/srv/irclogs.ubuntu.com/2011/11/29/#ubuntuone.txt

thisfredanyone know where the new developer tasks page on the canonical wiki lives?00:25
thisfredI had to reinstall00:25
thisfredeverything00:25
thisfredand now the internal chat won't work00:26
=== gatox_away is now known as gatox
thisfreddobey: thx00:40
thisfreddobey: the one I meant was for the PPA, but I found the PPA itself. The page seems to be gone though00:41
=== pfibiger` is now known as pfibiger
ryemorning06:35
mandelmorning all!09:43
JamesTaitGelukkig Dinsdag! :)09:48
alecuhello #ubuntuone!13:35
mandelok, my internet is crap :(13:35
mandelalecu, buenos dias!13:35
alecumandel, hello to you, sir!13:36
mandelalecu, I've got good news about the proxy settings, I'm done with the TestCase for both auth and nonauth, I just need to update u1trial to use --integration13:37
mandelbut for that I want to provide a test loader rather than using the default one and removing the integration tests13:37
alecumandel, oh, like a new test loader for trial?13:38
mandelalecu, yes, I think is a more elegant solution13:38
alecusounds... hard.13:38
mandelalecu, I'm looking at the unittest implementation, if it is too hard I'll simply purge out the integration tests from the tests collection13:39
mandelalecu, I need to go for lunch, if you see ralsina, tell him gatox is on a swap that he needs to approave ;)13:42
alecumandel, cool13:42
alecuand yes... it's a lovely day for some AC/DC13:50
alecuthunderstruck!13:50
dobeyhmm13:59
dobeymandel: simple solution is to do the same thing we do for qt vs gtk i guess14:00
* Chipaca looks around for a spare ralsina14:24
dobeyheh14:28
ralsinagood late morning!14:38
dobeyoh look, a ralsina14:39
ralsinahi dobey14:39
dobeyhola14:39
ralsinawhy did everyone load a kajillion admin requests on my day off? ;-)14:43
=== webm0nk3y is now known as jdobrien
dobeyi figured i would put in the rest of my time off for the year, before the year ends :)14:44
jdobriendobey: indeed.14:44
* dobey pbuilds a rhythmbox from trunk package14:49
mandeldobey, yes, that is also a good idea, simply pass those to be ignored14:58
mandelralsina, same here, I added all my holidays 'til the end :P14:59
mandelralsina, by the way, gatox needs a +1 for his swap day14:59
ralsinamandel: done14:59
ralsinamandel: same for all of your requests and dobey's15:00
dobeymandel: do those tests require squid?15:00
mandeldobey, yes, those are the ones.. I want to see if the test loader is the one that checks for a flag to decide if they are instantiated or no15:00
mandelnot*15:01
dobeymandel: :(15:01
dobeymandel: yes, the test loader pokes the test case hierarchy to see what service runners are needed15:02
dobeyno standup today too? yay!15:03
mandeldobey, yeah, I'm looking how the loader in done in python unittest to implement something nice for our use15:04
dobeymandel: we already implement a loader in u1trial15:04
dobeymandel: and i still don't think this sort of stuff belongs in u1trial15:05
mandelwell, we do this:  unittest.defaultTestLoader.loadTestsFromModule(module)15:05
mandelwhich is not really a loader, we poke stuff, look if we reqire services and then call the default loader15:06
mandelme15:06
mandelwe do have stand up? or not?15:06
dobeymandel: i think you're conflating loader, with the thing that discovers which tests to load; and you are wanting to do the latter15:08
mandeldobey, yes, I want to do the later, but the loader is the one that discovers the tests. I wanted to be able to tell the loader not to add a test to its test collection if it was an integration tests15:09
mandelI think is more elegant than loading ALL tests and then discard those that are integration tests in the runne15:09
mandelr15:09
dobeymandel: no; the thing that discovers the tests is the code in u1trial15:09
dobeymandel: the only sanely feasible way to do this, is to put integration tests under a different location in a different set of files15:10
mandeldobey, we have _load_tests , _load_tests: Load unit tests from a Python module with the given 'relpath' and _collect_tests: Return the set of unittests. which means that I have all the knowledge of which tests are to be ran, right?15:12
dobeymandel: there is one block of code that walks the filesystem tree for the passed-in path argument; there is another that walks over the list of test_foo.py files that function returns, and loads them into the test suite15:13
dobeyi presume _colelct_tests is the one that walks the fs tree, and thus the thing you want to write a different implementation of15:13
mandeldobey, yeah, I was looking at that part, I believe I have an idea of how to do it 'cleanly'15:21
dobeymandel: well, i don't think u1trial should have some special case for this15:22
mandeldobey, ideally it would be nice for it to be able to differenciate between integration tests and not integration tests.. should not be terribly hard, le me do some work and we cna take a look at the code15:23
dobeymandel: u1trial has no concept of what an integration test is; so an --integration flag doesn't make sense really15:23
dobeymandel: i can't even get developers to make that distinction15:23
mandeldobey, yeah, that is true, making the distintion between integration and unittests is hard..15:25
dobeymandel: and trying to do that programmatically will be wrong, if developers can't even make the distinction :)15:26
mandeldobey, oh, but let me try!15:26
dobeymandel: it's an exercise in futility15:26
dobeymandel: the only way you're going to make a distinction is by having the tests be in completely separate paths on disk; and then you can just do u1trial tests/ vs u1trial integrationtests/15:27
mandeldobey, that is what she said :P15:27
mandeldobey, for the previous sentence, ofcourse hehe15:28
mandeldobey, but yeah, in the worst case scenario, using diff paths is a simple way to do it15:28
ralsinanote: alecu is without power, he'll show up eventually15:28
dobeyheh15:30
dobey"foo is without power. * foo has quit"15:31
dobey[dobey@lunatari:banshee]: apt-cache search monkey brain15:32
dobeyttf-aenigma - 465 free TrueType fonts by Brian Kent15:32
dobeynot helpful.15:32
dobeymight help if i search for the right thing though15:33
=== m_conley_away is now known as m_conley
j0nrrye: any progress on the music id3 problem?16:06
mandeldobey, do you know it there are lipsoup bindings for pyhton that do not use gi, I'd like to add some tests to see that my proxy works as expected :)16:11
mandeldobey, once that tests is done I'll show you how I got the --integration thing working16:11
dobeymandel: no, there aren't16:11
mandeldobey, que mierda!16:12
mandeldobey, 'cause ulr just works with noauth proxies :(16:13
dobeyulr?16:14
mandeldobey, ups, sorry typing fail, urllib.. but urllib2 does, right?16:15
dobeymandel: i don't think either one implicitly handles proxies16:19
dobeyhrmm, what to get for lunch16:22
=== beuno is now known as beuno-lunch
dobeybbiab, lunch16:34
mandelralsina_, ping16:39
ralsina_mandel: pong16:39
mandelralsina_, I have the squid tests finished, but there is part of the code that is ugly... atm I'm doing a sleep to ensure that squid has enough  time to load the config and start, do you know of any command that can return if squid started?16:40
mandelralsina_, specially on a port16:41
ralsina_mandel: let me check16:41
mandelralsina_, 'cause with that, I'm done :)16:41
ralsina_mandel: usually squid will start in less than a second16:41
mandelralsina_, yeah, but I dont want to be foing a time.sleep(1) is very very ugly16:41
ralsina_mandel: you can use -d 316:42
ralsina_mandel: and grep the output for somethig16:42
ralsina_mandel: but it's much more complicated than sleep(1) ;-)16:42
ralsina_mandel: Or, you can loop connecting to the port, and then it's running16:42
mandelralsina_, what about this: squid -k check -f squid.conf16:43
mandel??16:43
ralsina_mandel: well....16:44
ralsina_mandel: you would still have to sleep and loop16:44
mandelralsina_, yeah, which is ugly.. bu nicer than a grep :)16:44
ralsina_are you doing squid -z?16:44
mandelralsina_, no, what is that?16:46
ralsina_mandel: -z builds the whole folder structure for the cache16:46
ralsina_mandel: should be run at least once16:47
mandelralsina_, yes, that is ran before we start squid16:47
ralsina_mandel: OTOH, it may be a good idea to run it every time, to make tests independent16:47
mandelralsina_, is done every testcase, not test since it would be too much16:47
ralsina_ok16:47
ralsina_is the order of tests randomized?16:47
* ralsina_ is nitpicky sometimes16:47
ralsina_but in this case it's overkill16:48
ralsina_anyway, yes, you can use -k check16:48
ralsina_in fact, if you are just changing configs from test to test, you could do -k reconfigure16:49
ralsina_the only bad parts with going that way is a possibility of negative DNS caches or similar stuff16:50
=== eu is now known as Guest2622
mandelralsina_, good point.. I guess I can get that done for version 2 :)17:07
mandelralsina_, we need something asap so that alecu can start writing integration tests at least on windows17:07
ralsina_mandel: of course17:08
mandelralsina_, dobey so I'm thinking about this: lp:~mandel/ubuntuone-dev-tools/proxy-testcase17:08
mandelralsina_, dobey that branch has two things, a SquidTestCase, and a --integration-tests flag. if you do a ./run-tests all tests should pass but if you do u1trial -c ubuntuone --integration-tests you will see one fail, since the integration tests fails always17:09
mandelralsina_, dobey you can also find a testcase that uses the squid nontauth proxy under ubuntuone/devtools/testcases/tests/test_squid_testcase.py, but for that to not be skipped you need to have in the system squid and htpasswd since I'm using a very basic auth for the auth proxy17:10
mandeldobey, I added you as a reviewer for th branch with alecu since you are the two ones I want to hear about :)17:16
mandel@ping17:17
ubot4pong17:17
Chipacamandel: <dobey> i guess i should get some lunch17:19
ralsina_mandel: I see nothing obviously wrong with the branches' squid config, if that makes you happy17:19
mandelChipaca, I know, but I hope he reads it when he is back :)17:20
mandelralsina_, have you tried running the tests?17:20
ralsina_mandel: nope17:20
mandelralsina_, and it has merge issues :(17:20
mandelralsina_, mierda!17:20
ralsina_mandel: it happens17:21
mandelralsina_, I'll fix does and EOD17:21
ralsina_mandel: ack17:21
dobeyhi17:28
dobeyhmm17:30
dobeyscary17:30
dobeyoh hi firefox; why are you eating 600M of RES?!17:31
Chipacadobey: because it can :)17:36
ralsina_dobey: CAN I HAZ MEMRY?17:36
=== beuno-lunch is now known as beuno
dobeyswap swap swap17:38
dobey2G of ram and still have to swap :(17:38
mandeldobey, you should use windows vista.. then you will stop complaining ;)17:44
mandeldobey, ralsina_ ok, EOD for me, I pushed the devtool branch one rev to fix the conflicts issues, feel free to complain about everything ;)17:45
mandelsee you all tom!17:45
* mandel gets ready for rugby17:45
dobeymandel: i think i'll upgrade to Slackware 9517:48
mandeldobey, or mint you to piss off people hehehe17:48
dobeyi'll just buy a mac and put the dock on the left17:50
dobeyor just leave it as it is, and say it's elementaryOS17:50
mandeldobey, use the mac in a diff orientation, works the same way :)17:52
mandelok, now I'm really leaving, laters!17:52
dobeyheh17:52
ralsina_dobey: how do you feel about some lucid-bug-verifying? ;-) https://bugs.launchpad.net/ubuntuone-client/+bug/89779917:59
ubot4Launchpad bug 897799 in ubuntuone-client "Nautilus crashes all the time I try to publish a file (affects: 1) (heat: 6)" [Undecided,New]17:59
dobeylucid? bah18:02
=== elopio_ is now known as elopio
dobeyvds: upgrade! :)18:03
vdsdobey, apt-get update or dist-upgrade ? :)18:04
dobeyvds: to ubuntu 11.10 :)18:05
dobeyvds: also, "apport-collect 897799 -p ubuntuone-client" please18:06
vdsdobey, apport on it's way18:07
dobeyhmm, though that apport info might not be useful18:10
dobeyvds: is there a nautilus crash file in /var/crash/ ?18:10
vdsdobey, nope18:11
dobey:-/18:11
dobeyneed a backtrace18:13
vdsdobey, I keep getting LP errors18:17
dobeyvds: yeah i think lp is being a bit buggered at the moment :(18:19
dobeybah, lp recipe builder is broken :(18:26
vdsdobey, apport explodes...18:28
vdsdobey, http://paste.ubuntu.com/753939/18:29
dobeyhuh; lovely. it seems you don't have firefox, but something is configured to use firefox as your default browser?18:31
dobeycome on twisted test suite; finish already19:00
=== m_conley is now known as m_conley_away
dobeylater all22:39
=== eu is now known as Guest51181
=== gatox is now known as gatox_away
ulidtkohey all23:17
ulidtkoi have a reproducible upload failure23:18
ulidtkowhat should i do?23:18
karniduanedesign: ↑ perhaps you could help my friend?23:39
duanedesignhello ulidtko23:39
ulidtkoduanedesign: hi23:40
duanedesignulidtko: what seems to be happening?23:40
ulidtkoi placed a folder of mp3 files into my u1 folder23:40
ulidtkothen i watch u1sdtool --status and u1sdtool --current-transfers23:40
ulidtkoit starts uploading the files, progresses till some point, and then just stops23:41
ulidtkotried 4 times already (restarting via u1sdtool -q && --start), every time the same scenario23:41
duanedesignok. Would you mind sending me your logs so I can look into what might be going on?23:41
karniduanedesign: thanks, /me coocks dinner23:41
karni*cooks23:42
ulidtkowhere do i get the logs?23:42
duanedesignthank you karni23:42
karniduanedesign: ~/.cache/ubuntuone/log23:43
duanedesignulidtko: http://pastebin.com/c5vKjRQ523:43
duanedesignulidtko:  i will PM you my email23:44
ulidtkotar.gz is ready23:45
ulidtkosent23:47
* duanedesign loking23:48
duanedesignlooking*23:48
duanedesignulidtko: you are being affected by bug #87292423:50
ubot4Launchpad bug 872924 in ubuntuone-client (Ubuntu Precise) (and 5 other projects) "Resuming a failed upload crashes with ValueError: I/O operation on closed file (affects: 27) (dups: 3) (heat: 153)" [Critical,Fix released] https://launchpad.net/bugs/87292423:50
duanedesignit has been fixed you should be seeing that fix in your updates soon23:50
ulidtkogood!23:51
ulidtkowill the updates arrive via ubuntu's update manager?23:51
ulidtkoi.e. don't I have to wait till next resease?23:51
duanedesignulidtko: no no23:56
duanedesignThis fix will be backported23:56
duanedesignulidtko: can you open the software center and select Software Sources23:56
ulidtkovery nice23:56
ulidtkodo you want me to enable oneiric-proposed?23:56
duanedesign:)23:57
ulidtkoi think i understand already :)23:57
duanedesignok great23:57
ulidtkoand... is there some kind of a workaround?23:58
duanedesignbe sure to restart the syncdaemon after you get the fix. You can do that with a disconnect/connect in the Control Panel or u1sdtool -q; u1sdtool -c   in Terminal23:58
duanedesignulidtko: let me see23:58

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!