| tomwardill | Total: 56 tests, 1 failures, 49 errors, 0 skipped in 1 minutes 48.597 seconds. | 10:05 |
|---|---|---|
| tomwardill | that went... well | 10:05 |
| SpecialK|Canon | but 6 succeeded right? | 10:14 |
| SpecialK|Canon | gotta celebrate the wins | 10:14 |
| SpecialK|Canon | `def test_foo(): pass` | 10:14 |
| SpecialK|Canon | I had a colleague who used to do that actually - said he liked knowing that however much he'd broken stuff, at least something went right | 10:15 |
| tomwardill | I've been known to break tests like that | 10:15 |
| tomwardill | if you sufficiently mangle the test setup, there's no guarantee even that will work | 10:16 |
| SpecialK|Canon | Oh absolutely | 10:16 |
| SpecialK|Canon | It's not invulnerable, but it's a nice baseline | 10:17 |
| tomwardill | lp.services.scripts.base.SilentLaunchpadScriptFailure: 1 | 10:17 |
| tomwardill | now, how can that just ... not be silent? | 10:17 |
| * tomwardill breaks out pdb | 10:17 | |
| cjwatson | Yeah, garbo likes to do that | 10:18 |
| cjwatson | Good opportunity to make your test add logger output as a test detail | 10:19 |
| cjwatson | (Though TestGarbo.setUp is supposed to do that ...) | 10:20 |
| === SpecialK|Canon changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | Firefighting: - | Critical bugs: <98 | Support and spam reporting: https://answers.launchpad.net/launchpad | ||
| tomwardill | cjwatson: does using <allow> in a configure.zcml allow write? | 14:51 |
| tomwardill | getting a ForbiddenAttribute trying to set an attribute on OCIFile | 14:52 |
| pappacena | maybe you need a <require> with set_schema? | 14:55 |
| tomwardill | yeah,t hat's what I was wondering | 14:55 |
| tomwardill | just slightly surprising that an explicit <allow> doesn't... allow | 14:56 |
| tomwardill | which makes me think I've got something else wrong somewhere | 14:56 |
| cjwatson | tomwardill: <allow> is only for gets, not sets | 15:02 |
| cjwatson | As pappacena says, you need <require set_attribute=> or <require set_schema=> for sets | 15:03 |
| cjwatson | Honestly I think removeSecurityProxy makes more sense than completely open set permissions | 15:04 |
| tomwardill | hmm, yeah, that would work | 15:04 |
| * tomwardill does that | 15:04 | |
| cjwatson | OK, I think my buildmaster tests all pass again. Next step is to make them make sense | 15:06 |
| * tomwardill knows that feeling | 15:07 | |
| tomwardill | thought i was done,t hen ran some more tests | 15:07 |
| tomwardill | SURPRISE NOPE! | 15:07 |
| tomwardill | mock_oci_datetime.noew = lambda: now | 15:13 |
| tomwardill | well, found a bug | 15:13 |
| pappacena | mock_oci_datetime.meow = lambda: cat | 15:14 |
| tomwardill | hah | 15:14 |
| tomwardill | there's a matching DB patch for this, but could do with someone checking I've not missed anything/done something silly in: https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/384121 | 15:40 |
| tomwardill | garbage collection for OCIFile | 15:40 |
| SpecialK|Canon | silly q - why did you move IOCIFS? | 15:43 |
| tomwardill | so it was in a more appropriate place in the file (next to IOCIFile) | 15:44 |
| tomwardill | as I had a 'I know I wrote this,w here the heck is the code' moment | 15:44 |
| tomwardill | (blame past tom for putting it in a weird place to start with) | 15:44 |
| cjwatson | I don't see anything obviously wrong at a first skim-read, but my head is mostly full of buildd-manager at the moment so I'll give it a proper look later. Other people should feel free to review too | 15:58 |
| tomwardill | "delivery area not accessible, follow up instructions on the way" no piano today then :( | 16:00 |
| SpecialK|Canon | tomwardill: do these tests share state? | 16:12 |
| SpecialK|Canon | ...I guess yes, shared testdb across tests | 16:12 |
| SpecialK|Canon | no, I just can't count, ignore me | 16:12 |
| tomwardill | SpecialK|Canon: no, the db is rolled back | 16:12 |
| tomwardill | jaj | 16:12 |
| tomwardill | *hah | 16:12 |
| tomwardill | matching DB patch: https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/384129 | 16:16 |
| cjwatson | Right, we roll back the db between every build, either by transaction.abort() if there've been no commits, or by restoring a fresh DB from a template | 16:16 |
| cjwatson | *between every test | 16:16 |
| cjwatson | And attack any case of state sharing between tests with fire | 16:17 |
| SpecialK|Canon | nod, cheers, I've now regained my ability to count/read and noticed the second makeOCIFile call... | 16:17 |
| cjwatson | SpecialK|Canon: https://code.launchpad.net/~cjwatson/launchpad-buildd/+git/launchpad-buildd/+merge/384140 | 19:30 |
| cjwatson | Very rough, but gets it out of my head | 19:31 |
| pappacena | {"errors":[{"code":"MANIFEST_INVALID","message":"manifest invalid","detail":{}}]} | 20:13 |
| pappacena | Thanks for such detailed message, DockerHub. That helped a lot... :unamused: | 20:13 |
| pappacena | https://hub.docker.com/repository/docker/pappacena/test-image 🎉 | 21:54 |
| pappacena | It kind of worked! :-) | 21:54 |
| pappacena | https://usercontent.irccloud-cdn.com/file/OBSogPP9/its-something.png | 21:55 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
