=== Ursinha is now known as Ursinha-afk | ||
=== Ursinha-afk is now known as Ursinha | ||
=== fginther is now known as fginther|away | ||
=== fginther|away is now known as fginther | ||
=== Ursinha is now known as Ursinha-afk | ||
=== Ursinha-afk is now known as Ursinha | ||
=== chihchun_afk is now known as chihchun | ||
=== roadmr_afk is now known as roadmr | ||
=== roadmr is now known as roadmr_afk | ||
=== roadmr_afk is now known as roadmr | ||
jibel | pitti, I made some progress on a nova setup script for the ssh runner. I'd need to add a wait_for_ssh in adt-virt-ssh because nova returns when the VM is active but it takes a moment before ssh is really listening. Are you okay with that or do you think of another way to do it? | 07:29 |
---|---|---|
pitti | jibel: no, that's a good idea; it can affect LXC as well | 07:29 |
pitti | jibel: http://anonscm.debian.org/gitweb/?p=autopkgtest/autopkgtest.git;a=shortlog;h=refs/heads/adt-virt-ssh is up to date with what I have | 07:30 |
pitti | (except the test cases) | 07:30 |
jibel | pitti, ok, that's the branch I checked out | 07:30 |
* pitti AFK for a bit | 07:31 | |
pitti | hm, I'm getting sh: 1: Syntax error: Bad fd number | 07:42 |
pitti | for a tar copy | 07:42 |
pitti | jibel: did you happen to run into something like that? http://paste.ubuntu.com/7709955/ | 08:09 |
pitti | jibel: that's the closest I got so far | 08:09 |
pitti | jibel: i. e. with multi-line scripts it fails | 08:10 |
pitti | $ bash -x /tmp/runcmd sh -ec 'exec 3>&1 >&2; whoami' | 08:10 |
pitti | that works fine | 08:10 |
pitti | jibel: so it's getting confused by line breaks somehow | 08:10 |
jibel | indeed quoting is completely different: sh -ec $'exec 3>&1 >&2\nwhoami' for the 1rst case and sh -ec exec\ 3\>\&1\ \>\&2\;whoami for the second | 08:14 |
jibel | I never noticed that before | 08:14 |
pitti | jibel: I'm ssh'ing into LXC, not sure if that makes any difference | 08:15 |
pitti | but I had assumed that pretty much any test would need to run scripts | 08:16 |
pitti | $ ./run-from-checkout -B ~/ubuntu/tmp/testpkg// --- ssh -s ssh-setup/adb | 08:16 |
jibel | pitti, it is not related to lxc, because if you execute it directly it's the same errpr | 08:16 |
pitti | but this stilll works fine | 08:16 |
jibel | sh -c "sh -ec $'exec 3>&1 >&2\nwhoami'" | 08:16 |
jibel | sh: 1: Syntax error: Bad fd number | 08:16 |
pitti | jibel: oh, the >3 bits are only done from build_source() | 08:18 |
jibel | pitti, re wait_for_ssh, I'll add a retry argument to VirtSubproc.check_exec(), what do you think? | 08:19 |
pitti | jibel: ah, I would just do that loop in virt-ssh's setup, and call VirtSubproc.execute_timeout until it succeeds | 08:20 |
jibel | pitti, ok | 08:21 |
jibel | pitti, autopkgtest in a nova instance: ./run-from-checkout --setup-commands="apt-get update" -d libpng --- ssh -s ssh-setup/nova -d -l ubuntu -- -f 1 -k j-lallement -i 138cdd41-e72d-4f98-b394-db2b66ba24af | 09:24 |
jibel | \o/ | 09:24 |
pitti | yay! | 09:24 |
pitti | jibel: still wondering how you can evade that quoting issue; maybe it behaves differently if you have sudo in between? | 09:25 |
pitti | jibel: but anyway, after wresting with lots of various shell trickeries, I resigned and changed adt-run to never give commands with newlines | 09:25 |
pitti | jibel: what's that magic -i number? | 09:33 |
* knome never knew "c" (etc) is a number | 09:34 | |
jibel | pitti, -f is the flavor (m1.tiny ...) -i the id of the image | 09:34 |
jibel | -k the name of your key in the keypair you registered | 09:35 |
pitti | jibel: ah, of course | 09:35 |
jibel | pitti, I tried in canonistack, it probably work in hpcloud but didn't try | 09:35 |
pitti | test_no_script_no_root (__main__.SshRunner) | 09:36 |
pitti | no setup script, no root ... ok | 09:36 |
pitti | test_password (__main__.SshRunner) | 09:36 |
pitti | no setup script, no root, password auth ... skipped 'ssh password auth is not implemented' | 09:36 |
pitti | \o/ | 09:36 |
pitti | jibel: ok, I committed two fixes to master (one of them the \n → ; change, to avoid the %q madness), and rebased the adt-virt-ssh branch | 09:38 |
pitti | jibel: does git pull just work for you, or does that complain? | 09:38 |
jibel | pitti, it seees to be happy, just complained that I changed adt-virt-ssh too | 09:39 |
jibel | seems | 09:39 |
pitti | jibel: ah, so "git stash save", "git pull", "git stash pop" | 09:39 |
jibel | pitti, yup, | 09:40 |
jibel | pitti, i'm adding a cleanup and will send you the nova setup script and my changes for review | 09:41 |
pitti | jibel: would you mind pulling again, and see how far this gets? ADT_TEST_LXC=adt-utopic tests/adt-run SshRunner | 10:03 |
pitti | jibel: I have a suspicion it'll fail because your tests/home/.ssh/id_rsa is not 0600, but 0644 or so (then ssh complains); git doesn't seem to maintain permissions that way | 10:04 |
* pitti makes it robust against that anyway | 10:04 | |
pitti | ah yes, even a simple checkout master / checkout adt-virt-ssh makes it fail | 10:04 |
pitti | jibel: so hold on | 10:05 |
pitti | jibel: ok, fixed; can you please give it a try? | 10:07 |
jibel | pitti, ADT_TEST_LXC=adt-trusty-amd64 tests/adt-run | 10:28 |
jibel | Ran 95 tests in 517.676s | 10:28 |
jibel | OK (skipped=32) | 10:28 |
pitti | jibel: splendid! You can run only the ssh ones with ADT_TEST_LXC=adt-trusty-amd64 tests/adt-run SshRunner | 10:29 |
jibel | (amongst which sshrunner) | 10:29 |
jibel | pitti, they all ran and passed | 10:29 |
jibel | correction 1 skipped | 10:30 |
pitti | jibel: right, expected; I wrote a test case for "password but no ssh key", and then realized that this isn't implemented yet | 10:30 |
pitti | (but we should at some point, by faking an SSH_ASKPASS) | 10:31 |
pitti | jibel: open and setup are really redundant, right? I'd consolidate this to use open only to correspond to the adt-virt protocol | 10:38 |
jibel | pitti, agreed | 10:38 |
pitti | jibel: WDYT? http://paste.ubuntu.com/7710515/ | 10:41 |
pitti | jibel: that's a ssh-setup/SKELETON file | 10:41 |
jibel | pitti, LGTM | 10:42 |
pitti | jibel: pushed that and the setup/open unification | 10:43 |
jibel | pitti, I'm nearly done with nova. I'll finish it after lunch | 10:43 |
pitti | jibel: so I suppose you need to adjust nova now | 10:44 |
=== roadmr is now known as roadmr_afk | ||
jibel | pitti, right, that's why I'm nearly done essentially :) | 10:44 |
pitti | jibel: ah, so for an external lxc setup script I'll also need the "state" string that open() can generate; you already did that, right? | 10:48 |
pitti | ok, running + lunch, bbl | 10:48 |
jibel | pitti, yes, I added an 'extraopts' key that is passed back to the script | 10:48 |
jibel | pitti, the form is extraopts="key1=val1 key2=val2 [...]" | 10:49 |
jibel | pitti, then there is a function to return the value for a key | 10:49 |
pitti | jibel: ah, is that easier to handle than extraopts=-d /foo -i myid ? | 10:50 |
pitti | or just a string which is passed back verbatim as --state | 10:50 |
pitti | jibel: sure, works as well; that function needs to be in every setup script then, right? | 10:51 |
jibel | pitti, right, your approach is more generic, I'll change that | 11:33 |
elopio | ping ubuntu-qa: can somebody review this please? | 12:50 |
elopio | https://code.launchpad.net/~elopio/ubuntu-calendar-app/fix1332173-swipe_to_create_new_event/+merge/224735 | 12:50 |
elopio | la_juyis: I'm here for when you want to pair program. | 12:51 |
la_juyis | elopio, great! I think I'll be ready in ~1hr | 12:51 |
=== _salem is now known as salem_ | ||
pitti | jibel: shockingly, writing tests for "with setup script" revealed another bug :) fixed/tests pushed | 13:05 |
pitti | jibel: (I created that bug, so I'm just proving myself wrong) | 13:06 |
jibel | pitti, it's impossible, the bug must be in the test ;P | 13:06 |
=== fader_` is now known as fader | ||
jibel | pitti, http://paste.ubuntu.com/7711161/ setup script for nova + extraopts + small fixes | 13:30 |
jibel | pitti, tested on canonistack, it should work the same on hpcloud, maybe the detection of the IP will be a bit different depending on the output of nova show | 13:31 |
pitti | jibel: landed click support in master now! | 13:49 |
* pitti tests virt-ssh with click then | 13:49 | |
pitti | jibel: adt-virt-ssh branch rebased against master with click, so we can test the two together now | 13:52 |
jibel | pitti, I filed bug 1335134 to not lose the diff for nova | 13:54 |
ubot5 | bug 1335134 in autopkgtest (Ubuntu) "add nova setup script for adt-virt-ssh" [Undecided,New] https://launchpad.net/bugs/1335134 | 13:54 |
* jibel updates master | 13:54 | |
pitti | jibel: I have it in a tab, but thanks | 13:54 |
pitti | jibel: sorry, haven't looked yet, was busy with rebasing and fixing some conflicts | 13:54 |
jibel | pitti, no problem, that's why I filed a report. | 13:55 |
jibel | Yay autopkgtest 3.0 \o/ | 13:56 |
pitti | hehe | 13:56 |
pitti | One Runner to Rule Them All! | 13:56 |
jibel | pitti, I'll teach the nova runner to create snapshot so it doesn't always have to create a new instance, and that'll open the revert capability | 13:58 |
pitti | jibel: nice | 13:58 |
pitti | jibel: also, I had the idea of speeding up virt-ssh by using ssh connection sharing; I'll file that as a bug too | 13:58 |
pitti | bug 1335136 | 13:59 |
ubot5 | bug 1335136 in autopkgtest (Ubuntu) "adt-virt-ssh: Use connection sharing for speedup" [Wishlist,Triaged] https://launchpad.net/bugs/1335136 | 13:59 |
jibel | pitti, did you try click on the phone already? | 14:01 |
pitti | jibel: not yet, currently writing a test for click + ssh | 14:02 |
pitti | jibel: and BAM!, bug found; I hate test suites! :-) | 14:03 |
pitti | yeah, passing --setup-commands through the ssh runner is again that quoting issue | 14:04 |
pitti | jibel: ok, this is still broken in some ways, that's something for next week I'm afraid | 14:05 |
pitti | jibel: I need to change the way --setup-commands works, to avoid running the commands through the auxverb :( | 14:06 |
pitti | or we need a proper fix for quoting of commands containing \n | 14:08 |
pitti | jibel: http://paste.ubuntu.com/7711353/ | 14:14 |
pitti | jibel: !! | 14:14 |
pitti | now, *drumroll*, putting it all together | 14:15 |
jibel | pitti, !!!! | 14:15 |
jibel | :) | 14:15 |
pitti | jibel: http://paste.ubuntu.com/7711366/ | 14:16 |
pitti | jibel: we somehow need to export the phone user's environment to that runner | 14:16 |
pitti | jibel: i. e. in the setup script | 14:16 |
pitti | so with "su" I get the env | 14:17 |
pitti | hm, also with ssh | 14:18 |
jibel | pitti, isn't it because we do a sudo su and lose the env at some point? | 14:18 |
pitti | but presumably not throgh sudo | 14:18 |
pitti | phone, brb | 14:18 |
jibel | try with sudo -E | 14:18 |
pitti | jibel: it's probably not on that side, we need su -l or so | 14:38 |
pitti | root@ubuntu-phablet:~# su -c 'env|grep UPSTART' phablet | 14:39 |
pitti | root@ubuntu-phablet:~# su -l -c 'env|grep UPSTART' phablet | 14:39 |
pitti | UPSTART_SESSION=unix:abstract=/com/ubuntu/upstart-session/32011/1312 | 14:39 |
jibel | pitti, ah it's the same -l that solved the cd . permission denied. it makes sense | 15:03 |
pitti | jibel: it just breaks other stuff :/ | 15:03 |
balloons | ubuntu-qa, can someone review these pep8 fixes. Should be really quick check; https://code.launchpad.net/~nskaggs/ubuntu-terminal-app/new-pep8-fixes/+merge/224856, https://code.launchpad.net/~nskaggs/ubuntu-rssreader-app/new-pep8-fixes/+merge/224855, https://code.launchpad.net/~nskaggs/ubuntu-clock-app/new-pep8-fixes/+merge/224853, https://code.launchpad.net/~nskaggs/dropping-letters/new-pep8-fixes/+merge/224857 | 15:18 |
elopio | balloons: me _o/ choose me!!! | 15:18 |
balloons | elopio, :-) I think that's the rest of the core apps that needed it | 15:18 |
la_juyis | balloons, i can do at least some | 15:18 |
* elopio cedes the job to la_juyis | 15:19 | |
elopio | go and earn some nice karma. | 15:19 |
la_juyis | elopio, :D | 15:20 |
elopio | balloons: can you review this one? https://code.launchpad.net/~elopio/ubuntu-calendar-app/fix1332173-swipe_to_create_new_event/+merge/224735 | 15:21 |
balloons | elopio, oO.. is this what I think it is? ;-) | 15:23 |
elopio | balloons: probably not. I went a little crazier on this one and it got big. | 15:25 |
* balloons sees an EventBubble class.. Leo went ham! | 15:27 | |
jibel | pitti, last but not least a man page http://paste.ubuntu.com/7711679/ | 15:35 |
* tachyons_ is away: Away , ദൂരെ | 16:07 | |
balloons | elopio, so you don't interact with the the time and date spinners | 16:22 |
balloons | I assume because of the AP date/time issues we encountered? | 16:22 |
balloons | we should link that bug.. let me find it | 16:23 |
elopio | balloons: I didn't do it in this branch because the original test wasn't doing it. | 16:26 |
elopio | I haven't even tried yet. | 16:27 |
balloons | elopio, yea, that code must have been completely gutted. I had code to do it, I clearly didn't leave it as a comment I guess. It used the new helper | 16:27 |
elopio | balloons: what's next after this branch should be to add an event for the future. | 16:29 |
elopio | I'm not sure if I will be able to do it next week though. | 16:29 |
balloons | elopio, https://bugs.launchpad.net/autopilot/+bug/1328600 | 16:29 |
ubot5 | Ubuntu bug 1328600 in Autopilot "test_new_event autopilot test fails on device (r315)" [High,Confirmed] | 16:29 |
elopio | balloons: oh, right, that one. | 16:29 |
elopio | we need the autopilot guys to fix it first. | 16:30 |
balloons | elopio, I would simply note the bug and give a small comment about setting times | 16:30 |
balloons | elopio, yes, but that way we don't forget and try and add the time selection | 16:30 |
elopio | balloons: in progress... | 16:30 |
elopio | balloons: pushed. | 16:33 |
balloons | elopio, shouldn't class DayView inherit the Page class? | 16:38 |
balloons | actually I suppose many pages might make sense to import from page.. I realize they don't use the methods atm... Also should clean up the imports, still using emulators as toolkit_emulators | 16:42 |
elopio | balloons: I think that page class should be removed. | 16:51 |
elopio | you shouldn't just drag a page up or down. | 16:51 |
elopio | you should drag it to make something visible, and I think it's clearer if you call that something.swipe_into_view() | 16:51 |
balloons | elopio, that's also an option | 16:51 |
elopio | so, the Page object will be eventually without methods, and then it makes no sense to add it to the class hierarchy. | 16:52 |
elopio | we do need a page helper. But I think it will have something like a reference to the previous page, and a go_back method that opens it. | 16:52 |
balloons | I'm happy with cutting it.. but if it stays you should use it | 16:53 |
elopio | balloons: oh, I didn't notice it's not longer in use. | 16:53 |
elopio | so if you are happy with what I said above, I'll remove it. | 16:54 |
balloons | elopio, yes when you mentioned the methods I was thinking why they existed. | 16:54 |
balloons | elopio, it's used in test_yearview I see at least | 16:55 |
elopio | balloons: humph, I can't find where that self.page is defined | 16:57 |
elopio | balloons: I think that means test_yearview will fail in 4 days. | 16:59 |
balloons | elopio, :-) | 16:59 |
balloons | well we caught it in time | 16:59 |
elopio | balloons: I'll report the bug. That can be done with swipe_into_view if the element is loaded, of with flickable.click_element if we have to scroll for it to be created. | 17:03 |
=== roadmr_afk is now known as roadmr | ||
elopio | https://bugs.launchpad.net/ubuntu-calendar-app/+bug/1335241 | 17:08 |
ubot5 | Ubuntu bug 1335241 in Ubuntu Calendar App "test_yearview will fail after june" [High,New] | 17:08 |
balloons | hmm.. I set my month locally to a later date | 17:10 |
balloons | still seems to be fine | 17:15 |
balloons | ahh, broke it :-) | 17:16 |
elopio | balloons: when you have time, remind me to talk about patching home and about reminders. | 17:19 |
balloons | elopio, I haven't forgotten. | 17:19 |
balloons | elopio, let's chat after lunch on it eh? | 17:20 |
balloons | elopio, I think for the merge as-is, I would like to see Page removed. We can fix the scroll issue in another mp | 17:20 |
balloons | I'll approve | 17:21 |
balloons | top approved, thanks | 17:21 |
=== chihchun is now known as chihchun_afk | ||
elopio | thanks balloons | 17:46 |
=== rsalveti_ is now known as rsalveti | ||
=== la_juyis` is now known as la_juyis | ||
=== tachyons_ is now known as tachyons | ||
=== salem_ is now known as _salem |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!