11:59:39 #startmeeting requirements 11:59:40 Meeting started Wed Aug 24 11:59:39 2016 UTC and is due to finish in 60 minutes. The chair is prometheanfire. Information about MeetBot at http://wiki.debian.org/MeetBot. 11:59:41 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 11:59:43 The meeting name has been set to 'requirements' 11:59:47 #topic rollcall 11:59:56 hi 12:00:05 o/ 12:00:24 o/ 12:00:26 o/ 12:00:36 sigmavirus: you too 12:00:45 o/ 12:00:53 there you go :D 12:01:57 ok, more people than normal 12:02:07 #topic controversies in the Queue? 12:03:01 prometheanfire: we need to be careful with osc 3.0.1 12:03:09 gate's slow, openstacksdk needs a bump but we are waiting on upstream for a new release 12:03:19 but with the gate as wedged as it is we're kinda stuck 12:03:21 ya, osc too 12:03:36 we can't get the u-c changes in befoer the bot triggers :( 12:04:15 ya 12:04:58 anything specific about osc? 12:05:18 just that it's a really big chnage and has the potrntial to break a lot 12:05:36 puppet are working with RDO to verify that side is cool 12:05:42 ah 12:05:54 once the ACK we just need buy in from shade 12:06:16 shade uses osc? 12:06:18 k 12:06:22 nevermind, not important 12:06:39 next? 12:06:44 +1 12:06:57 #topic Requirements freeze R-5 (Aug 29 - Sept 02) 12:07:17 so, how does this work, with gate slow and all 12:07:31 prometheanfire: badly :( 12:07:44 you prioritize against the slow gate 12:08:33 work with changeset authors to figure out what needs to get in before the freeze, some things won't be as high a priority as others and so you make the hard decision and work the important ones through the gate 12:08:34 what I mean is, hard or soft cutoff? 12:08:36 sigmavirus: sure 12:09:16 all changes submitted before/during that time frame can be merged 12:10:04 and should we make an announce to the list (though it's on the schedule 12:10:08 Perhaps all by 29th anything later are we can't make a promise 12:10:23 ya 12:10:25 prometheanfire: Yeah we do need an email to the list as well 12:10:26 +1 12:10:46 who wants to do that? 12:10:54 * tonyb will 12:11:16 *nods* 12:11:28 #action tonyb email the list about the Requirements freeze R-5 (Aug 29 - Sept 02), Perhaps all by 29th anything later are we can't make a promise 12:11:35 next? 12:11:48 +1 12:12:06 #topic Barcelona Design Summit 12:12:12 who's going? 12:12:15 * prometheanfire is not 12:12:28 * tonyb will be there 12:12:31 not I 12:13:34 ya, don't think many are going 12:13:43 phooey 12:13:56 * prometheanfire wanted to go, but can't :( 12:14:08 space requirements? 12:14:13 o/ 12:14:21 number80: you going? 12:14:26 prometheanfire: what rooms do we need (if any) 12:14:27 prometheanfire: yes 12:14:53 tonyb: I think a quick status update meeting would be nice 12:15:04 I was thinking 1 fishbowl and maybe a meetup but that seems like overkill based on who'll be there 12:15:05 beyond that, not sure if any are needed 12:15:11 ya 12:15:28 A work Session is always good 12:15:38 Main issue with Barcelona is space though 12:15:53 I agree we need at least one session 12:16:02 (space/slots/time) 12:16:36 dirk: We can ask. 12:16:57 ok, next? 12:17:05 I think it was just about submitting your wishes 12:17:05 DO we really think we'll get value out of a meetup? We could just have lunch/dinner/beer 12:17:26 Yeah, I think either a fishbowl or a work session has more vakue 12:17:28 I think an informal thing would be nice 12:17:35 dirk++ 12:17:39 Will we actually work on anything as a team? 12:17:42 +1 for informal meetup 12:18:18 okay I'll request a fishbowl 12:18:33 k 12:18:35 tonyb: knowledge transfer is always good 12:18:58 I know that I can ask lots of questions to fill up the slot if wanted ;-) 12:19:00 dirk: yeah. 12:19:09 dirk: :) 12:19:13 who want's to make sure we have a fishbowl? 12:19:29 prometheanfire: I will 12:19:42 prometheanfire: I have an email to reply to from ttx 12:19:44 #action tonyb make sure we have a fishbowl at the conf 12:19:51 next? 12:19:54 +1 12:20:06 #topic mascot 12:20:11 #link https://etherpad.openstack.org/p/requirements-team-mascot 12:20:23 our options were picked 12:20:49 Well scorpion was/is available but I'm not sure that's the image we want 12:20:50 waterfall would be funny, given it's dev connotations 12:20:56 "the sting in the tail" 12:21:32 So I was thinking we could just mark up the options and go with that. 12:22:11 You can pick more than one option but only vote once for a given option 12:22:39 I'm not sure any of the options really screams "requirements" to me 12:23:13 something that's dependant upon for the community 12:23:20 so, earth, or a field of grass 12:23:22 or something 12:23:54 Riverbed perhaps? 12:23:59 http://famouswonders.com/wp-content/uploads/2009/03/serengeti-landscape.jpg 12:24:17 dirk: do you want another quantum, because that's how you get another quantum :P 12:24:53 prometheanfire: ;-) 12:25:02 prometheanfire: interesting idea 12:25:55 added to list 12:26:02 cool 12:26:27 give it say 24 hours and then we'll get back to Heidi 12:26:33 k 12:28:09 next? 12:28:12 +1 12:28:28 #topic Tasks from Etherpad 12:28:39 #link https://etherpad.openstack.org/p/requirements-tasks 12:29:22 I've been working with other project trying to get 20 solved 12:29:26 prometheanfire: Thanks again for checking up on the bot failures 12:29:35 tony has helped :D 12:30:04 anyone else working on anything from the list? 12:30:41 I'm workign on 18 but only slowly 12:31:07 and 2 12:32:01 k, it'd be nice to have 2 done before the summit :P 12:32:06 next? 12:32:13 +1 12:32:26 +1 12:32:39 #topic Volunteer for next 2 meetings 12:32:44 Aug 31 - coolsvap 12:33:01 Sep 14 - ??? 12:33:26 prometheanfire: what happend to Sept 7th? 12:33:43 :D 12:33:48 oh, missed that 12:33:53 tonyb: september 7th is not on my calendar 12:33:55 Sep 07 - ??? 12:34:05 sigmavirus: you volunteering? 12:34:11 way to be seen sir 12:34:25 No I'm not. I generally don't make it to these any longer 12:34:53 any takers? 12:34:57 I can do either the 7th or 14th 12:35:02 but would rather not take both 12:35:22 prometheanfire: ok you take one and put me down for t'other 12:35:42 ok, I'd rather take the 14th if possible 12:35:56 prometheanfire: works for me 12:35:59 k 12:36:40 #action tonyb run the sept 7 meeting 12:36:49 #action prometheanfire run the sept 14 meeting 12:36:59 #topic optional-requirements 12:37:04 #link https://etherpad.openstack.org/p/optional-requirements-draft 12:37:08 number80: ? 12:37:49 yes 12:38:08 first, I'd like to hear what is blocking us, it's not clear for me 12:38:37 If there's nothing against it, I may just push a review and ask feedback on the list 12:38:58 how it'd integrate with other projects 12:39:15 number80: I'm not sure I understand the point. Doesn't it just complicate all out tooling? 12:40:02 How is optional req different from global requirements except that it allows duplication? 12:40:08 tonyb: the idea is to track libraries that are not essential for runtime but provides additional goodies (performance boost, etc.) 12:40:16 how is the behavior of it's use differ 12:40:29 would we just check against both files when updating? 12:40:37 dirk: that could be a way to summarize it, the goal is to relax rules when we can 12:40:38 Is this just that we know which requirements are more central to meet than others? 12:40:41 can we just comment gr/uc instead? 12:41:08 dirk: yes, and it would encourage people to use abstraction layers e.g anyjson 12:41:09 number80: which rules need be relaxed? 12:41:19 sigmavirus: no feature duplication 12:41:30 Yeah I'd rather do it in g-r either in seperate section or with a local comment 12:42:00 but fundamentally we still ned to ensure co-installability so we still need to do the same work 12:42:03 IMHO, what we want is not to require dozen of libraries doing the same thing to run an openstack cloud 12:42:22 tonyb: that's a good compromise (the section proposal) 12:42:51 number80: right 1 or 2 with *very good* reasons but not dozens 12:43:09 +1 the section idea 12:43:17 number80: I'm not sure that's easy to achieve though 12:43:42 Like e.g. pymysql vs pgsql 12:43:43 tonyb: yep, the goal is that additional goodies should remain non-essential but still tracked and tested 12:43:55 Which one is core which one is optional? 12:44:46 dirk: I think this falls under the backend category and when we should favor any of them 12:44:53 dirk: the first is "core" the additional ones would become extra 12:45:10 dirk: we'd only allow a duplicate library if it had a *really good* reason 12:45:18 my $0.02 12:45:27 but we could enforce that a specific mysql python driver must be supported 12:45:47 somehow, not sure how 12:45:52 and allow another one if it brings performance improvement for instance 12:45:58 gating only on one set would help maybe 12:46:09 To be honest I prefer one more library over one that exists but that has 200 dependencies 12:46:59 dirk: that could fall under the "very good reasons" to accept it 12:47:04 One of the things we shoudl do is facilite movement from $old to $new if we create that situation 12:47:17 I understand the reason that we want to keep things 'small' and 'manageable' bit that's a complex thing, not just in/out imho 12:48:07 Right, I see value in documentation $old->$new 12:48:25 E.g. add a deprecated-by comment to g-r deps? 12:48:46 +1 12:48:58 are we removing the old? 12:49:01 dirk: we need to do more than document 12:49:25 dirk: that's how we end up with libraries that are stuck on 2year old releases 12:49:48 I think this could be a cross-project session topic: how to get rid of requirements crufts 12:49:54 if we don't facilitate the move then we're making life hard for everyone 12:49:57 tonyb: yeah, we also need to push by calling out deadlines 12:50:18 tonyb: +1 12:50:27 dirk: and what do we do if $projects don't move? 12:50:28 so, less optional, more better? 12:50:45 prometheanfire: Yeah 12:51:03 prometheanfire: well "better for soem people" 12:51:09 tonyb: 'it depends' :-) 12:51:14 well, for the ones we already added, they are truely optinal 12:51:23 they are try wrapped import statements 12:51:29 prometheanfire: for example *json some uses may not care about speed and prefer pure python 12:51:32 so they will get used when we add them 12:51:55 as for getting people to switch fully... 12:52:14 ya, that'll be fun :P 12:54:10 ok so we're going to add comments/ move thing to seperate section in g-r 12:54:24 but not add an additional file for tracking these things 12:54:27 who wants to submit that? 12:54:33 number80: ? 12:54:59 we should still be very selective about adding funcational duplicates 12:55:07 very 12:55:27 tonyb: I'll do it 12:56:21 #action number80 to submit review moving existing optional-reqs packages to their own section in gr 12:56:26 not sure about uc 12:56:41 ther *must* be only one uc 12:56:41 ok 12:56:50 only one, yes 12:56:54 but in it's own section 12:57:03 #topic open floor 12:57:24 no need for a seperate section in uc 12:57:35 * dirk needs to leave, cya 12:57:37 it's machine generated 12:57:42 true 12:57:55 so, that it? 12:58:22 * tonyb is done 12:58:33 #endmeeting