opendevreview | OpenStack Proposal Bot proposed openstack/horizon stable/wallaby: Imported Translations from Zanata https://review.opendev.org/c/openstack/horizon/+/848047 | 03:24 |
---|---|---|
opendevreview | Merged openstack/horizon stable/wallaby: Imported Translations from Zanata https://review.opendev.org/c/openstack/horizon/+/848047 | 05:30 |
opendevreview | Radomir Dopieralski proposed openstack/horizon master: Add setup and teardown to the volumes pagination tests https://review.opendev.org/c/openstack/horizon/+/847985 | 07:12 |
opendevreview | Radomir Dopieralski proposed openstack/horizon master: Add setup and teardown to the instances pagination tests https://review.opendev.org/c/openstack/horizon/+/847979 | 08:06 |
zigo | Hi there! Is there anyone working on Django 4.x compat already? | 09:03 |
zigo | Horizon is broken in Debian Unstable because of it... | 09:04 |
zigo | I get 52 failed unit tests... some probably not in Horizon though. | 09:06 |
zigo | Oh, wrong version of django-compressor... | 09:08 |
zigo | Fixing. | 09:08 |
zigo | One single failure in TemplateTagTests.test_site_branding_tag now ... | 09:12 |
zigo | Much better ! :) | 09:13 |
vishalmanchanda | zigo: hello, as of now noone is working on djagno 4.x compatibliy with horizon. | 09:20 |
vishalmanchanda | zigo: if you already fixed or working on some issue to make it work with djang04.x | 09:21 |
vishalmanchanda | zigo: we will be happy to review it. | 09:21 |
zigo | vishalmanchanda: I wish I had plenty of time to do the upstream work on all the 500+ Python package I maintain in Debian for OpenStack, though realistically, this is impossible ... :P | 09:22 |
zigo | As much as I can see, it looks like we mostly need django-compressor 4.x, which I did, and Horizon (yoga) seems happy with it. | 09:22 |
zigo | I'll tell you in a bit when I go further in unit testing. | 09:23 |
vishalmanchanda | zigo: nice. | 09:23 |
vishalmanchanda | zigo: Please open a bug if face any issues and add more details as possible and I will check how can we help you. | 09:24 |
zigo | vishalmanchanda: Do you know how I can black list a single unit test with pytest? | 09:24 |
zigo | I'm currently doing: | 09:24 |
zigo | python3 -m coverage run -a -m pytest horizon/test/ -n $$(nproc --all) -v --ds=horizon.test.settings -m "not selenium and not integration" | 09:24 |
zigo | (though it'd be nice if one day, I could find the time to run the Selenium tests too...) | 09:25 |
zigo | I tried adding "and not test_site_branding_tag" but pytest still runs it... :/ | 09:26 |
zigo | Oh, probably my mistake is -m vs -k. | 09:26 |
zigo | Trying ... | 09:26 |
vishalmanchanda | zigo: for skiping test please see https://github.com/openstack/horizon/blob/master/openstack_dashboard/test/integration_tests/tests/test_floatingips.py#L45 | 09:27 |
zigo | It's easier for me to just fix the command line in d/rules (no need to patch your upstream code this way). | 09:28 |
zigo | vishalmanchanda: https://bugs.launchpad.net/horizon/+bug/1980214 | 09:36 |
zigo | 5 times AttributeError: module 'django.utils.http' has no attribute 'is_safe_url' in openstack_auth | 09:38 |
zigo | Once AttributeError: module 'django.middleware.csrf' has no attribute 'REASON_BAD_TOKEN' | 09:38 |
zigo | These it's easy for me to fix, I'll try. | 09:38 |
zigo | vishalmanchanda: How can I test the in-use Django version in Python ? | 09:43 |
vishalmanchanda | zigo: sorry I didn't get your question. | 09:46 |
zigo | vishalmanchanda: I'd like to do something like this: | 09:46 |
zigo | if django >= 4: | 09:46 |
zigo | What's the syntax ? | 09:46 |
rdopiera | django.__version__ | 09:48 |
rdopiera | but it's a string, so >= won't work the way you want | 09:48 |
zigo | I can use from distutils.version import LooseVersion right ? | 09:51 |
zigo | distutils will be removed in Python 3.12, and LooseVersion is now in its own module. Is it ok to use that? | 09:53 |
zigo | try: from distutils.version import LooseVersion except: from looseversion import LooseVersion ? | 09:54 |
zigo | vishalmanchanda: rdopiera: Your thought? Can I do that, or do you recommend a better way to check for Django's version? | 09:57 |
rdopiera | no idea what that looseversion thing is | 09:58 |
rdopiera | whatever works for you | 09:58 |
zigo | https://pypi.org/project/looseversion/ | 09:59 |
amotoki | zigo: I haven't replied to your emaial, but as far as I checked with Django 4.0 yesterday, I saw only two patterns of failures. the one is in openstack_auth and the other looks related to django translation proxy. | 11:02 |
amotoki | zigo: the first one apparently depends on django internal and I think what we can do is just to catch up with Django changes (I haven't identified all new attributes in csrf reasons should be added or not.) | 11:03 |
amotoki | zigo: regarding checking Django version, "django.VERSION >= (4, 0)" should work (after "import django") | 11:03 |
amotoki | django.VERSION returns a tuple of the version fields, so you can esaily compare using a tuple. | 11:04 |
opendevreview | Merged openstack/horizon master: Fix Create Container issues https://review.opendev.org/c/openstack/horizon/+/846424 | 11:24 |
zigo | amotoki: Thanks. | 11:34 |
zigo | I believe I solved the csrf_reason thingy... | 11:35 |
amotoki | zigo: nice | 11:38 |
amotoki | zigo: btw, does debian unstable always track the latest django version including non-LTS? | 11:38 |
zigo | amotoki: Usually, Chris Lamb (our former Debian leader who's maintaining the Django package) tracks things in Experimental, and waits before uploading to Unstable. | 11:39 |
zigo | I'm not sure if he does the non-LTS versions, but in Unstable it's always LTS. | 11:39 |
zigo | I'll add some autopkgtest in Horizon, so he gets less reasons to break me with his uploads (ie: Experimental pseudo-excuse page will show failures in Horizon, so he can check that before uploading to Unstable). | 11:40 |
zigo | BTW, looks like Horizon is doing // unit test testing now, right? | 11:40 |
zigo | It used to be not possible ... | 11:41 |
rdopiera | what? | 11:41 |
amotoki | zigo: what do you mean? i cannot understand what "doing // unit test testing". // ???? | 11:41 |
zigo | parallel | 11:42 |
zigo | sorry ... | 11:42 |
rdopiera | unit tests can be parallel, the other kinds of tests can't | 11:42 |
amotoki | horizon UTs consist of four UTs. IIRC only plugin UT cannot run in parallel. Others can be parallel. | 11:44 |
zigo | That's still a very good improvement. | 11:44 |
rdopiera | oh, I didn't look into plugin unit tests, sorry | 11:46 |
opendevreview | Thomas Goirand proposed openstack/horizon master: Django 4.x: fix csrf reasond list https://review.opendev.org/c/openstack/horizon/+/848100 | 11:59 |
zigo | There you go. | 12:00 |
zigo | Should I just except that single reason that changed, or is my list approach prefered? | 12:00 |
opendevreview | Merged openstack/horizon stable/yoga: Make existing empty metadata properties optional https://review.opendev.org/c/openstack/horizon/+/845834 | 12:00 |
zigo | https://docs.djangoproject.com/en/4.0/releases/4.0/ says: | 12:04 |
zigo | django.utils.http.is_safe_url() is removed. | 12:04 |
zigo | What should be used instead? | 12:04 |
zigo | I hate Django release notes. They say "foo is removed". They don't say "foo is removed, please use bar instead" ... :( | 12:04 |
zigo | Oh, found it: | 12:06 |
zigo | To avoid possible confusion as to effective scope, the private internal utility ``is_safe_url()`` is renamed to ``url_has_allowed_host_and_scheme()``. | 12:06 |
* zigo tries replacing it | 12:06 | |
zigo | Looks like that was it ! :) | 12:29 |
zigo | I'm sending that 2nd patch. | 12:29 |
zigo | Oh, it's upstreamed already, great ! :) | 12:31 |
amotoki | zigo: most usage of is_safe_url() usage was cleanup in commit 00def145de2498816255959d1aefd285cd5583bf (as part of yoga), but... | 12:39 |
amotoki | zigo: I forgot to clean up one remaining one. it was clean up in commit 33efe3179d11316776e9efaa0aaccc505f771ac7 which is only in zed. | 12:40 |
amotoki | zigo: do you see more missing usage of isa_safe_url()? | 12:40 |
zigo | amotoki: No, that's fine, thanks. I uploaded horizon 22.1.0-2 to unstable with the fixes. | 14:17 |
vishalmanchanda | #startmeeting horizon | 15:00 |
opendevmeet | Meeting started Wed Jun 29 15:00:07 2022 UTC and is due to finish in 60 minutes. The chair is vishalmanchanda. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
opendevmeet | The meeting name has been set to 'horizon' | 15:00 |
vishalmanchanda | hello, anyone around for horizon weekly meeting. | 15:02 |
rdopiera | o/ | 15:02 |
rdopiera | Tatiana is on pto | 15:02 |
vishalmanchanda | rdopiera: ok | 15:02 |
vishalmanchanda | then I guess only two of us. | 15:03 |
vishalmanchanda | ok let's start the meeting. | 15:03 |
vishalmanchanda | #topic Notices | 15:03 |
vishalmanchanda | A casual reminder about schedule. | 15:03 |
vishalmanchanda | This week is R-14 week | 15:03 |
vishalmanchanda | For more info about schedule please refer https://releases.openstack.org/zed/schedule.html | 15:04 |
vishalmanchanda | agenda of meeting can be found here https://etherpad.opendev.org/p/horizon-release-priorities#L39 | 15:04 |
vishalmanchanda | moving to next announcement | 15:05 |
vishalmanchanda | release team purposed a patch to cut a new release of horizon for stable/wallaby branch. | 15:06 |
vishalmanchanda | As of now, there is no open patch for horizon stable/wallaby branch. | 15:06 |
vishalmanchanda | https://review.opendev.org/c/openstack/releases/+/847908 | 15:06 |
vishalmanchanda | that's all announcment from my side for this week. | 15:07 |
vishalmanchanda | moving to next topic | 15:07 |
vishalmanchanda | #topic Release priorities | 15:07 |
vishalmanchanda | Xstatic angular migartion to 1.8.2 | 15:07 |
vishalmanchanda | there are many horizon plugins in which npm job is failing with error | 15:08 |
vishalmanchanda | https://zuul.opendev.org/t/openstack/build/680a0fe17bcb468db43aa17d4e774c7f/log/job-output.txt#3308-3314 | 15:09 |
vishalmanchanda | rdopiera: any idea why job is failing with syntax error in horizon plugins? | 15:10 |
rdopiera | looks like it's using ` instead of ' to close the string | 15:13 |
rdopiera | I can make a patch for that | 15:13 |
rdopiera | no, wait, maybe that's intentional, I don't know js/angular enough to tell | 15:14 |
rdopiera | ah, it's a template literal | 15:15 |
rdopiera | https://betterprogramming.pub/javascript-how-backticks-work-de269e0fb8ba | 15:15 |
rdopiera | maybe that job is running with too old version of javascript to understand it? | 15:15 |
vishalmanchanda | I am just wondering why we see this error in horizon plugins side but not in horizon | 15:17 |
rdopiera | node version? | 15:18 |
vishalmanchanda | 14 I guess atleast in case of octavia-dashboard and senlin-dashboard | 15:19 |
rdopiera | I see they are using PhantomJS 2.1.1 to run their tests | 15:19 |
rdopiera | we use firefox... | 15:20 |
rdopiera | we can rewrite that to not use that syntax | 15:21 |
vishalmanchanda | rdopiera: +1. | 15:21 |
vishalmanchanda | that's all from my side for this week. | 15:22 |
vishalmanchanda | I purposed patches for few horizon plugins to make it work with angular 1.8.2.2 | 15:23 |
vishalmanchanda | If you have time, please take a look. | 15:23 |
vishalmanchanda | You can find more details here https://etherpad.opendev.org/p/Fix_Horizon_Plugins_With_Angularjs_v1.8.2.2 | 15:23 |
rdopiera | will do | 15:24 |
vishalmanchanda | rdopiera: thanks. | 15:24 |
rdopiera | I'm still working on making the integration tests a bit more reliable | 15:24 |
vishalmanchanda | rdopiera: yeah i saw your patch but didn't look at it yet. | 15:24 |
rdopiera | I have three patches up that add setup/teardown and deleting of any leftover data -- even just running the tests for them cleared our gate | 15:25 |
rdopiera | we don't even need to merge them | 15:25 |
rdopiera | we can just do a recheck on them each time the gate is stuck ;-) | 15:25 |
vishalmanchanda | rdopiera: ohh cool | 15:25 |
rdopiera | though of course I would prefer to merge them eventually | 15:26 |
rdopiera | there are still some random failures that I can't quite diagnoze | 15:26 |
rdopiera | I get them on my laptop too, but only with firefox | 15:26 |
vishalmanchanda | rdopiera: sure, let me know once they are ready to review. | 15:26 |
rdopiera | all except the volumes should be ready | 15:27 |
rdopiera | the failures are random | 15:27 |
vishalmanchanda | ok | 15:27 |
vishalmanchanda | you use chrome to run integration test in your local env? | 15:28 |
rdopiera | I have also discovered that though you can set things like the public network name or the demo project name in the config file for the tests, most of those values are ignored as they are hardcoded in the tests | 15:28 |
rdopiera | vishalmanchanda: I have also proposed a patch to add an option to run the tests using chromium or chrome | 15:28 |
vishalmanchanda | rdopiera: this one https://review.opendev.org/c/openstack/horizon/+/847763 | 15:29 |
rdopiera | right | 15:29 |
rdopiera | it will use chrome if you set an environment variable | 15:29 |
rdopiera | it will use either chrome or chromium, depending on which chromedriver you have in path | 15:30 |
vishalmanchanda | rdopiera: ok will try your patch. | 15:30 |
rdopiera | you can also try the phantomjs | 15:31 |
rdopiera | I didn't, because I don't have it packaged for fedora | 15:31 |
vishalmanchanda | rdopiera: one question on this, Doesn't it going to use the browser we pass in karma.conf.js | 15:31 |
vishalmanchanda | https://github.com/openstack/horizon/blob/master/horizon/karma.conf.js#L123 | 15:32 |
rdopiera | I think that just tells zuul what to install in the vm | 15:32 |
rdopiera | horizon has firefox hardcoded | 15:32 |
vishalmanchanda | rdopiera: yes | 15:33 |
vishalmanchanda | rdopiera: ok thanks for the detailed info. will test your patch asap. | 15:34 |
vishalmanchanda | moving to next topic | 15:34 |
vishalmanchanda | #topic open-discussion | 15:34 |
vishalmanchanda | rdopiera: Do you have any other topic to discuss? | 15:34 |
rdopiera | not really, I'm focusing on the integration tests right now | 15:35 |
vishalmanchanda | ok then let's end this meeting. | 15:35 |
rdopiera | I will probably not be able to upgrade jquery | 15:35 |
rdopiera | in this release | 15:35 |
vishalmanchanda | rdopiera: no worries, I will take a look once angularjs migration is completed. | 15:36 |
vishalmanchanda | thanks for joining. | 15:36 |
vishalmanchanda | See you next week. | 15:36 |
rdopiera | thanks, bye | 15:37 |
vishalmanchanda | #endmeeting | 15:37 |
opendevmeet | Meeting ended Wed Jun 29 15:37:15 2022 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:37 |
opendevmeet | Minutes: https://meetings.opendev.org/meetings/horizon/2022/horizon.2022-06-29-15.00.html | 15:37 |
opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/horizon/2022/horizon.2022-06-29-15.00.txt | 15:37 |
opendevmeet | Log: https://meetings.opendev.org/meetings/horizon/2022/horizon.2022-06-29-15.00.log.html | 15:37 |
opendevreview | Vadym Markov proposed openstack/horizon master: Enable redirect URL check https://review.opendev.org/c/openstack/horizon/+/830410 | 16:02 |
tobias-urdin | vishalmanchanda: sorry I missed the meeting today, but I saw from last meeting that you brought up https://review.opendev.org/q/topic:keystone-endpoint-type and had some questions about it | 16:50 |
tobias-urdin | unfortunately the change mentioned in there that caused a behaviour change caused some issues for us since we really need the traffic from horizon -> keystone to go over the internal endpoint | 16:51 |
tobias-urdin | which that changed, so it kind of became a regression for us since we cannot change it and keep other endpoints to use public url | 16:51 |
vishalmanchanda | tobias-urdin: hi, dropped a comment on your patch https://review.opendev.org/c/openstack/horizon/+/844574 | 19:28 |
vishalmanchanda | tobias-urdin: Please take a look once around. | 19:28 |
opendevreview | Tobias Urdin proposed openstack/horizon master: Add OPENSTACK_KEYSTONE_ENDPOINT_TYPE config opt https://review.opendev.org/c/openstack/horizon/+/844574 | 23:51 |
tobias-urdin | ack thanks, fixed in above PS | 23:52 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!