20:00:00 #startmeeting Ansible Windows Working Group 20:00:00 Meeting started Tue Nov 23 20:00:00 2021 UTC. 20:00:00 This meeting is logged and archived in a public location. 20:00:00 The chair is nitzmahone. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 20:00:00 Useful Commands: #action #agreed #halp #info #idea #link #topic. 20:00:00 The meeting name has been set to 'ansible_windows_working_group' 20:00:04 bam 20:00:09 #chair jborean93 20:00:09 Current chairs: jborean93 nitzmahone 20:00:12 'allo 20:00:17 hiya! 20:00:18 yo 20:00:21 hello! 20:00:24 #info agenda https://github.com/ansible/community/issues/581 20:00:36 welcome back mkletz 20:00:41 thank you! 20:00:49 Since the composite DSC PR got merged, I assume we don't need to talk about that anymore? 20:00:56 Yea I mostly covered that with https://github.com/ansible-collections/ansible.windows/pull/313#issuecomment-974910937 20:01:03 yeah we just had the CI ignore the issue 20:01:33 Essentially the way that PowerShell parses that particular file required OMI to generate the MOF (or something). Since this is really a file just for testing we just skipped that from pslint 20:01:33 I had a question about the changelog fragments 20:01:44 nice 20:01:46 Go for it 20:01:57 #topic open floor 20:02:00 So the changes for composite resources are likely going to take a significant refactor 20:02:18 I'd like to keep PRs as small as possible so they're reasonably reviewable 20:02:36 How to reflect just a refactor in the change log? 20:02:53 it didn't seem appropriate to be a feature or a bug 20:02:58 I think usually, the changelog is for user-facing stuff, so may not need an entry at all 20:03:01 unless changes behavior 20:03:09 agreed 20:03:09 ok 20:03:11 so in https://github.com/ansible-collections/ansible.windows/pull/315 20:03:11 yep 20:03:16 I should remove the changelog fragment? 20:03:41 Can go either way for that 20:04:10 kk 20:04:11 We'll still often include things like that just so people know there might be a destabilizing change to a module, but it's 50/50 20:04:16 got it 20:04:33 that particular one I'd remove personally, unless maybe the error messages will be changed by it 20:04:52 nope I left all the errors the same 20:04:56 I just de-duped the exit logic 20:05:12 technically changing error messages could be "breaking" changes even though they aren't treated like it, because we sometimes base behavior and conditionals on module message output 20:05:35 right 20:05:49 * nitzmahone grumbles about a core team discussion just this morning on that topic ;) 20:05:55 lol 20:06:03 heh, I presume the relevant xkcd was posted 20:06:18 https://xkcd.com/1172/ 20:06:36 hahaha 20:06:38 indeed it was 20:06:44 (well, referenced at least) 20:07:21 I was going to work on 1 more PR before I start refactoring to standardize parameter blocks and quotes. I assume that's the type of change we wouldn't bother with a changelog on? 20:08:37 Yeah, so long as it's not likely to cause user-visible changes in behavior, that's usually fine 20:08:45 just as an FYI we try to avoid style changes unless it's something to satisfy pslint 20:09:24 gotcha, I saw some discussion around making the quotes a default rule in PsScriptAnalyzer but if that's the case I will avoid doing so 20:09:59 easy enough to resolve when the linter starts complaining about it 20:10:01 would be interesting to see how well PSSA works with that rule, there are many cases when you would want either a double or single quote so hard to really enforce that 20:10:24 let me see if I can track it down 20:10:27 what are the specifics of this rule? 20:10:38 sounds like another one of those potentially overly-pedantic rules 😐 20:10:50 ^ am thinking the same 20:11:16 I'm find with those rules if 1. there's an automatic formatter to fix violations, 2. it isn't full of edge cases 20:11:25 yep 20:11:33 At least with python `'` and `"` are mostly the same 20:11:53 If there end up being a ton of ignores/exceptions for legit reasons, that rule's a good candidate to be disabled 20:12:02 https://github.com/PowerShell/PSScriptAnalyzer/pull/1470 20:12:10 looks like it's been an opt in rule for awhile 20:12:58 but yeah the gist of the discussion seems to be there is "technically" a performance difference but it's like more disruptive and pedantic than it's worth 20:13:29 yea I would struggle to justify such a rule in PowerShell personally 20:13:46 the main benefit I could see with that rule, is that does have better performance for me, cognitively 😅 20:13:54 lol 20:13:55 I feel like I do less mental processing of the string when I see it single-quoted 20:14:08 I agree but I won't die on that hill 20:14:16 oh absolutely agree 20:14:33 at least it does have automatic formatting to fix it :) https://github.com/PowerShell/PSScriptAnalyzer/blob/master/Rules/AvoidUsingDoubleQuotesForConstantString.cs#L90 20:14:39 I tend to use single quotes nearly everywhere I can unless I'm following convention in existing code, but yeah, not going to go crazy over it 20:14:47 * nitzmahone has to slap hand numerous times every time he switches back to a curly-bracket-scoped language where single quotes don't mean the same thing 20:15:19 Heh, I wonder how bug-ridden that auto-formatter is... 20:15:38 I've actually only ever had it break a file once after using it quite a bit over the years 20:15:44 (we found a couple "fun" bugs recently in other rules) 20:16:03 The bug that drives me nuts 20:16:18 is that the unused variable detection can't see into child scopes 20:16:24 I'm hoping to try and fix it sometime this week once I finish my current thing 20:16:46 So if you have a parameter that is only used within a where-object or something 20:16:48 nah, it autoformats `@{foo='bar'}` to `@{foo = 'bar' }` (the extra space at the end) 20:16:55 You might see some "creative" spacing until they fix it 20:17:06 --^ that 20:17:12 https://github.com/PowerShell/PSScriptAnalyzer/issues/1742 20:17:17 `is that the unused variable detection can't see into child scopes` heh I just naswered a question about this on stackoverflow recently 20:17:25 and then closed another one as a duplicate a day or 2 later 20:17:59 https://stackoverflow.com/a/69847506/3905079 20:18:17 mkletz: any other discussion on next steps for the win_dsc stuff? 20:18:37 I'd like a sanity check on my general idea 20:19:16 so since a composite resource is many resources 20:19:34 I was thinking a bit of logic that basically determines how the objects to invoke are supplied 20:19:42 and have it end iterating through them at a single point 20:19:47 instead of two distinct workflows 20:19:50 if that makes sense? 20:20:34 +1 from me in one form or another; for my vote, how cleanly the impl can separate the composite/not cases and if there are any new deps required will determine whether it's immediately accepted as a change to win_dsc in ansible.windows or a new module in community.windows first. That sounds like a pretty sane way to do it. 20:20:58 I figure that way we can also slowly introduce the change over a few PRs 20:21:03 yea to me the devil is in the details and it's hard to comment without seeing the actual work 20:21:07 ditto 20:21:22 right, it's sort of a loose idea in my head at the moment. I'm still getting familiar with the win_dsc code 20:21:35 it's definitely one of the more gnarly ones out there 20:21:44 getting the module invocation wasn't fun 20:21:52 yeah, and we might actually have a customer use case for this now also 20:22:13 so more motivation to work on it 20:22:22 I wouldn't be surprised if I end up using the implementation at work as well :) 20:22:25 Though it still sounds like the majority of the code is applicable to both cases, so it's really just "do we need to do the composite MOF compile step, and how many resource apply's are we doing, 1 or N?" 20:22:36 yeah basically 20:22:46 this is how the thing goes 20:22:50 does it go once or a lot 20:23:32 So if the impl can very cleanly separate those cases and keep the signal path clean for the existing stuff, I'd be +1 for just pulling it in to the supported module, but the code will have to do the talking ;) 20:23:43 yup for sure 20:24:07 I think I agree with what Matt said 20:24:09 (but something as simple as adding a loop that might be 1 is much preferable to duplicating the code) 20:24:42 especially the win_dsc code 20:24:51 yeah, I just need to spend some time getting more familiar with that code needs to be in said loop 20:25:00 what code* 20:25:27 I'm confident it can work, just how pretty it is remains to be seen 20:25:40 We're waiting with bated breath ;) 20:25:44 lol 20:26:34 Anything else for today then? 20:26:42 not from me 20:27:22 nor me 20:27:37 Cool- til next week then! Have a great Thanksgiving, for the US folks... 20:27:42 I'm all good 20:27:49 #endmeeting