15:31:59 #startmeeting Ansible Documentation Working Group 15:31:59 Meeting started Tue Nov 20 15:31:59 2018 UTC. 15:31:59 This meeting is logged and archived in a public location. 15:31:59 The chair is acozine. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:31:59 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:31:59 The meeting name has been set to 'ansible_documentation_working_group' 15:33:03 I don't have much experience with the mechanics of IRC meetings, so today will be an experiment 15:33:13 #chair 15:33:13 Current chairs: acozine 15:33:23 #chair samccann gundalow bcoca 15:33:23 Current chairs: acozine bcoca gundalow samccann 15:33:29 who else is around? 15:33:39 o/ 15:33:51 lurkers will appear when they have input 15:34:16 * acozine sets up searchlight 15:35:15 today's first topic is the loop docs 15:36:00 we've got an open PR from our meeting a few weeks back that updates the main loop page, and a few open PRs that change specific examples elsewhere in the documentation from `with_*` to `loop` 15:36:21 main loop page update PR: https://github.com/ansible/ansible/pull/47895 15:37:13 full list of PRs with "loop" in the title: https://github.com/ansible/ansible/pulls?utf8=%E2%9C%93&q=is%3Apr+is%3Aopen+label%3Adocs+loop+in%3Atitle 15:39:35 there's a comment on PR 47895 that hasn't been addressed - and I think if we can address it, we can decide on all the other PRs 15:40:43 in other words, once we can define and recognize use cases that should continue to use `with_*`, we can either merge or close the rest of the PRs that update `with_*` to `loop` 15:41:36 the comment is: https://github.com/ansible/ansible/pull/47895/files#r229865255 15:41:41 thoughts? 15:42:02 #topic documenting loops 15:44:59 okay, let's look at one of the specific examples - should this one be changed from `with_items` to `loop`? https://github.com/ansible/ansible/pull/47215/files 15:47:11 * gundalow waves 15:47:17 in another meeting 15:47:35 there's a warning in the main loop page update that says be careful changing with_items to loops because of the implicit flatten 15:48:08 I'd be inclind to say we shouldn't change this in PR 47215 because we don't know what's in foo. 15:48:17 (inclined) 15:48:37 yeah, and I don't entirely understand that - it would be great to have an example task, contrasting what happens if you use `with_items` and what happens if you use `loop` 15:49:11 anybody have a good example? 15:50:12 didn't someone talk about that a bit back, with an example? that if foo was a list of families, and it ended up something like [samccann, [acozine,mrMan,MrCat],gundalow] 15:50:33 that with items flattens that to 5 items. but loop treats it as only three? 15:50:45 * samccann barely knows what she's talking about here...just so's ya know 15:50:55 that sounds like a great example - was that here in IRC? 15:51:22 i'm not sure. I just know someone else said something like that 'somewheres' and it made sense to me 15:51:53 might be time to just experiment offline w/ a task that does something like that and displays the result to show the difference? 15:51:56 gotcha - well, it's enough info for us to construct an experiment from 15:52:06 samccann: exactly 15:52:29 okay, let's see if any of the other proposed changes use something other than `with_items` 15:53:42 gosh they are mostly replacing `with_items` 15:53:45 hm, nope, they're all `with_items` 15:54:02 I wonder if we have examples of any of the other `with_` lookups? 15:54:29 my nickel - we do our experiment, update the main loop page with the results, and close these other prs with a pointer to that to show why we don't want to change 15:54:59 unless we can see in each of those other PRs some specific evidence that the with_items is a single list, not..erm.. nested?? (not using the correct terms here) 15:55:03 agreed - though we might be able to change one or two of them, depending on what kind of data they're likely to use 15:55:18 yeah, that's a better way of saying it :-) 15:55:28 okay . . . progress 15:56:13 still leaves us w bcoca 's original comment on the main loop page 15:56:30 https://github.com/ansible/ansible/pull/47895/files#r229865255 15:56:33 #agreed samccann and acozine to experiment with silly-but-easily-understandable data 15:56:52 which flies so far over my head I'm not even feelin the breeze as it goes by :-) 15:57:50 bcoca: could you restate the comment, or offer an example? 15:58:50 what would someone be doing that would require multiple lookups in a loop? 16:01:51 it's mighty quiet here today 16:02:35 Turkey week in the US :-) 16:03:12 yeah, lots of folks away, plus whatever other meeting gundalow is in going on - let's make this week's DaWG meeting a short one 16:03:19 #topic open floor 16:03:38 anybody been lurking so far, with a burning topic to discuss? 16:04:30 hi Delph_ 16:05:20 okay, let's hold over the other agenda items for next week 16:05:37 happy Thanksgiving to those who celebrate it! 16:06:16 hearing no other topics . . . 16:06:29 #endmeeting