14:02:05 <mark-burnett> #startmeeting airship
14:02:06 <openstack> Meeting started Tue Sep 25 14:02:05 2018 UTC and is due to finish in 60 minutes.  The chair is mark-burnett. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:09 <hogepodge> Hi
14:02:10 <openstack> The meeting name has been set to 'airship'
14:02:31 <mark-burnett> I didn't really prepare an agenda, though: https://etherpad.openstack.org/p/airship-meeting-2018-09-25
14:02:33 <mattmceuen> hey mark-burnett hope you're feeling a bit better :)
14:02:54 <mark-burnett> yeah, a bit better, still a bit of a cold though
14:02:59 <mark-burnett> :)
14:05:25 <mark-burnett> thanks everyone for adding items to the agenda
14:06:12 <mark-burnett> #topic Meeting TIme
14:06:30 * roman_g o/ and left
14:06:42 <mark-burnett> Looks like no change here - we had one or two new additions to the doodle, which look good for the current time.
14:07:13 <sthussey> Let's close it
14:07:25 <mark-burnett> Yeah, we can't keep this issue open every week for months
14:07:33 <sthussey> This thing is just floating and the people asking for it aren't speaking up
14:07:40 <sthussey> If they really care, they'll ask again
14:07:45 <mark-burnett> Agree
14:07:49 <mark-burnett> Ok, let's move on
14:08:05 <mark-burnett> #topic -2 Reviews
14:08:14 <mark-burnett> @portdirect asked that we discuss this again this week
14:08:53 <mark-burnett> I don't know whether he's here?
14:10:20 <mark-burnett> well, it seems like we aren't going to have much discussion around this
14:10:25 <mattmceuen> lol
14:10:32 <mattmceuen> Channeling my inner pete
14:10:34 <mark-burnett> two meetings in a row and nothing to say on the topic
14:11:04 <portdirect> Sorry been held up
14:11:14 <mattmceuen> "-2 reviews are typically reserved for strong responses when a PS doesn't just need to be changed, it needs to be abandoned"
14:11:15 <portdirect> Give me 5 if you can
14:11:49 <mark-burnett> well, we can move on and maybe come back
14:12:17 <mark-burnett> #topic reducing gerrit bot noise
14:12:37 <mark-burnett> We discussed this briefly last week, but attendance had diminished before we could reasonably vote
14:13:00 <b-str> So, what would be the options to vote on?
14:13:02 <mark-burnett> This was originally roman's suggestion, and there didn't seem to be much objection to removing the PS upload notifications
14:13:03 <mattmceuen> sorry, I missed that one - what are the knobs we can turn down for the bot?
14:13:15 <mattmceuen> ah I see
14:13:18 <mark-burnett> I'm not sure whether it's possible to set it to merges only or not
14:13:22 <evrardjp> Is that a problem?
14:13:30 * mark-burnett shrugs
14:13:40 <evrardjp> I mean you can configure your client if you don't care.
14:13:52 <evrardjp> : )
14:14:08 <mark-burnett> If everyone configures their client to hide those, then why have them?
14:14:09 <mattmceuen> FWIW they don't bother me, I find them informative around "what's going on at the moment".  But I can easily respect it being distracting / annoying for other folks
14:14:21 <evrardjp> mark-burnett: not everyone, I like those :)
14:14:33 <b-str> I don't know if it's possible, but I think a digest summary would be nice, at some interval.
14:14:40 <evrardjp> mattmceuen: +1
14:15:00 * portdirect sneaks in through the back door, sits down and hope no one notices him knock over a coffee mug
14:15:01 <mark-burnett> I wonder if roman/others would like to try client-side configuration
14:15:01 <sthussey> I believe the knob is available: https://docs.openstack.org/infra/system-config/irc.html#gerritbot
14:15:08 <evrardjp> b-str: you can ignore the notifications in your irc client, and ask for a daily digest on gerrit email notifications maybe ?
14:15:32 <mattmceuen> aha thanks sthussey
14:15:32 <sthussey> Because of how Gerrit works, I'd be good with turning off PS upload notifications
14:15:40 <b-str> sure, they don't bother me per se, but I can see where getting 4 notifications for 4 "oops" is a little aggressive reporting.
14:15:51 <evrardjp> oh it's technically possible, and the feature was added indeed
14:16:06 <mark-burnett> yeah, hard to separate the "substantial update" vs "4 typo fixes in 1 hour on a single change"
14:16:45 <sthussey> I think Roman's concern is if someone asks a question and then in 10m it is scrolled off the screen due to PS notifications, it might be hard to get traction with support
14:16:47 <evrardjp> well yeah, there is no difference between new patches and updates
14:16:58 <b-str> TBH, put me in the camp of "don't care - not a problem for me"
14:17:13 <mark-burnett> Yeah, I think that's a valid one, though maybe addressable for him with just ignore
14:17:19 <sthussey> I personally don't care, but don't find IRC that informative. Gerrit has a web interface, I can easily see what is going on there
14:17:31 <mark-burnett> That is more digestable, imo too
14:17:50 <sthussey> I would table it until Roman is here to state his case
14:18:14 <mark-burnett> I think that's fair
14:18:25 <mark-burnett> I'll mention the suggestion to him, and he can try it
14:18:33 <mark-burnett> Should we move back to -2 reviews?
14:19:05 <b-str> go for it
14:19:14 <portdirect> sure, this was a comment based on some behavior that we had in the early days - though its fading out now a bit i think
14:19:16 <mark-burnett> #topic -2 Reviews
14:19:55 <portdirect> but its always been my understanding that -2 can essentially be read as "this is a bad idea, we really should never merge this"
14:20:24 <portdirect> so we should probably be a bit less liberal with them?
14:20:28 <evrardjp> it is the intent -- but it doesn't prevent a merge
14:20:55 <evrardjp> -2 should be kept for extremes reason in openstack, so I suppose airship could take this as inspiration
14:21:05 <aaronsheffield> That seems a bit extreme.  Shouldn't folks just read them as 'this will break something.' So is blocks the merge until that is addressed.
14:21:08 <portdirect> this is my thought
14:21:08 <sthussey> Gerrit says 'Do not merge'. Seems pretty straightforward.
14:21:28 <sthussey> People can read into it what they want.
14:21:33 <portdirect> i'm just offering the perspective of where ive come from in the wider community
14:21:40 <hogepodge> -1 means something is wrong and you don’t think it should be merged, but needs work
14:21:53 <hogepodge> -1 workflow will prevent a merge
14:21:58 <evrardjp> this please have a look at https://docs.openstack.org/project-team-guide/review-the-openstack-way.html
14:21:58 <portdirect> `-1` would seem pretty approprate for you ned to update or respond to this before merge?
14:22:27 <portdirect> `The purpose of the -2 vote is to indicate to the submitter that any further time they spend on the change will almost certainly be wasted. `
14:22:51 <portdirect> ^ I suppose this is the nub of my argument
14:22:57 <aaronsheffield> -1 WF doesn't persist between updates to the PS.
14:23:13 <sthussey> Then Gerrit should be updated to reflect that -2 indicates the PS should be abandoned
14:23:14 <mattmceuen> That's because you can fix things with PS :)
14:23:20 <evrardjp> I have certainly used the -2 before for procedural reasons around a release too.
14:23:21 <b-str> I've been bit by the -1 thing
14:23:46 <mark-burnett> Yes, sometimes -1's get ignored
14:23:47 <evrardjp> let's not talk about abandonning, but talking about channeling energy: if needs updates -> -1
14:24:00 <mark-burnett> Perhaps that's the root issue?
14:24:07 <portdirect> evrardjp: +++
14:24:19 <evrardjp> I would be happy if airship would follow openstack procedures, as it makes things clearer for contributors
14:24:21 <sthussey> If a PS will break existing functionality, I think a -2 is warranted
14:24:31 <b-str> I review after a -1 that didn't address all the -1's things - don't see an issue, but the -1 person sidechannels me about "why not my -1's"
14:24:35 <openstackgerrit> Kaspars Skels proposed openstack/airship-treasuremap master: Set Keystone admin endpoint to match public  https://review.openstack.org/604947
14:24:46 <evrardjp> sthussey: I do not agree, as this would not pass gating.
14:25:09 <mark-burnett> in a perfect world that's true
14:25:09 <sthussey> That may be true when zuul gating can support running larger tests
14:25:41 <evrardjp> this is further away of the initial conversation :)
14:25:51 <openstackgerrit> Merged openstack/airship-shipyard master: Update to Airflow 1.10  https://review.openstack.org/603927
14:25:51 <mark-burnett> it's not too far i think
14:25:54 <evrardjp> if you definitely do not want the patch to get in, add -w.
14:25:59 <portdirect> so I propose one of two things
14:26:04 <sthussey> Workflow is cleared w/ every PS on a cS
14:26:09 <portdirect> 1) we adopt the spirit of: https://docs.openstack.org/project-team-guide/review-the-openstack-way.html
14:26:16 <portdirect> or 2 we write our own
14:26:51 <evrardjp> sthussey: on purpose: "do not merge in this case, you're breaking x." Then the person fixes its patch to not break x, and review happens again. Why taint the patch?
14:26:56 <portdirect> otherwise the different style will make it hard for people who have worked on openstack and other projects to come in and participate
14:27:13 <sthussey> What is tainted?
14:27:18 <mattmceuen> I would prefer we adopt the openstack way, as we plan to collaborate / integrate with that project quite a bit, and the PTG proved that there is an overlap in user / developer bases
14:27:18 <sthussey> You remove the -2 and life goes on
14:27:20 <mark-burnett> I think -2 has been used in order to avoid -1's being ignored
14:27:25 <evrardjp> portdirect: agreed, one or the other, but I think it is worth documenting
14:27:32 <mark-burnett> They're easily removed, as sthussey says
14:27:37 <mark-burnett> But they aren't automatically removed
14:27:37 <evrardjp> sthussey: I will never review a -2.
14:27:47 <sthussey> Okay
14:27:52 <portdirect> neither would i for any project other than here ;)
14:27:57 <mark-burnett> Does a change need more reviews at that time?
14:27:58 <evrardjp> well I mean objectively, why would I?
14:28:01 <mark-burnett> It clearly doesn't
14:28:07 <mark-burnett> so it's an accurate indicator in this case as well
14:28:56 <sthussey> I would happily replace -2 with the idea that if a CS gets a -1, it won't be merged until the person giving the -1 provides a +1 or +2
14:29:01 <hogepodge> Sometimes explaining why you’re ignoring a -1 in a merge is also polite. It’s the final decision of the cores, but clear communication is better for everyone
14:29:08 <mattmceuen> If people are ignoring -1s, that's a problem we need to fix
14:29:16 <evrardjp> agreed with hogepodge
14:29:22 <portdirect> ++
14:29:40 <portdirect> there will be times were a diff of opinion comes in
14:29:43 <evrardjp> here the idea is: if you want to bring more people to the community be clear at every time. In your contributor's guide (say which practices you have)
14:29:47 <evrardjp> or in your reviews
14:29:55 <portdirect> eg: https://review.openstack.org/#/c/604428/5
14:30:06 <sthussey> but there have been cases that a -1 review is left and then the author responds to it and the CS is approved and merged before the original reviewer even gets a chance to re-review
14:30:32 <mattmceuen> We have to weigh:
14:30:44 <mattmceuen> 1) the chance that things get merged badly if we're not careful
14:30:55 <mattmceuen> 2) the fact that everyone with experience with -2's interprets them harshly
14:31:16 <mattmceuen> If we define a non-conventional meaning for -2, we'll be alienating potential contributors
14:31:16 <evrardjp> 1) git revert
14:31:26 <mattmceuen> Let's work to be careful with merges
14:31:31 <mark-burnett> isn't a revert also impolite?
14:31:35 <mark-burnett> that's the issue here, politeness
14:31:40 <openstackgerrit> Felipe Monteiro proposed openstack/airship-pegleg master: Remove Pegleg stub logic from CLI and engine  https://review.openstack.org/605091
14:31:47 <mark-burnett> or hurt feelings, or however we want to say it
14:32:00 <sthussey> I'll happily follow whatever is decided, but I come from a perspective of trying to get things done. Feelings don't enter into the equation.
14:32:05 <evrardjp> 2) as long as there is a policy it would be clearer -- currently there is only an implicit policy which is not the clearest one if I understand the conversation
14:32:13 <sthussey> Feel free to give my PS a -5 if that is what is warranted
14:32:21 <aaronsheffield> Zuul can't run full Airship gates, so we need something else to block patch sets, -2 is the only tool available for that at this time.
14:32:39 <evrardjp> I don't agree
14:32:43 <portdirect> no - a choesive core team is
14:32:48 <evrardjp> portdirect: +1
14:32:52 <portdirect> *cohesive
14:32:59 <evrardjp> it's the +w that decides a merge
14:33:32 <evrardjp> if the -2 -1 discussion is talking about merging, I think we are doing it wrong
14:33:43 <portdirect> Ive seen several cases where there have been 2*+2 and a -1 from a core, often that core who -1 is also the same person to +wf
14:33:50 <mattmceuen> I don't want to lose contributors for the sake of keeping code out.  It's a tradeoff and it's worth the slight risk.
14:34:01 <portdirect> as they accept that even though they dont like it, others do
14:34:03 <evrardjp> -2 -1 is about code quality -> -1 update it, -2 go back to drawing board?
14:34:09 <evrardjp> at least in openstack ters
14:34:11 <evrardjp> terms*
14:34:13 <jamesgu__> according to openstack guideline, -2 should be accompanied by a comment explaining the reason that the change does not fit with the project goals, so that the submitter can understand the reasons and refocus their future contributions more productively.
14:34:22 <hogepodge> It can be confusing because some rules are cultural (two +2s) and some enforces by the gate (only +1 w needed to merge)
14:34:55 <openstackgerrit> Felipe Monteiro proposed openstack/airship-pegleg master: Remove Pegleg stub logic from CLI and engine  https://review.openstack.org/605091
14:35:06 <sthussey> So it seems like the action item here is likely 'Write the guidelines'
14:35:13 <evrardjp> I agreed on both assertings of jamesgu__ and hogepodge (again)
14:35:14 <sthussey> It won't be decided in IRC
14:35:33 <evrardjp> sthussey: it will be discussed in a review , guess what will happen? :p
14:35:46 <evrardjp> hahaha :)
14:35:56 <evrardjp> but yeah maybe worth moving to another topic...
14:36:29 <sthussey> As stated, it needs to be documented anyway. I believe there is an ongoing effort for governance documentation
14:36:39 <sthussey> This can just follow under it - Code of Conduct type of thing
14:37:09 <sthussey> I really doubt it is that big of a thing
14:37:11 <b-str> I do think that if -1's persissted, it would be clearer.
14:37:28 <sthussey> In 1000+ reviews of Airship PS, less than 10 are -2 according to stackalytics
14:37:49 <openstackgerrit> Felipe Monteiro proposed openstack/airship-pegleg master: Remove Pegleg stub logic from CLI and engine  https://review.openstack.org/605091
14:38:15 <mattmceuen> are we moving on mark-burnett or wrapping up discussion on this?
14:38:45 <mark-burnett> Yeah, some good points have been discussed, but it's clear we're not going to settle on something here i think
14:39:09 <openstackgerrit> Scott Hussey proposed openstack/airship-in-a-bottle master: [WIP] Add apiserver w/ webhook  https://review.openstack.org/604918
14:39:19 <mark-burnett> #topic Marketing Deadlines
14:39:58 <mark-burnett> Please note the deadlines and links in the etherpad: https://etherpad.openstack.org/p/airship-meeting-2018-09-25
14:40:07 <mark-burnett> I think I should have put this in announcements :>
14:40:13 <mark-burnett> Ok, on to new business
14:40:20 <mark-burnett> #topic Pegleg breaking changes
14:40:59 <mark-burnett> So the initial pegleg sketch was mostly a stop gap, and frankly full of dubious technical decisions
14:41:25 <mark-burnett> Felipe has been trying to improve it, and has included some breaking changes that are discussed here: http://paste.openstack.org/show/730723/
14:41:58 <mark-burnett> Looks like there's room for discussion around managing breaking/changes, but it feels like it just fits in generally with a versioning conversation
14:42:13 <mark-burnett> I'm not sure we're at a place where we can solidly version these tools yet, tbh -- surely some of them
14:43:15 <mattmceuen> I am a big fan of the changes, just a little bit of growing pains - nice work felipe
14:43:40 <mark-burnett> Yeah, I'm not sure there's much more to talk about here - just an effort to raise awareness about the newer changes
14:44:22 <mark-burnett> Ok, let's finish up
14:44:25 <mark-burnett> #topic roundtable
14:44:30 <mark-burnett> Anything we missed?
14:45:22 <sthussey> I think today has gone long enough
14:45:37 <sthussey> For next meeting possibly open the topic of a versioning strategy for the components
14:45:46 <mark-burnett> Sure
14:46:07 <mark-burnett> Thanks all
14:46:09 <mark-burnett> #endmeeting