15:01:46 #startmeeting Ansible Docs Working Group aka DaWGs 15:01:46 Meeting started Tue May 14 15:01:46 2019 UTC. 15:01:46 This meeting is logged and archived in a public location. 15:01:46 The chair is acozine. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:46 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:01:46 The meeting name has been set to 'ansible_docs_working_group_aka_dawgs' 15:01:59 #chair samccann 15:01:59 Current chairs: acozine samccann 15:02:06 who else is around? 15:02:09 #chair kmaxwell 15:02:09 Current chairs: acozine kmaxwell samccann 15:03:26 felixfontein: dag decentral1se Pilou shaps Xaroth zoredache alongcha_ andersson007_ are you folks running with the Ansible DaWGs today? 15:03:41 Not today unfortunately 15:04:08 alongcha_: okay, hope to see you next time! 15:05:48 I'll start with a reminder: 15:06:00 AnsibleFest Call for Proposals closes on May 28th 15:06:02 https://ansiblefest2019.eventpoint.com/cfp 15:06:15 if you submit a talk proposal and it's accepted, you get free registration 15:06:27 we will also be having a Contributor Summit on the Monday of Fest week 15:07:07 it would be great to see a lot of documentation champions in Atlanta 15:07:14 I hope everyone on this list can come! 15:07:24 \o/ 15:07:27 #topic AnsibleFest details 15:07:43 #link https://ansiblefest2019.eventpoint.com/cfp 15:07:58 DEADLINE: May 28 15:08:05 #info AnsibleFest Call for Proposals closes May 28th 15:08:10 thanks samccann 15:08:22 sometimes I remember irc commands! 15:08:31 have you a link to last year's talks handy? 15:08:32 more often than I do, apparently 15:09:03 for inspiration :) or to judge the level 15:09:20 kmaxwell: good question - I found some videos here: https://www.ansible.com/resources/videos/ansiblefest-austin-2018 15:10:44 ty! I remember watching some of those 15:10:48 #link https://www.ansible.com/resources/videos/ansiblefest-austin-2018 to get inspiration from last year's event 15:10:50 and here's a blog post from Jeff Geerling about the Contributor summit: https://www.jeffgeerling.com/blog/2018/things-i-learned-ansiblefest-austin-2018-contributors-summit 15:11:13 #link https://www.jeffgeerling.com/blog/2018/things-i-learned-ansiblefest-austin-2018-contributors-summit for details on the Contributor Summit that same week 15:11:24 * samccann on a link roll today! 15:11:44 * samccann tho prefers sushi rolls 15:12:14 * acozine is getting hungry now, and lunch is still hours away 15:13:08 last year's schedule seems to have disappeared into the Internet of the Past - all the links i can find point to the 2019 CfP 15:13:57 #topic PR review and stats 15:14:58 in a conversation with dag yesterday, I discovered that some PRs that update the module docs have the label `docs` . . . we are working hard to review all documentation-related PRs and merge as many as we can 15:15:37 so if you see a PR that is docs-related, and it doesn't have the `docs` label, add it if you can, or post a link here in the #ansible-docs channel 15:15:46 or bring it to this meeting for consideration/review 15:16:08 grrr. . . edit fail 15:16:23 Some PRs that update the module docs DO NOT have the label `docs` 15:16:34 so keep an eye out for those 15:16:41 #info not all docs PRs have the docs label if it's module docs. If you see any, add the docs label or post a link here in #ansible-docs channel 15:16:59 speaking of stats, we are down to 61 open PRs for docs! 15:17:21 hooray! though with ^^^ it may be a few more than that 15:17:51 #topic The Version-Changer 15:17:59 take it away, samccann ! 15:18:30 woo hoo! So I finally have a version changer mostly working. You can see it 'in action' on the test site - http://docs.testing.ansible.com/ansible/latest/index.html 15:19:25 keep in mind that when you go 'back' a version, you may hit a 404 if the page didn't exist in that release. also, there are some older version changer hacks on the older test sites (devel, 2.6) that you may see. 15:19:55 Once we merge this, we'll have to do a fast backport to 2.8, 2.7, 2.6 to get them all working together so to speak 15:20:08 WIP pr is https://github.com/ansible/ansible/pull/55655/files 15:20:17 it looks great samccann 15:20:30 up in the top section, right where you'd expect to find it 15:20:32 (ignore the version.html changes - an older fix I have to back out of the PR now that I have it working at the top of the left-hand navigation) 15:20:50 I like it! 15:21:02 and I'd love reviews on the WIP PR - my javascript/html knowledge is straight from Prof. Google 15:21:43 (struggling not to mention that somehow latest says devel just above the box, that's maybe something else) 15:21:55 yeah that's worth reminding me of for sure! 15:22:11 I 'think' it has to do with how I published to the test site, but I need to verify that 15:23:00 #info version-changer PR is https://github.com/ansible/ansible/pull/55655/files and needs reviews. Ignore version.html as that needs to be removed from the PR 15:23:23 #info - also - determine why the test site shows devel as the version when it is 2.7 15:23:31 just so I don't lose that thought! 15:23:49 kmaxwell: don't struggle - all comments and observations are welcome 15:24:10 better if I just asked what stage are you at? I like the direction you're going 15:24:42 happy to try some detailed testing whenever you're ready 15:24:45 I need to clean up the PR. Then do a 'dummy' backport to older versions to make sure the back n forth work 15:24:53 so https://github.com/ansible/ansible/pull/55655/files#diff-0b7e420875907c4ef7d35eac6fa3e2deR247 will eventually need to read `2.6 2.7 latest devel`, right? after 2.8 is released? 15:25:02 that would be great! 15:25:23 samcann: have you decided how to handle pages that don't exist in ealier versions? 15:25:32 #info versions need to be 2.6 2.7 latest devel to catch up with the upcoming 2.8 release 15:25:37 yep, need to fix that as well 15:25:42 I think the `devel` next to latest is coming from https://github.com/ansible/ansible/pull/55655/files#diff-46b6ec95f1914057dc059bd9e2ea6f7bR158 15:26:00 the example I stumbled into is http://docs.testing.ansible.com/ansible/2.6/scenario_guides/cloud_guides.html 15:26:10 oh, or maybe that's from the Jenkins job . . . 15:26:45 maybe for the first implementation of the version changer you don't need fancy handling 15:27:18 at an earlier meeting we agreed that having a better 404 page is a pre-requisite for the version-changer 15:27:28 there's an issue for it: https://github.com/ansible/ansible/issues/51439 15:27:43 I thought I remembered as much, glad you're on top of it 15:27:55 any chance you'd like to pick that one up kmaxwell ? 15:29:01 just having a read of it now 15:29:59 will come back to your question, if we you'd let me just flesh out my understanding a bit more 15:30:30 so we add a custom 404 to devel as per issue description 15:30:42 which means devel / 2.9 will have custom 404s 15:31:04 does that docsite change then get backported to 2.6 2.7 2.8? 15:31:22 I'd like to backport both the version-changer and the custom-404 to all supported versions 15:31:47 preferably the 404 would be "version-aware" 15:32:11 so if you hit a missing page on the 2.6 docs, it would give you links to the 2.6 index page, or section index, or something 15:32:56 once 2.8 is released (this week), the supported versions will be 2.6, 2.7, 2.8, and the devel branch 15:33:08 I think I was reading today that support is 3 versions / c. 12 months 15:33:26 yes 15:33:27 you beat me to that ^^^ :-) 15:33:37 Ansible 2.5 was released about 18 months ago 15:33:40 heh 15:34:41 I will comment on the issue to add some of this detail, and if I can make the time try an implementation 15:34:51 I think all those stable branches have docs infrastructure that's recent enough that we could backport a custom 404 and the version-changer, though the proof is in the eating, as they way 15:34:57 s/way/say/g 15:35:05 kmaxwell: awesome, thank you! 15:35:10 be sure to note https://github.com/ansible/ansible/issues/51439#issuecomment-481331515 15:35:14 ;) 15:35:24 acozine: to answer your question in a pretty non-commital way :-) 15:35:33 hey, you're a volunteer 15:36:07 I don't even get to volun-tell staff, let alone community folks 15:36:11 kmaxwell: just add anything you learn/think of along the way to the 404 issue. that way someone else can pick it up if you don't get the chance 15:36:22 is the version changer going to block on the custom 404? 15:36:30 nope, it's the other way around 15:36:41 * acozine re-reads that 15:36:55 yes, the version changer is blocked by the custom 404 15:37:06 404 has to come first is another way of saying the same thing 15:37:07 we already have a bad UX with missing pages 15:37:13 let's not make it more visible 15:37:21 or higher-traffic 15:37:40 kmaxwell: yeah, I'm just hopped up on adrenaline and caffeine 15:37:45 it's release week!!! 15:38:06 :-) 15:38:11 skim everything and leap to conclusions! 15:39:04 samccann: the version-changer is looking awesome, thank you for all your hard work on it 15:39:16 that Prof. Google, she's a pretty good teacher 15:39:39 * bcoca dances with versions 15:39:40 heh 15:40:01 does it redirect you to 'upgrade now' page when selecting 1.9? 15:40:13 bcoca: heh 15:40:43 it should redirect to a very special cowsay picture in that case 15:40:49 with a large . . . pile 15:40:54 +1000K 15:41:14 #topic open floor 15:42:13 I thought dag and Pilou would be here today and we'd have a full review of https://github.com/ansible/ansible/pull/45805 15:42:29 maybe next week . . . 15:42:37 anyone have other stuff to bring up? 15:42:41 PRs to review? 15:42:49 Issues to highlight? 15:43:04 acozine: I am here 15:43:15 #chair dag 15:43:15 Current chairs: acozine dag kmaxwell samccann 15:43:23 dag: you've been very, very quiet 15:44:07 do you want to give a short summary of the arg_spec-to-module-docs PR? 15:45:15 acozine: I have been very, very busy ;-) 15:45:37 well, that PR was not finished because it would not be accepted 15:45:45 ah, okay 15:46:19 because some hidden roadmap had foreseen changes to this area 2 versions later 15:46:59 hm 15:47:02 Maybe you know more about this ? 15:47:33 maybe? it's not immediately springing to mind 15:47:44 since then I know there had been internal discussions whether the arg_spec needed to be (completely?) in documentation, or documentation needed (completely?) in the arg_spec 15:47:55 dag: that discussion never ends ... 15:48:02 has been 'vaporware' since 2.1 15:48:02 bcoca: ok :-) 15:48:20 I don't think we can have really good documentation that lives entirely in the argspec 15:48:24 why i just did plugin config ... and ran 15:48:25 right, but I was told this would never be accepted because of that 15:48:39 acozine: agreed 15:48:51 acozine: why plugin config uses the docs 15:48:57 acozine: that's why I still think my approach is best, even if we move something around later something like this will be needed 15:49:04 so my concerns are more about how to integrate the parts that *are* in the argspec with the parts that are not 15:49:45 right now we make it easy for people to correct mistakes in the documentation, and I want to preserve that ease-of-use 15:49:59 tangentially related - the network team has used a 'resource model' that will autogenerate both the documentation and the arg_spec 15:50:06 we may be able to adopt dag's change *and* preserve the ease-of-use 15:52:04 acozine: basically required/default/type become 'programing' while description and rest stay 'pure docs' 15:52:12 ^ choices also 15:52:20 dag: do you have time to get the PR updated (passing CI) so we can have a full discussion about it at a WG meeting next week or the week after? 15:52:54 acozine: but it would not only be docs issue, that really affects core more than anything 15:53:21 bcoca: what do you mean? 15:53:37 it changes how modules are 'programmed' 15:54:20 it does? I thought it pulled existing information from the argspec to produce parts of the documentation 15:54:43 bcoca: so what would be the right venue/group for reviewing the PR, then? 15:55:02 but it makes a decision on how the programming is supposed to work 15:55:07 core meeting 15:55:33 okay 15:55:55 I don't habitually go to the core meetings, but I'd like to be there for this one 15:57:15 bcoca: so would dag's PR require changes to most (or all) of the existing modules? 15:57:47 iirc all 15:58:06 (logs of core meeting where this PR was discussed: https://meetbot.fedoraproject.org/ansible-meeting/2019-03-19/ansible_core_irc_public_meeting.2019-03-19-19.05.log.html#l-137) 15:58:11 since you are moving the docs from the 'inline yaml' into argspec 15:58:36 it doesn't let us override what's there as an interim step? 15:58:46 Pilou: thanks for the link 15:59:05 i forget, have not looked at that in months 15:59:22 `19:40:56 +1 to intent, but -1 to implementation` 15:59:33 ^ i would probably stand with that again 15:59:35 well, we're not going to make much progress in the one minute we have left in this meeting 16:00:10 acozine: my PR does not require changes to module docs 16:00:39 acozine: it would use the information from the argspec, unless it is also in the documentation (it prioritizes docs over arg_spec for good reason) 16:00:46 dag: it falls back to current docs? 16:00:52 ah, k 16:00:58 i didnt remember exactly 16:01:11 the reason for this is that in some cases people want the docs to be different then the implementation 16:01:27 e.g. type=raw for backward compatibility, but we prefer people to see type=list 16:01:57 ah, that's a classic "do what I say, not what I do" 16:02:03 or there is no default in the arg_spec, but there is a default in the documentation 16:02:26 because there is a default in the implementation (may be conditional, etc...) 16:02:42 so the arg_spec is always correct, but not always wanted 16:02:57 dag: that makes sense 16:03:45 so do you have time (and the desire) to update the PR (again, I know)? 16:03:49 sure 16:04:26 okay, ping me when it's ready for the next round and I'll add it to the Docs WG agenda and suggest it for the Core agenda also 16:04:36 https://github.com/ansible/ansible/pull/45805#issuecomment-422767850 16:04:54 The only thing we need to modify is how we use mock in the code 16:05:27 as per comments from @abadger 16:05:46 well, it needs a rebase - I think that's why it's failing CI 16:06:11 the support for suboptions is another problem, AFAIK suboptions are also currently not support by validate-module 16:06:28 3 conflicts 16:06:37 they should be pretty easy 16:07:20 okay, we're 7 minutes over time for this week 16:07:42 thanks kmaxwell bcoca dag Pilou samccann for another great DaWGs meeting 16:07:45 #endmeeting