19:01:04 #startmeeting Ansible Community Meeting 19:01:04 Meeting started Wed Mar 1 19:01:04 2023 UTC. 19:01:04 This meeting is logged and archived in a public location. 19:01:04 The chair is felixfontein. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 19:01:04 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:01:04 The meeting name has been set to 'ansible_community_meeting' 19:01:04 #topic Agenda https://github.com/ansible/community/issues/679 19:01:10 acozine, andersson007_, anwesha, baptistemm, bcoca, briantist, cidrblock, cyberpear, cybette, dericcrago, dmsimard, felixfontein, geerlingguy, gotmax, gundalow, gwmngilfen, ikhan_, jillr, jtanner, lmodemal, mariolenz[m], markuman, maxamillion, misc, nitzmahone, oranod, resmo, russoz, samccann, thaumos, zbr: The Ansible community meeting is starting now! 19:01:15 The ping list is stored at https://kutt.it/meeting-people. Feel free to add or remove yourself. 19:01:18 #info Agenda: https://github.com/ansible/community/issues/679 / Topics: https://github.com/ansible-community/community-topics 19:01:21 #topic Updates 19:01:26 o/ 19:01:34 o/ 19:01:39 #chair samccann cybette_ 19:01:39 Current chairs: cybette_ felixfontein samccann 19:01:51 o/ 19:02:00 I'm sort of here but I might disappear 19:02:09 #chair briantist 19:02:09 Current chairs: briantist cybette_ felixfontein samccann 19:02:10 Howdy 19:02:13 #chair wbentley15[m] 19:02:13 Current chairs: briantist cybette_ felixfontein samccann wbentley15[m] 19:02:50 wbentley15[m]: btw, do you want to appear on the ping list? 19:03:12 Sure, I will join when I can. 19:03:44 #info Ansible Contributor Summit 2023.02 playlist: https://www.youtube.com/playlist?list=PL0FmYCf7ocrY-uMXYlnbCI2jZz8rDM8zP 19:03:52 .hello2 19:03:53 maxamillion: maxamillion 'Adam Miller' 19:04:00 #chair maxamillion 19:04:00 Current chairs: briantist cybette_ felixfontein maxamillion samccann wbentley15[m] 19:04:10 #info #info Ansible track @ Cfgmgmtcamp 2023 playlist: https://www.youtube.com/playlist?list=PL0FmYCf7ocrZkU2-l-dxveX9woxZUe1ic 19:04:13 wbentley15[m]: https://github.com/ansible-community/community-topics/commit/97de463ed58336280c8b806b09cdd60f59dfc472 19:05:55 #info gwmngilfen-work blogged on the Ansible community strategy for 2023: https://ansible.github.io/community/posts/state_of_the_community_2023.html 19:06:29 #info Ansible 7.3.0 has been Released! https://groups.google.com/g/ansible-devel/c/t9ohIIqRkLU 19:06:41 #info ansible-core 2.14.3 and 2.13.8 have been released: https://groups.google.com/g/ansible-project/c/yokousquPLU/m/6u_uCPNFAQAJ 19:07:09 #undo 19:07:09 Removing item from minutes: INFO by felixfontein at 19:06:41 : ansible-core 2.14.3 and 2.13.8 have been released: https://groups.google.com/g/ansible-project/c/yokousquPLU/m/6u_uCPNFAQAJ 19:07:13 #info ansible-core 2.14.3 and 2.13.8 have been released: https://groups.google.com/g/ansible-project/c/yokousquPLU 19:07:16 shorter URL :) 19:07:24 woo! so much awesome stuff :) 19:08:11 hi 19:08:19 .hi 19:08:20 gotmax23: gotmax23 'Maxwell G' 19:08:27 * gotmax23 has some Fedora packages to update 19:08:28 if nobody has a preferred topic, I'd suggest a quick discussion about some docs features (semantic markup and private plugins) 19:08:32 #chair jtanner gotmax23 19:08:32 Current chairs: briantist cybette_ felixfontein gotmax23 jtanner maxamillion samccann wbentley15[m] 19:09:22 #topic Semantic markup for Ansible documentation 19:09:31 felixfontein : we'd like to briefly ask for feedback about the strategy posts, it can be after your docs features discussion 19:09:31 #info Discussion: https://github.com/ansible-community/community-topics/issues/53 19:09:48 cybette_: oops I just started the topic, let's do it afterwards then :) 19:09:58 #info Specification: https://hackmd.io/VjN60QSoRSSeRfvGmOH1lQ 19:10:02 no problem! 19:10:38 #info PRs: https://github.com/ansible/ansible/pull/74937 https://github.com/ansible-community/antsibull-docs/pull/4 19:10:44 I've done my best to inform other impacted projects about semantic markup and private plugins. So the communication part has happened anyway :-) 19:10:46 #info example PR: https://github.com/ansible-collections/community.dns/pull/23 19:10:52 samccann: cool, thanks! 19:10:59 o/ 19:11:05 felixfontein: are you planning on merging core PRs before the 2.15 branch pull? 19:11:05 is anyone not somewhat familiar with the history and wants a short recap? 19:11:10 * acozine finished lunch finally 19:11:14 #chair acozine 19:11:14 Current chairs: acozine briantist cybette_ felixfontein gotmax23 jtanner maxamillion samccann wbentley15[m] 19:11:18 (for semantic markup) 19:11:38 samccann: I can't merge :) I can only merge the antsibull-docs PR, and I'd like to do that assuming the SC agrees 19:12:25 since there hasn't been much feedback yet I want to start a community/SC vote on whether we should try to get the PRs into ansible (for ansible-doc) and antsibull-docs for ansible-core 2.15 / Ansible 8 19:12:53 fwiw I like it :) 19:12:53 ah ok cool. that answers my 2nd question as well (whether there will be a steering committee vote on both of these) 19:13:30 I have no idea how good the chances are to get the ansible-core PR merged, but I guess having a positive vote on it will be better than not having that 19:13:58 does anyone have second thoughts / thinks this should wait / ...? 19:14:30 the longer we wait, the more rebasing we have to do, and the fuzzier it gets in everyone's minds 19:14:45 I'm not super involved in docs stuff, but It seems like a good idea. I appreciate consistent formatting and anchor links and other things we can do with this data. 19:14:46 plus, we've been talking about semantic markup for years 19:15:15 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:15:15 o/ 19:15:30 acozine: it's more than two years by now 19:15:40 I say let's vote and do our best to get it merged, and if that doesn't work / meets significant resistance, we drop it and move on 19:15:43 yep 19:15:49 I think we started in oct/nov/dec 2020... someone would have to look at the meeting logs to figure out when exactly we started :) 19:16:59 cool. since nobody seems to object, I'll create a vote later today :) 19:17:00 I think it's a good idea, I want to implement it, but at this point I'm also tired of talking about it and of the will-we-won't-we back and forth 19:17:34 Are there any actual objections to this besides the issues with having to sync with other projects? 19:17:38 acozine: assuming the SC doesn't vote against it (or doesn't manage to have enough votes), I think we can get this rolling 19:18:39 gotmax23: I was reticent to merge because it will impact other projects (galaxy/ng for sure) and didn't want garbage to start showing up in their module docs output 19:18:46 but yeah, time to move forward imo 19:19:09 #topic How (and whether) to mark private plugins in collections 19:19:11 as I recall, core wasn't against semantic markup but it's been a couple of years.. sooner the better on the merge 19:19:16 another docs topic ;) 19:19:31 #info Discussion: https://github.com/ansible-community/community-topics/issues/154 19:20:21 this was a discussion started some months ago, there were some folks from community (including me) who want to have a mechanism to hide certain modules/plugins in collections 19:21:18 the proposal was to mark them (via filename, metadata, ...) so that they don't show up on the collection's docsite or in `ansible-doc --list` 19:21:54 the core team didn't want such a feature in core, so `ansible-doc --list` will always show all plugins/modules, but we can still try to get other tools to hide them 19:22:03 (like antsibull-docs for the docsite :) ) 19:22:16 I think figuring out how we want to record this data (with _ prefixes or otherwise) and hiding this from the docsite is a good start. We can keep [INTERNAL PLUGIN] or something in the description for other tools (e.g. ansible-doc) that don't respect this metadata. 19:22:18 for that we (the community) have to decide a) whether we want to do this without core, and b) how exactly we want to do this 19:22:55 Perhaps ansible-lint can also get on board and report usage of private plugins from other collections as an error 19:23:03 gotmax23: yeah, I personally would use all possible indications (metadata, description, ...) to mark plugins as hidden anyway 19:23:06 the drawback of doing it w/o core or other projects - inconsistent docs. (meaning what you say, docsite won't show it, but ansible-docs will.. galaxyNG probably will etc) 19:23:34 galaxy[_ng] won't be able to hide them if ansible-doc doesn't hide them 19:23:43 that's true. but I guess it's better if *some* tools don't show them, instead of all showing them 19:24:07 felixfontein: I tend to agree 19:24:12 I saw a comment somewhere about 'it breaks the premise of auditability' ...but I don't know the full context, just relaying what I heard on the streets... :-) 19:24:21 jtanner: depends on how the marking is done - if we would say (for example) a leading underscore means private, galaxy[_ng] can easily do it 19:25:05 samccann: hmm, it would be interesting to know more about that. I haven't heard that argument before 19:26:16 yeah sorry I don't have details 19:26:39 maybe it's that someone could still USE the private plugin, right? 19:26:41 would be nice if the persons who know the details could comment them in https://github.com/ansible-community/community-topics/issues/154 :) 19:26:57 I mean even if we make it invisible to the docs etc it doesn't stop someone 'in the know' from using it anyway? 19:27:12 Yes, I think that was the idea 19:27:33 i think it's in galaxy's best interest to abide by whatever ansible-doc is doing and not to apply convention over top of it that we'll have to maintain as the consensus changes over time 19:27:34 true, you can still use them, since core does not prevent that, but if something is marked as private, not easily visible, and say ansible-lint marks usages of it as a problem, I think that's acceptable 19:28:03 jtanner: I fully agree, that's why I would prefer if ansible-doc also allows to do this the same way as other tools do 19:28:10 can someone help me understand why I am getting this error ? https://pastebin.mozilla.org/pdYgPtsy 19:28:14 :nod: 19:28:39 jtanner: Well, the SC will come to a consensus if we conduct a vote 19:28:39 So I like the idea of an ansible-lint rule being able to warn about using a private plugin for sure 19:28:47 aderium: you forgot `community.crypto.` 19:28:47 imo, you all need to sell core on this idea first 19:29:10 I'm pretty sure it's a firm rejection from core already 19:29:41 aderium: We're in the middle of the weekly Community Meeting. You're welcome to join the meeting and discuss our current topic, but these questions are better suited to the #ansible channel. 19:30:01 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:30:38 felixfontein error now is `FAILED! => {"msg": "lookup plugin (community.general.file) not found"} ` 19:30:40 jtanner: we tried, core discussed it internally and decided against it, apparently without looking at the community discussion 19:30:59 hence the word "sell" 19:31:02 aderium: please ask in #ansible (IRC) / #users:ansible.com (Matrix) 19:31:17 aderium : please wait until after the meeting for your questions or join the #ansible channel. thanks! 19:31:28 This is the core response if people want to see their views - https://github.com/ansible/ansible/pull/79370#issuecomment-1404251346 19:31:28 Sorry ! and Thank you 19:31:48 dunno if we captured their feedback anyplace else 19:32:09 * gotmax23 hates issue locking, but that's another discussion... 19:32:16 samccann: that was the only public feedback we got 19:33:45 It does sound like implementing this in the docsite and other tools like Galaxy or ansible-lint but not ansible-doc is in line with what core suggested 19:34:15 as I read that feedback, they aren't against hiding things in docs, but are against changing ansible-doc output. 19:34:19 I don't think hiding these from ansible-doc -l should be a blocker 19:35:27 the gotcha from what jtanner said - we couldn't hide it from Galaxyng either 19:36:18 And with GalaxyNG showing module docs now, I expect a group of our useers will be looking over there, not at docs.ansible.com. I dunno how big a group over time...but docs embedded in GalaxyNG now is a big win for users 19:36:41 true 19:36:56 Yeah, once that's actually supported for community collections :) 19:37:16 jtanner: Is the only objection on your side that ansible-core doesn't want to implement this? 19:37:26 well, and once plugins are "private", folks will want "private" module args too https://github.com/ansible/ansible/issues/79695 19:37:45 'private' module arguments already exist 19:38:08 (and are often necessary if you combine modules with action plugins) 19:38:30 Do you have an example? 19:38:32 the problem right now is that you need to turn of certain sanity checks for them, instead of being able to turn them of only for specific options 19:38:48 this may come down to - do the benefits of marking things private outweigh the drawbacks of this only showing up on the docsite? 19:38:54 mariolenz[m]: if you have an action plugin that calls the module (of the same name) but passes in some extra things from the controller, it needs to use a module option 19:39:05 gotmax23: my primary objection is that galaxy[_ng] uses ansible-doc for the source of truth and we don't want to put shim code on top for establishing a convention we'll have to continuously maintain over top of the ansible-doc results. 19:39:16 mariolenz[m]: but that's a module option that shouldn't be publicly documented, because if the user uses it, the action plugin will overwrite the user-specified value 19:39:25 jtanner: Slippery slope fallacy 19:39:38 (hmm assuming mariolenz[m] meant me and not jtanner :D ) 19:39:53 it's not a slipperty slope ... we already have that scenario with requires_ansible and runtime.yml 19:39:59 * samccann driveway was a slippery slope this am.. til the ice melted 19:40:20 jtanner: the thing about private module args is that ansible-doc doesn't show them already now. the problem is mainly with sanity checks, not with ansible-doc 19:40:25 I don't think skipping modules that start with _ is that hard to implement yourself 19:41:21 samccann: that can be 'fun' :) 19:42:03 well, it seems we've reached an impasse... 19:42:23 it would be nice to have more opinions (especially from SC) on this 19:43:07 I'm +1. Even if we all agree on this, we don't have any say on what other projects besides antsibull-docs/the docsite do. 19:43:13 is there a fixed proposal from the people who support this? 19:43:41 once the approach is fixed, then yeah, put it up for SC votes and see what happens 19:43:53 samccann: not yet, since so far it was core who basically decided how it should be implemented, until they said they won't have it - now that question is back open 19:44:22 core didn't want the _ prefix in the module/plugin name, but instead some metadata (as in meta/runtime.yml) 19:44:23 I'm okay with having private modules, would prefer to use metadata to mark them (rather than relying on the Python convention) for the same reasons briantist mentioned in the discussion issue. 19:45:00 @cybette:ansible.im cyb-clock chimes every 15 minutes during the community meeting 19:45:14 Easier for non-Python-experts to understand, which means supportive of a healthy community of both users and developers. 19:45:35 I guess I could start a vote on how it should look like, i.e. filename convention vs. metadata 19:45:57 Ack 19:46:45 the question is whether the vote is about two concrete proposals (`_` prefix for filename, `private: true` in meta/runtime.yml), or more abstract (filename convention, 'some metadata') 19:47:28 (one could for example also put something in the module/plugins's DOCUMENTATION - but that makes hiding the docs in listing harder to implement) 19:48:25 I'm not sure if this is the best approach. Instead of having private in meta/runtime.yml, can't we have it in the module itself? 19:48:50 we could do 'informal voting' in the existing discussion (metadata, underscore, something in text, etc) and after we have a consensus, turn that into a formal vote 19:48:57 I would prefer it in the module itself as well 19:49:00 Like version_added or similar. 19:49:14 mariolenz[m]: having it in the module itself means that the module's docs have to be loaded to determine whether it's hidden - which means that listing all modules (or plugins) would be a LOT slower 19:49:25 I thought there was some reason that couldn't work in the original 19:49:26 ah that 19:49:42 well, runtime.yml is still better than underscore imo 19:49:50 so from an implementation POV, it's a good idea to have it somewhere else - like in meta/runtime.yml (that's easy to load) or in the filename (that's also easy to load) 19:50:27 I like an _ prefix 19:51:22 felixfontein: do you want to start a formal vote on `_` or runtime.yml? 19:51:26 I'm happy with both, but I also understand that `_` is very Python specific (or more generally, programming specific; it's also used in JavaScript for example), and not obvious for non-programmers 19:51:40 gotmax23: I think I'll do that if nobody objects / has another idea 19:52:03 I could also add the third option (something in DOCUMENTATION) and add the reason why I wouldn't do that (inefficient listing) 19:52:14 wdyt? 19:52:25 Well, it shouldn't need to be obvious to non-programmers. Plugin descriptions should still have an [INTERNAL PLUGIN] prefix. 19:52:26 it doesn't seem like anyone wants the third option though right? or I missed it 19:53:25 felixfontein: Good point. 19:53:57 gotmax23: to me it's not about programmers or not, even if several languages use it it's not really "programming" specific, and it has no specific meaning in Ansible from an end-user perspective. You and I get it just fine, but we're not representative of the wider user base, and it's not an obvious meaning to most 19:54:25 briantist: you and mariolenz[m] said you like it (until I brought the listing argument) :) 19:54:54 Let's move this to the issue and start open floor? 19:55:07 There's 5 minutes left 19:55:21 felixfontein: oh right, I thought you meant free-form text in a description or something, don't mind me 19:56:01 I'd like to squeeze in a topic at the end, mainly for info and you can leave comments in the issues (that I'll provide) 19:56:13 ok let's move on 19:56:14 cybette_: please use #topic :) 19:56:18 #topic Ansible Community Strategy 2023 19:56:21 #topic Open Floor 19:56:28 #topic Ansible Community Strategy 2023 19:56:42 (hmm or should I have used #undo?) 19:56:46 #info The Ansible Community Team has been sharing and discussing (at events and online) some of our thoughts and ideas around the state of the community and what we'd like to do 19:56:47 #link https://ansible.github.io/community/posts/state_of_the_community_2023.html 19:56:48 Hah 19:56:53 hehe 19:57:10 We'd like to hear some feedback, if you've attended Greg's talks at Cfgmgmtcamp/Contributor Summit, or read his blog posts. 19:57:10 Sorry, I was late and don't know if you already talked about this: https://github.com/ansible-community/community-topics/issues/201 19:57:11 #info Please add your thoughts to the two main proposals in their respective issues 19:57:13 #link DNS & Website https://github.com/ansible-community/community-topics/issues/201 19:57:13 If you didn't, please have a look at it and comment. 19:57:19 #link Discourse Forum for project-wide discussion https://github.com/ansible-community/community-topics/issues/202 19:57:30 thanks mariolenz this is exactly what I'm hoping for :) 19:58:00 :) 19:59:26 Feel free to add your thoughts in the 2 issues, or ping Gwmngilfen , myself, anyone in the community team you feel comfortable chatting with to discuss. Thanks! 20:00:08 #topic open floor 20:00:16 does anyone have something else to discuss in the last ~minute? 20:03:13 looks like not :) 20:03:17 #endmeeting