09:02:42 #startmeeting Contributors Summit Hackathon 09:02:42 Meeting started Mon Mar 30 09:02:42 2020 UTC. 09:02:42 This meeting is logged and archived in a public location. 09:02:42 The chair is gundalow. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:02:42 Useful Commands: #action #agreed #halp #info #idea #link #topic. 09:02:42 The meeting name has been set to 'contributors_summit_hackathon' 09:02:53 yay \o/ 09:03:05 #chair cybette phips felixfontein resmo dmellado 09:03:05 Current chairs: cybette dmellado felixfontein gundalow phips resmo 09:03:17 I'm be bouncing online/offline a bit this morning 09:03:40 Feel free to `#action We should do foo` for anything that we need to update in docs, examples, CI, etc, etc 09:04:24 gundalow: do you mind if I merge #66? 09:04:46 felixfontein: mind putting the whole link? 09:04:59 folks, I'm feeling a bit unwell today, but will be around to help if needed ;) 09:05:26 dmellado: https://github.com/ansible-collections/community.general/pull/66 09:06:00 felixfontein: merge at will 09:06:28 Morning all 09:06:33 hi all 09:06:40 gundalow: will do, thanks! 09:07:33 #action gundalow add `settings.yml` to all Community repos 09:07:35 hey shaps 09:07:38 and everybod else 09:08:06 hi sshnaidm! 09:08:31 gundalow, what is settings.yml? 09:13:45 I'm going to refactor all the github modules today! \o/ 09:16:29 cool! 09:16:48 \o/ 09:16:50 morning 09:17:03 sshnaidm: Another sausage making detail I guess ;) But I am also interested in learningabout it. 09:30:21 morning 09:32:25 morning cybette 09:35:47 when importing from module_utils now, should I do ..module_utils or ansible.collections.community.general.module_utils ? 09:40:57 ok, I'm fully around now! 09:41:13 shaps: I'm not sure whether the relative import will work with Ansible 2.9 09:41:55 right, so I'll just stick with the full path, that works fine on 2.9 right? 09:42:29 I think so. maybe relative imports also work in 2.9, I haven't tested that 09:46:43 I would like to not use full path. I wonder if e.g. doc_fragments would work with relative path 09:46:50 hmm, every PR in community.general with a changelog fragment will trigger a *full* CI run. that's really annoying. 09:47:16 resmo: I don't think doc fragments will work with relative paths, since that's not a Python import 09:48:20 felixfontein: yes I know, but haven't tested it yet. 09:51:42 what bother me is, collections are not easy to "fork" (e.g. to implement a quick fix unless the fix is in upstream collection). as a result, there might be many collections for the same purpose because they can't be merged back. I hope not but that is my concern. 09:52:52 I know what you mean 09:52:56 Morning all 09:53:05 morning gwmngilfen! 09:53:17 o/ gwmngilfen 09:59:50 hi gwmngilfen 09:59:50 also, regarding collections, there's a thing that you might need to consider, which is its requirements 09:59:58 besides the reqs con collections 10:00:33 we're thinking about how to handle reqs when a collection required x>=1 but another dependant would need x<1 and such ;) 10:00:46 basically we were thinking about using a global constraints file 10:00:59 but there are still so much stuff to consider 10:01:25 andersson007_: can we merge https://github.com/ansible-collections/community.general/pull/58 ? 10:01:51 dmellado: best strategy: don't have reqs ;) 10:03:04 So what have I missed? General hacking, or are we gathering on topics? 10:03:23 gwmngilfen: I think right now everyone has their own topics 10:04:01 yeah, mostly general hacking 10:04:15 felixfontein: i'd better wait for bmalynovych a couple of days, he's a WG leader 10:04:24 #action every PR in community.general with a changelog fragment will trigger a *full* CI run. that's really annoying. 10:04:44 Super. Well if anyone wants to work on data things let me know :) 10:04:47 andersson007_: ok, then I'll add a changelogs/ directory manually 10:05:01 ok 10:05:11 gundalow: I prepared https://github.com/ansible/ansible/pull/68550 and https://github.com/ansible-collections/community.general/pull/71 so that we can turn that off temporarily :) 10:05:30 dmellado: https://github.com/ansible-collections/grafana/blob/master/.github/settings.yml is configuration for https://github.com/probot/settings to enforce various GitHub settings such as branch protections 10:05:36 sshnaidm https://github.com/ansible-collections/grafana/blob/master/.github/settings.yml is configuration for https://github.com/probot/settings to enforce various GitHub settings such as branch protections 10:05:46 sorry, forgot who asked, ignore that dmellado 10:06:17 resmo: would you be able to expand on "what bother me is, collections are not easy to "fork" (e.g. to implement a quick fix unless the fix is in upstream collection)." 10:07:13 #chair dmellado andersson007_ gwmngilfen shaps resmo 10:07:13 Current chairs: andersson007_ cybette dmellado felixfontein gundalow gwmngilfen phips resmo shaps 10:07:14 felixfontein: lol, I wish I would be able to do that xD 10:07:17 no worries gundalow 10:08:36 gundalow: mind if I merge https://github.com/ansible-collections/community.general/pull/71 if tests pass, and then start telling people to add changelog fragments to their PRs? 10:08:48 is there a way I can run sanity/integration locally without having to have community.general in ansible_collections/community/general? 10:09:01 felixfontein: haha, was just looking for a PR that triggeres changelogs, and I see you have a likely fix in https://github.com/felixfontein/ansible/commit/2b540f4be436195f4aa1e7283c20582f6fcb7201 10:09:05 also, not sure where that ansible_collections/ should be in the first place? 10:10:23 shaps: I have community.general checked out into `~/git/work/ansible_collections/community/general` 10:10:25 shaps: not that I know. it should probably be in a path where ANSIBLE_COLLECTIONS_PATHS points to 10:10:58 gundalow: it is a minimal fix ;) but it should work for changelog fragments 10:11:00 `ansible_collections/community/general` is so ansible-test knows what to do 10:11:48 felixfontein: yup, feel free to merge #71 if CI looks good 10:12:17 mattclay: is https://github.com/felixfontein/ansible/commit/2b540f4be436195f4aa1e7283c20582f6fcb7201 the right fix so changelogs don't trigger a full CI run? 10:12:41 #action revert https://github.com/ansible-collections/community.general/pull/71 once changelog detection has been fixed 10:12:51 I'll test it once that's merged, by creating a PR which adds changelogs/fragments/.empty :) 10:12:59 mattclay: correction, PR is https://github.com/ansible/ansible/pull/68550 10:13:58 all discussions about collection development are only in this channel (not in devel)? 10:14:33 andersson007_: in general no, but right now it's easier to ask here I guess 10:14:38 gundalow: yeah, my OCD kicks in for that ansible_collections directory heh 10:15:11 But I guess I’ll just have to do what ansible-test tells me to 10:15:27 shaps: ansible itself needs that as well 10:15:45 (which I guess is the main reason why ansible-test enforces it, since it internally uses ansible code all over the place) 10:16:28 So every ANSIBLE_COLLECTIONS_PATH has to have ansible_collections as “top level dir”? 10:17:06 I think so 10:17:08 yuck, none of the other *_PATH's work like that, do they? 10:17:17 No 10:17:23 shaps: also it is PATHS with a trailing S 10:17:29 nor ANSIBLE_LIBRARY 10:17:43 * cyberpear still needs to open a PR to alias the signular PATH 10:17:45 felixfontein: yes typo 10:17:57 cyberpear: please do that :) 10:18:03 shaps: just wanted to make sure you use the right variable :) 10:18:15 @shaps: yes, `ansible_collections` is mandatory 10:18:48 Yep, I guess it’s because of the “ansible_collections.” namespace 10:19:21 yes 10:19:47 this is how the collections loader works 10:21:30 cool, thanks 10:22:16 i'm going to spend a bit of time tweaking the Contributor Summit survey from Atlanta for all the lovely participants yesterday - if anyone would like to suggest questions that should be on there, fire away :) 10:22:59 cybette: is the recording already available somewhere? 10:25:38 regarding changelogs was a decision already made to use fragments? 10:26:25 gwmngilfen: Excellent. I wonder if it would be worth asking more questions around the virtual side (did the video work for you. Would you like like captioning). Did you use Video and IRC. Is English your second language (maybe useful to help weight the understaning/ability to flow) 10:26:49 zbr: AFAIK still in RFC 10:26:51 i am asking because I find it as a PITA, especially for occasional contributors, many of them will forget to bother to add the fragment, and the PR will endup not being merged.. 10:27:26 there will be reminders 10:27:35 also it is a PITA when there is no changelog 10:28:16 if you do 2-3+ contribs per week it is no longer an issue, as it becaomes a habit. but for others is not like that and lowers the success rate (merge). 10:28:49 do you have a better alternative? (and no, using the git commit history is not an alternative.) 10:29:06 I am not proposing no-change log either. 10:29:34 gundalow for sure, we need to strip out all physical attendance stuff. Tooling questions make sense. Will prototype something shortly 10:29:38 felixfontein: read about https://github.com/release-drafter/release-drafter and how it works. 10:29:57 i agree that git-changelog is not user-friendly, sometimes is not even developer friendly. 10:30:02 (Although my laptop just froze, yay) 10:30:13 release notes != git changelog, we all know that. 10:30:14 zbr: that tool uses git history, right? 10:30:28 zbr: nice hint 10:30:29 we do not want "attempt 666 to fix ci" 10:30:39 acozine: samccann FYI we are talking about changelogs 10:31:06 resmo: videos are being processed, we will ping here once they are on YouTube 10:31:14 gundalow: cool 10:31:14 I think right now we're repeating the same discussion as last time 10:33:51 what are the options about changelogs? 10:34:10 gundalow: is it possible to turn it off that shippable does a full (?) CI run after every PR merged? 10:34:11 zbr: IMHO changelog fragments would be recommended to get embedded in ACD changelogs but the collection can make any form of changelog (e.g. openstack uses reno release notes) as long it is linkable. 10:34:27 andersson007_: Afternoon. See https://github.com/ansible-collections/overview/issues/18 for a summary of where we are upto 10:34:43 felixfontein: hum, is that because we are burning CI capacity? 10:34:48 * gundalow looks at Shippable config 10:34:52 gundalow: afternoon, thanks 10:35:01 resmo: zbr: collections can also use their git history if they want. but it's not an option for community.general. 10:35:28 gundalow: it is blocking tests for other PRs for quite some time 10:35:45 * resmo -> lunch 10:35:50 resmo: en guete! 10:36:08 felixfontein: äbeso ;) 10:36:14 resmo: merci! 10:36:56 andersson007_: https://github.com/ansible-collections/overview/issues/18 10:37:04 oh 10:37:09 I just see gundalow was quicker :) 10:37:19 felixfontein: yep, thanks! 10:37:34 felixfontein: lets see if https://github.com/ansible-collections/community.general/pull/73 works 10:38:11 gundalow: can we just merge it? otherwise we'll only find out in one hour or so ;) 10:38:47 I ll drop mine too :) https://github.com/ansible-collections/community.general/pull/57 felixfontein is looking already in some errors but I have a sanity issue related to symlink for ansible-doc. Not sure how to fix that though. :( 10:39:48 felixfontein: Do you have powers to see https://app.shippable.com/github/ansible-collections/community.general/dashboard I don't see a new run on `master` so I *think* that worked 10:39:55 gundalow: looks like it works, at least shippable didn't start another build for master 10:40:11 * gundalow kills the existing `master` CI runs 10:40:12 gundalow: I have, also one more level up (which is also very useful) 10:41:29 gundalow: for some reason, CI for https://github.com/ansible-collections/community.general/pull/72 was also shown as master 10:42:04 felixfontein: hum, not sure what Shippable does with PRs in flight 10:42:24 it might be that the branch it shows is the branch where the PR is supposed to be merged into 10:42:48 did you just caccel it again? 10:42:51 *cancel 10:43:13 yup, I emptied out https://app.shippable.com/github/ansible-collections/community.general/dashboard then closed/opened your PR 10:43:23 ah, the close probably cancelled my rerun 10:43:47 close/reopen CI tigger was done by ansibot IIRC, so I have to restart it manually 10:43:58 or did you just do it? 10:44:05 I just did it 10:44:13 ok :) 10:44:31 it should be a short run anyway, if https://github.com/ansible/ansible/pull/68550 works 10:44:35 now, I realized that we never seen ansibot and gundalow in the same room, maybe ansibot is gundalow secret identity 10:44:43 hum, and I'm still seeing `context: master`, so I don't know anymore 10:45:09 misc: good theory! :D 10:45:22 gundalow: I guess context is the brancht he PR is supposed to be merged into, then 10:45:42 misc: I liked this tweet from jpmens yesterday https://twitter.com/jpmens/status/1244252849774759937?s=19 10:46:03 gundalow: looks like it is working! \o/ 10:46:42 misc: though technically jtanner is the voice of ansibulbot 10:47:58 gundalow: the PR's CI went through in 3 minutes! 10:48:32 \o/ 10:48:36 nice work felixfontein 10:49:12 I'll merge that PR, mention changelog fragments in the other PRs, and then after lunch I'll add changelog fragments for the already merged PRs (when necessary) and look at Federico87_'s PR 10:49:28 (if someone wants to look at why tests fail for Federico87_'s PR before that, feel free!) 10:59:58 acozine: samccann FYI for community.general we are asking people to add `changelogs/fragments` (following Ansible's standard), this is so we can merge PRs. If we decide in the future to change the format we can rewrite those files 11:00:59 * gundalow -> lunch 11:05:05 * felixfontein -> lunch 11:08:49 resmo not yet, hope to get them up today/tomorrow 11:10:10 heya. If I could get a ping when recordings are available I'd like to listen in - missed way too much of yesterday in the end :-( 11:32:00 jhawkesworth: I'll be uploading them to this channel so you can subscribe to it: https://www.youtube.com/channel/UCThD8eHpOt7YdJlzpQRiAvQ 11:32:48 Thank you cybette 11:33:37 #info Contributor Summit recordings will be uploaded to https://www.youtube.com/channel/UCThD8eHpOt7YdJlzpQRiAvQ 11:34:01 jhawkesworth: btw have you filled in this attendee survey https://forms.gle/Uv9Zdaz7aTAAwtsQ8 11:41:21 I have now cybette although I doubt I qualify as 'attending' as it turned out. I do appreciate the work that you folks have put in to make this happen though. Thank you all 11:47:45 jhawkesworth: every bit counts :) we also realise that Sunday may not work for everyone. going forward we plan to have more virtual contributor sessions like this accommodating to different time zones and days. 11:47:47 * felixfontein is back 11:47:59 * shaps food 11:48:31 +q 11:48:37 > <@cybette:matrix.org> jhawkesworth: I'll be uploading them to this channel so you can subscribe to it: https://www.youtube.com/channel/UCThD8eHpOt7YdJlzpQRiAvQ 11:48:37 * +1 11:50:35 * gwmngilfen food and childcare, back in a few hours 12:00:42 jhawkesworth: you very much do count :) 12:17:50 anybody please let me know how can I join to hackthon? is it over? I am sorry for late. 12:21:28 gundalow: AFAIK there are 3 kinds of collections: supported, partner and community. Is there a way to find out what kind of collection it is? 12:21:45 on galaxy? 12:23:31 lillianphyoe: the hackathon is still going on! right now, there's no specific procedure for anything, feel free to work on anything you're interested in 12:25:21 Federico87_: sorry for CI, that wasn't intended 12:25:40 np, is running now :) 12:25:56 Federico87_: ah, great :) 12:26:37 felixfontein thank you. have to submit ansible roles which I want to contribute on ansible galaxy ? 12:31:03 Federico87_: sanity tests now pass! \o/ 12:31:46 lillianphyoe: I didn't understood your question. are you asking where to submit such roles on galaxy? 12:32:18 yes that is! 12:32:48 integration failed due to a too early sunday morning code change. On it now :D 12:33:29 lillianphyoe: hmm maybe someone here knows more about that? I only did that once and that was a longer time ago 12:33:46 Federico87_: isn't it Monday? 12:34:10 correct, but I changed the code yesterday ;) 12:35:31 lillianphyoe: did you look at https://galaxy.ansible.com/docs/contributing/index.html , in particular https://galaxy.ansible.com/docs/contributing/creating_role.html and https://galaxy.ansible.com/docs/contributing/importing.html ? 12:42:37 Integration passed locally...let's see what happens on git 12:57:23 I got these errors for integration: 12:57:30 OPENSUSE: Could not find a module for zypper 12:57:59 FREEBSD: Could not find a module for pkgng 12:58:01 on CI? 12:58:05 yes 12:58:18 let me take a look 12:58:36 Can't run on local as I get some container error when I try to run integration on local 12:58:40 ah 12:58:49 you didn't include the requirement in meta :) 12:59:28 I added a suggestion 12:59:30 I reverse eng from other test so most probably yes :) 12:59:35 that is needed until ansible/ansible can do routing 13:00:05 (i.e. until https://github.com/ansible/ansible/pull/67684 is merged) 13:00:18 that will happen soon, and until then we need this hack :) 13:00:21 ok. do you have a specific container version for local freebsd and opensuse? 13:00:35 no, it is not related to the container version 13:01:06 the problem is that zypper and pkgng are now in community.general, and no longer in ansible/ansible itself, but the package installer detection doesn't know that yet (and so `package` tries to load them from ansible/ansible) 13:01:53 so is that the reason why I get this erro when I try to run locally 13:02:01 ERROR: Command "docker exec -i 62033a12a7d3499cb90c68ad6aead57584f4d1a0798447cba6ec9bd5fc449c75 dd of=/root/docker.sh bs=65536" returned exit status 1.>>> Standard ErrorError response from daemon: Container 62033a12a7d3499cb90c68ad6aead57584f4d1a0798447cba6ec9bd5fc449c75 is not running 13:02:10 ? 13:02:28 no, that's unrelated I think 13:08:42 felixfontein: investigating failing tests for cloudstack 13:09:34 resmo: thanks! 13:09:56 felixfontein ERROR! the role 'setup_pkg_mgr' was not found 13:10:45 can't find that role anywhere under community.general 13:10:56 Federico87_: if you have that locally, you probably need to rebase so you get changes after you branched your branch 13:11:02 on CI it ran through 13:14:52 All checks passed \0/ 13:15:08 great :) 13:15:12 Thanks for your time. 13:15:23 When potentially would be merged? 13:19:02 Federico87_: let me do some reviewing 13:19:58 thanks 13:24:35 hum, not sure what heppened then. What messages did I miss? 13:25:29 15:19:25| felixfontein> Federico87_: let me do some reviewing 13:25:30 15:20:21| Federico87_> thanks 13:25:41 (that's it) 13:26:35 misc: Thanks :) 13:30:43 gundalow: https://github.com/ansible-collections/community.general/pull/76 13:30:58 should I create a similar PR for f5? 13:33:37 felixfontein: if you could that would be great 13:33:56 ok, will do! 14:48:28 shaps: how's github looking? 15:01:07 * gundalow starts removing network stuff from community.general 15:06:03 \o/ 15:06:11 then I won't touch f5 for now (since that's network) 15:09:08 felixfontein: yup, thanks 15:11:58 @gundal 15:11:58 fail 15:11:58 gundalow: where's it going? we should add those repos to the contributor metrics 15:18:38 * geerlingguy "gundal" sounds like a character in an old western 15:29:40 gundal... with his 10gal had and six shooter hangin off a hip... 15:30:50 gundalow: not too bad, got to finish updating the webhooks module and get some tests written up 15:37:29 can someone review please CI patch for ansible podman collection? https://github.com/containers/ansible-podman-collections/pull/15 15:37:41 used a generated one as a base 15:40:03 sshnaidm: quick question—any reason why all the `run` commands are separated line per arg? e.g. https://github.com/containers/ansible-podman-collections/pull/15/files#diff-ef213732f0f3ba7e84c4fdc772bc2e38R99-R104 15:40:27 Seems like something like that reads better (and keeps file more compact) if written like `python -m pip install --user ${{ matrix.ansible-version }} virtualenv` instead 15:40:47 geerlingguy, well, no idea, it was generated like that by migrate tool, I think it tried to keep line shorter(?) 15:41:10 geerlingguy, yeah, agree 15:41:41 heh especially like this https://github.com/containers/ansible-podman-collections/pull/15/files#diff-ef213732f0f3ba7e84c4fdc772bc2e38R108-R109 15:43:12 other than that, seems correct. Maybe a few more test scenarios in the matrix than I would necessarily care about (just because that means test runs will take a long time) 15:50:08 geerlingguy, yeah, next step is reduce it.. the first version included in 3 times more I think :)) 15:50:44 hmm could we increase the number of processing jobs for community.general a bit? 15:51:15 is this gh workflow totally free for opensource? or there are some limits 15:52:04 sshnaidm: yes and maybe, though will be cheaper than shippable 15:52:40 sshnaidm: correction yes free. No limits 15:52:50 gundalow, nice 15:53:15 GitHub actions with private repos cost 15:57:09 you still have something like 2k hours (or units?) free for private repos 16:07:38 2,000 Action minutes/month for public repo https://github.com/pricing 16:09:44 and then ci turns into pumpkin.. 16:11:18 community.general will eat up those minutes quickly 16:13:31 yep 16:13:47 a day already has 1440 minutes 16:14:04 and community.general has 100 CI nodes :) 16:22:21 yeah, that's not going to last long is it? :D 16:22:46 if you're really lucky, a day maybe? ;) 17:20:37 sshnaidm: geerlingguy felixfontein shaps So according to https://github.com/organizations/ansible-collections/settings/billing (which you will not be able to see) it implies GitHub actions for free repos are unlimited 17:21:38 oh, I wonder if `2,000 Free for public repositories` is badly worded. Free for public. 2,000 for private repos 17:21:40 I wonder for which number of unlimited they start asking questions :) 17:21:49 That was my understanding too. But the pricing page seems different to me 17:22:15 I exported the billing log, GitHub Actions are in there 17:22:45 gundalow: yep that’s what I expected by looking at the billing settings (of some of my own repos) 17:22:58 Do the actions also show a price or are they 0? 17:23:44 I think we are good 17:25:39 There is a Javascript error (ERROR TypeError: "n.infoData.apb_metadata.metadata is undefined") here: https://galaxy.ansible.com/ansible/django-template, is that expected ? The "progress circle" never disappear. The same error occurs with other roles. 17:25:56 gundalow, I suppose it should be same for https://github.com/containers/ansible-podman-collections ? 17:26:51 gundalow, but anyway we have cirrus ci and zuul as well, so should be fine too.. 17:32:03 gundalow: heh, wait until we have like 10 runners going full bore 24x7, we'll see if a GitHub rep pings someone 17:33:02 I'm kind of amazed Travis CI's never throttled me back... every few weeks I push up changes that trigger 200+ builds (with 800+ individual environments) and I can actually make an impact on the graphs here :D https://www.traviscistatus.com/#system-metrics 17:33:26 geerlingguy: they used to throttle us tons, why we moved to shippable 17:33:33 at least one reason 17:52:05 felixfontein, would you like to look as well? https://github.com/containers/ansible-podman-collections/pull/15 17:57:19 sshnaidm: looking 18:00:58 sshnaidm: why do you run ansible-test unit and sanity twice, once in venv and once in docker container? isn't one of them enough? 18:02:02 felixfontein, well, I wonder if something can be different? tbh I can't think about something that will differ 18:02:26 sshnaidm: for sanity tests, nothing should differ (if it does, it is probably a bug) 18:02:41 sshnaidm: for unit tests, it should only depend on the Python version (or on requirements installed) 18:02:49 felixfontein, afaik what matters is only python version, yeah 18:03:37 felixfontein, so better to use containers only? 18:05:02 sshnaidm: I'd stick to containers (resp. ansible-test's default container) 18:05:49 also leaves more resources for integration testing :) 18:09:09 also the default container provides a full list of ansible versions for unit tests (for sanity tests the version doesn't matter) 18:09:58 felixfontein, actually I found some differences in various ansible-test versions 2.9 and devel, but I'll use the latest 18:10:16 like different results for sanity tests 18:10:35 sshnaidm: Yes, there are new/different sanity tests in devel vs stable-2.9. 18:11:09 mattclay, ack, so choosing latest version is the best approach I suppose? 18:11:16 yes, but that's not a difference in Python version or environment in which that version of ansible-test is run :) 18:12:02 sshnaidm: That depends what you're trying to do. If you have a collection that you want to work on Ansible 2.9 and the upcoming 2.10 (devel) you probably want to test against both -- but if you're short on test resources then test against devel for now. 18:14:51 mattclay, I see, so let's check both.. 18:28:55 We're currently testing kubernetes collection just against devel—there are a few things that behave different in 2.9 so it's easier to not test there for now 18:29:10 (versus trying to maintain the config to test in both successfully) 18:33:27 community.general is currently also only tested against devel; I guess we might also want to run some tests against 2.9 eventually, but right now getting tests against devel work has top priority :) 18:42:19 hum, that's a fair point 18:42:33 I hadn't really thought about 2.9 and collections 18:43:27 Well at least for now, community.general doesn't have additions/fixes that aren't in 2.9, so it's more of a moot point 18:43:39 there are already some people using the kubernetes collection with 2.9, so we might add tests for it at some point 18:44:12 geerlingguy: how do you know there are people using 2.9+kubernetes collection? 18:44:46 2 things: https://galaxy.ansible.com/community/kubernetes shows 25,000 downloads (so far) 18:45:00 and the operator-sdk project is already using it instead of builtins 18:45:15 nice 18:45:19 yer a popular guy 18:45:25 hey it ain't me 18:45:33 well, you organized it 18:46:21 Well yeah, but I've only touched maybe 2% of the actual code :P — but it does do better when you have both good code and some organization! 18:53:30 a clear objective and a plan do wonders ... 18:59:34 geerlingguy: there are already PRs for community.general which add new stuff (or maybe we already merged one? I'm not sure anymore) 19:06:33 eek! I'd hold off on that until it's stable / passing tests for everything (did we make it to that point yet?) 19:07:30 tests are mostly stable 19:07:56 there are still failures, but they don't hide anything serious anymore 19:08:16 (one needs to look at what exactly is failing for sanity and units) 19:10:14 i would include 'ignored' tests in that 19:10:24 since many of those are cause of missing deps 19:11:00 you mean skipped unit tests? 19:11:35 I don't think they were running in ansible/ansible either, since the same requirements are around 19:13:09 requirements is other collections, not talking about librarires/utilitites 19:13:21 all requirements are there 19:13:37 not for xfs 19:13:38 except where the requirements are broken (f5 modules) resp non-existant (netapp modules) 19:13:46 ah, right, that needs `mount` 19:13:47 they are actually commented out, all fo em, in galaxy.yml 19:13:58 felixfontein: one example, but many more 19:14:02 CI installs them anyway 19:14:25 tests/utils/shippable/shippable.sh is currently the "real" source of truth :) 19:15:35 * bcoca starts ddrinking to wipe that out from his memeory before it fixates 19:15:41 also, I think almost all requirements are for testing, not for the modules/... themselves 19:15:44 lol 19:15:56 for xfs, they are required at runtime 19:16:05 there's too much in flux right now; I'm waiting for the network modules to vanish (which will kill most dependencies) 19:16:09 pretty sure the crypto and netcommon are also required at runtime 19:16:19 felixfontein: but those are moving into a network collection right? 19:16:27 same for google modules (since they use the common module_utils from the 'supported' ones) 19:16:32 geerlingguy: I think so. gundalow is working on that AFAIK 19:16:43 (seems like a lot of networking ones are where there is more pain, especially for deeper testing) 19:17:08 geerlingguy: indeed! 19:17:33 felixfontein: if collection was only needed for testing there would have been a requirements.yml in tests/ with that list 19:17:40 migration script accounted for that 19:17:59 https://github.com/ansible-collections/community.general/blob/master/tests/requirements.yml 19:18:01 bcoca: there is, but currently ansible-test doesn't use it 19:18:03 ^ those are 'testing only' 19:18:09 so shippable.sh has to install them manually 19:18:11 i know, but we were asked to create it 19:18:24 anything in galaxy.yml is RUNTIME 19:18:24 eventually ansible-test will use them (I hope) 19:18:38 we generated the list, if people wish to ignore .... 19:19:18 felixfontein: Having ansible-test install collection dependencies is something I plan on working on. I just haven't had time to get to it yet. 19:19:40 file format in test/ might not be ideal though 19:19:53 Yeah, I don't know what the file format will be yet. 19:19:54 i would have made 'comments' vs sections on integration vs unit 19:20:05 https://github.com/ansible-collections/community.general/blob/master/tests/requirements.yml <= the one we already generated i mean 19:20:21 you should have required collections there already (at migration time .. once time passes ...) 19:20:34 mattclay: no worries 19:21:22 mattclay: since we really had nothing to go on, that is what we ended up with, but feel free to set your own format(s) , this hopefully will make that task easier for these collections 19:21:57 felixfontein: bcoca geerlingguy plugins/modules/network/* (and related plugins) are moving out of community.general into (new) community.network collection. It's getting there 19:22:18 gundalow++ 19:22:49 gundalow: still unrelated to the xfs issue .. but good! 19:23:07 besides network dependencies, there are only three runtime dependencies: ansible.posix (one module), community.kubernetes (kubevirt modules), google.cloud (google modules) 19:23:35 (the one module is aix_filesystem) 19:23:38 bcoca: oh, didn't see context 19:24:09 felixfontein: no, as i said, everything in galaxy is runtime dep 19:24:58 bcoca: yes, but they are almost all for network modules, and they are moving out of the repo 19:25:00 most of them are cloud/networking modules that were ignored by vendors which took ownerhsip of others + module-utils common to both 19:25:08 netowrk, yes, but still also cloud 19:25:43 and still unrelated to my issue, which is ansible.posix related 19:25:56 yes 20:02:49 so for installing from git it's enough to do like: git clone https://github.com/namespace/collection ~/.ansible/collections/ansible_collections/namespace/collection 20:02:56 right? 20:03:19 for most issues, yes 20:03:43 right now there are slight diffs between MANIFEST.json (installed package) and galaxy.yml (raw repo) 20:03:56 but should not impact runtime 20:04:26 great, thanks 07:18:39 Morning all 07:29:40 how's removing the network modules going? 07:30:11 also, should we merge https://github.com/ansible-collections/community.general/pull/76 ? 07:36:32 felixfontein: yup, I think so. I've approved the PR 07:37:29 #action we need ansible-test to validate meta/routing.yml in collections (do files exists) 07:38:42 #action validate-modules deprecation plugin test needs updating for collections to use meta/routing.yml to check if the file is deprecated. 07:40:20 mattclay: 1) what would be the right way for ansible-test to know if it should run extra tests against a collection vs ansible/ansible checkout? 2) what do you think could be the conditional to check for a collection in validate-modules? 07:41:23 gundalow: We already have a conditional in validate-modules to change tests based on whether or not a collection is being tested. 07:42:03 gundalow: cool! I've pressed the big green button :) 07:42:42 gundalow: https://github.com/ansible/ansible/blob/2b4ca69bc249b27d853085125a6cb091fd11409e/test/lib/ansible_test/_data/sanity/validate-modules/validate_modules/main.py#L2114 07:43:35 mattclay: ah excellent. Thanks 07:43:40 morning all 07:43:51 Hi andersson007_ and mattclay 07:44:15 felixfontein: I'm not sure about this error https://app.shippable.com/github/ansible-collections/community.network/runs/3/2/console 07:44:17 morning andersson007_! and hi mattclay! (probably more like afternoon/night?) 07:45:03 hi gundalow felixfontein mattclay: ansible officially supports 2.7+ and 3.5+ python versions, correct? 07:45:25 gundalow: looks like (also) an internal ansible-test error to me 07:45:41 andersson007_: Yes, and 2.6 on modules/module_utils. 07:46:00 ok, thanks 07:46:31 mattclay: btw, units/2.6 doesn't seem to run any unit test on community.general. do you know whether that is intentional? 07:47:36 gundalow: it could be that ansible-test gets confused because essentially the whole collection is added in that PR. maybe merge the PR and see how it behaves afterwards? 07:48:56 felixfontein: Sounds like an ansible-test bug. I'm checking. 07:50:35 felixfontein: OK, not an ansible-test bug. Unit tests are in the wrong place. They should be under `tests/unit/plugins/{modules,module_utils}/` in collections. 07:51:25 They are under the old path of `tests/unit/{modules,module_utils}` which is where they are in the ansible/ansible repo. The unit test dir structure is intended to match the code under test, which is why they differ between ansible and collections. 07:51:53 Morning chaps! @felixfontein if you are around, would you mind to have a look to my answers to https://github.com/ansible-collections/community.general/pull/57 so I can work on that today and hopefully push the last commit :) 07:51:53 hmm, interesting. that means that the layout in community.general is also wrong 07:51:55 ? 07:52:00 and potentially in all migrated collections 07:52:04 So that's something that was overlooked in migration. All migrated collections with unit tests will be affected. 07:52:06 gundalow: ^ 07:52:32 Well, with module/module_utils unit tests anyways. 07:52:41 Federico87_: I will have time later today 07:52:52 Thanks :) 07:53:19 I could detect that in ansible-test and add a warning for that situation I suppose, if anyone thinks it would be helpful. 07:53:42 so both modules and module_utils need to be moved under plugins 07:54:07 Yes 07:54:48 gundalow felixfontein: That ansible-test traceback looks interesting, but nothing obvious comes to mind. I'll take a closer look. 07:57:04 mattclay: gundalow: I've created PRs for community.general and community.crypto to move the unit tests 07:57:26 gundalow: I see something else in the logs for community.network that should be fixed: `Skipping matrix check on repo "ansible-collections/community.network" which is not "ansible-collections/general"` 07:57:53 mattclay: ah, thanks. I must have missed a reference. 07:59:41 gundalow: Ah, found the problem. The PR contains `tests/output/.tmp/` -- need to add that to `.gitignore` 08:01:14 gundalow: The traceback itself is still a bug that should be fixed, so feel free to open an issue for that. But the issue with the PR is the unwanted files. 08:02:43 It's caused by a symlink in the PR that is an absolute path that does not exist. 08:03:36 does anyone know which licencing types are acceptable/preferable for community.general module_utils files? 08:04:40 good question :) I'd say GPLv3+ should also be ok now. gundalow? 08:06:14 :) gundalow: you're gonna made a historical decision right now 08:06:58 s/made/make 08:11:56 Time for me to get some sleep. 08:12:32 morning 08:12:56 #info Playlist for contributor summit recordings: https://www.youtube.com/playlist?list=PL0FmYCf7ocraJzcnE3-VwVozQ0Zt7vm7z 08:13:42 mattclay: night. Thanks for your help. 08:14:20 andersson007_: licensing hasn't changed from Ansible/ansible, same rules apply 08:16:22 good night mattclay! 08:16:32 gundalow: ok, thanks 08:16:34 gundalow: so still BSD only for module_utils? 08:18:30 #action template repo needs to include gitignore and galaxy.yml (with galaxy tags) 08:22:40 andersson007_: https://docs.ansible.com/ansible/latest/dev_guide/developing_modules_in_groups.html#submitting-a-group-of-modules 08:24:15 gundalow: thanks, i've already seen it, just wasn't sure about collections 08:34:17 morning all 08:40:55 Morning all 08:43:43 morning shaps and gwmngilfen! 08:44:00 #action DOCS make it clear that (at least) Ansible managed/co-owned (what are we calling them) collections MUST be under the same licenses as ansible/ansible was 08:44:28 can someone review https://github.com/ansible-collections/community.general/pull/77 and https://github.com/ansible-collections/community.general/pull/74 ? 08:46:20 * gwmngilfen should probably spend today making various stats projects public 08:50:51 felixfontein: +1d both 08:51:06 I'm not precious about showing my code - but I start a new git repo for every report or app, so there's already 40-50 of them. Can't really spam ansible-community with all that :) 08:51:15 felixfontein: I've merged the changelog. I don't have any MacOS knowledge so I can't review the first one 08:51:27 gwmngilfen: hehe, do you think you can combine them into one repo? 08:51:43 (even if that's one repo with 50 project directories?) 08:52:15 Terrible idea, I think. I'm leaning towards having them under my username, and migrating any that start getting contributions 08:52:17 gundalow: it's mainly changing the location of the .socks file, no mac knowledge needed (I don't have much either ;) ) 08:52:41 DING DING DING Reminder to do `#action foo` to throw stuff in the todo list. I'll review that list, create/update issues and add to https://github.com/orgs/ansible-collections/projects/1 08:52:47 gwmngilfen: hehe, fair enough 08:53:24 #action DATA publish all appropriate R code and link to it from relevant places 08:53:35 :) 08:53:47 Keep me honest gundalow :) 08:54:24 Likewise :) 08:54:35 I run Gitea here in the house so I can backup nonpublic stuff, so I think it's just a matter of configuring sync 08:59:18 gundalow do I correctly recall an action to add more repos to the contributor stats? I feel that came up on Sunday 09:02:30 gundalow: here's an interesting question: https://github.com/ansible-collections/community.crypto/pull/12 should module docs examples use FQCNs or not? 09:14:11 gundalow: I'll create a PR to remove all network modules from community.general 09:19:50 felixfontein: Last time I asked this on ansible-devel, the consensus was that yes, examples should FQCNs. 09:20:19 tadeboro: I think that was us two discussing it, or was another person involved? 09:20:46 felixfontein: gundalow also chimed in and this when I considered things settled ;) 09:20:54 ah ok 09:21:07 I just remember talking about it with one person 09:21:13 memory limit reached ;) 09:21:48 Year, my brain is becoming a (very short) FIFO also ... 09:37:37 felixfontein: I've already got a draft branch to `git rm` network content from community.general. FYI it's more complex that `git rm -rf plugins/modules/network` 09:38:37 gwmngilfen: https://github.com/ansible-collections/community.network/ is to be added. I didn't add it to the GoogleSheet yet. Is it possible for it to match `lib/ansible/modules/network/` 09:39:31 morning 09:39:40 gundalow: 09:39:46 EMORETEA 09:40:15 gundalow: yeah, it's a lot more complex :) 09:40:22 gundalow: yeah that'll work, the lib/ansible/modules is implied so a match of "network" should be all you need. I can add it if you wish, but it would be good to check that it works ok for you :) 09:40:31 gundalow: if you want to finish it, feel free! 09:41:14 "yes, examples should FQCNs." ... this is true. FQCNs are a feature that should be considered the new standard, not a bug. 09:47:24 1. Modules that were in `ansible/ansible` before the mass migration can continue to be used by the short form thanks to [routing.yml](https://github.com/ansible-collections/community.general/blob/master/meta/routing.yml) 09:47:24 2. New modules (ie ones that have only existed in a Collection) will not work in the short form unless Playbook authors add the `collection: ` form 09:47:24 3. `EXAMPLES` for new modules MUST have FQCN. This means Playbook authors can copy-paste examples without having to set `collections:` 09:47:24 4. FQCN and namespaces are a feature, we want to make more people aware of them. Though we MUST NOT break 2.9 Playbooks 09:47:24 5. For smaller collections like Crypto I think it's OK to bulk update all examples to use FQCN 09:47:24 6. EXCEPTION: I'm not sure if it makes sense to bulk update EXISTING modules docs in community,general. I expect we will continue to move content out of community.general into smaller Collections (like we've just done with Network, rabbitmq, etc). Adding in FQCN may confuse that matter. 09:47:58 jtanner: does ^ sound fair (thanks for #4). Got to keep on repeating that 09:49:04 6 sounds good 09:49:10 sure, as long as we're not also implying we're going to keep expanding the list of "worked like 2.9" modules in routing.yml 09:49:46 I guess the migration to use FQCN should be gradual, but we should end up with *all* examples using FQCNs 09:50:20 I think so too. everything that is stable can start using FQCNs 09:50:42 (i.e. that won't be moved around ;) ) 09:51:40 resmo: btw, how's your moving of some of the cloud modules going? 09:51:43 I am a bit undecided whether to use fqcn or not for examples 09:51:57 resmo: for which collections? 09:53:31 7. The list of entries in `routing.yml` was frozen in time when `pre-ansible-base` was tagged. This means NO MORE modules or plugins will be added. Hence #3 09:53:41 from the users view, it would make sense. from a dev view, "forking" (as I already mentioned yesterday) is harder --> when forking is hard, real forks will pop up for collections 09:53:59 jtanner: I hadn't heard any expectation that routing.yml would be updated. Though you are right. It's good to make this explicit 09:54:19 me either, buy my spidey sense tingles 09:54:29 but* 09:55:13 hehe, likewise. Could be viewed as a backdoor way to keep pushing collection adoption down the path 09:55:55 * gundalow goes to put ^ in an issue 09:56:21 felixfontein: so, vultr and exoscale should not be ready today, cloudstack is a bit more work, 09:56:31 gundalow: jtanner: I guess incorrect entries of routing.yml can still be updated? 09:56:47 probably 09:57:10 some entries are pointing to non-existing collections 09:57:22 and that'll continue to be an issue 09:57:40 which is another reason it doesn't make total sense to map out collections in core 09:58:39 I'm not sure where else it should be 10:01:59 long term, collections need to be independent of core. it could be a mixture of server side code in galaxy and the client, or meta in the collections or other things... but we need to make it so core isn't trying to duplicate the functions of a package repository 10:02:36 routing.yml is a holdover feature imo to give folks that "2.9 feel", but it shouldn't be permanent 10:03:28 felixfontein: jtanner hum, sorry, but frozen I mean no new entries added. 10:03:43 I've already got it on the list that we need have tests for routing.yml 10:04:20 jtanner: yep, I'd also say routing.yml in core is mainly for keeping backwards compatibility 10:04:28 not disagreeing. just saying even the frozen/static list should eventually go away 10:04:29 so that existing playbooks/roles still work 10:04:47 for 3.0? :) 10:05:01 :canofworms: 10:05:05 lol 10:06:04 once there is a good tool to convert roles/playbooks/... to FQCNs (or `collections:` adding) it will be easier to convince people to convert 10:06:23 I guess routing.yml could go away in 3.0 (or 2.13 or whenever ansible is not bound anymore by compatibility promise). 10:06:25 (sivels prototype for FQCN is already working quite well I think, at least for the two files I've tested it with yet ;) ) 10:06:47 tadeboro: there has to be a deprecation warning first, and then four versions 10:07:12 yeah, when the ecosystem of tools / blogs / reddits /etc eventually make FQCNs "easy", routing.yml is definitely ready for the chopping block 10:07:14 and already adding a deprecation warning to 2.10 might scare too many people 10:07:46 but that's something someone else has to decide :) 10:11:40 Aye, one thing at a time 10:13:04 yep. let's get 2.10 working first :) 10:16:35 hm, I now started getting errors on the docs fragments, is this a known issue or am I doing something wrong? 10:17:53 shaps: what's the error? 10:18:12 `ansible.errors.AnsibleError: unknown doc_fragment(s) in file plugins/modules/source_control/github/github_deploy_key.py: github` 10:18:41 I can definitely see the fragment in `plugins/doc_fragments/github.py` 10:19:48 shaps: `community.general.github` is the correct name of the docs fragment 10:21:01 that sounds reasonable 10:21:04 testing now 10:23:10 yep that looks better now. Thanks felixfontein 10:23:34 yw 10:26:16 Documented the FQCN discussion in https://github.com/ansible-collections/overview/issues/40 10:34:30 #action DATA make the config for public apps visible and open for PRs 10:34:48 * gwmngilfen is reviewing notes from SUnday 10:50:25 Are there any plans to create a portable Ansible Development kit, where everything you need to get started is bundled, like the correct version of python, pip, virtalenv, tox, ansible/yaml liniting, molecule etc? 10:51:35 Kind of Like the Puppet Development kit (PDK), so for people like me who work inside PCI bubbles can take what they need in one hit and it be supported by Red Hat/Ansible 10:56:06 yes, we are likely to create something along those lines, but no, we're unlikely to package up that entire ecosystem of stuff and build it through redhat's systems and put a "supported" label on it 10:57:55 we don't support playbook development (only bugs), so capital S "support" on content development is very very unlikely 10:58:22 OK, that was a big ask :) With the "Support" 11:00:03 if you need some offline tooling to help with content dev, it would be great if you could either file a support ticket outlining your needs, or contact your redhat account rep and have that person feed the info to our product management 11:00:31 the latter usually has more weight 11:01:25 will do, right away, thanks @jtanner 11:02:00 np 11:07:20 belfast77, it could be community supported container, for example 11:07:56 in some contrib/ folder 11:12:18 #action DATA take a stab at differentiating unmaintained from stable repos, possibly via time-to-1st-comment survival or issues-to-PRs ratio 11:12:45 * gwmngilfen feels a logistic regression coming on for that... 11:14:53 if unmaintained, there's bound to be logical regressions =P 11:20:37 * felixfontein -> lunch 11:21:10 jtanner: quit trolling me, you know logistic != logical :P 11:21:30 trolling is my job 11:21:49 yes, but you're supposed to hide behind ansibot, right? :P 11:21:56 oh oops 11:44:01 * shaps -> lunch 11:57:14 @sshnaidm thanks, sorry I was on a call discussing this. we are going to raise it with our account manager 12:03:33 re 12:10:01 re 12:23:23 Somepoint pointed me to https://www.w3.org/2001/12/zakim-irc-bot.html which is what w3 use to run meetings. The ability to add items to a queue looks interesting 12:25:14 wow, that bot seems almost old enough to order drink over irc in some part of the US 12:25:54 lol 12:26:24 but yeah, I think it would be useful to have something better than meetbot 12:26:33 (who is kinda unmaintained, IIRC) 12:27:44 I kinda think that could be a interesting internship for RH, some coding, some interview to find what people need for meetings 12:34:46 fyi, /rebuild_failed is live on ansible/ansible 12:34:54 \o/ 12:35:03 resmo: felixfontein agaffney ^ 12:35:41 how do I use it? I don't think I've ever actually used a bot command before 12:35:58 also, will that eventually apply to the collection repos, as well? 12:36:41 jtanner: nice 12:36:50 agaffney: just add a single comment `/rebuild_failed` on a PR 12:37:10 tested it, it works! 12:37:15 When we rebase the collections branch for the bot we will get that feature 12:37:17 https://github.com/ansible/ansible/pull/66920 (see last comment) 12:37:37 apropos, when will we get ansibot on community.general? 12:38:03 felixfontein: somepoint next month, hopefully earlier 12:38:24 It's in the list once other content has moved has moved out and CI is good 12:39:13 * agaffney twiddles his thumbs waiting for the bot to pick up his `/rebuild_failed` comment 12:43:13 it looks like it's working 12:43:57 \o/ 12:45:41 `Showing 1,934 changed files with 0 additions and 344,418 deletions.` 12:50:58 yay \o/ 12:53:11 so in community.network we have subdirectories, like in community.general 12:53:30 and symlinks in plugins/modules 12:53:31 makes sense 12:54:02 though getting `ERROR: plugins/modules/avi_alertscriptconfig.py:0:0: missing: __metaclass__ = type` even though got `plugins/modules/network/avi/avi_alertscriptconfig.py validate-modules:doc-missing-type` 12:55:39 in community.network? 12:55:45 are you sure it is a symlink, and not a copy of the module? 12:57:04 headesk 12:57:06 thanks felixfontein 13:01:53 gundalow: I think you want to rebase your PR https://github.com/ansible-collections/community.general/pull/84 to current master, because f.ex. many unit tests have been moved, and the ignore.txt files have been updated 13:02:14 (and you hvae a lot of dead symlinks to remove :) ) 13:15:53 * gundalow finds this handy oneliner `find . -xtype l -delete` 13:16:28 yep, that one is very useful :) 13:16:45 unfortunately I always forget about xtype shortly after I use it... 13:17:20 I didn't know about it till now 13:21:59 Omg I think I’m done with the github changes 13:23:09 I wanted to add some integration tests, but init sure how to test (mostly needs a repo and a token) 13:23:20 *not sure 13:23:46 shaps: Excellent 13:24:22 shaps: The existing `git` integration tests use some GitHub repos to talk to 13:24:47 Yeah, but most of the github modules are “destructive” 13:25:10 (create/remove webhooks, deploy keys, etc.) 13:26:02 I have been using a repo under my username for my own testing, but I don’t really want to receive an email for a dummy deploy key being added when CI runs :D 13:26:20 resmo: the CS error is really strange, your additional setup now runs, but it doesn't help: https://app.shippable.com/github/ansible-collections/community.general/runs/318/98/console 13:30:19 shaps: For testing, is it OK that we use github.com 13:31:09 Yes that’s what it is currently using 13:31:22 Give me 10 min I’ll send the PR, makes it easier 13:34:47 gundalow: a couple of unit tests still need to be deleted, and a huge amount of ignore.txt entries, but besides that CI looks good for your PR :) 13:35:19 gundalow: btw, shippable runs two CI runs for your PRs (maybe only when it is a branch in the repo itself, and not a branch in a fork?) 13:41:30 felixfontein: hopefully my latest commit should "fix" cloudstack ci 13:42:45 resmo: good luck! do you have any idea why this is "suddenly" failing? 13:44:03 yes, the tests were split up into 2 groups, in one of the groups the "creating a VM" fails because of the slow startup of the test container. 13:44:37 it is more a cloudstack issue rather than a ansible issue 13:45:15 (we were just lucky in the past not hitting this issue) 13:46:00 gundalow: you also removed all cloudscale integration tests (cs_*) in your PR 13:46:17 resmo: ah, so it was essentially pure luck :) 13:47:45 gundalow: Right, have a look at the github_webhook test as an example - https://github.com/ansible-collections/community.general/pull/85/files#diff-4aa737321a2dd6a6936e3a0c18b9c8deR1 13:48:43 https://github.com/ansible-collections/community.general/pull/85 <- if anyone wants to have a look/review, please be nice :) 13:50:44 the fun bit will be when I have to "backport" this to 2.9 :D 13:50:53 Not even sure it's doable/worth 13:53:25 shaps: community.general will most likely NOT work with 2.9 13:53:42 I personally don't think it's worth the time to do stuff with 2.9 + collections 13:54:01 yeah, but I've done all those changes because I need them at $JOB :D 13:54:11 #action DOCS Collections and 2.9: not really a priority. ansible-base (2.10) already exists 13:54:12 So one way or another I'm going to have to make it work for 2.9 :P 13:54:17 shaps: because work needs 2.9? 13:54:23 gundalow: what do you think will block using community.general (or at least parts of it) with 2.9? 13:54:39 probably because work needs 2.9 at least until 2.10 is there :) 13:55:14 gundalow: I upgraded to 2.9 a couple weeks ago, still had to support stuff (poorly) written for v2.4 or something like that 13:56:10 also work uses tower, so until I get a stamp from RH Support that I can upgrade to 2.10 + collections, it's not going to happen 13:58:50 btw, shaps, if you have more time for reviewing: https://github.com/ansible-collections/community.crypto/pulls/ :) 13:59:07 yep 13:59:25 any one in particular or just all of them? 13:59:56 #6, #8, #10 should be easiest 14:00:13 and #12 probably too 14:00:51 btw, if you wouldn't have replaced ' with " in the github modules, the diff would be a lot simpler to read :D 14:01:19 hah, blame "black" for it :) 14:01:34 what's "black"? 14:01:43 (I hate to google for that, it's too generic ;) ) 14:02:22 felixfontein: gofmt for python ;) 14:02:29 ah 14:02:43 yep, what resmo said 14:03:22 'blame black' googling can be pretty nsfw 14:03:29 lol 14:03:35 shaps: please undo the x -> _x renames, that doesn't work in collections anymore 14:04:09 the filename you mean? 14:04:29 yes 14:04:35 k sure 14:04:45 also you're missing a changelog fragment :) 14:05:13 What is the best way to send a sequence of commands with `run_command`? i.e. `cd`, `ls`, `touch` If I pass all of them in a single list the terminal believes that `ls` is the folder where it has to `cd`I tired to add `;` like `cd` `;` `ls` `;` `touch` 14:05:19 I can put each command in a separate list but and pass each list with a for loop to `run_command` but I was wondering if there is a more elegant way to do it 14:05:21 felixfontein: yep, I've put [WIP] for a reason! 14:06:54 bcoca: yeah, I don't even want to think about what the results would look like 14:14:30 shaps: thanks for reviewing! 14:14:42 resmo: do you mind if I merge https://github.com/ansible-collections/community.crypto/pull/12 ? 14:15:17 resmo: if someone wants to fork it, they currently have to do a big search+replace anyway 14:15:51 (already just for docs fragments, where there's no relative way to write them AFAIK) 14:16:45 Felden: I don't think run_command is thought to run a series of commands. in your terminal, that only works because the shell essentially sees your line as a program and runs that program by executing the individual steps 14:16:52 oh, sorry 14:17:11 this should have gone to Federico87_ 14:18:21 hum, so do directories under `plugins/` need `__init__.py`? Looking at collections we seem to be inconsistent 14:19:30 #action DOCS and CI test for `__init__.py` existing (or not?!??!) under collection's `plugin/` and `tests/utils` directory structure 14:19:44 DING DING DING! DaWGs (Documentation working group) meeting in #ansible-docs in 10 minutes. We have a fun filled agenda this week! - https://github.com/ansible/community/issues/521#issuecomment-600166114 14:23:02 felixfontein: meeeeerge ;) 14:24:12 resmo: thanks! 14:24:46 felixfontein: regarding 2.9 vs 2.10 + collections. 1) Routing & tombstoning will not be backported to 2.9, so will have to use FQCN. 2) Deprecated modules may not work (same PR) 3) We aren't testing against 2.9, therefore have to assume it is broken 14:24:54 hmm, I think the cisco.aci collection updated and broke the community.general CI: `ERROR! module community.general.aci_interface_policy_fc missing documentation (or could not parse documentation): unknown doc_fragment(s) in file /root/ansible/ansible_collections/community/general/plugins/modules/aci_interface_policy_fc.py: cisco.aci.modules` 14:25:51 gundalow: 1) true, but it should be ok to use some new features from it; 2) that's probably the worst part; 3) that's something we could change ;) but let's definitely focus on 2.10 first :) 14:26:32 #action Collection + 2.9 testing (backlog) 14:26:34 resmo: can you take a look at https://github.com/ansible-collections/community.crypto/pull/11/files ? 14:27:41 resmo: btw, are you still interested in co-maintaining the acme_* modules? (or even some more crypto modules?) 14:30:06 shaps: any thoughts on https://github.com/ansible-collections/community.crypto/pull/7 ? (the general idea, not that it now has conflicts :D ) 14:38:54 @gundalow: they should not need __init__.py 14:40:52 gundalow: ansible has a custom python loader which makes the __init__.py unnecessary for python2 (it's used in python3 as well but python 3 would already find them... They're python 3 namespaces packages) 14:41:18 abadger1999: Thanks for the info 14:55:14 felixfontein: I don't think I've ever been against the change 14:55:57 shaps: I don't think anyone really is, but if nobody mentions anything, nothing will happen :D 14:57:30 yes, that was to say I'm going to +1 it :) 14:57:39 shaps: hehe ok, thanks! :) 15:01:06 left a comment saying I'm +1, I'll review the actual changes in a bit 15:02:10 thanks! 15:24:24 hmm can anyone enlighten me, why does collection build fail by looking into a path cloudstack/cloud in my cloudstack collection? 15:24:26 https://github.com/ngine-io/ansible-collection-cloudstack/runs/549147037 15:24:50 (path does not exist because the modules are right in plugins/modules 15:25:43 ehn never mind 15:38:06 what was the irc handle of the other greg? 15:38:26 me? 15:38:28 :D 15:38:41 ?me <- stats greg 15:38:49 gwmngilfen: ah k ;) yeah 15:38:51 aaaaand I fail at typing 15:39:22 gwmngilfen: do you already have a stat of collections showing up per day on galaxy? 15:39:30 gwmngilfen is the greg with the name I neither can remember to pronounce or to spell... 15:39:33 i do not 15:39:53 felixfontein: gun-gil-ven 15:40:28 resmo: that's not a bad idea, although it's been a while since I messed with the galaxy API 15:40:41 gwmngilfen: thanks! I'm afraid I won't remember that tomorrow anymore, though :) 15:40:45 #action DATA graph new collections over time on Galaxy 15:41:03 felixfontein: just call me "gwim" then, everyone else does :P 15:41:05 gwmngilfen: cool thx 15:41:24 lol ok :) 15:41:30 or use tab completion ;) 15:41:35 (at least here in IRC) 15:41:39 resmo: if you're not already aware, stats.eng.ansible.com/apps is my playground. I'll let you know if I get anything working :P 15:44:00 gwmngilfen: nice 16:38:38 @sshnaidm: FYI I've added some post-mortem comments on the podman collections PR # 15 16:49:29 felixfontein: Yes, double-runs is one issue with opening PRs using a branch in the repo instead of a fork. Shippable can be configured to not do that, but then you have to whitelist or blacklist branches to make that work. Between that and non-fork branches polluting other people's forks, we've tried to avoid PRs without using forks. 16:50:31 webknjaz, thanks, looking 16:55:37 mattclay: thanks for the explanation! 16:59:08 mattclay: I'm currently trying to add basic Ansible 2.9 CI to community.crypto, and first thought I'd use the group for that (we only have one group). but that obviously doesn't work for integration tests (as I've realized now :) ). is it possible to add more things to the node name after another '/'? like having nodes 'units/2.7/1/devel' and 'units/2.7/1/2.9'? 17:01:35 felixfontein: I'd probably add it before the group, not after. Either way, you probably want to modify `shippable.sh` to handle that and strip off the prefix/suffix before passing it to the existing job scripts, so they won't need any modifications. 17:02:45 felixfontein: Maybe look at it as a prefix and a suffix and see which appears to be more readable. 17:03:09 mattclay: ok, I assumed I have to do that anyway. I was mainly concerned with shippable having problems with this, I don't know how flexible it is :) 17:03:16 thanks, I'll try it out 17:04:27 felixfontein: If the labels are too long they may get truncated in the UI or otherwise make the matrix difficult to read, so play around with it a bit to see what works. 17:05:05 Shippable itself doesn't care what the labels look like. That's entirely up to the shell scripts we run in CI. 17:05:59 ok, thanks! 17:18:48 I think prefix works pretty well, thanks for that suggestion: https://app.shippable.com/github/ansible-collections/community.crypto/runs/45/summary/console 17:20:56 #action gundalow contact all collection owners to update Unit test directories; Python imports; ignore.txt 17:21:31 gundalow: which python imports do you mean? 17:22:49 felixfontein: I assume if the unit tests import other unit tests they may need updating? 17:23:10 maybe that isn't the case ever. 17:23:14 actually probs never 17:24:25 you mean import other unit tests from other collections? or you mean from community.general? 17:24:42 ah, you mean after renaming the unit test directory 17:24:46 (I think?) 17:26:01 hmm, btw, some parts of community.general definitely do not work with Ansible 2.9; for example the archive module uses an Ansible-2.10-only feature (optional path argument to load_file_common_arguments) 17:26:53 'mostly works' 17:26:59 yep 17:28:02 it's not impossible to work around this, but I guess not really worth most of the time :) 17:28:10 one question for colections in general, do you add 'compat code' so you work across more ansible versions or set 'required' ansbile version and people need to downgrade to get compatible (going forward, 2.8/2.9 barely support collections) 17:28:38 felixfontein: you 'can' work around it by catching exception and 'reattempting' args w/o the 'path' 17:28:55 i would not advise doing that, but 'possilbeish' 17:30:46 I just did that for one place in community.crypto's module_utils :) I'd really like to support 2.9, at least as long as 2.10 is not released 17:31:10 hello guys, do you have any example how can I grab a Bearer token using ansible? 17:33:17 ok, some longer time ago I apparently used archive+unarchive to do what copy can also do... 17:34:54 felixfontein: probalby worked LOTs faster 17:34:59 also ... syncrhonize ... 17:35:24 rmarcand: dont know of any examaples, but uri/get_url modules should be able to 17:35:24 lol 17:35:48 @bcoca thank you 17:36:12 I'm even more scared of synchronize, since I found out that it uses internals of the docker connection plugin 17:36:45 felixfontein: soo many wrong things there ... 17:37:05 yeah, I heard about it, but seeing it with your own eyes makes it even worse ;) 17:37:29 that explain the quote of Dune at the start of the file 17:40:01 i would have gone with Dante's Inferno quote ... 17:59:38 mattclay: I think change detection for collections in Ansible 2.9's ansible-test is broken: https://app.shippable.com/github/ansible-collections/community.crypto/runs/47/6/console when running with ci_complete, it works fine: https://app.shippable.com/github/ansible-collections/community.crypto/runs/49/6/console 18:04:56 felixfontein: Yeah, there have been bugfixes for that in devel that I haven't backported to 2.9. I'll try to track them down and get them backported. 18:20:16 mattclay: thanks! 18:45:29 There have been some good updates on FQCN in examples, other feedback is welcome https://github.com/ansible-collections/overview/issues/40 18:52:54 Anyone else still hacking? 18:53:01 wondering if I should `#endmeeting` 18:53:29 I'm currently eating dinner, but I'll do a bit later maybe :) 18:53:33 I guess #endmeeting is fine 18:54:19 #chair 18:54:19 Current chairs: andersson007_ cybette dmellado felixfontein gundalow gwmngilfen phips resmo shaps 18:54:26 Thank you everybody! 18:54:33 Lots of cool stuff has been done 18:54:36 #endmeeting