[00:27] ok, all done for today… === lifeless_ is now known as lifeless [07:37] mandel: you about ? === j0nr_ is now known as j0nr === tetris4_ is now known as tetris4 [08:43] Happy Friday, folks! :) [09:13] mandel: do you happen to know the revision 3.0.2b is built upon? === teknico_ is now known as teknico [11:22] good morning [12:00] hello, all! [12:00] alecu, hi === urbanape` is now known as urbanape === yofel_ is now known as yofel [13:30] Good morning, sorry I am so late, had some rain-relate issues at home :-( [13:31] ralsina, hi [13:52] ralsina, raining inside your home? === slank` is now known as slank === slank is now known as Guest97004 [13:56] gatox: some small bits of it ;-) [13:56] bummer :S [13:57] gatox: np, just the laundry room, and just because water pooled on the roof [13:57] gatox: so I had to go unclog a drainpipe in the pouring rain at 6AM [13:57] gatox: NOT FUN [13:57] agree... not fun [14:13] gatox: looks like the sync indicator is merged... wonder if it's released yet :-) [14:13] :P [14:16] hi folks [14:17] mmcc, hi! [14:23] no mandel today? [14:32] Afternoon! [14:32] mandel, hi [14:33] howdy mandel! [14:35] gatox, when are you going to nz? [14:35] mmcc, hello! I was late lost night talking with jfunk, looks like we have a number of new bugs :( [14:35] mandel, on tuesday [14:35] mandel, i arrive at nz on thursday [14:36] mmcc, I want to take a look at the ipc one, I wonder why it happens... [14:36] gatox, I thought it was earlier [14:39] ouch, that is a long trip, gatox you are lucky you are not too big hehe [14:39] mandel: sounds like a plan. I saw your conversation yesterday, he found a few good ones. [14:40] mandel, but on my way back... i leave nz the 8th of September, and i get back to argentina the same day jejeje [14:41] mmcc, indeed, I think there is a good way to make the ipc more resistant to the stale broker problem [14:42] gatox, you better sell well ninja and bring some hobbits back ;-) [14:42] mandel, will do! [14:42] mandel: great - what causes it, anyway? [14:44] Mmcc, it usually means that the remote reference you have in the client is old and pb is trying to use a python obj that does not longer exists in the server [15:00] me [15:00] me [15:01] me [15:01] me [15:01] me [15:02] DONE: refactoring playlists web api TODO: consolidate all couchdb acces in web api base, s/couch/u1db/ for playlists BLOCKED: no NEXT: briancurtin [15:02] DONE: fiddling with output from a subprocess stdout [15:02] TODO: write some util function that'll make this work on 2 and 3, it keeps breaking one or the other [15:02] NEXT: gatox [15:02] ralsina, alecu, mmcc: eh? guess you get to mock yourselves on twitter [15:02] DONE: [15:02] Worked on the Share Links Tab [15:02] TODO: [15:02] Finish fixing some tests and some parts of the code that are not behaving properly. [15:02] BLOCKED: [15:02] No [15:02] Madkiss, go [15:02] crap [15:02] Yes I do! [15:02] mandel_, go [15:02] me === mandel_ is now known as mandel [15:03] me [15:03] mandel, go [15:03] ok, dobey go [15:03] DONE: team call, finished releases [15:03] TODO: reviews, icon generating magic, music store [15:03] BLCK: None. [15:03] alecu: go [15:03] DONE: discussions on dash apis, vala+oauth+webservice calls [15:03] TODO: more on the dash [15:03] BLOCKED: no [15:03] NEXT: mmcc [15:03] DONE: more SMJobBless, launchd daemon checkin code, sent build to QA [15:03] TODO: launchd integration, test, propose [15:03] BLCK: no [15:03] next: mandel? [15:04] DONE: lots of calls about the dash, team call, more calls, and I called people. Administrivia. Helped around, reviews. TODO: fix windows test failure on jenkins BLOCKED: no [15:04] Sorry, got disconnected [15:04] mandel: from reality? [15:04] alecu, I think that was a while ago [15:05] Mockery of the day: https://twitter.com/ralsina/status/239015414393339906 [15:05] He, is possible [15:05] DONE: talk with QA about the Mac app. There are a bungee of big bugs to fix. Mainly adding folders from control panel, stale brokers and the app crashing fur to json decode problems [15:05] TODO: look at the stake broker bug and the json one. [15:05] Blocked, over 3g because they are updating my network [15:09] ralsina, did my best with the connection I have.. [15:09] mandel: good enough :-) [15:09] * mandel is also very sleepy [15:10] in the words of the french knight in Monty Python's Holy Grail "now leave before I mock you again!" [15:11] alecu, ralsina, I plan to make the ipc more resistant to those stale brokers by catching the exception and trying to connect to a new broker will make it shower in CSS of an error but should recover from it, what do you think? [15:12] mmcc: my mac dev env is screwed. Where are the latest instructions, and are they current? [15:12] Ups... Autocomplete [15:12] "shower in CSS"? [15:12] mandel: are you typing on your phone? [15:13] ralsina: the google doc is close to current. you can help me make it so now! [15:13] mandel: don't you have hotspot functionality on the phone? So you can go back to the computer :-) [15:13] mmcc: awesome [15:13] mmcc: will follow it, and ping you on failure [15:13] mmcc: got the URL handy? [15:13] ralsina, yes, it seems I cannot use irc over 3g on the pc [15:14] ralsina: https://docs.google.com/a/canonical.com/document/d/1f7xaDT-hblCKIXrKXZjpKtFRsBRnionDix5NojwdgmE/edit === Guest97004 is now known as slank [15:14] mmcc: awesome === slank is now known as Guest26013 [15:14] ralsina, Vodafone is crap and blocks it... Also is a PITA cause the rest works [15:15] And I keep getting raising instead of ralsina.. [15:15] ralsina: I've updated the instructions now... [15:16] I removed the windows-specific stuff for clarity. that doc is the OSX dev setup doc anyway :) [15:19] ok, need to get lunch. bbiab [15:19] ralsina, I'm going to request a holiday, there is no way I can work like this.. [15:19] mandel: agreed :-( [15:20] ralsina, I'll try to catch up I've the weekend.. [15:21] mandel: if you do that, don't ask for the day off [15:21] mandel: off is off [15:26] ralsina: OK, I actually updated the whole thing now. ping me with any questions… [15:43] mmcc: will do. I intend to be able to help with the UI issues very late today / early monday depending on luck [15:44] ralsina: great. when you get to the end, I'm very curious what you see when you run the u1cp tests - you're on 10.7 right? [15:44] mmcc: yes [15:44] eeeexcellent === salgado is now known as salgado-lunch [15:50] * gatox lunch === gatox is now known as gatox_lunch [15:52] ralsina, I'm getting an error when accessing canonicaladmin, does it work for you? [15:53] mandel3g: wfm [15:53] ack [15:58] ralsina, request made [16:03] ralsina, have you installed qt via brew already? I'm not sure how much of this you have to repeat… [16:04] mmcc: yes [16:04] mmcc: or at least brew says it has it already [16:05] mmcc: BTW, lovely how everytime a configure script checks for javac a dialog pops up telling me to install it ;-) [16:05] mandel_: I'll approve it in a bit. So, you have the rest of the day off, go rest :-) [16:06] ralsina: heh. I guess I installed java already. [16:09] ralsina: I just remembered that you need to get a qt with no X11 deps so it will run on stock 10.8 [16:09] can you do 'brew cat qt | grep X11' for me? [16:09] mmcc: sure [16:10] if that comes back with X11 flags, then you need to reinstall qt. [16:10] mmcc: I get -L/usr/X11/lib -I/usr/X11/include [16:10] ok, how do I reinstall it? [16:11] mmcc: but that's just if I want to do a bundle to share? [16:11] mmcc: since rebuilding Qt is bound to take a few hourds, I would rather do that on the weekend if at all possible [16:12] ralsina: for me it downloaded a prebuilt binary - was done in a minute [16:12] % brew install https://raw.github.com/mxcl/homebrew/master/Library/Formula/qt.rb [16:12] mmcc: ok, will do it then :-) [16:13] mmcc: I get "undefined method 'version'" maybe my brew is too old? [16:13] hrm [16:14] maybe. try 'brew update' [16:14] also if you already had qt 4.8.2, that'll be a problem and you'll need to brew uninstall qt before running that install from the URL [16:15] it's telling me I have local changes to the qt recipe, probably something alecu did before hegave me this mini [16:15] I have 4.8.0 so it's not that [16:16] ok… well you can save off the locally-changed qt by doing 'brew edit qt' and saving that somewhere else [16:16] I'm assuming it gives you the option to overwrite the changed file? [16:17] mmcc: no, I have to go git stash it [16:17] mmcc: user friendly it isn't [16:17] giiiiiiit [16:17] mmcc: ok, it's working now, thanks! [16:17] the number of projects that decided *git* was a user interface… the mind boggles [16:18] ralsina, mmcc: I had to use a fix for the qt brew. I added it to the osx dev setup doc, and it was this: https://github.com/tmf/homebrew/commit/f34897104399f48b36e1e86a776c70dd2d882c5a [16:18] mmcc: starting with git ;-) [16:18] alecu: we are using a bottle now, so that should not be needed [16:18] ralsina: right [16:21] great now my xcode is too old [16:21] alecu, that bug is indeed fixed in the latest qt.rb, just differently :) build systems, man, build systems [16:22] Xcode too old? what's telling you that? I thought it shouldn't matter… [16:22] ralsina: don't worry about the "too old". [16:22] mmcc: brew [16:22] alecu: ok, stopping xcode update then [16:22] ralsina: which brew command? qt? [16:23] mmcc: yes [16:23] mmcc: tells me to PLEASE install 4.4.1 because 4.3.2 is too old [16:25] ralsina: it complained the same about 4.4.0, but I just ignored it. [16:32] mmcc: does ubuntu-sso-client give a clean bill of health with pyflakes already? [16:33] dobey: I don't know, but I doubt it [16:33] dobey: answer is no. it complains about gettext's '_' [16:34] and some other minor things [16:34] right; commented on your merge [16:36] crikey, why are mandel's fsevents branches so freakin huge. and why has nobody reviewed them yet [16:36] * mmcc thought those already went through [16:36] (mine, I knew mandels were rotting there) [16:37] dobey, you answered your own question. I have reviewing them as a TODO for today thoguh [16:38] why isn't https://code.launchpad.net/~mandel/ubuntuone-fsevents-daemon/sbjson-link/+merge/116445 landed? it has 2 approves [16:39] I don't know. we should probably land it. [16:39] * mmcc flipped it [16:40] mmcc: you have first a command to install protobuf via brew and later the doc says "install protobuf following these instructions" is that so or should I ignore the second one? [16:41] ralsina: looking… [16:43] if you have protoc in your path then ignore the second one [16:44] mmcc: ack [16:44] mmcc: looks like I do [16:45] ls -l `which protoc` should show a symlink to /usr/local/Cellar/protobuf/blah [16:46] mmcc: missing step somewhere in the middle: install qtreactor [16:46] that should be in the buildout, i thouhgt [16:47] mmcc: looks like it isn't or it failed to install. I'll check [16:47] did you get an error in running tests, or were you testing from the interpreter? [16:48] mmcc: it's in the links but is not listed as a requirement, it seems [16:48] mmcc: when running tests for u1cp [16:49] mmcc: oh, wait it's in the development section of the buildout [16:49] mmcc: I'll figure it out [16:49] it should be there, it's one of the eggs… [16:49] if you just try to import it from the interpreter, does that work? === salgado-lunch is now known as salgado [16:49] mmcc: latest qt4reactor tries to use PySide!!!! [16:50] um [16:50] that'll teach us for using /master in a buildout :\ [16:50] mmcc: got it, when updating qt, pyqt broke [16:51] mmcc: and qt4reactor tries pyside if pyqt fails to import [16:51] oh, shoot. sorry, should've remembered that [16:51] mmcc: it's ok [16:52] mmcc: so now it will rebuild pyqt from source, right? [16:52] yes [16:52] but that's not slow === gatox_lunch is now known as gatox [16:59] dobey, re pyflakes on darwin, pylint has a lot more spew. It looks like it can't find dev-tools even though the u1lint wrapper definitely has it in the path [17:00] mmcc: possibly another missing step, clientdefs.py has to be moved to the right place before you can run tests [17:00] mmcc: right. probably because of the missing patch issue [17:01] right, now I remember that [17:02] mmcc: and also, storage-protocol seems to need to be installed [17:03] ralsina: for the sso tests, or for the control-panel tests? [17:03] mmcc: aaaaaand I am getting the error messages when running tests :-) [17:03] mmcc: u1cp tests [17:03] ralsina: ack. [17:03] pretty much everything has to be 'installed' for u1cp tests to run [17:04] ralsina: yeah the buildout doesn't do any of the build steps for the source reps, it just bzr branches them for you [17:04] source deps, not source reps [17:04] mmcc: ok [17:05] mmcc: on windows this used to work by just setting PYTHONPATH and not installing [17:05] mmcc: but it doesn't really matter [17:08] mmcc: now I get "can't import name fseventsd" [17:09] mmcc: probably because I haven't installed ubuntuone-client how do you do that on mac? [17:11] aha. ok, you hit a gap in the docs. good [17:11] fseventsd is from mandel's branch. you need to grab both that and the fsevents.py, let me look for URLS [17:12] into parts/ do bzr branch lp:~mandel/ubuntuone-fsevents-daemon [17:13] actually, I'll just put this in the doc [17:13] mmcc: cool [17:13] I get invalid transport because bzr probably doesn't have the launchpad plugin or something [17:14] mmcc: shouldn't that be lp:ubuntuone-fsevents-daemon ? [17:14] and lunch [17:14] yes. d'on [17:29] and that's still not open source, so I need to install credentials [17:30] ping alecu - re: checking version of the fsevents daemon. I have code doing that as a result of launchd integration, not 100% sure where the best place to put it is [17:30] ralsina: yes we need to sort that out. there's still a "TODO" comment about licensing in the source [17:38] mmcc: now I get "no module fsevents". I assume I need to install macfsevents? [17:38] yes. I updated the doc - you need to git clone and build it [17:42] mmcc: missing dependency dirspec [17:42] mmcc: oh, wait, I need to install it [17:42] ralsina: what was missing dirspec? the buildout should have put it in parts/ , and it should be on sys.path of the wrapped interpreter [17:43] so 'python -c "import dirspec" ' should work [17:44] mmcc: yes, forgot to source the env [17:44] aha. ok [17:44] I am logged in :-) [17:45] thanks for wading through that, glad the doc is updated now [17:46] although I am not getting any folders other than ~/Ubuntu One [17:47] yes, there appear to be UDF bugs… [17:47] are shares showing up? [17:47] mmcc: I get a SSL error about it not finding the godaddy cert [17:49] hrm. it should see the cert files in ubuntuone-storage-protocol/data … oh. I think that code is wrong. I don't think it handles running from source correctly [17:50] no, it should be right. [17:51] mmcc: it's trying to load it from the installed location [17:51] mmcc: because I installed storage-protocol [17:51] oh. I see. I never installed storage-protocol [17:52] was expecting it to be run from the source tree [17:53] mmcc: then it has to be added to the PYTHONPATH [17:53] mmcc: not installing and putting it in PYTHONPATH works and I get all the shares listed :-) [17:53] good. [17:53] mmcc: which means I see the scrolling bug, which means I can get to work on it [17:55] very good [18:06] mmcc: can you verify for me that adding "-graphicssytem raster" to the command line fixes the scroll? [18:07] ralsina: I haven't seen the bug, I don't have enough shares to scroll [18:07] mmcc: want some? ;-) [18:07] ugh: https://launchpadlibrarian.net/113439378/buildlog_ubuntu-quantal-i386.ubuntuone-control-panel_4.1%2Br348-53~quantal1_FAILEDTOBUILD.txt.gz [18:07] you bet i do [18:07] mmcc: you can just add folders in your system [18:07] lots of failures in there; and logging messages to console amidst the tests :( [18:08] dobey: "constants module not available." is strange [18:09] indeed [18:09] dobey: does it do the same in your q box? [18:09] i haven't run it locally; but this is happening on oneiric and precise as well, for the nightlies [18:10] dobey: I can do a run on precise to verify === salgado is now known as salgado-afk [18:12] dobey: I got a segfault! [18:12] stable-4-0 was fine for me yesterday at least [18:13] dobey: same here [18:13] dobey: as in "was fine for me yesterday" [18:14] trying trunk now [18:15] we need to fix the on_download_finished exceptions… I'm at #300 and counting and it's stalling the control-panel UI [18:16] or am I out of date and they're fixed in trunk? [18:16] hrmm; trunk ./run-tests passed fine here :-/ [18:16] * mmcc checks [18:16] mmcc: any news on that download priority issue we talked about yesterday? [18:17] chaselivingston: thanks for reminding me. [18:17] mmcc: np [18:17] \ [18:17] mmcc: no, they still exist [18:18] mmcc: they break windows jenkins too [18:18] guys, chaselivingston was noticing that he wasn't getting files downloaded to his mac while uploading a ton of data. is that a bug or expected bad behavior? [18:18] mmcc: if you try to add folders, does the "select a folder" dialog go away on its own sometimes? [18:19] ralsina: I'm still stalled on the loading overlay :( [18:19] mmcc: that's probably that you need to start syncdaemon manuall [18:20] really? [18:20] mmcc: happened to me a few minutes ago [18:20] mmcc: if u1cp can't find it / start it it freezes [18:21] is that the "select some folders to sync to the cloud" page hanging? [18:21] dobey: no, this is in u1cp proper [18:21] no, it was starting it just fine - I saw my precise VM pop up a notification about downloading new .DS_Store after I started it [18:21] hrm, well quitting and restarting with U1_DEBUG=1 got me to a working folders screen [18:21] hrmm [18:22] not good: I canceled the add_folder dialog and got this in the log: "ubuntuone.controlpanel.qt.addfolder - INFO - on_add_folder_button_clicked: user requested folder creation for path u''. [18:22] " [18:23] mmcc: pong re: "checking version of the fsevents daemon" [18:23] mmcc: that's ok [18:24] mmcc: it just means "hey, you didn't select anything. Ok!" [18:24] ralsina: ok. [18:24] mmcc: although I agree it's a scary message ;-) [18:25] ralsina: yeah, combined with the controlpanel status still saying "file sync starting…" I thought it was busily trying add something random [18:25] alecu - pong your pong, just a sec [18:25] no prob [18:29] mmcc: do qt plugins work on mac? If they don't I need a workarund for my scroll workaround :-( [18:29] ralsina, they should. can you give me more context? do you need to add one? [18:30] mmcc: basically, the workaround I found for the scrolling is using -graphicssystem raster and that's a plugin [18:30] mmcc: or at least I think it's one [18:31] alecu - re the daemon version issue: http://paste.ubuntu.com/1164920/ [18:31] ralsina: you can check with the QT plugin env var.... /me looks [18:31] QT_DEBUG_PLUGINS=1 [18:34] mmcc: cool, not a plugin [18:34] mmcc: so now I just need to inject that option only on darwin and it's fixed [18:35] brb [18:36] "Jobs' Bless"!!!! People praying to Saint Steve already? [18:36] heh, they've been doing that for years [18:37] lol [18:39] alecu: where would be a good place in u1cp to put a platform-specific constant I need in main? [18:39] ralsina: in u1cp's main? [18:39] mmcc: +1 to "all the version checking can be done in controlpanel, and SD doesn't need to know about launchd" [18:39] alecu: yes, I need to inject two args into sys.argv when we are running on darwin [18:39] alecu, sounds good. [18:40] alecu: I *could* do it in the bin/ script but that's icky [18:40] ralsina: inject args into sys.argv? why? [18:40] alecu: because of a bug in Qt: https://bugs.launchpad.net/ubuntuone-control-panel/+bug/1018918 [18:40] Ubuntu bug 1018918 in Ubuntu One Control Panel "Scrolling shares on darwin does not work" [High,In progress] [18:40] alecu: I can workaround it by passing "-graphicssystem raster" [18:40] ralsina: you just need to add it to the args passed to Qapplication, right? [18:40] mmcc: right [18:41] see line 88 of gui/qt/main/__init__.py [18:41] we're already doing some arg tweaking there [18:42] mmcc: yes, but it's not platform-specific [18:43] mmcc: although a bit lower we do have a darwin-specific bit [18:43] ralsina: we are even checking for sys.platform a few lines below [18:43] alecu: yes, just saw it [18:43] ok, I'll do it there then [18:43] ralsina: also, at the top of the file we have some "win32" and "darwin" checking for dbus... [18:43] alecu ack on your +1, proceeding thusly [18:44] alecu: yes, so that file is dirty already :-) [18:44] ralsina: right. Go ahead, I'll look to the southeast. [18:45] alecu: I can file a bug about refactoring the platform-speciicness out of it [18:45] ralsina: just saw the bug you mentioned about the dialog disappearing. I believe it is bug #1040905 [18:45] Launchpad bug 1040905 in Ubuntu One Control Panel "Cannot “Add a folder from this computer” more than once per session" [Undecided,Confirmed] https://launchpad.net/bugs/1040905 [18:45] ralsina: that sounds great. [18:45] mmcc: right [18:45] mmcc: and re: "CP kills old installed daemon * TODO:", I was supposed to create a bug for that, where we can discuss the best way to do it. [18:46] alecu, ralsina: that'll be a tricky refactor, due to the dependencies between qt4reactor, reactor, QApplication and QCoreApplication, etc… [18:46] mmcc: yes [18:46] mmcc: worse or better: there is a hack to install the reactor in the bin/ubuntuone-control-panel-qt script as well [18:46] mmcc: I propose that we postpone that refactor till after the rushes [18:46] alecu,mmcc: yes, this is for next cycle, probably [18:46] but I am filing so it's there [18:47] bug #1041333 [18:48] Launchpad bug 1041333 in Ubuntu One Control Panel "main/__init__.py is platform-specific" [Medium,Triaged] https://launchpad.net/bugs/1041333 [18:48] ok [18:49] mmcc: so, my idea is that the root daemon gets a message telling it that it will be updated, so it notifies every running SD of that. Then each SD shuts down, and the root daemon too. The new version of the root daemon is started, and then somehow (*insert magical idea here*) the new version of the SD is started for every user that had it running. [18:51] alecu - upgrading other users' syncdaemon is probably not workable [18:51] mmcc: perhaps we can have a SD wrapper script that checks the error code of the SD process, and on a given value of that error code, the wrapper process waits a few seconds, and then restarts it. [18:51] mmcc: it's the system's root daemon we are upgrading, so every SD version should match! [18:52] alecu, I agree! but syncdaemon itself is going to be in another user's directory, unless they installed the app in a system-visible place [18:52] and if we upgrade SD inside an existing .app, we break the code signing, etc [18:53] might be the best we can do is teach CP how to tell the user that they need to upgrade [18:53] mmcc: I assumed everybody installed these things in /Applications .... :P [18:53] and that they won't be able to sync until they do [18:53] alecu: can't make any such assumptions [18:53] mmcc: right :-( [18:53] many people will end up running it out of ~/Downloads, or never moving it out of the install DMG [18:54] mmcc: do we have a "system registry" of sorts in osx? [18:54] alecu, that depends on what you're trying to do :) [18:55] there isn't a single binary grab-bag database for config info, but we do have the filesystem [18:56] mmcc: I would assume that the user that has power to install the root daemon, is the one that has admin power. And by installing (or upgrading) the root daemon, we would write the path to the system's Syncdaemon.app [18:56] mmcc: right, we have some equivalent of /etc [18:56] like /etc ? :) [18:56] right [18:56] :-) [18:57] so the idea is to move syncdaemon out of the app package, and put it in a global place so it's possible to upgrade it? [18:57] * mmcc thinks [18:57] * alecu starts up his mac, to cut on the amount of guessing. [18:58] I don't think we can do that without shipping two copies of Qt [18:58] ugh [18:59] because CP needs qt to start so it can install SD, and SD needs qt, but can't point to CP because we don't know where the user has CP [19:00] mmcc: what if we symlink from the "canonical SD place" (say, /etc/ubuntuone/syncdaemon) to the SD from the last installed root daemon? [19:01] not sure I understand "SD from the last installed root daemon" - the daemon is just one executable file and doesn't include SD… [19:01] mmcc: right, but it's installed by *some* copy of the control panel, right? [19:01] mmcc: and that control panel should have a SD next to it [19:02] next to it, yes. but in some user's directory [19:02] and to be clear, no code is really installed, it all stays in the .app wrapper - SD is just UbuntuOne.app/blah/blah/Syncdaemon.app/ [19:03] mmcc: right. And with the SD being in the user's directory, we hit permission issues if some other user tries to run it... [19:04] right. [19:04] mmcc: so, the only thing that's really installed is the root daemon executable. [19:04] yes [19:04] (and some .conf files, but that's not important right now) [19:05] we *could* do this: [19:06] I upgrade my CP, it detects an old daemon, upgrades that, which tells my CP that there are other SDs listening. [19:06] So my CP puts a copy of the whole new U1.app in a system-visible place [19:06] (or in each users' drop box folder) [19:07] and then when their SD notices the issue, it looks in the known place for a totally new U1 app to upgrade to [19:07] upgrading piecemeal will break code signing, but upgrading the whole thing works [19:08] and we already have it downloaded, so it'd be smooth for the other user to just upgrade [19:09] (that's it - what do you think?) [19:13] ~/Users/Shared/ would be one place we could put the downloaded .app for upgrading [19:17] mmcc: it sounds good [19:17] mmcc: is ~/Users/Shared writable by all users? [19:18] heh, I meant /Users/Shared, and yes [19:18] mmcc: I would like for only the user entering the sudo password to be able to write there... [19:18] why? [19:18]