19:01:25 #startmeeting ansible core public irc meeting 19:01:25 Meeting started Tue Jan 15 19:01:25 2019 UTC. 19:01:25 This meeting is logged and archived in a public location. 19:01:25 The chair is bcoca. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:25 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:01:25 The meeting name has been set to 'ansible_core_public_irc_meeting' 19:01:29 @dag? 19:02:36 k, skipping his issues i believe he has conflicting meeting now 19:03:04 #topic https://github.com/ansible/ansible/pull/50705 19:03:15 federico olivieri? 19:03:45 and then 19:03:56 * gundalow waves 19:04:50 well, unsure what pettitioner wants and also seems more appropos for network meeting 19:06:05 #topic https://github.com/ansible/ansible/issues/50503 19:06:07 fun one 19:06:19 @sdoran, want to make case? 19:07:06 This is tricky. 19:07:25 guessing that is why you wanted it brought up here? 19:08:29 I put it on the agenda because it was eating up triage time. 19:08:37 And we should discuss it further. 19:09:18 I think warning is definitely a good idea. Since right now it silently transforms the data. 19:09:43 I lean that way as well 19:09:50 i think config toggle to a) stay as current, b) warn c) error by making it strict conversion 19:10:04 Definitely not a failure- if we want that, it should be a different type, eg, `stringnoreallyimeanit` 19:10:06 I’m not sure on the solutions mentioned. A command line flag was brought up as a way to enable it. 19:10:16 module_string_conversions: transparent|warn|strict 19:10:22 Would a config option be “too sticky”? 19:10:37 -1 to the command line flag, but I could get behind a config opt-out 19:10:45 well, i would say requiring a cli option makes it too variable and plays wont work the same 19:10:54 ETOOMANYCLIOPTS 19:10:58 I was thinking that too. 19:11:02 hi everyone 19:11:05 while config will make it 'stable' 19:11:08 * bcoca waves 19:11:14 Plus another option to (maybe) our in the Tower UI. 19:11:14 felix! 19:11:37 So a config option makes the most sense. 19:11:51 Hopefully someone won’t set it and forget about it. :) 19:12:09 we seem to have tacit agreement on making this a toggle 19:12:14 Defaulting to warning on conversion to string makes sense to me 19:12:29 and .. i was going to say .. we can now start fighting about the default 19:12:45 +1 for warning by default 19:12:52 error by default will probably break way too much 19:12:54 The people that need it most are the ones that will never opt-in to checking 19:12:55 @bcoca So the config option has three values? 19:13:13 i would say +1 for 'as is', change to warning in future version and to strict in even more future version 19:13:23 Just want to make sure I understood what you described earlier. 19:13:37 sdoran: that is my thought, many wont care about warning, we have a lot and no good ways to switch off 19:13:39 basically ignore/warn/error 19:13:49 yep 19:13:56 ^ i like those values better 19:13:57 Currently we have “ignore”. 19:14:02 I'm not sure what the benefit of waiting to make warn the default is though 19:14:21 in particular if this one's easy to turn off with the correct config value 19:14:24 nitzmahone: minor, just ensure toggle works before we make it the deafult 19:14:31 i can be swayed on that, easily 19:14:58 We've got enough time before 2.8 freezes that if we do it now, we can revisit/fix if there are problems 19:15:20 roger, sdoran want to add ticket/project with what we decided here? 19:15:43 I’m trying to think what that’ll look like to most users. Trying to avoid warning fatigue. 19:15:44 or before that, any opposition to solution? a) add toggle with ignore/warn/error, b) default to warn 19:15:58 I'd be much more conservative about that if we were defaulting to `error`, but with warn-as-default, worst case it's an annoyance 19:16:06 that is my only concern, we went from 'not enough info' to 'nagging 19:16:27 but that is very subjective threshold 19:16:34 Can they change something easily to make the warning go away? Quote the value maybe? 19:16:37 But I don't think delaying changes that- most users aren't digging through the config base looking for new knobs to turn ;) 19:16:42 Exactly 19:16:50 Ok. 19:16:56 Quoting the value will make it a string, so it's a very easy fix 19:17:07 maybe set it to warning on devel, and directly after branching stable-2.8, set it to ignore on there? so that at least on devel it's warning all the time? 19:17:15 As long as we explain the fix in a well crafted warning, I’m ok with warm being the default. 19:17:26 It'd be harder if we were going the other way (eg, warning on string to int conversion) since a number of places have to do that by default 19:17:34 That seems helpful since it’s brining to light something they may not have been aware of. 19:17:35 add config entry to warn message to 'turn off' and i think we are golden 19:18:00 *warn 19:18:16 (Warm being the default would be an interesting setting) 19:18:17 eg `this message can be disabled by setting BLA to "ignore" in ansible.cfg` 19:19:05 and as felixfontein points out, we can change default before release if we see it causes too much distress 19:19:08 maybe it would also help if that feature could be toggled per task/include/...? or is that too much complexity? 19:19:22 If the config entry includes a var, that would work 19:19:41 it could, since its task level and vars are present 19:19:42 (well, and if we look it up in host context) 19:19:53 but i think a warnings: on|off task level feature might also work 19:20:01 that would allow people to convert their stuff role by role / task by task, depending on how many warnings they get 19:20:44 I'd go for small (define a var on the config entry) for now rather than new play/task synatx 19:21:14 That seems too wordy. If you’re togging warnings at the task level, just fix the task. You’re already there. 19:21:25 Let’s start with a global config setting. 19:21:43 im ok either way, i do see var as bit of overkill but im fine with it 19:22:13 going to clsoe this as 'fix by toggle' we can bikeshed on teh fix itself once we have PR 19:22:30 Can I write this up in a WIP PR? Or is there a better context? 19:22:32 #action sdoran to add ticket to 2.8 project for 'string conversino warning toggle for module optoins' 19:22:39 works4me 19:22:44 K. 19:25:17 #topic https://github.com/ansible/ansible/pull/49138 19:25:26 yahoo 19:26:36 so im going to guess, you want someone else to review so new module and category can be added? 19:27:11 im not sure how the process works, but that sounds about right 19:27:46 i see felixfontein already gave it attention ... anyone have time to give this a once over? 19:28:06 alb0t: we are down a 'cloud team' so its been hard to get time for core to look at things 19:28:56 isnt this more a dtabase than a cloud thing? 19:28:57 bcoca: no worries, I can wait. We're already using it in production so reaping the benefits, so I'm not in a rush… but I think it would be useful to a lot of other people. 19:29:07 Yeah it's for a NoSQL database. 19:29:33 sorry its correclty under /database/ ... i must be seeing things, could swear it was /cloud/ 20s ago 19:29:43 * bcoca looks for coffee IV 19:29:48 :) 19:30:08 head in the clouds :P 19:31:10 k, if no one else i'll add to my list .. but its a long list, i would try as felixfontein suggests and ping users during community meetings 19:31:29 #topic open floor 19:31:42 it looks like gunadlow added an approval 19:31:54 I think this PR is looking for core reviews: https://github.com/ansible/ansible/pull/50335 :) 19:32:13 gundalow qualifies .. last i looked 19:32:41 he should have rebuild_merge or shipit, approve wont get it merged by bot 19:33:03 new modules aren't automerged anyway, I thought? 19:33:17 so rebuild_merge 19:33:44 ah, that one does? nice :) 19:34:35 just cause stale CI 19:34:40 forces new ci run 19:34:55 I mean merge things ansibot usually doesn't automerge 19:35:02 thanks everyone :D 19:35:38 felixfontein: yep, you should be able to use that for docker stuff now 19:37:10 haven't really had a chance to try it out yet ;) 19:40:46 alb0t: congratulations for your first module :) 19:41:26 (assuming there are no surprises ;) ) 19:42:27 felixfontein: thanks for all your help. I've learned a lot :) and will be brushing up on my python more to make this more stable 19:43:54 you're welcome! 19:45:58 ok, if no new business in next min, closing meeting for today 19:46:11 *wave* 19:46:33 ok, I got something 19:46:41 https://github.com/ansible/ansible/issues/50319 19:47:13 most docker modules have various options which require minimum API versions and minimum docker-py versions. we've collected a lot of them so the modules do checking on startup 19:47:32 it would be nice to also have that in the documentation, but without doing the work twice (i.e. have two different datasets) 19:47:53 does anyone has a good idea how to do that? 19:52:04 * cyberpear waves 19:54:02 * felixfontein waves back 19:54:07 any time for https://github.com/ansible/ansible/pull/50327 ? 19:54:39 thought we had already discussed that 19:54:54 k, didn't know if we needed followup 19:55:07 I made the changes discussed 19:55:15 i'll add to my review list, unless someone else wants to tackle first 19:55:50 thx 19:56:02 #endmeeting