[09:19] <paradroid> i copied a 10 MB file to the Ubuntu One directory in Nautilus about two hours ago and it still isn't showing up on the website - any ideas why?
[11:13] <paradroid> i just tried copying across an empty file called 'test' which i created directly in Ubuntu One via Nautilus and it's there already. still no sign of the 10 MB file (a tar)
[11:27]  * gnomefreak never did get my .tar.gz to show on the UI. i tried coping it to the ~/ubuntu-one and tried the upload function on the UI 
[11:28] <paradroid> right, i'll file a bug on launchpad then
[11:28] <gnomefreak> paradroid: thanks can you drop the link here
[11:28] <paradroid> will do
[11:28] <gnomefreak> thanks
[11:32] <paradroid> gnomefreak: hmm, they're showing up if i upload through the web UI. i'm going to do some more testing
[11:32] <gnomefreak> paradroid: i will try again but the 2 i tried never showed up
[11:33] <paradroid> i'm uploading the original 10 MB file i tried, this time through the web UI. will take a few mins on my connection
[11:39] <gnomefreak> im going to try it in a few
[11:42] <paradroid> hmmm. it's appeared now. i'll try going through the client again
[11:47] <paradroid> gnomefreak: it's working now. must have been some kind of server hiccup earlier, i don't know
[11:49] <gnomefreak> paradroid: thanks im off for a bit for coffee smoke and a few other things
[11:49] <paradroid> yw
[11:53] <paradroid> well, i don't know what went wrong earlier but this all seems to be working fine now. looks like i'm done here
[12:05] <gnomefreak> what is the max space for ubuntuone?
[12:13] <CardinalFang> gnomefreak, For free, I *think* it's 1GiB.
[12:13] <gnomefreak> CardinalFang: thanks
[12:16] <CardinalFang> gnomefreak, I'm wrong.  It's 2GB free, 10GB for theoretically paid users.
[12:16] <gnomefreak> CardinalFang: cool thanks
[12:17] <CardinalFang> gnomefreak, this may change.  I'd be surprised if it goes down, though.
[12:20] <CardinalFang> aquarius, heya.  I was thinking about desktopcouch.records.server.get_records() .  Shouldn't it not emit "deleted" documents?
[12:34] <gnomefreak> i lost my ubuntu-one dir. how do i get it back
[12:44] <aquarius> CardinalFang, erm? not sure I understand the question
[12:45] <CardinalFang> aquarius, We have a utility function that gives a cursor-like object for retreiving rows/documents.
[12:45] <aquarius> *nod*
[12:46] <CardinalFang> aquarius, We also have functions to "delete" documents.  This really adds an attribute to a document, not removes it.
[12:46] <CardinalFang> aquarius, Should the getter function pay attention to deletedness, or should we work that out in the app?
[12:46] <aquarius> CardinalFang, yep, which is a bodge to get around not having history yet :(
[12:47] <aquarius> I think the getter function should not return deleted documents, myself. They've been deleted. (That they haven't *actually* been deleted is an artifact of the implementation.)
[12:49] <CardinalFang> aquarius, agreed.  Okay, so how should we apply this change?  A new function?  The terrible thing is we store a document that contains javascript.  We don't want to delete and add every time.
[12:50] <CardinalFang> aquarius, I wonder if we can have additional values in the design document.  add  view["version"] = 2
[12:50] <CardinalFang> Check for that at execute time.
[12:50] <aquarius> you can add whichever extra values you want to a design doc, afaik
[13:02] <CardinalFang> aquarius, the python API tries to be too friendly.  :(
[13:03] <aquarius> can't be *too* friendly. :)
[13:04] <CardinalFang> aquarius, Well, it knows what design documents look like.  I'm wrong if I want something different.
[13:04] <aquarius> oh, I see what you mean
[13:06] <SteveA> save
[13:07] <thisfred> aquarius: I don't know if you saw, but ssh is working, but not all the way: push replication yes, pull replication no. The fact that one works, would lead one to believe it can't be that hard to fix the other.
[13:08] <thisfred> (and I know it's your day off, so will bother you as infrequently as I have to ;)
[13:09] <aquarius> thisfred, I did not see that...how's it doing it? ssh?
[13:09] <aquarius> how are we using ssh? confused
[13:12] <thisfred> aquarius: turns out, if you call https:.../_replicate, it works, iff you add one line to bin/couchdb (they had ssl support, just didn't switch it on)
[13:13] <thisfred> aquarius: but only when the target is remote, and the source is local. Adam is looking into that
[13:14] <CardinalFang> That's probably why it's not switched on.
[13:16]  * CardinalFang goes afk.
[13:17] <thisfred> CardinalFang: actually, I think I may have been the first person to actually test this. I couldn't find any references in the mailing list archives or on the wiki, surprisingly enough, but you may be right, someone may have added half working support and then given up. I think it just works, because of erlang functional magic, but they broke the magic somewhere.
[13:18] <thisfred> aquarius: I put something on the wiki, but I'll send you a mail with how to set up a test env, which is quite easy. (once I found out how to configure nginx correctly)
[13:30] <aquarius> blimey, it *has* ssl? coolio!
[13:37] <Chipaca> aquarius: who?
[13:37] <aquarius> Chipaca, couchdb has ssl. This is a bit of luck for us
[13:37] <Chipaca> ah :)
[13:39] <aquarius> Chipaca, but less interesting than you might have thought :)
[13:39] <Chipaca> maybe.
[14:12] <statik> moin
[14:22] <aquarius> thisfred, does desktopcouch work on jaunty?
[14:22] <aquarius> (do we have backports?)
[14:23] <thisfred> aquarius: nope
[14:24] <thisfred> aquarius: which is tunring out to be problematic for gwibber
[14:24] <aquarius> kk. So, if someone wants to try out DC, they have to be running karmic?
[14:24] <thisfred> aquarius: for now at least
[14:24] <aquarius> I'm putting together my developer week talk
[14:24] <thisfred> aquarius: backporting shouldn't be rocket science, I think we don't have karmic only deps
[14:25] <aquarius> couch
[14:25] <thisfred> but maybe we do
[14:25] <thisfred> yeah
[14:25] <thisfred> which pulls in some things itself
[14:25] <aquarius> so..."preparation" for participating in the talk is...run karmic. heh.
[14:25] <aquarius> or just watchthe talk and then wait for karmic. kk.
[14:25] <thisfred> k, cool, when/where is your talk?
[14:26] <thisfred> ubuntu-classrooms?
[14:26] <aquarius> yep, and on wednesday
[14:32] <statik> aquarius, pushing developers to run karmic in order to hack on desktopcouch is probably a good thing, we need help testing and finding any integration problems with our packages on karmic
[14:32] <statik> i should have the new snapshot done in a few minutes
[14:32] <aquarius> statik, I agree, I just wanted to check that someone wasn't surreptitiously backporting everything :)
[14:33] <statik> i have no objection to backports but have no intention of doing any atm
[14:34] <statik> CardinalFang, since i'm preparing a new couchdb upload i'm going to go ahead and fix the 'should not delete couchdb user during package purge' bug
[14:37] <urbanape_> Morning, all.
[14:39] <statik> hi urbanape! think we will have bindwood support for oauth today?
[14:39] <statik> hey rodrigo_, just trying to figure out timing for couchdb oauth - when do you think you will have a package of couchdb-glib that does oauth?
[14:40] <rodrigo_> statik: I think I just need a couple of days or so
[14:40] <urbanape> statik, I'll try to get that done
[14:40] <rodrigo_> statik: so I could submit a package with the polling changes (2 branches submitted already) and the oauth stuff by the end of the week
[14:41] <rodrigo_> statik: but isn't it going to be disabled by default in dc?
[14:41] <aquarius> rodrigo_, as part of my desktopcouch talk for developer week I'm going to have a section on how to use couchdb-glib. How would you like to, er, write that section for me? :-)
[14:42] <statik> rodrigo_, that sounds good, please let me know if it won't be uploaded this week. it won't hurt anything to have couchdb-glib oauth-signing requests and couchdb ignoring them, right? I'm trying to clear the path for enabling mandatory auth in couchdb
[14:42] <rodrigo_> aquarius: yes, I can do it, ping me tomorrow and I'll prepare some quick howto
[14:42] <rodrigo_> statik: ah, ok
[14:43] <rodrigo_> aquarius: your talk is on Wed, right?
[14:43] <statik> urbanape, thanks!
[14:46] <statik> thisfred, based on my reading of the backlog I think there should be a bug filed against the couchdb package in ubuntu and linked to a couchdb jira bugtracker issue about enabling SSL. does that sound right?
[14:48] <CardinalFang> statik, Roger.
[14:49] <aquarius> rodrigo_, yep
[15:00] <jblount> MEETING BEGINS
[15:00] <jblount> "me" then "DONE / TODO / BLOCKED"
[15:01] <teknico> me
[15:01] <CardinalFang> me
[15:01] <jblount> me
[15:01] <dobey> me
[15:01] <urbanape> me
[15:02] <statik> me
[15:02] <teknico> ok
[15:02] <teknico> DONE: proposed a branch with the skeleton for the contacts CRUD web ui, and a branch to improve the scripts in utilities/, helped jdo with reviews
[15:02] <teknico> TODO: flesh out the details view for the contacts CRUD web ui
[15:02] <teknico> BLOCKED: none
[15:02] <teknico> next: CardinalFang
[15:02] <CardinalFang> DONE: code for 419969; need reviews.
[15:02] <CardinalFang> TODO: finish 419973 and 419975.
[15:02] <CardinalFang> BLOCKED: None
[15:02] <CardinalFang> jblount!
[15:02] <jblount> DONE: Review day, setup new (yes another one) laptop over the weekend.
[15:02] <jblount> TODO: Figure out why lp:~jblount/ubunet/overlays-skin is failing through PQM, work on human readable names for computer
[15:02] <rodrigo_> me
[15:02] <jblount> BLOCKED: Nope
[15:02] <jblount> dobey: rocknroll
[15:02] <dobey> ☭ DONE: Reviews, Uploaded poauth trunk
[15:02] <dobey> ☭ TODO: Release/package poauth, client bug fixes
[15:02] <dobey> ☭ BLCK: None.
[15:03] <dobey> urbanape: your roll
[15:03] <urbanape> DONE: Pushed a branch establishing a dependency on our modified zipstream python package.
[15:03] <urbanape> TODO: Bindwood OAuth integration
[15:03] <urbanape> BLOCK: None
[15:03] <urbanape> statik: your turn
[15:03] <statik> DONE: Quick fix to ubuntuone-client to have apport stop attaching syncdaemon log files by default (in response to user complaints about privacy due to filenames).
[15:03] <statik> TODO: New snapshot package of  couchdb 0.10.x with a handful of bugfixes including cascading auth. Talk with james_w to learn more about bzr-builddeb and unify sourcepackagebranches for bindwood and desktopcouch to match ubuntuone-client.
[15:03] <statik> BLCK: not-blocked-just-yet but anxiously awaiting versions of desktopcouch, bindwood and couchdb-glib which use oauth by default.
[15:03] <teknico> rodrigo_, up to you
[15:04] <rodrigo_> • DONE: More tomboy certs work. Submitted changes polling branches.
[15:04] <rodrigo_> • TODO: Start upstream discussion for adding social services accounts config to about-me. Talk to Ara about writing mago tests for evo-couchdb. Propose couchdb-glib/evo-couchdb for GNOME 2.29. Store UUIDs for postal addresses. Conflict resolver tool in pair tool. oAuth authentication and signing of all couchdb-glib requests. Look at becoming a MOTU (https://wiki.ubuntu.com/UbuntuDevelopers). Tomboy syncing fixes
[15:04] <rodrigo_> • BLOCKED: none
[15:04] <rodrigo_> and nobody else, right?
[15:04] <jblount> rodrigo_: right :)
[15:04] <jblount> MEETING ENDS
[15:04] <statik> did everyone see the announcement that macaco contacts is going to align with desktopcouch?
[15:04] <CardinalFang> Rawk!
[15:05] <jblount> One small step for desktopcouch...
[15:05] <statik> http://www.themacaque.com/?p=248 was announced yesterday as "being part of a bigger idea"
[15:06] <statik> aquarius, i know you are on bank holiday today but this is a very cheer-inducing blog post ^
[15:06]  * aquarius reads
[15:07] <rodrigo_> statik: cool
[15:10] <rodrigo_> statik: since you already have an account there, could you add a comment about the 'title', 'comments', 'webs' and 'ims' fields he's storing under app_annotations/macaco_contacts? we have those fields now in the top level, so he doesn't need to store them there.
[15:10] <rodrigo_> statik: also, instead of preferred_address having all the duplicated data, he should just use the uuid of the address he wants
[15:11] <statik> rodrigo_, not that i'm lazy, but i think it would be very encouraging to the author if you commented on his blog directly, even though it's a tiny hassle to create an account and wait for your comment to be moderated
[15:11] <rodrigo_> statik: if he fills all fields in evolution, he should be able to see how we store all of them (except IMs, which is still missing)
[15:11] <rodrigo_> statik: ok :)
[15:30] <statik> aquarius, thisfred, CardinalFang: pitti just uploaded the new couchdb snapshot to karmic for us
[15:30] <statik> it should be on your desk in a couple of hours
[15:31] <CardinalFang> Sweet.
[15:34] <thisfred> statik: woohoo!
[15:37] <jan____> w00t
[15:41] <kenvandine> statik, does that new couchdb snapshot fix the oauth problem?
[15:59] <urbanape> dobey, got a sec?
[16:01] <jan____> kenvandine: which one?
[16:01] <dobey> urbanape: what's up?
[16:04] <dobey> brb, gonna reboot
[16:07] <urbanape> d'oh. I'll wait til you get back
[16:09] <dobey> back
[16:09] <dobey> and grrrrrr
[16:09] <dobey> nautilus moved all my desktop icons *AGAIN*
[16:10] <urbanape> boo and/or hiss
[16:11] <urbanape> So, I was wondering if you could briefly walk me through how we conventionally maintain package branches. If I had to guess, I'd say it's a branch kept parallel but in sync with trunk that also has the packaging stuff.
[16:11] <urbanape> we never merge it to trunk, but we keep it up to date WRT trunk?
[16:12] <dobey> no
[16:12] <dobey> source package branch contains the contents of the release tarball that is being packaged, and the debian dir
[16:15] <urbanape> ah, okay, so, still. All the packaging info belongs there and outside the actual source repo. Does it typically live as its own project, then?
[16:15] <dobey> no
[16:16] <dobey> lp:~ubuntuone-control-tower/ubuntu/karmic/ubuntuone-client/karmic is where the current karmic packaging is, for example
[16:17] <dobey> for ubuntuone-client
[16:18] <dobey> ubuntuone-storage-protocol is similarly named (but replace the obvious project name)
[16:18] <urbanape> how is that not a separate project from ubuntuone-client?
[16:18] <dobey> it's not a project
[16:18] <dobey> it's a branch of the "ubuntu" distribution, "karmic" series
[16:19] <dobey> or something
[16:19] <dobey> but project is the wrong term for it
[16:19] <urbanape> okay, point taken. But it's not just a branch of the project that it packages.
[16:19] <dobey> no
[16:23] <urbanape> so, do I need to join the control tower team to push this package branch?
[16:24] <urbanape> Presumably, we'd make ~ubuntuone-control-tower/ubuntu/karmic/bindwood/karmic with the packaging info.
[16:24] <dobey> urbanape: you can just push it to your own for now
[16:24] <urbanape> k
[16:25] <dobey> urbanape: hrmm
[16:26] <dobey> urbanape: https://code.edge.launchpad.net/~ubuntuone-control-tower/bindwood/ubuntu
[16:26] <dobey> urbanape: i guess we should move that to the correct spot, and then people can propose merges into it
[16:26] <dobey> urbanape: i don't know whwat's up with that specific branch though. i'm guessing statik put it there
[16:29] <statik> the package branches are confusing right now, i want to organize them this week
[16:30] <urbanape> k, I'll ditch the packaging stuff from my debugging branch and re-propsoe.
[16:30] <urbanape> I just want to get that out of the way, as it's hanging out in limbo.
[16:30] <statik> urbanape, for the bindwood oauth work you can just propose normal branches to bindwood/trunk, and we'll tackle the packaging in a second step
[16:33] <statik> cool
[16:34] <urbanape> yeah, that's all I meant. Want to get stuff moving.
[16:36]  * dobey updates his review
[16:37] <CardinalFang> review desktopcouch fixes, anyone?  https://code.edge.launchpad.net/~cmiller/desktopcouch/pairing-fixups/+merge/10926
[17:00] <dobey> lunch times
[17:04] <aquarius> CardinalFang, not a proper review, but...what's up with the configparser changes in local_files?
[17:04] <aquarius> did they change the name of it in 3.0?
[17:05] <thisfred> aquarius: CardinalFang while I (may) have your ears: do we create databases if they don't exist on replication (hint: I think this would be a good idea, and would solve #399645)
[17:05] <aquarius> yes
[17:05] <aquarius> we have to
[17:06] <aquarius> otherwise the first time you pair it won't work
[17:08] <urbanape> aquarius, thought you were vacationing today
[17:08] <CardinalFang> aquarius, Yes, name changes in py 30.  Yes, databases created at replication time.
[17:08] <aquarius> urbanape, I am
[17:08] <aquarius> urbanape, just can't tear myself away :)
[17:09] <urbanape> sad, sad, sad
[17:09] <aquarius> well, I'm hacking, and so was on irc anyway
[17:09] <aquarius> and deliberately *not* coming in here just because it's a bank holiday is ultra-petty :)
[17:10] <CardinalFang> aquarius, as replication daemon is only peer-to-peer so far, Bug#399645 is not fixed.
[17:10] <urbanape> so, I looked through older emails, but didn't find anything. What's the easiest way to turn on oauth for our desktop couch, so I can make Bindwood fail miserably?
[17:11] <aquarius> [couch_httpd_auth]
[17:11] <aquarius> require_valid_user = true
[17:11] <aquarius> in the ini file
[17:13] <urbanape> rock
[17:15] <urbanape> cool, denial, sadness and despair
[17:35] <urbanape> aquarius, so long as you're just hanging out. got a sec for a quick skype call?
[17:35] <urbanape> (if not, no worries)
[17:35] <aquarius> urbanape, yeah, why not. I'm just about to bail, but I always have time for you. Heh. :)
[17:36] <dobey> wink wink, nudge nudge
[17:36] <aquarius> urbanape, go for it
[17:38] <CardinalFang> thisfred, what goes into replicating d-c dbs to U1?  Anything I need to know about?
[17:40] <thisfred> CardinalFang: if all goes well, the process from the client POV should be identical to LAN replication
[17:41] <thisfred> we need to set up a proxy to unwrap the SSL request, and forward it to (the correct[*]) couchdb
[17:41] <thisfred> [*] #422083
[17:47] <thisfred> CardinalFang: the only difference to the client would be that it replicates to or from an https:// url
[17:48] <CardinalFang> thisfred, Okay.  I'm coding it up now.
[17:48] <CardinalFang> Thx.
[17:48] <thisfred> wowsome!
[17:55] <dobey> oi
[17:56] <dobey> oh man
[17:57]  * CardinalFang calms dobey.
[18:01] <dobey> CardinalFang: try to create a new ssh key in seahorse...
[18:03] <CardinalFang> dobey: Ah, broken I guess.  "ssh-keygen" for you.  :(
[18:04]  * CardinalFang writes his first  trueval if condition else falseval  expression in Python.
[18:05] <CardinalFang> Er, we assume Python 2.6, right?
[18:05] <CardinalFang> Oh, or 2.5
[18:09]  * dobey files an upstream bug for seahorse
[18:12] <urbanape> man, that still seems so weird.
[18:13] <urbanape> I really would have preferred the more ubiquitous ternary operator.
[18:19] <dobey> hrmm, ssh-keygen is confusing
[18:26] <kenvandine> thisfred, CardinalFang: i found a problem
[18:26] <thisfred> kenvandine: ah
[18:26] <kenvandine> getting the port from the log (in dc)
[18:26] <CardinalFang> kenvandine, No!  Go away!
[18:26] <kenvandine> not working well when couch isn't working
[18:26] <kenvandine> it gets the old port
[18:26] <kenvandine> :)
[18:27] <thisfred> kenvandine: ouch, that is my fault
[18:27] <kenvandine> so getPort succeeds and returns a port from ages ago
[18:27] <CardinalFang> :(
[18:27] <CardinalFang> Probably my fault too.
[18:27]  * kenvandine files a bug
[18:28] <CardinalFang> kenvandine, see if there's already a bug.
[18:28] <thisfred> kenvandine: though I don't really know how to solve this, other than deleting the logfile on shutdown, which seems a hysterically bad idea
[18:28] <CardinalFang> I thought I saw something like that in email.  I might have dreamed it.
[18:29] <kenvandine> 420911
[18:29] <CardinalFang> thisfred, If we have to revert, I think my "netstat" was prettier than "lsof".
[18:29] <kenvandine> not the same
[18:30] <kenvandine> all the lsof stuff was removed before 0.1 was relleased :)
[18:32] <dobey> so
[18:33] <kenvandine> i think what we had before was decent :)
[18:33] <kenvandine> bug 422127
[18:39] <kenvandine> statik, we set 420911 and 422127 to alpha5 blockers
[18:42] <kenvandine> statik, welcome to being on the install CD :)
[18:45] <CardinalFang> thisfred, so, for u1--dc replication, I give the username somehow, yes?
[18:46] <urbanape> thisfred, do you know how I can get aquarius' branch to establish the oauth bits in my keyring?
[18:47] <thisfred> CardinalFang: no, we use the oauth credentials to get the username from the service jdo built.
[18:47] <thisfred> urbanape: the oauth bits should be put in the keyring by the filesync, I think
[18:48] <CardinalFang> thisfred, right, but that username is in the u1-end database name somehow, yes?
[18:48] <urbanape> and if I just manually start up desktopcouch-service from his branch, do I need to prompt that initial filesync?
[18:50] <thisfred> urbanape: I am unsure, and may be spouting manure: CardinalFang: you generate oauth tokens for the pairing, or is each machine using the U1 credentials even when talking over the LAN?
[18:51] <thisfred> CardinalFang: it's in a database on the server yes, and the couchdb's url is computed from it.
[18:58] <CardinalFang> thisfred, aquarius has it so there's one set of oauth tokens per machine, and we use these to communicate with everything.
[19:01] <thisfred> CardinalFang: right, and these are generated right? urbanape: sorry I had it all backwards: the credentials for replicating with ubuntuone are the same as for filesync, but each other machine generates its own and pairing shares them with any other machine that wants to replicate with it.
[19:01] <CardinalFang> thisfred, Yes, generated locally.
[19:03] <urbanape> using the same code that dc.records uses, though, I get no matches in my keyring, using his new oauth branch. wonder if I need to prime the pump, so to speak, with a sync.
[19:08] <thisfred> urbanape: not sure, I think the pairing would do it, not the replication action, but Monseigneur Fang may have more valuable insights there.
[19:09] <statik> bug 420911
[19:09] <statik> bug #422127
[19:10] <urbanape> so, what I'm trying to do is integrate Bindwood into the new OAuth way of things. What I need to do is ensure my system is doing all its work with OAuth, and I need to be able to fetch the secret bits from the keyring. I'm tentatively using aquarius' desktopcouch branch as the backend for this.
[19:18] <thisfred> statik: JIRA issue 491 (ssl support) filed: should I file a corresponding bug in LP, linking to that, and if so, against desktopcouch, or couchdb proper.
[19:37] <statik> thisfred, thanks! file against either desktopcouch or ubuntu (couchdb sourcepackage), and then you can do 'also affects' the other project so the same bug shows up in both places
[19:37] <thisfred> statik: perfect, will do
[19:41] <urbanape> statik, I have the plan of attack, and know pretty much what to do for oauth-integration for bindwood, but I'm not sure it'll get finished today.
[19:41] <urbanape> likely tomorrow
[19:42] <statik> urbanape, ok, thanks for working on that and letting me know
[19:45] <urbanape> it's really straightforward, but involves a black box in the middle I haven't yet cracked.
[20:07] <CardinalFang> thisfred, do you mind terribly if I revert the get_port change from last week?
[20:07] <thisfred> CardinalFang: no, I see I did not think it through as much as I thought I had, and no good way to solve it.
[20:08] <CardinalFang> Me either.  I reviewed it.  :(
[20:08] <thisfred> I wish it would just give back the port some other way than a log file.
[20:09] <thisfred> Also: I still don't see why we don't randomly choose a port, and pass it.
[20:09] <CardinalFang> Yeah.  There's a couchdb flag that shuts down the running daemon, synchronously.  It must communicate.  It should also have a stat-dump option.
[20:09] <thisfred> (I think aquarius told me, and I just forgot, mind)
[20:10] <CardinalFang> thisfred, 1) race condition.  2) it's just about as hard to search for an unused port as it is to search for a running program.
[20:11] <thisfred> true, but does port=0 solve these?
[20:11]  * CardinalFang contiplates SO_RESUSE and a while loop checking result of popen.
[20:12] <CardinalFang> thisfred, zero is a magic value that tells the kernel it can choose anything it wants.  It solves 1 perfectly, and it pushes 2 into kernel space where it's cheap to search for ports.
[20:14] <thisfred> CardinalFang: so there can only ever be one couch at a time that is started with 0? I thought the evidence suggested otherwise, and most of our trouble stemmed from preventing multiple couches being started.
[20:17] <CardinalFang> thisfred, it doesn't stop multiple couchdbs.  It doesn't stop anything, unless all ports are used.  That's the point of "0".  There must be another way to do that.  Atomic operations are usually forced using files and O_EXCL|O_CREAT open() flags.
[20:17] <CardinalFang> It could be we need a wrapper that does this, if couchdb doesn't do it.
[20:18] <thisfred> CardinalFang: ah sorry, I'm thick, I get it: the race condition being accidentally starting two on the same port...
[20:19] <CardinalFang> Or, another program jumps in and steals the port.
[20:24] <CardinalFang> thisfred, Oh, and two different processes can't bind to the same port.  One can bind and and then fork into more than one process, though.
[20:26] <thisfred> right, I geddit slowly
[21:06] <aquarius> CardinalFang, thisfred: ping
[21:06] <CardinalFang> aquarius, j0!
[21:06] <thisfred> aquapong
[21:06] <aquarius> just looking at DC
[21:07] <aquarius> desktopcouch-service calls desktopcouch.get_port()
[21:07] <aquarius> but get_port() doesn't call get_pid(), it just looks in the log file
[21:07] <aquarius> so it doesn't bother to check if DC is actually running or now
[21:07] <aquarius> not
[21:07] <aquarius> which means that the d-bus api doesn't work.
[21:07] <aquarius> um.
[21:07] <aquarius> this is less than ideal :)
[21:07] <CardinalFang> aquarius, We're on it.
[21:09] <thisfred> aquarius: we've discovered at least two problems there, and are reverting the whole get port from logfile thing. Well, CardinalFang is.
[21:10] <thisfred> aquarius: by trying to make it less crap, I made it more crap. Go figure.
[21:10] <aquarius> heh
[21:10] <aquarius> what's wrong with getting the port from the logfile?
[21:10] <CardinalFang> Ah, but it was very fast crap.
[21:10] <thisfred> if couch fails to start, it's still found
[21:11] <thisfred> unless the logfile isn't there, and then it craps out too
[21:12] <aquarius> yeah, it needs to block until couch starts, and then read the logfile.
[21:12] <thisfred> aquarius: that's what I *thought* I had done
[21:12] <aquarius> we needed to do that anyway -- whichever approach we take has to wait until couch starts up and then ask it for a port, whether we read the port from the logfile or from /proc walking or from lsof or whatever
[21:12] <aquarius> you have not done it ;-)
[21:13] <aquarius> am reassured to discover that I trip over this problem and you guys are already fixing it, though. I am behind the times :)
[21:14] <aquarius> the startup procedure isn't writing out the bookmark file either afaict?
[21:17] <CardinalFang> Yes.
[21:17] <CardinalFang> Yes, no bookmark file.
[21:18] <dobey> aquarius: well you do have an Android phone. they are so last year :)
[21:19] <thisfred> hey!
[21:20]  * thisfred just bought a G1, about a day before the mytouch came out, and stands by that decision
[21:20] <dobey> wtf is a mytouch?
[21:20] <thisfred> it was cheap, and it works, and I hate touch screens
[21:21] <dobey> uh, the g1 is a touch screen...
[21:21] <thisfred> dobey: apparently the new T-mobile android without a keyboard, which the g1 has
[21:21] <thisfred> dobey: I know, but it has a "real" keyboard, which my fat and clumsy fingers prefer
[21:22] <dobey> so does my Pre :)
[21:22] <thisfred> I know, but my android was cheap, and the data plan was cheap. I'm Dutch, I care more about price than features ;)
[21:24] <dobey> what's your total monthly bill then?
[21:26] <aquarius> dobey, heh :)
[21:27] <aquarius> dobey, screw Palm anyway. They think it's a good idea to refuse to comment on when their phone will be available. So, they lose a customer for being unfriendly.
[21:27] <dobey> heh
[21:28] <dobey> thisfred: or did you get the crappy voice plan?
[21:28] <statik> re
[21:29] <dobey> hi statik
[21:29] <statik> dobey, so for the ubuntuone-client sourcepackagebranches, are you using a patchsystem at all? or do you generate a new .orig.tar.gz for every code-related fix?
[21:32] <thisfred> dobey: yeah, the cheapest voice plan, no text msges + unlimited data (or I *think* that's what they mean by unlimited web browsing and email. I'm not streaming anything before I see my first bill ;)
[21:33] <dobey> statik: no, i've only been doing tarball releases
[21:33] <dobey> statik: maintaining patches to our own code, in packages, is kind of silly
[21:36] <statik> dobey: yeah, i agree. but for the patch i made on saturday night, i wasn't sure how to accomplish it with a sourcepackagebranch. I guess the sourcepackagebranch way would be to get my branch landed on trunk of ubuntuone-client, merge it to a branch of the sourcepackagebranch for ubuntuone-client, generate a new upstream release tarball, and request a sponsor for the updated pacakge with the new orig.tar.gz that had the 2-line patch?
[21:37] <statik> sourcepackagebranches were supposed to make life easier than doing it the bad old way with patchsystems and debdiff, i wonder if i'm missing something
[21:37] <dobey> well there are more changes than yours in trunk
[21:37] <statik> sure
[21:37] <statik> but if i want to get a focused change uploaded to ubuntu to fix a security or privacy issue quickly
[21:37] <dobey> i don't think source package branches make it easier than using patch systems
[21:39] <dobey> statik: the quick way is to do the patch in the source package branch as if somoene else is upstream, and we don't control the code
[21:39] <dobey> statik: of course, this will break building trunk packages in the PPA, since the code is already fixed in trunk (and the patch will then fail to apply)
[21:39] <aquarius> erm, where's the couchgrid stuff? I thought it was in quickly?
[21:40] <statik> aquarius, it's in desktopcouch
[21:40] <aquarius> oh.
[21:40] <aquarius> heh.
[21:40] <aquarius> :-)
[21:40] <dobey> statik: but generally the packages should be fixed first in that case, then we land the fix and do a new tarball release, and update the packaging for the new tarball (removing the patch stuff from it)
[21:41] <statik> dobey: this is all making sense now
[21:42] <dobey> thisfred: so you're paying like $70/mo just to get unlimited data? doesn't seem so cheap to me :)
[21:42] <dobey> but GSM in the US isn't exactly cheap
[21:43] <dobey> CDMA is definitely cheaper
[21:43] <thisfred> dobey: something like that but for 2 people
[21:43] <thisfred> (but only one dataplan)
[21:44] <thisfred> I think, may be less even
[21:44] <dobey> well the cheapest voice plan i see is 29.99, and data is 39.99
[21:44] <dobey> so that's ~70, plus taxes
[21:45] <dobey> so around $80 total i guess, plus the second voice plan
[21:47] <dobey> and it sucks that i can't push source package branches for projects not already in ubuntu :-/
[21:49] <aquarius> dobey, how much is yours per month then?
[21:49] <CardinalFang> Wait, wtf.  localhost:67855/_utils  -- that's not even legal.
[21:52] <dobey> aquarius: $100/mo (+ taxes), and that's unlimited everything (data, voice, text)
[21:52] <aquarius> a hundred bucks a month/
[21:52] <aquarius> ?
[21:52] <dobey> yes
[21:52] <aquarius> stone me.
[21:52] <dobey> that's like 5 GBP or something
[21:53] <aquarius> I pay £20/month. I don't get unlimited voice for that, mind. I do get unlimited data and text.
[21:53] <aquarius> $100 == £61
[21:53] <dobey> aquarius: t-mobile is $100/mo just for unlimited voice (no text or data)
[21:53] <aquarius> I don't make anything within *range* of £40 of calls every month, let alone £40 over the 100 free minutes I get.
[21:53] <dobey> which is why most people don't have unlimited voice on t-mo
[21:54] <aquarius> Man, you're being ripped off somethnig vicious. Move to England. :)
[21:54] <dobey> not really
[21:55] <dobey> though i don't have GSM
[21:55] <dobey> i could get a cheaper plan that doesn't have unlimited voice, but sometimes i do go over the minutes limit for that plan
[21:56] <dobey> *shrug*
[21:58] <dobey> man, the syncdaemon state machine is way too complex :(
[22:06] <rmcbride> FWIW, statik is upgrading his IRC Proxy server. THose that connect through it are temporarilly offline
[22:08] <dobey> aquarius: how much is unlimited voice on o2?
[22:08] <aquarius> dobey, don't know
[22:09] <rmcbride> of course now that I have an irrigation system for the containers, the daily rains start up again and I won't need it for months
[22:10] <dobey> aquarius: you are on o2 right?
[22:10] <aquarius> dobey, t-mobile
[22:10] <dobey> ah
[22:10] <dobey> o2 looks expensive :)
[22:11] <aquarius> iirc, unlimited voice/data/text was about £35, so a bit over $50/month
[22:20] <dobey> hrmm, not actually unlimited it seems, but probably sufficient enough that most people won't hit the limits
[22:32] <dobey> gah, mocker. :(
[23:36] <dobey> later all!
[23:36] <statik> see ya
[23:44] <aquarius> I should probably go to bed too :)
[23:44] <aquarius> ah, days off spent in front of the computer anyway. :)
[23:50] <thisfred> aquarius: ah you are there. I sent comments in mail
[23:50] <thisfred> and now I will have to unstarve myself
[23:51] <thisfred> overall it looks good, I may have forgotten to say, but meant to imply by lack of scathing remarks ;)
[23:52]  * aquarius grins
[23:52] <aquarius> I shall read the email
[23:53] <aquarius> ah, all good suggestions. thanks!
[23:53] <thisfred> yw
[23:53] <aquarius> I'll do them tomorrow, though, since I turn into a pumpkin in seven minutes :P
[23:54] <thisfred> now dinner, and then Retired Racing Greyhounds for Dummies, chapter 3. I shit you not!
[23:55] <thisfred> ttyl
[23:58] <CardinalFang> Good night all.