED: no
[15:08] <rodrigo_> jblount: go
[15:08] <jblount> DONE: Not much, got distracted trying to diagnose some js weirdness and trying to fix some broken tests.
[15:09] <jblount> TODO: talk to the urbanape about his inline-edit branch, get my js minifying and css minifying proposed
[15:09] <jblount> BLOCKED: Nope
[15:09] <jblount> dobey: dude, yeah, man
[15:09] <dobey> ☺ DONE: Work estimation, Review of "no notifications" branch, Discussion, Triage
[15:09] <dobey> ☹ TODO: Finish work estimates, Backport fixes to stable-1-0, Prepare an SRU
[15:09] <dobey> urbanape: roll out
[15:09] <dobey> ☹ BLCK: None.
[15:09] <urbanape> DONE: Tried to get some help from jblount with new lazr-js composition branch, but it's currently harder to build than it needs to be.
[15:09] <urbanape> TODO: Get a pre-built version of lazr-js into our own project space and pull it from there via sourcedeps until we have a proper package built.
[15:09] <urbanape> BLOCK: None
[15:09] <urbanape> vds: you're up
[15:09] <vds> DONE: face duty, discussed status of mobile sync with chipaca and teknico, started two branches to port funambol v8 in sourcedeps with teknico
[15:09] <vds> TODO: land the two branches
[15:09] <vds> BLOCKED: nope
[15:09] <vds> that's it!
[15:09] <CardinalFang> Thanks all.
[15:09] <urbanape> (oh, and get my part of dobey's estimate done)
[15:12] <jblount> urbanape: Any idea where we stand in the bindwood web ui stuff? Is that a few weeks away?
[15:15] <dobey> blah, amazon won't let me add mp3s to the shopping cart
[15:19] <rodrigo_> dobey: wait for our music store to buy music :D
[15:20] <dobey> rodrigo_: i have a $25 certificate for amazon, but it only works for stuff i can stick in the cart :(
[15:22] <thomastp> dobey, kenvandine: well the problem I additionally had was that the ubuntuoneclient package has arch-specific parts
[15:22] <urbanape> jblount: yeah, I think we're gonna want the manifest stuff to land first, so we have a structure to work from.
[15:23] <jblount> urbanape: Right on, it just came to mind so I thought I'd poke you about it.
[15:25] <dobey> thomastp: that's an rpm problem i think. you just have to build arch-specific rpms
[15:26] <thomastp> dobey: yep, and so recently rpm was changed so that you could have your main package arch, but the rest noarch if you want
[15:27] <thomastp> the reverse doesn't work, so  I worked around it by creating a separate ubuntuone-client-gnome arch package
[15:27] <dobey> thomastp: well that's nice
[15:27] <thomastp> and everything else noarch
[15:27] <thomastp> it's a hack, but hey, it works for now
[15:27] <dobey> well the concept of a "main" package is sort of silly
[15:27] <dobey> there's source, and then there's the binary packages it creates
[15:28] <dobey> but alas, rpm can be silly often
[15:31] <dobey> yay firefox is back down to 282M after closing all the amazon tabs :-/
[15:35] <aquarius> I am seriously debating switching to chrome ;(
[15:35] <dobey> yeah
[15:35] <dobey> i need to try chromium some more
[15:37] <thomastp> dobey: so is there a ubuntuone-client 1.0.2 official tarball
[15:37] <thomastp> ?
[15:37] <thomastp> aquarius: do you have bindwood for chrome?
[15:38] <aquarius> no
[15:38] <dobey> thomastp: yes, it's on the +download page, but it's at the bottom i think, because it's from the stable-1-0 series, and not trunk
[15:40] <mandel> aquarius: hello, do you mind if I propose to merge the contact wrapper I have done to desktopcouch, I do not want people to depend on my app to be able to use it, and you already have the module name ;)
[15:41] <mandel> aquarius: Is basic, but it adds some nice stuff to easy the pain of using desktopcouch
[15:42] <thomastp> dobey: ok, found it.  Any reason for not actually promoting it to 'latest' ?
[15:44] <dobey> thomastp: i believe the lp ui is buggy and the "latest" is the "latest" from the trunk series
[15:44] <dobey> thomastp: and 1.0.1 from trunk was before i branched stable-1-0 and did 1.0.2
[15:44] <thomastp> oh, ok, ouch
[15:44] <dobey> yeah :-/
[15:45] <thomastp> ok, packages done, now to figure out all python issues I still have
[15:45] <dobey> i need to see if that's filed, and file if not
[15:45] <dobey> thomastp: could you file a bug for the __init__.py conflict?
[15:45] <thomastp> dobey: sure
[15:46] <dobey> thanks
[15:46] <rtagger> Is there any reason why bindwood does not store favicon image blob along with the bookmark itself?
[15:48] <dobey> facundobatista: is bug 476632 not fixed with r73 of storageprotocol?
[15:49] <dobey> rtagger: couchdb isn't really suited for storing binary blobs
[15:50] <dobey> don't know if there are other reasons too...
[15:51] <thomastp> in this case it'd be fine, it's just a small e.g. base64 blob
[15:52] <dobey> maybe
[15:52] <rtagger> dobey: this makes synced bookmarks really stand-out, like they are broken. We got so used to seeing the nice icons by now and this will definitely raise many questions later. I.e. is it a Google Search or help forum topic...
[15:52] <dobey> rtagger: well, you should also give your bookmarks meaningful titles :)
[15:53] <rtagger> dobey: yes, will tell this to the makers of page <title>s :)
[15:53]  * dobey thinks Mozilla Evangelists should be doing that...
[15:54] <aquarius> rtagger, interesting idea, though. Can you file a wishlist request about that? It would be a good idea
[15:54] <aquarius> mandel, heya! Yes, definitely propose it
[15:55] <mandel> aquarius: sweet, just did, but I have not added any type of copyright, I do not know which one to use :P
[15:56] <aquarius> mandel, propose for review with a note saying you're not sure which copyright to add, and we can work out which one is needed
[15:56] <rtagger> Like for "About Us" link that is now in my Desktop Couch folder. Who is "Us"?... The mozilla head nearby would definitely resolve this issue :)
[15:57] <mandel> aquarius: ok, I'll do so, right now I'm writing some code as an example in a comment
[15:58]  * dobey kicks firefox... just because i maximized one window, doesn't mean i want all new ones to be too!
[15:59] <facundobatista> dobey, no... r73 fixed an specific not-handled-correctly issue... but there are more
[15:59] <dobey> facundobatista: ok
[16:26] <dobey> thomastp: btw, did you see my last comment on the MANIFEST bug?
[16:30] <thomastp> https://bugs.launchpad.net/desktopcouch/+bug/487353 <- anyone seen either failure ? I still don't have my desktopcouch replicating to ubuntuone
[16:30] <joshuahoover> thomastp: i'm looking into this now
[16:31] <thomastp> joshuahoover: awesome, thanks
[16:32] <joshuahoover> thisfred: bug 487353 ...any thoughts?
[16:34] <thomastp> dobey: yeah, commenting now
[16:36] <thomastp> dobey: so I'm not sure.  can you give me a new tarball to test ?
[16:37] <joshuahoover> aquarius: ping
[16:37] <dobey> thomastp: sure
[16:37] <aquarius> pong
[16:38] <dobey> thomastp: http://www.gnome.org/~dobey/ubuntuone-storage-protocol-1.0.0.tar.gz
[16:39] <dobey> thomastp: although, interestingly enough, it looks like sdist didn't stick the MANIFEST in the tarball...
[16:39] <thomastp> dobey: well, the MANIFEST in there doesn't have pems :)
[16:39] <dobey> thomastp: what MANIFEST?
[16:39] <joshuahoover> aquarius: can you look at 487353 which thomastp filed?
[16:40] <dobey> thomastp: i guess this is another sdist bug :(
[16:40] <thomastp> dobey: oh, so the MANIFEST isn't actually in the tarball
[16:40] <thomastp> dobey: got it
[16:40] <thomastp> dobey: so sdist generates it in place or something ?
[16:40] <dobey> thomastp: it's in the checkout where i ran sdist, yes
[16:41] <dobey> it's generated from the MANIFEST.in and the standard stuff that gets included
[16:42] <thisfred> joshuahoover: will have a look
[16:42] <joshuahoover> thisfred: ok, i asked aquarius since i wasn't sure if you were around or not, thanks!
[16:43] <aquarius> wtf? sock is None?
[16:44]  * aquarius does the baffled look
[16:46] <dobey> thomastp: hrmm, we have 0.4.0 in ubuntu karmic
[16:46] <thisfred> aquarius: ok, sounds like you have it
[16:46] <thisfred> :P
[16:49] <aquarius> thomastp, I don't think this is httplib2
[16:49] <aquarius> thomastp, unless I'm confused. /usr/lib/python2.6/httplib.py is in the python2.6 package
[16:50] <aquarius> python-httplib2 is a separate package
[16:50] <dobey> aquarius: the trace looks like httplib2 is calling httplib, and couchdb is using httplib2
[16:51] <aquarius> dobey, agreed
[16:51] <aquarius> baffled as to why there wouldn't be a socket, though. I mean, if sock is none, how does anything work?
[16:51] <aquarius> we are using httplib2 0.4.0, though
[16:51] <dobey> i don't know. maybe angela lansbury can answer it
[16:52] <aquarius> the timeout error in httplib2 0.4.0 is a weird one, too, thomastp
[16:58] <aquarius> thomastp, can you find out what self.timeout *is* in that situation, given that it's not a float?
[17:02] <thomastp> aquarius: it looked like a None, which didn't make sense to me with all the if timeout is not None checks
[17:02] <thomastp> aquarius: I'd have to roll back, I can do that
[17:02] <aquarius> thomastp, yeah -- we're using httplib2 0.4.0
[17:02] <thomastp> ah, ok
[17:02] <thomastp> so yeah, I'll roll back and try
[18:49] <dobey> thomastp: so i'm not really sure what to do about the MANIFEST not existing in the tarball exactly
[20:35] <dobey> maybe i should just override bdist_foo in setup.py to just error out
[21:15] <dobey> apachelogger: around?
[21:15] <apachelogger> dobey: like 20%
[21:15] <dobey> apachelogger: you're doing the c++ kde ui right?
[21:16] <apachelogger> dobey: yes
[21:16] <dobey> apachelogger: great. we should chat a bit, but maybe would be better when you're more around :)
[21:18] <apachelogger> dobey: tomorrow sometime?
[21:18] <dobey> apachelogger: yeah. what part of the earth are you located at? :)
[21:19] <apachelogger> dobey: central europe
[21:20] <dobey> apachelogger: ok. i will try to ping you at a better time for you tomorrow :)
[21:20] <apachelogger> dobey: I should be around the better part of pm UTC
[21:20] <dobey> great
[21:20] <apachelogger> ok :)
[21:38] <chewit> Is Ubuntu One working on improvements to the web interface
[21:39] <chewit> I'm having to use the web interface on the windows computers at University, when I want to upload files to ubuntu one, I have to do it one at a time.
[21:39] <dobey> as in immediately, or in general?
[21:39] <chewit> well, when ever, just wondering if you are
[21:39] <chewit> cause i really would like multi uploading
[21:39] <chewit> like in dropbox
[21:40] <dobey> yes, the one-at-a-time from the web is a known problem that is being looked into
[21:40] <dobey> dropbox lets you upload multiple files via the web?
[21:40] <chewit> yeh, thats what i am using atm
[21:40] <dobey> or you mean, as in "they have a windows client"?
[21:40] <chewit> no, i just need a web client i can access in windows which will allow me to do multi uploading
[21:41] <chewit> i am using drop box cause of that feature, but if ubuntu one can, that will be brilliant
[21:41] <dobey> how does dropbox do it?
[21:42] <chewit> when you choose the files to upload, it allows you to select more than one
[21:42] <chewit> i could host a screenshot showing you, maybe easier for you to understand
[21:43] <dobey> thanks
[21:45] <chewit> http://edhewitt.co.uk/dropbox/
[21:46] <chewit> not sure how much you can get from it
[21:46] <dobey> ah, interesting
[21:47] <chewit> if you click on 'Basic Upload' (as seen in screenshot 1), that uploader works like ubuntu one
[21:47] <dobey> cool
[21:47] <chewit> any ideas when you are planning to improve the web interface?
[21:48] <dobey> the web ui is able to get ongoing improvements, as it doesn't have to abide by the ubuntu release schedule. so we can change the UI whenever. i work on some of the desktop client stuff, so i probably won't be doing anything to the web ui personally.
[21:49] <dobey> but i know there are definitely improvements in the works, though i'm not sure where multi-file upload is prioritized in all that work :)
[21:50] <chewit> ok, well I am glad its a known problem :D
[22:02] <dobey> i'll bug some people and see if it's in the works to be fixed soonish
[22:03] <dobey> will probably have to wait until tomorrow to bug them though, since it's pretty much end of day now :)
[22:03] <dobey> and it's time for me to head off a bit as well
[22:03] <dobey> later, and thanks for the feedback :)