22:00:20 <danwent> #startmeeting
22:00:21 <openstack> Meeting started Tue May  8 22:00:20 2012 UTC.  The chair is danwent. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:00:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
22:00:27 <danwent> who's here?
22:00:42 <edgarmagana> I am here
22:00:46 <danwent> jkoelker: ping?
22:00:57 <edgarmagana> I missed it last time... did not want to happend again  :-)
22:01:02 <garyk> gary's here
22:01:03 <rkukura> me
22:01:06 <danwent> edgarmagana: much appreciated :)
22:01:08 <s0mik> Hello folks
22:01:21 <_cerberus_> dietz is here
22:01:32 <SumitNaiksatam> Hello!
22:01:43 <danwent> _cerberus_: ah, good.  I forgot your handle had an underscore, so tab wasn't finding you :)
22:01:48 <jkoelker> hi
22:01:48 <danwent> http://wiki.openstack.org/Network/Meetings
22:01:52 <markvoelker> o/
22:01:54 <_cerberus_> danwent: totally on purpose
22:02:03 <danwent> _cerberus_: very smart :)
22:02:22 <danwent> Ok, so first topic is reviews.  Summary: we've got a lot of them.
22:02:31 <danwent> https://review.openstack.org/#/q/status:open+project:openstack/quantum,n,z
22:02:59 <danwent> and more importantly, the reviewing workload is being handled by a very small portion of the community
22:03:12 <danwent> special thanks to the many new people who have stepped up.
22:03:44 <danwent> later on we'll be talking a bit about core-dev status, both promoting new folks and looking at cleaning up the list for people who are no longer participating
22:04:11 <danwent> I'm not aware of any reviews that require special attention, but I wanted to give people a chance to call anything out.
22:04:37 <danwent> Ok...
22:04:51 <danwent> So we have 2 weeks until we branch for F-1.  That's not a lot of time.
22:05:25 <danwent> We've made a lot of progress on the codebase in the past few weeks, but we're still cutting it somewhat close for two of our high priority issues.
22:05:39 <danwent> https://launchpad.net/quantum/+milestone/folsom-1
22:06:01 <danwent> particularly the quantum + melange API changes (and backend implementation).
22:06:14 <danwent> jkoelker: I believe you've been making progress on this behind the scenes, care to report?
22:06:18 <jkoelker> si
22:06:20 <jkoelker> http://etherpad.openstack.org/quantum-v2-api
22:06:30 <jkoelker> that's our etherpad for the API
22:06:43 <jkoelker> and http://etherpad.openstack.org/quantum-v2-melange-integration
22:06:58 <jkoelker> if for the Database schema to support that api
22:07:17 <_cerberus_> Also, somewhat tangential, but http://etherpad.openstack.org/quantum-authnz
22:07:19 <_cerberus_> Since there's really no auth today
22:07:29 <danwent> _cerberus_:  yup, that's the second bp I wanted to talk about :)
22:08:04 <danwent> can we link those etherpads from the BP in launchpad, so all of this can be found from the main quantum page?
22:08:09 <jkoelker> they are
22:08:15 <jkoelker> just updated it a bit ago
22:08:21 <danwent> ah, perfect :)
22:08:35 <danwent> I see, its on the whiteboard (I was looking at the spec link)
22:08:47 * jkoelker sucks at launchpad
22:08:58 <danwent> hehe, in many circles, that's a badge of honor :P
22:09:32 <jkoelker> BP updated, now with proper fields ;)
22:09:43 <danwent> ok and _cerberus_ , does that BP include updating the client_lib?
22:09:49 <danwent> to "speak keystone"?
22:10:02 <danwent> i forget if we decided to separate that out.
22:10:20 <_cerberus_> danwent: that's the intent, yes
22:10:35 <_cerberus_> We've got Vek looking at that on our end
22:10:42 <_cerberus_> (who isn't in the room atm)
22:11:19 <danwent> _cerberus_:  ok, thanks.  and I assume tr3buchet's work to redo the quantum + nova integration will start using the real client, not one pasted into nova, so we should be able to leverage that work there as well.
22:11:53 <danwent> ok… so I will definitely make sure I get thoughts on the etherpad here, as this is the big think ttx will be me up on if we aren't to the implementing phase by next tuesday :)
22:12:31 <danwent> are there any other BPs we need to discuss for F-1 at this point? most of the other stuff seems to be humming along, with a lot in code-review already.
22:13:32 <danwent> ok, getting the API squared away in F-1 will be really important, as there are a bunch of things in F-2 that depend on it.
22:13:53 <danwent> Alright, the last topic I wanted to bring up was membership in the quantum core dev team.
22:13:58 <_cerberus_> Not it
22:14:08 <danwent> :)
22:14:29 <danwent> vish sent an email to the main nova list about doing a "spring cleaning" on the nova core team.
22:15:03 <danwent> when quantum started, we actually let anyone who volunteered be a core dev, with the idea that we'd do a spring cleaning at some point too.
22:15:10 <danwent> now that we're core, it seems like a good thing to do.
22:15:31 <danwent> note, the goal here is not to force anyone out, just to make sure the team membership is an accurate representation of who's doing work.
22:15:47 <danwent> https://launchpad.net/~quantum-core/+members#active
22:16:01 <danwent> if you look at the list, there are some people that I suspect have never done a single quantum review :)
22:16:29 <edgarmagana> dan: Ying could be removed
22:16:40 <danwent> my thinking is that I'll just email people who haven't been involved, and ask them if they plan on continueing to be involved to the degree that they should be considered core.
22:16:49 <danwent> edgarmagana: good to know, thanks.
22:17:16 <danwent> we as a community can decide exactly what level of engagement that means, and how we decide on it.
22:18:08 <danwent> but my rough feeling is that a "core" dev should be doing 2-3 hours of reviews a week on average.  I'm happy to hear input from others though, as raw hours of reviews are not the only way people can contribute (answering email on the list, etc. is also important, as are other community tasks)
22:18:16 <danwent> what do other people thinks?
22:18:58 * jkoelker crickets
22:19:00 * markvoelker is ok with this plan
22:19:02 <garyk> i am pretty new to this - i feel that sometimes it takes quite a while to get a review.
22:19:02 <mnewby> hard to measure
22:19:05 <edgarmagana> danwent: what about also asking to be attending this meeting as well
22:19:15 <jkoelker> +1 to dan's suggestion
22:19:23 <danwent> edgarmagana: agreed, i think reasonable attendence should be an expectation
22:19:26 <cdub> i think that core is about review, not just community involvment
22:19:36 <danwent> garyk: yes, reviews are taking too long now.
22:19:50 <danwent> cdub:  that's actually a good point to discuss.
22:19:57 <cdub> somebody who's great at answering list questions != someone who is also therefore implicitly trusted to have good judgement on code reviews
22:20:20 <cdub> but it's definitely a very useful community contribution
22:20:22 <danwent> cdub: definitely agree, but at the same time, we want to encourage developers to be helping the community in many ways
22:20:43 <mnewby> danwent: encourage, yes.  not sure how being part of core encourages that though.
22:20:47 <cdub> honorary plaque?  /me ducks
22:20:50 <danwent> cdub: so i agree that it can't be a reason to promote someone to core (that should be based purely on review + code skills)
22:20:55 <danwent> cdub: :)
22:21:23 <danwent> cdub: ok, i'd be fine if we don't consider that in the criteria, though I'd love to have some way to encourage people to do such thankless tasks
22:21:40 <cdub> shameless beer bribery
22:21:55 <danwent> cdub: if you're paying….
22:22:00 <garyk> :)
22:22:02 <cdub> free as in...
22:22:09 <cdub> damn, doesn't work that
22:22:10 <cdub> way
22:22:12 <mnewby> ensuring that reviews are processed in a timely fashion requires core devs that are diligent at monitoring gerrit emails and performing the reviews that appear
22:22:28 <cdub> perhaps soemthing on the list that highlights people's activity
22:22:45 <jkoelker> perhaps doing a review day once a week that rotates would help
22:22:52 <danwent> ok, so I think its fair to say that review skills + familarity with code are primary criteria, as well as attendence of team meetings (not mandatory, but they should make an effort to attend)
22:22:59 <mnewby> hmmm
22:23:06 <mnewby> cdub: i like that idea.
22:23:33 <med_> meritocracy by making merit more obvious.
22:23:43 <mnewby> tracking who is reviewing could be a good way of reminding people who haven't been participating as much that they could be doing more
22:23:49 <s0mik> I think something akin to Nova Review days would also help spread the load..
22:23:57 <danwent> med_:  highlighting who is doing the reviews (perhaps based on lines of code reveiwed) would be nice
22:24:07 <_cerberus_> s0mik: or just do them on the same day
22:24:30 <danwent> my concern with reviews days is that it can lead to people leaving reviews hanging until review day...
22:24:49 <danwent> i'm curious to hear experience from people on Nova wrt review days
22:24:53 <danwent> did they work?
22:24:56 <s0mik> Nova reviews days has an assignee everyday
22:25:18 <danwent> s0mik: ah, i see, its not a team day, its a set of devs assigned to review that day?
22:25:23 <s0mik> its akin to review "guru", on a particular day of the month, certain person is responsible to at least doing initial revies
22:25:38 <danwent> I think something like that could work, once we have more core devs participating…
22:26:01 <danwent> The second half of this discussion is how do we start promoting the people who have been reviewing, but aren't yet core.
22:26:05 <_cerberus_> danwent: I will say that, as a Nova core, I'm much more likely to review on that day versus others
22:26:25 <danwent> _cerberus_: ok, and how frequently do you have them?
22:26:33 <danwent> you personally
22:26:36 <_cerberus_> Every 2 weeksish
22:26:50 <markvoelker> Note also that the review days schema requires a bit of maintenance (http://www.mail-archive.com/openstack@lists.launchpad.net/msg11233.html)
22:27:16 <danwent> markvoelker: yeah, saw that as well, which is why i'm a bit suspicous that it has been working
22:27:26 <s0mik> quantum being a smaller project, would require a little more commitment with the limited set of cores we have..
22:27:57 <danwent> Ok.  right now I feel like everyone would have a review day twice a week.
22:28:18 <danwent> my main goal is first to get existing core devs reviewing more, and to add more deserving peopel to the core team.
22:28:24 <_cerberus_> There are other problems with the nova review days that are difficult to address
22:28:32 <_cerberus_> the minimum number sets a bar you have to meet
22:28:38 <_cerberus_> Instead of necessarily wanting to review
22:28:45 <danwent> do people think instituting review days at this point would get more people reviewing?
22:29:22 <cdub> question is what is the barrier?
22:29:23 <danwent> _cerberus_: I also don't like just measuring # of reviews… as then people avoid large reviews.
22:29:29 <s0mik> danwent: I think at the minimum, review days will ensure that reviews don't go unnoticed for long
22:29:35 <_cerberus_> danwent: and that's the other primary issue
22:29:39 <cdub> is it time commitment?  focus? tools? lack of expertise?
22:30:07 <danwent> cdub: my sense if that there are only a handful of peopel regularly checking
22:30:14 <cdub> (it's why i don't like gerrit, but that's just me :)
22:30:21 <danwent> and that the total number of changes are more than they can handle
22:30:33 <danwent> cdub: you can get notifications via email
22:30:44 <danwent> that is what I meant by checking.
22:31:13 <danwent> perhaps we should say that all core devs need to be signed up to get those notifications?  they can always ignore then, but its a good reminder
22:31:18 <mnewby> cdub: why don' tyou like gerrit?
22:31:27 <danwent> oh boy....
22:31:36 <cdub> because of the review process (i'm crusty here, sorry)
22:31:54 <cdub> but i think it obscures the process over a old-school mail list mechanism
22:32:04 <mnewby> cdub: obscures it how?
22:32:18 <cdub> one advantage of having the reivews in your face is that you see and watch other reivews even if you don't participate
22:32:28 <danwent> unless you're signed up for a review, you don't get further notifications on that review.
22:32:31 <cdub> mnewby: it's another tool that you have to go fish for
22:32:35 <cdub> things to review
22:32:52 <mnewby> cdub: and there is no advantage to using a web tool vs a mailing list for review?
22:33:19 <cdub> mnewby: to me?  no.  but i'm happy to consider myself the minority there ;)
22:33:23 <danwent> cdub: are there practical things we can do here?
22:33:42 <cdub> danwent: it's why i asked what you (collective) think the barriers are
22:33:43 <s0mik> can we have quantum-core as  a reviewer and make gerrit send out a quantum review to the entire core list?
22:33:57 <danwent> s0mik: i already get emails for each new review
22:34:04 <rkukura> danwent: how do we turn on email notifications?
22:34:05 <cdub> if it's just people aren't spending time...well, email, gerrit, anything won't fix that
22:34:06 <danwent> and was proposing that all core devs must do the same
22:34:07 <mnewby> s0mik: each developer can sign up for notifications in gerrit
22:34:13 <danwent> rkukura: in gerrit
22:34:33 <mnewby> preferences -> watched projects
22:34:37 <danwent> #todo #danwent, send out instructions on enabling email notification in gerrit
22:34:37 <rkukura> danwent: I'd have done it long ago if it was intuitive
22:34:48 <danwent> mnewby: you win :)
22:34:51 <_cerberus_> cdub: no matter what, I think it all comes back to time
22:34:52 <cdub> i signed up a few months  back and got none, annoying
22:34:56 <mnewby> oops.   Settings -> Watched Projects
22:35:13 <danwent> _cerberus_: I tend to agree.  we need more people reviewing regularly.
22:35:14 <s0mik> I guess thats why liked review board ;)
22:35:28 <cdub> _cerberus_: that's what it typically is ;)
22:35:40 <mnewby> Ah… I think I have an idea.
22:35:48 <mnewby> We haven't been using groups.
22:36:09 <mnewby> We should document that for any new quantum review, the only reviewer to add is 'quantum-core'.
22:36:17 <mnewby> Heck, maybe we can get it set by default.
22:36:40 <cdub> mnewby: what would that do?  send notification to all core devs?
22:36:45 <mnewby> Then there won't be a need for developers to individually configure themselves to receive notifications.
22:36:47 <mnewby> cdub: yes
22:37:19 <s0mik> mnewby: the concept of groups is what I was referring to earlier.. I think that would be great to have by default
22:37:40 <mnewby> s0mik: let's ask openstack-ci.
22:37:45 <danwent> mnewby: if you want to explore that, I'm for it.  I would prefer that it is automatic.  Otherwise, I think we should just have core devs sign up to "watch".  That's pretty straightforward in my opinion
22:38:09 <danwent> ok, so we've agreed that core devs will get lots more notifications for reviews.
22:38:33 <garyk> when does one assign a reviewer and when does one hope it will be picked up?
22:38:36 <mnewby> danwent: I'm on it.   Will make for more awareness of reviews, at least.
22:38:45 <danwent> I beleive we're also ok with a "spring cleaning", as no one will be forced out (they can always say that they will start reviewing…)
22:38:49 <mnewby> garyk: Hopefully we can have all core reviewers assigned.
22:38:58 <danwent> my final area that I want to explore is the bar at which new people should become core.
22:39:11 <danwent> we have some new people who aren't core, but none the less are doing a lot of very useful reviews
22:39:19 <garyk> i can use a drink at a bar now
22:39:22 <danwent> yong and garyk come to mind
22:40:08 <danwent> how important is it that people have been contributing code to the project for a while, vs. simply having been proved a reliable reviewer of code?
22:40:30 <garyk> i think it is too soon for me. i need some more time to be more familiar with things
22:40:43 <mnewby> danwent: important.  reviewing can be just vetting python without any idea of implicit design considerations.
22:40:46 <s0mik> I like the process of lazy voting where a core or the member self nominates after having achieved enough expertise..
22:40:55 <cdub> reviewing is a distinct skill from code contributions (not mutially exclusive or anything like that)
22:40:56 <mnewby> s0mik: +1
22:41:16 <danwent> cdub: I agree tha reviewing is a distinct set, but there's something to be set for "context" while reviewing
22:41:47 <danwent> s0mik: I agree, but I want to make sure people understand the bar, so they understand when they are ready to promote themselves.  Otherwise people may wait too long.
22:42:07 <danwent> people tend to be overly conservative with self-promotion, as they don't want to get shot down.
22:42:18 <cdub> yeah, i think people demonstrating competent review skills w/out many contributions should be considered, that's waht i mean
22:44:02 <garyk> cdub: +1
22:44:04 <danwent> Ok, so I don't think there should be any official bar in terms of code contributions, but the person should feel confident that they have a "big picture" understanding of quantum architecture.
22:44:30 <edgarmagana> danwent: +1
22:44:36 <mnewby> A simple vote on the ml should suffice, so that if anybody has any concerns they can be addressed.
22:44:37 <cdub> danwent: yeah, can we help with that?
22:45:55 <s0mik> I agree with mnewby if there are concerns it can be handled by ML, a litmus test for core is very hard..
22:46:06 <danwent> Ok, so I think we're roughly on the same page here.  The quantum community is growing and changing quickly, which is a good think, so its important that we discuss how to handle this growth.
22:46:57 <danwent> s0mik: I agree.  I more want people to know the criteria by which they will be judged in such a vote.  Otherwise, they are unlikely to promote themselves.
22:47:18 <danwent> Ok, any other comments on core status?
22:47:58 <danwent> do people think targeting 2-3 hrs per week for a core-dev review time (on average) is reasonable?
22:48:15 <danwent> to high / to low?
22:48:42 <danwent> too tired to answer?
22:48:48 <rkukura> danwent: sounds about right
22:48:49 <edgarmagana> lol!
22:48:55 <garyk> i was hoping for 2-3 hours a day :)
22:49:09 <edgarmagana> danwent: sounds good!
22:49:10 <cdub> hmm, seems reasonable-ish (i'd of thought for more, but considering overall lack of reviews, it's realistic)
22:49:11 <danwent> garyk: I should be clear.. that is for the low bar of being a core dev
22:49:15 <med_> On average garyk,  you can offset some benchsitters.
22:49:20 <cdub> lol
22:49:47 <danwent> yes, many of us obviously spend a lot more time than that, but this is the low bar, below which you should probably be asked to be removed from the core team.
22:50:05 <danwent> Ok, any other items for Open Discussion?
22:50:34 <danwent> ok, thanks folks!
22:50:37 <danwent> #endmeeting