15:01:08 #startmeeting Docs Working Group aka DaWGs 15:01:08 Meeting started Tue Apr 27 15:01:08 2021 UTC. 15:01:08 This meeting is logged and archived in a public location. 15:01:08 The chair is acozine. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:08 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:01:08 The meeting name has been set to 'docs_working_group_aka_dawgs' 15:01:13 #topic opening chatter 15:01:14 woot! 15:01:20 who's around to talk docs? 15:01:22 #chair samccann 15:01:22 Current chairs: acozine samccann 15:02:34 andersson007_: dericcrago dmsimard aminvakil briantist cyberpear felixfontein jmn kindlehl madonius tremble you folks talking docs today? 15:03:20 official agenda begins with https://github.com/ansible/community/issues/579#issuecomment-818898282 and continues with https://github.com/ansible/community/issues/579#issuecomment-827671912 15:03:30 gundalow: I missed you in the roll-call above 15:03:47 sorry, no time today... 15:03:56 felixfontein: we'll miss you! 15:05:03 if I didn't call your name, it means either 1) you haven't participated in the past, or 2) you have participated and I've forgotten, or 3) I just missed your name for no good reason at all 15:05:35 everyone is welcome, first-timers especially! 15:05:59 so if you're here and you want to participate, you can say hi or wave like this: 15:06:03 o/ 15:06:18 possibly a quiet day today? 15:06:24 seems like it, yeah 15:06:29 Mr. Badger is on PTO 15:06:37 so is Ms. Malivert 15:07:05 Mr. Badger sounds like a kids book! Mr. Badger's Adventures in Ansibleland! 15:07:07 Mr. Fontein is out 15:07:13 heh 15:07:26 okay, I'll start posting news items 15:07:33 cool 15:07:38 with handing infos 15:07:50 s/handing/handy/g 15:08:04 #info ansible-core 2.11 was released yesterday (April 26) 15:08:35 #info Ansible 2.9.21rc1 and 2.10.9rc1 are being cut today 15:09:17 #info last week's DaWGs Mini Docs PR Day reviewed and merged or close 12 PRs, thanks to all who participated! 15:09:52 We're a bit behind on publishing the core docs. But you can see the docs that will be 2.11 as the devel docs for now 15:10:19 #info core docs are now at https://docs.ansible.com/ansible-core/devel/index.html 15:10:45 We are also hard at work backporting all the docs PRs that have been merged into `devel` back into `stable-2.11`. 15:11:08 15:11:12 heh 15:12:19 we've updated the version-switcher on the /ansible/ docs so only three versions are available - `devel`, `latest` (which is 3), and `2.9` 15:12:53 When Ansible 4 is released in a few weeks, we expect to remove 3 and add 4 as `latest` 15:13:03 I'm thinking once 2.9 goes EOL someday, then we'll only have devel and latest, right? 15:13:37 If folks decide to form a maintenance team for Ansible 3, we can keep both 3 and 4 in the rotation, but for now, we're not planning on it 15:13:56 ^^ we should discuss as well. 15:14:19 samccann: yes, when 2.9 goes EOL there will only be 2 options in the /ansible/ version-switcher, unless folks step forward to maintain other community versions 15:14:21 in the sense that anyone who forms maintenance for it also supports docs backports etc as needed 15:14:53 we can merge the backports etc, but we can't maintain the older docs 15:15:21 true, docs maintenance will be part of the maintenance burden 15:15:41 maybe i should change the topic . . . 15:16:05 #topic versions and maintenance 15:16:15 :-) 15:16:31 #info version-switcher for /ansible/ docs now only shows `devel`, `latest` (which is 3) and `2.9` 15:16:48 #info once Ansible 4 releases, it will become `latest` and Ansibl 3 docs will be EOL 15:17:23 #info if a maintenance team steps forward for Ansible 3, we can add `3` to the mix, but maintainers will need to maintain docs (with backports) as well as code 15:18:12 anything else on versions and related topics? 15:18:35 tadeboro: you around to chat about docs? 15:18:37 thinking 15:19:00 * tadeboro reporting for duty 15:19:03 no I don't think there's anything else on version fun 15:19:16 #chair tadeboro 15:19:16 Current chairs: acozine samccann tadeboro 15:19:26 we're a small group today, several folks are out 15:19:35 #topic Sphinx upgrade 15:19:58 tadeboro: what's your rank? 15:20:24 General? Lieutenant? Sergeant Major? 15:20:41 Right now, I am a ship cook ;) 15:20:42 They are the very model of a modern Major General... 15:20:45 heh 15:21:01 send some ship's biscuits and grog over, will you? 15:21:43 #info WIP PR to upgrade the Sphinx version is open: https://github.com/ansible/ansible/pull/74372 15:22:22 if we remove the limitation, folks may end up with different versions of Sphinx 15:22:40 on the one hand, that means we don't have to worry about getting more and more behind 15:23:11 on the other hand, we risk things breaking if there are changes in Sphinx that we were not expecting 15:23:25 I think overall, in requirements files, we don't fix a version or range unless there's a need to, right? 15:23:41 yeah, and I think that's the right approach here too 15:23:41 as in we have that risk in all the packages we depend on in Ansible 15:23:45 true 15:23:58 and yeah it will bite is in the backside *cough* docutils *cough* 15:24:03 yep 15:24:16 Having no limit is good and bad at the same time: for seasoned contributors, this means they can help fix issues; for first time contributors, change in the Sphinx itself can make contributing a bit harder. 15:24:35 I'm just feeling a bit tender in that part of my (technical) anatomy from recent bites 15:24:49 tadeboro: yep 15:25:08 do you have a preference for which risk we should run? 15:25:19 yeah wondering how we track changes better from the upstream projects we depend on? 15:25:21 But from my personal experience, staying behind the upstream is usually a bigger problem and would thus remove the upper bound as done in the PR. 15:25:41 okay, sounds like the three of us are all in agreement 15:26:03 We have the nightly builds that will show if there is a catastrophic failure 15:26:14 How often do we regenerate the devel docs? If the frequency is more than once a month, that should reveal problems soon enough. 15:26:17 #info if any folks have spare time this week, please test the latest Sphinx version with our docs 15:26:18 but oddities in output it won't find *cough* docutils *cough* 15:26:32 tadeboro: the devel docs run on an automated job 15:26:39 it checks 4x per day for changes 15:26:48 and publishes the docs if there are any changes 15:27:04 that includes changes to the code, not just to the docs (it looks for merged PRs, basically) 15:27:18 so if it fails, we know about it quickly 15:27:31 yeah the problem is what happened when docutils updated, broke the theme we depend on.. and things looked just a bit 'off'. People noticed etc and we patched around it. 15:27:32 but, as samccann says, sometimes the job succeeds, but there are other problems in the docs 15:28:02 So I guess the question is - is that an okay feedback loop? That devel docs are cutting edge and if things go bad, community will notice and let us know? 15:28:10 I think so 15:28:29 we can't expect perfection 15:28:31 Breakages will always happen, so I think we just embrace them and let users report bugs when they find them. 15:28:53 the only other option I can think of is if we start tracking upstream packages more closely so we can anticipate when a release is coming up there, then keep a closer eye on the impact in devel 15:29:11 ok I can live with that :-) 15:29:59 I have been generating docs using latest stable Sphinx for almost two years now for our collections and I think my build broke once in all that time. 15:30:14 tadeboro: that's a good record 15:30:26 looks like there's a sphinx-docs channel here in freenode IRC 15:30:38 I might join that, see how much traffic it gets 15:30:42 oh interesting 15:30:55 https://www.sphinx-doc.org/en/master/ 15:32:24 #agreed we'd rather risk breakages with updates to Sphinx than face the problems of pinning to a known older version 15:33:26 anything else on Sphinx? 15:33:38 oh 15:33:46 We have that PR somewhere on using the ansible theme the community created 15:33:58 afaik that's waiting on fixes to that theme before we can go further. 15:34:16 #info Sphinx has a channel called sphinx-doc here on freenode IRC, maybe one of the DaWGs can join that and keep tabs on upcoming changes 15:34:30 #topic ansible Sphinx theme 15:35:00 oh gosh now I gotta find that PR Lol! 15:35:07 #info WIP PR: https://github.com/ansible/ansible/pull/74318 15:35:27 oh you are on FIRE today! thanks! 15:36:33 #info the ansible theme is pretty close to what we need, but have some requests in for changes to match our current look and feel and to support ansible-core docs with theme changes 15:36:38 theme needs to support (or at least allow) multiple "look and feel" options, so we can differentiate the `ansible-core` docs from the `ansible` docs 15:36:53 samccann: jinx 15:37:00 heh 15:37:06 * samccann hands over a soda 15:37:15 * acozine chugs 15:38:51 #action samccann to test sphinx 4.x beta as it has 'breaking changes' warning on it 15:39:25 #info theme changes for `ansible-core` introduced in https://github.com/ansible/ansible/pull/74356; current ansible-core docs are published as https://docs.ansible.com/ansible-core/devel/ 15:39:44 wow, we're going to go from 2.x to 4.x in a few weeks? 15:39:49 heh 15:40:01 we're living in the future! 15:40:15 more like let's see if things blow up horribly with sphinx 4. If so, we may need to pin it to < 4 until we can figure out what's going on. 15:40:18 heh 15:40:42 any other Sphinx wisdom? concerns? ideas? 15:40:49 #info core theme changes support a different look so readers know they are not on the Ansible (package) docsite 15:41:36 #topic searching and sitemaps for the `ansible-core` docs 15:41:59 this topic may need a few weeks of discussion 15:42:29 a lot of documentation is duplicated on `docs.a.c/ansible/` and `docs.a.c/ansible-core` 15:42:40 which is confusing for search engines, and makes for bad SEO 15:43:02 so far the site search has only been turned on for `/ansible/` 15:43:12 and we only have a sitemap for `/ansible/` 15:43:56 there are a few pages that ONLY appear on `/ansible-core/` docs, mainly the core roadmaps 15:44:27 the porting guides are different on `/ansible-core/` but the core items are included in the larger `/ansible/` porting guides 15:44:42 so the question is . . . 15:45:15 how much do people want to use the site search on the `/ansible-core/` documentation? 15:45:31 right now, if they do, the results will take them to `/ansible/latest` 15:45:34 There are a couple of options we can go with 15:45:47 we can turn off the swiftype search for core docs and go back to opensearch. 15:46:15 OR we could create a new search engine target specific for ansible-core so the search would only pick up core docs 15:47:21 I must admit I never used the search functionality on docs.a.c, I just limit the search directly in the search engine. 15:47:46 And I have a feeling that I am not alone here. 15:47:51 heh 15:47:57 yeah, most of our traffic comes from google 15:48:03 very true. 15:48:12 So making sure SEO is right is probably more important than getting search functionality right. 15:48:34 so messing up the google results for the sake of better local sitesearch seems like a step backwards 15:48:53 not sure how that happens? ^^ 15:48:59 what are you envisioning there acozine? 15:49:34 if we publish a sitemap for ansible-core, google will index it and might start returning those pages instead of `/ansible/latest` 15:49:40 cyb-clock sez we are 50 min into the meeting 15:49:43 heh 15:49:58 yeah agreed. I think we do NOT do a sitemap for ansible-core 15:50:19 but we could still replace the site search with 'something better' that would presumably have no impact on the googles of the world 15:51:02 Is there a non-technical reason behind the dupplication of content between ansible and ansible-core sites? 15:51:26 There were two reasons I remember: 15:51:33 1 - core releases separately from Ansible 15:51:51 2 - Folks felt it was important to have a COMPLETE website 15:51:56 also 15:52:05 as in, we couldn't have ansible-core on one site, and al lthe collections on another site 15:52:11 3 - core maintenance is different from Ansible maintenance 15:52:26 ansible-base 2.10 is still maintained, even though Ansible 2.10 is not 15:53:51 Hmm, when I search for common stuff (like playbook keywords), I would expect search engine to serve me results from ansible-core. 15:54:27 would you? instead of from the latest version of Ansible (which is based on an older ansible-core version)? 15:54:45 But maybe I am not the right person to design this for since I know too much about the ansible-core vs. ansible and I can navigate my way around. 15:55:05 that's another difference, there is no `latest` for ansible-core 15:55:19 heh 15:55:27 we have that same problem 15:55:55 I can't really imagine what the site looks like to someone who doesn't already know her way around 15:56:00 first, how are downstream packaging it? are they still using the 'ansible' package, are they offering a 'core' package or both? 15:56:18 neither afaik 15:56:21 ^ big chunk of users will consume from OS package 15:56:32 samccann: so they stopped offering ansible as OS package? 15:56:35 oh downstream as in other OS's 15:56:47 yes 15:56:54 RHEL is TBD. I don't know if it will be in RPMs etc 15:57:09 as for something like Debian, dunno. maybe relrod knows what they are doing? 15:57:10 RHEL is special case 15:57:30 debian, ubuntu, freebsd, centosalternativescauseusersaremiffedwithRH 15:57:45 amazon linux ... 15:58:06 For Red Hat, all I found is "here is how you should use execution environments". For other distros, I am not really sure since my method of installing ansibleis venv. 15:58:33 this is ubuntu - https://launchpad.net/~ansible/+archive/ubuntu/ansible 15:58:38 and looks like it's base only 15:59:08 Gentoo has ansible and ansible-base. 15:59:12 well, RHEL assumes 'paying customer' and AAP .. so again, special case 15:59:12 I see both (on the upper right) 15:59:30 https://packages.gentoo.org/packages/search?q=ansible 15:59:33 (both ansible and ansible-base on the ubuntu page) 15:59:40 ^ that ppa which 'WE' control, but not what OS ships with 16:00:23 acozine look at the actual package numbers for ubutu - 'ansible' is 2.9 only 16:00:31 right now most distros are still shipping 2.9 16:00:48 heh 16:00:54 why i ask if ther eis any info on what they plan for > 2.9 16:01:08 cyb-clock sez we are 1 hr into the meeting 16:01:10 cause that will 'taint' how we want to present info for users in the end 16:01:16 okay 16:01:20 the PPA is being maintained by the Community team going forward. So ansible-core hasn't landed there yet. Though the Core team still currently drops 2.9 and 2.10 updates there. 16:01:22 we are not going to solve this today, that's for sure 16:01:59 let's keep it on the agenda for next week 16:01:59 sorry what's the PPA? lost track on that 16:02:00 relrod: why i quoted 'we' 16:02:17 samccann: Ubuntu/Debian packages 16:02:20 PPA is an auxiliary package, an add-on 16:02:21 samccann: page linked above, its 'repo available to distro but not part of distro (ubuntu)' 16:02:23 ok thanks 16:02:50 though they work with every debian derivative, its mostly an ubuntu sponsered '3rd party stuff here' repo facility 16:02:58 #info need to keep this discussion open - what are the distros doing with ansible vs ansible core etc. Gentoo seems to have both for example 16:03:15 let's do a quick open floor 16:03:28 #topic open floor 16:03:40 all questions, comments, suggestions, ideas welcome 16:04:11 if you have an open PR and need reviews, you can post it 16:04:31 if you have a question about using the docs, you can ask it 16:05:10 if you have a critique of a page, or a suggestion for improving the docs, you can discuss it 16:05:25 this meeting happens every week at this same time 16:05:49 anyone can add items to the agenda at the bottom of https://github.com/ansible/community/issues/579 16:06:05 and/or you can bring any topic up during the open floor 16:06:33 * acozine stops typing for a minute . . . 16:07:32 hearing no open floor items . . . . 16:07:49 thanks tadeboro samccann bcoca relrod 16:08:08 chat welcome in the channel at all times! 16:08:10 #endmeeting