19:12:13 #startmeeting ansible core public irc meeting 19:12:13 Meeting started Tue Jul 23 19:12:13 2019 UTC. 19:12:13 This meeting is logged and archived in a public location. 19:12:13 The chair is bcoca. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:12:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:12:13 The meeting name has been set to 'ansible_core_public_irc_meeting' 19:12:20 \o 19:12:30 #topic https://github.com/ansible/ansible/pull/58646 19:12:32 o/ 19:12:37 o/ 19:12:44 ^ acozine brought up last meeting , but we had no quorum 19:13:03 previous vote was to always show 'type: str' by default , even when abset (cause its the actual implied default) 19:13:24 but acozine worried this will show 'explicitly incorrect' docs for many modules (aroun 1400, 4000 options are mismatched) 19:13:43 and requested we vote to add the check, but hold off on 'default display to str' until modules were converted 19:13:49 s/converted/corrected/ 19:14:01 we had 4/0 last meeting, but need more peopel to weigh in 19:14:10 o/ 19:14:28 ^ feel free to update/correct 19:14:33 I still feel that incomplete docs are better than inaccurate docs 19:14:53 I'd rather not display the type than display the wrong type 19:15:27 "When the default argument type was changed to str, we did not change the behavior of ansible-doc to display type: str if no type is specified in the module documentation. Rather than requiring every string parameter be documented as such, we should insert type: str if it isn't present for a given parameter i" 19:15:30 seems logical 19:16:26 giving a fewmins for people to catch up, but starting vote 19:16:49 +1 to existing PR changing checks, but removing the default display part 19:17:01 jtanner: it is logical, but unfortunately "no type is specified in docs" != "type is actually a string" 19:17:40 +1 to checking, but not displaying any default until modules with missing docs are updated 19:20:12 jillr is probably right, i just don't think that they'll all get updated 19:20:42 that is my concern also, but matt clay has assured me people do go over ignore.txt and udpate modules to fix the issues 19:20:59 Yep, +1 to the changes, but don't display that value until we're sure it's accurate 19:21:05 i would love a 'warning' on ci that 'you changes to module A are good, but it has these minor outstanding issues : ' 19:21:17 ^ but that is a lot to ask 19:21:23 "hey $contributor, since you're in the neighborhood..." 19:21:28 exactly 19:21:56 Could be problematic for the Windows modules that haven’t been converted to the new arg spec 19:22:02 makign it an error seems like punishing contributors (unless we can match they are the ones that created the issue) 19:22:20 jborean93: agreed, and i would have voted otherwise if i had realized that was still an issue 19:22:38 but this vote should leave us with a check and a 'long list' of todo in fixing modules 19:22:43 Plus action plugins don’t have an arg spec 19:22:46 Yeah, adding the windows modules that haven't been updated to ignore would make sense 19:22:54 jborean93: that is what 'controller side argspec' is for 19:22:59 any day now ... 19:23:12 I thought that was implemented in 2.4 :P 19:24:00 s/2.4/since 1.4/ 19:24:32 I think we could be smarter and only error if type != string on modules we can validate the arg spec for 19:25:00 Currently validate modules will just skip the validation if it can’t read the arg spec (I.e. old PowerShell type) 19:25:46 i think that is how the previous and new check would still work 19:26:15 i basically merged 3 checks into one simpler one (args mismatch docs) 19:26:28 didnt do anything about general validate-modules execution rules 19:28:27 k, so factoring in last meeting i think i can say +6/0 .. going to consider it 'passing' 19:28:38 #topic open floor 19:32:52 k, since nothing has com up in 5m, ending meeting 19:32:56 #endmeeting