16:08:02 #startmeeting Documentation Working Group aka DaWGs 16:08:02 Meeting started Tue Nov 16 16:08:02 2021 UTC. 16:08:02 This meeting is logged and archived in a public location. 16:08:02 The chair is samccann. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 16:08:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:08:02 The meeting name has been set to 'documentation_working_group_aka_dawgs' 16:08:22 #topic Opening Chatter 16:08:26 Anyone around to talk the docs today? 16:08:36 o/ 16:08:56 o/ 16:09:00 #chair ariordan 16:09:00 Current chairs: ariordan samccann 16:09:15 #chair felixfontein 16:09:15 Current chairs: ariordan felixfontein samccann 16:09:27 andersson007_ dericcrago dmsimard gundalow tadeboro briantist cyberpear mrproper[m] Xaroth acozine you folks chatting docs today? 16:10:59 * samccann ponders if that set of pings even worked w/o the @ in front of each 16:12:29 on the IRC side it works fine :) 16:12:44 heh cool good to know. 16:13:33 dmsimard: gundalow acozine - a few extra matrix pings 16:13:39 Meanwhile, 16:13:42 #topic Action Item Review 16:13:58 #link https://github.com/ansible/community/issues/579#issuecomment-964347798 16:14:01 for the agenda 16:14:57 #info working on getting us to Sphinx 4.2 and updated requirements in general. Will sync with test updates coming from core soon and then be ready to merge 16:15:00 @smccann:matrix.org cyb-clock chimes Fifteen exciting, thrilling minutes have slipped by whilst we plot to take over the universe, one doc at a time! 16:15:14 heh 16:15:33 I'm halfway afk now, have to talk to a colleague 16:15:34 o/ 16:15:46 any other general action item updates before we get to the meat of the fun 16:15:55 thanks felixfontein 16:16:08 #chair briantist 16:16:08 Current chairs: ariordan briantist felixfontein samccann 16:16:09 as always, I'm also in another meeting and otherwise distracted 16:16:15 Welcome! 16:16:30 ok thanks for being here anyway! 16:17:19 #topic Docs survey feedback 16:17:56 #info making strides on the survey issues. Module docs should be showing first in the list of plugins soon. 16:19:03 #info acozine added a list of places where we use `host: all` so we can move forward on https://github.com/ansible/ansible/issues/75506 16:19:36 #info looking for volunteers to tackle some of the other survey issues that folks felt we needed to fix - https://github.com/ansible/ansible/issues?q=is%3Aissue+is%3Aopen+survey 16:20:37 hello, I'm somewhat around 16:20:38 #topic Semantic Markup 16:21:12 #chair dmsimard 16:21:12 Current chairs: ariordan briantist dmsimard felixfontein samccann 16:21:18 Welcome welcome 16:21:47 #info created a proposal for semantic markup https://github.com/ansible-community/community-topics/issues/53 so we can solicit comments/feedback 16:22:53 #info confirmed that this detail will be present in the `ansible-doc --json` output that many other projects use (galaxy-ng, ansible-navigator, AH, collection_prep script, possibly vscode plugin so we need to coordinate the change 16:23:46 sounds sensible to me, I like "Do this in such a way that we do not have to touch the code in ansible/ansible again if we decide we need to display this information in a different way." 16:23:49 I think other than begging people to comment on that proposal, we're kind of stuck for now. Much nagging to ensue 16:24:30 what kind of comments are we waiting for? 16:24:46 are there outstanding questions or decisions to be resolved? 16:25:09 basically we need the other teams that depend on `ansible-do` output to say either yes, they will support this, or no, but they can ignore it etc 16:25:29 got it, thanks 16:25:33 are those RH teams? 16:25:56 Basically once we turn this on, any collection can start using it, so if say Automation Hub/galaxy-ng don't pay attention to this change, we may see the new markup showing up in their output instead of being stripped off or converted to something else like bold etc 16:26:09 yep, RH teams 16:27:03 There's a batch of wip PRs queued up and waiting for these decisions. I'm hoping to bang the drum a lot louder now so we can either move forward, or drop it 16:27:43 sounds good, this has been around for a while, I think it's a great idea so I hope we can move forward with it soon 16:27:44 #topic Community-contributed examples 16:27:47 o/ 16:27:58 #info reviving this discussion https://github.com/ansible-community/community-topics/issues/39 16:29:03 #info Docs survey respondents asked for more examples in ansible. The idea is to create a repo with CI to accept and test community-contributed examples. We may link to them from docs.ansible.com or bring them into the docs themselves at some point 16:29:12 #chair Gwmngilfen 16:29:12 Current chairs: Gwmngilfen ariordan briantist dmsimard felixfontein samccann 16:29:53 We had an open question on what to name the new repo. The options were given here - https://github.com/ansible-community/community-topics/issues/39#issuecomment-906120708 16:30:01 @smccann:matrix.org cyb-clock chimes Fifteen exciting, thrilling minutes have slipped by whilst we plot to take over the universe, one doc at a time! 16:30:16 as either `community-examples` or `ansible-examples` 16:30:34 with the given that `ansible-examples` exists in ansible/ansible repo today. 16:31:00 I agree with calling it `community-examples` 16:31:30 I kind of prefer `community-examples` because it makes it very clear what they are and folks will hopefully understand we don't have any implied support for whatever shows up there 16:31:31 i assume it would exist under the ansible-community org? 16:31:43 Gwmngilfen: yes 16:32:10 in that case the ansible part is cared for, my only question is whether ansible-community/community-examples is repetetive? 16:33:50 * acozine just got out of a meeting 16:33:51 just scanned the repo and there is no examples there now 16:33:59 o/ 16:34:12 so yes, we could just call it `ansible-community/examples` 16:34:24 #chair acozine 16:34:24 Current chairs: Gwmngilfen acozine ariordan briantist dmsimard felixfontein samccann 16:34:32 welcome! 16:34:51 i have no real preference other than I already type the word "community" an awful lot ;) 16:34:51 * acozine reads the backscroll 16:34:53 Should we do a quick vote on `examples` vs `community-examples`? 16:34:59 heh 16:35:56 either one sounds fine for a repo name, `examples` may be easier for auto-complete 16:36:13 for folks who have a bunch of `community.collection` repos 16:37:28 the collection repos would typically live under ansible-collections org 16:37:40 re 16:37:41 sorry 16:37:50 if it's already in the ansible-community org, adding an extra "community" in there may be redundant 16:38:17 dmsimard: true, but locally, I have a bunch of repos in the same directory regardless of the org they came from 16:38:26 maybe that's atypical, though 16:38:44 The only argument against `examples` is that having a fork or clone called `examples` might be awkward. 16:38:54 acozine: its not atypical, its the default of `git clone` to use just the repo name and omit the org 16:39:13 acozine: most people check out the collection repos somewhere else than other repos, I think 16:39:43 I think the original purpose was to collect templating examples, right? could we use a name that reflects that? 16:39:46 at least I have them checked out in some ansible_collections/ tree, so that I can use them with Ansible; so community-examples would be not a problem 16:39:52 (also I know that I can pick the name of the checkout folder myself :) ) 16:40:23 acozine: originally templating examples, but based on the proposal, I think people are volunteering more example types 16:40:46 ah, I see 16:41:14 ariordan: what is your concern about the fork/clone? 16:41:14 actually I went back and read the proposal and it's still mostly templates 16:42:18 so we could call it `jinja-examples` to make it clear. That might also help if we come up with other example types (not that I can think of another category at the moment) 16:42:20 a repo called `template_examples` might be more findable (in searches) than one called `examples`, which could be anything 16:42:35 I think I'd still prefer 'community-examples' to emphasize that this aren't 'official' examples 16:42:37 yeah make sense 16:42:38 ah, yeah, `jinja-examples` is good 16:42:48 (even though it's inside the ansible-community GH org) 16:43:29 if we call it jinja-examples, things will get ugly when we want to add non-jinja examples :) 16:43:34 felixfontein: could we solve that with an obvious and bold statement in the repo readme file that these are community contributed and take with a grain of salt? 16:43:46 samccann: I think so 16:44:00 repo renames are always possible if the scope grows 16:44:11 so lets not overthink it too much ;) 16:44:12 yeah I like the flexibility of other example categories. 16:44:22 they are, but renames always mean a ton of links that need to be changed all over the place 16:44:25 heh as in bikeshedding the name? 16:44:50 both 'examples' and 'community-examples' don't have that problem :) 16:44:56 I'm leaning back toward `community-examples` and we create a directory for `templates` or whatever we want to call it 16:45:01 @smccann:matrix.org cyb-clock chimes Fifteen exciting, thrilling minutes have slipped by whilst we plot to take over the universe, one doc at a time! 16:45:15 * acozine hears a weird noise from the kitchen, BRB 16:45:20 time flies when you're painting that bike shed! 16:45:46 :) 16:45:54 not that we really started painting... ;) 16:45:54 heh 16:46:02 VOTE: We will create the examples repo in `ansible-community` and call it `community-examples` 16:46:04 #chair 16:46:04 Current chairs: Gwmngilfen acozine ariordan briantist dmsimard felixfontein samccann 16:46:27 +1 16:46:29 not worth an argument, +1 16:46:33 +1 16:46:34 +1 16:46:35 heh 16:46:36 ditto, +1 16:47:11 +1 (for create the damn thing, I don't care about the name 😁) 16:47:18 ok calling that agreement - 5 +1, 2 no answer 16:47:22 it's more important to get it started than to get the perfect name 16:47:37 #agreed e will create the examples repo in ansible-community and call it community-examples 16:47:51 ok one quick question and then we can open the floor 16:47:56 #topic Ansible 5 prep 16:48:15 We have an oddity this time, where Ansible 4 won't EOL until about a month after Ansible 5. 16:48:39 just 'examples' might be fine, otherwise you make people type community a lot 16:48:50 * bcoca does drivethrough bikeshed 16:49:02 So do we remove Ansible 4 from the version switcher when Ansible 5 releases, or do we keep it for that extra month and then remove and add EOL banner to Ansible 4 16:49:24 * samccann puts tire spikes under bcoca s wheels 16:49:39 I'm fine with both ways 16:49:43 bcoca: i already made that point and got outvoted ;) 16:49:45 whatever is less work :) 16:50:05 apparently writers like more typing, who knew? :D 16:50:21 my first thought was 'ie' as latinism 'for example' .. but thought that might be too lazy on my part 16:50:29 heh 16:51:20 if making the version-switcher change is easy, why not keep the docs reflecting "reality" . . . but for one month it's not worth more than an hour's work 16:51:45 bcoca: isn't it `e.g.` or `ex genera` for "for example? 16:51:48 it's one of those backport joys 16:52:10 if we keep 4 around, we backport this to 2.12, 2.11, and 2.9 16:52:12 acozine: yes, and i.e means "in other words" 16:52:12 acozine: probably, i tend to confuse both 16:52:19 but eg as repo name works4me 16:52:19 and then a month later, do it all again 16:52:47 folks - we painted that shed.. it's aquamarine... let's move on 16:53:02 meh, I'd just do the usual "it's time to move to 5, folks" thing 16:53:26 people who know how the versions work can always find the old docs 16:53:27 I think we had agreed on something on the version switcher a few meetings back 16:53:33 and it might take google a month to catch up anyway 16:53:55 dmsimard: that was in the community meeting, and I think it was "whatever docs team wants to do / is least amount of work" ;) 16:53:58 #info we will remove Ansible 4 from the version switcher when Ansible 5 releases 16:54:02 dmsimard: probably yeah. 16:54:02 suprised it was not automated already 16:54:06 ^ the version switching 16:54:26 ok gonna do a quick open floor 16:54:31 #topic Open Floor 16:54:32 felixfontein, samccann: +1 16:54:47 anyone have anything to share about docs or things we haven't covered? here's the time! 16:55:00 i have a question :) 16:55:13 for open floor: porting guide update for 5 https://github.com/ansible/ansible/pull/76303 :) 16:55:51 there's a piece of work i'd be interested in leading/helping/doing/whatever, regarding getting collection specific links on the docs for each collection, but I have zero idea where to start. who should I bug to flesh out my ideas? :) 16:56:10 dmsimard: don't we have feature freeze? 16:56:18 dmsimard: merged. thanks! 16:56:29 gwmngilfen-work: define collection specific links on the docs for each collection 16:57:02 dmsimard: there are quite a few collection version bumps that shouldn't have made it into 5.0.0b2 16:57:19 as a couple of examples of what I'm thinking of: 16:57:19 - home assistant does this on their integrations: https://www.home-assistant.io/integrations/ovo_energy/ 16:57:19 - nextcloud do it for their app store: https://apps.nextcloud.com/apps/forms 16:57:50 in either case scroll to the bttom and you will find links to the correct github repo, docs site, etc. my intention is links to repos and (where appropriate) chat locations 16:58:20 thats obviously going to involve some collection level metadata, so it needs to be thought through 16:58:51 felixfontein: the role doesn't pass --feature-frozen to antsibull 🤦, I will fix it for b3 16:58:57 (this was actually gundalow's idea from a while back, but I think it's worth at least figuring out how hard it is :P) 16:59:46 dmsimard: ok :) it will get even more interesting for rcN (N > 1), there not even bugfix releases are accepted except ones manually approved, that probably also needs a role adjustment 16:59:49 Gwmngilfen: it would be relatively easy to add a link from the collection page to the collection on Galaxy; adding consistent links to repos is hard, because we don't pull docs from the repos 16:59:59 Gwmngilfen: I'm a bit lost. Each collection on docs.ansible.com does have a link to galaxy, which then has links to the repos 17:00:19 gwmngilfen-work: there is already some amount of metadata available that galaxy uses, i.e https://github.com/ansible-collections/community.general/blob/d29aecad269a777aacaeabe30e0fb5fc8b674cd9/galaxy.yml#L12-L15 17:00:36 * samccann uploaded an image: (41KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/EDJjZGdZAHHMJiRYRvRUXYok/image.png > 17:00:43 samccann: at the module level, we do have links back to Galaxy 17:00:49 but at the collection level we don't 17:00:59 is that what you're talking about Gwmngilfen ? 17:01:08 ok, so the metadata exists - so why do we depend on Galaxy? why not consume that directly? 17:01:35 this is what I meant about just learning a bit more about what's involved :) 17:01:43 felixfontein: probably remembers best why we use galaxy and not the repo 17:01:45 collections can be hosted privately IIRC 17:01:51 or perhaps thats learning a bit more about why things are the way the are 17:01:56 they don't have to be on GitHub or even visible 17:02:07 gwmngilfen-work: I meant that galaxy uses it for rendering the links at the top right of https://galaxy.ansible.com/community/general 17:02:12 so we pull from Galaxy, not from the codebase 17:02:12 samccann: do you mean in building the docs? because we don't know where the repo lives (and it could move) 17:02:55 we are over the hour, and I know I'm opening a can of worms here. perhaps I should poke a few people sometime to ELI% to me :) 17:03:02 *5 17:03:15 Talking of pulling from Galaxy: I was having trouble building webdocs; coredocs built fine. Requests to Galaxy were failing. I eventually changed some settings in  `~/.antsibull.cfg` and `make webdocs` is working now. Here are the settings that worked for me: `thread_max = 3` and `max_retries = 15`. 17:03:47 ariordan: nice! how long does the full build take with those settings? 17:03:56 the flakiness of Galaxy and the inability to make meaningful change there is part of my concern, but not central to the idea of getting useful links in the hands of docs reader without multiple clickthroughs 17:04:18 I'd guess that the sphinx build time will dominate the overall build time 17:04:25 felixfontein: #info if you are having problems building docs locally for Ansible (make webdocs) experiment with `~/.antsibull.cfg`, specifically something like `thread_max = 3 and max_retries = 15.` 17:04:27 (note that "they can go to Galaxy and click" is insufficent IMO, people will likely not click at all) 17:04:44 not that we're able to get the weblogs to check :) 17:04:52 Thanks, acozine for sending a `pip list` so I could eliminate my mac setup from the scenarios. 17:05:01 samccann: if you do that without `felixfontein:` it will end up in the meeting notes :) 17:05:03 The build takes around 40 minutes now. 17:05:35 Gwmngilfen: i think that depends on who we thing the target audience is for the collection docs. If it is a user, then pointing to galaxy is the correct thing to do 17:05:56 As that is where the user goes, not the repo 17:06:02 ariordan: I'm still very curious why you (from your location) have so much more trouble... I guess it's somewhat related to galaxy / cloudflare / your internet connectionn / your region somehow... anyway I'm glad that you have found config settings to make it work :) 17:06:17 #info if you are having problems building docs locally for Ansible (make webdocs) experiment with ~/.antsibull.cfg, specifically something like thread_max = 3 and max_retries = 15. 17:06:18 heh thanks 17:06:24 ariordan: how long did the build take before? 17:06:36 (when it still worked without special settings) 17:06:43 Gwmngilfen: agreed about the not clicking, and making users go through galaxy is not ideal, but we don't have consistent data on repos and we cannot require it from collection maintainers 17:06:58 felixfontein: It took around the same time. 17:07:07 acozine: i figured that, but is it not possible to render optionally? 17:07:36 ariordan: ok, that's good :) 17:07:52 I didn't have any trouble building webdocs until last week. I don't know why it suddenly failed to work. 17:07:54 about links: we can also include the links from galaxy.yml / MANIFEST.json in the plugin index (like https://docs.ansible.com/ansible/latest/collections/community/general/) 17:07:55 possible? yes 17:08:01 samccann: understood for source, but is a link to Galaxy right for reporting an issue or wanting to ask a question? those examples work well because they have 3 buttons with clear purpose (even though in most cases all three go to github) 17:08:42 acozine: Gwmngilfen I'd be against replacing the current galaxy link with a repo link. I'm still open to being convinced that adding a repo link where available might be helpful, but we need to balance that with the size of that top-level info block as well since it's getting bigger and bigger 17:08:50 feels like we could guide users a bit better than "there's galaxy, shrug" 17:09:09 sure, that makes sesnse 17:09:17 we need to keep the galaxy link, for some collections that is all we can access 17:09:43 Gwmngilfen: the current link goes directly to that collection on galaxy. 17:10:06 i don't want to make this overrun too much. can i set something up to learn a bit more about how it's built and what the options might be? 17:10:08 gwmngilfen-work: if you are talking about the link at the top of https://docs.ansible.com/ansible/latest/collections/community/general/doas_become.html (`This plugin is part of the community.general collection (version 3.8.1).`), I don't think that's one we should change. 17:10:17 I don't know how much work it woudl be to add a repo link at the collection level (for example, https://docs.ansible.com/ansible/latest/collections/community/postgresql/index.html) if we have it . . . 17:10:24 also, what the performance hit would be 17:10:54 do we currently access data from galaxy.yml when we pull the docs? 17:10:57 acozine: as said above, not much work :) 17:11:05 acozine: the main question is how should these links look like 17:11:26 we are looking for MANIFEST.json / galaxy.yml anyway to extract the collection's version 17:11:53 i'm not especially precious about it being at the top, or replacing what is there - i'm just seeing other communities guiding users better than I think we do when they have issues or questions 17:12:25 so I want to figure out if we can borrow ideas. we do have places to ask or report things in many cases, after all 17:12:34 keeping the link to galaxy is fine, I understand the least common denominator thing, but I would also like to see a better link where possible, which will work in like 90%+ of all cases to bring people somewhere better than galaxy 17:12:45 better links would be great 17:12:47 we used to have them 17:13:16 briantist: what are you looking for when you 'somewhere better than galaxy?" 17:13:29 are you looking to report a problem? 17:13:58 can you qualify 'better'? in what aspects? 17:14:18 possibly; whatever the case, I almost always want to go to the collection's github. If I go to galaxy, it's just an intermediate step to get to GH 17:14:34 i think better means "more and more specific" to me 17:14:51 perfect world, you get both links 17:15:04 as some repos do not build docs 17:15:12 probably the best bet is to open an issue in antsibull with the proposal/idea 17:15:13 so you might need to go to galaxy anyways 17:15:33 then we can comment/decide asynchronously on where the repo link could live in the docs page 17:15:44 3 buttons to repo, repo/issues, and repo/discussions would still be better than just repo, because you can style those buttons. people give up looking for things really fast 17:16:00 that is a lot of buttons 17:16:27 i wouldnt go above 3, that tallies with what I see elsewhere, and I cant think of more places I want to go 17:16:43 where would repo/discussions lead? 17:16:54 source/report/chat would cover most things 17:17:11 ah, so basically to Matrix? 17:17:15 gwmngilfen-work: there's no link to repo/discussions available 17:17:30 we only have a link to the issue tracker, to the homepage, to the documentation, and to the repository 17:17:31 acozine: some kind of "ask a question" rather than "i found a bug", so Matrix/IRC, GitHub Discussions, whatever the collection maintainer(s) prefer 17:17:34 (and any of them might be missing) 17:17:49 yes, indeed 17:18:04 but this should be defined by collection maintainers 17:18:28 I have to go eat lunch before my next meeting 17:18:47 and I have to cook. apologies for causing a big overrun :) 17:18:51 we can allow our own set of links (or even collection's own list of links) in docs/docsite/extra-docs.yml (or some other file in that directory) 17:19:05 I like samccann's idea of opening a ticket somewhere with more details 17:19:07 samccann: an issue on antsibull you say? I can do that 17:19:14 and revisiting next week/in future 17:19:15 galaxy.yml / MANIFEST.json is very hard to adjust / extend, since too many cooks are involved :) 17:19:28 ok sounds good 17:19:36 gonna end this meeting if I don't hear screams 17:19:43 will do. thanks for the enlightenment, folks 17:20:25 thats https://github.com/ansible-community/antsibull/ yes? 17:20:30 yes 17:20:33 sgtm, I have to go too 17:20:39 cool. will sort that tomorrow 17:20:56 #endmeeting