07:20:01 #startmeeting requirements 07:20:02 Meeting started Wed Nov 22 07:20:01 2017 UTC and is due to finish in 60 minutes. The chair is prometheanfire. Information about MeetBot at http://wiki.debian.org/MeetBot. 07:20:03 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 07:20:06 The meeting name has been set to 'requirements' 07:20:19 ping tonyb, prometheanfire, number80, dirk, coolsvap, toabctl, smcginnis 07:20:25 #topic roll-call 07:20:27 o/ 07:21:21 Oh today? man I'm confused 07:22:50 tonyb: yep, todays wednesday right? 07:22:51 o/ 07:23:27 prometheanfire: Yeah somehow I had it in my head that we moved to Thursday I don't knwo why 07:23:44 Except for prometheanfire , I think he's still on Tuesday? 07:23:55 nah, 1:23 am here 07:24:06 ok 07:24:08 #topic Any controversies in the Queue? 07:24:26 I don't think so 07:24:41 Any comments on the lower-constraints stuff? 07:24:53 dirk is doing good work on the zuul stuff still :D 07:25:08 I think i screwed up the last revision but before it was passing 07:25:35 I would like to add a devstack run now against LC 07:25:37 dirk: I haven't looked yet :( 07:25:50 tonyb: we need your depsolver as well :p 07:26:36 tonyb: another, unrelated one for you: https://review.openstack.org/518448 07:26:48 See referenced bug report 07:27:00 prometheanfire: It is true, It's on my todo list (to clean it up) but it wont happen this month 07:27:04 but as far as the lower-constraints stuff, looks ok 07:27:06 I haven't tested that change myself yet 07:27:11 tonyb: christmas present then :D 07:27:24 prometheanfire: :) 07:27:58 dirk: okay I'll look at that. I confess I'm confused why this is a problem now, IIRC it only came up with $project was only half doign requirements management 07:28:56 tonyb: it might be a Problem for some longer time already 07:29:31 It's just that a coworker now discovered it, but Am almost sure it was lurking for a while 07:29:56 One thing I noticed is that we don't add constraints for py36 07:30:13 we don't test it either iirc 07:30:15 dirk: but doesn't pulling stuff from git violate the main idea of requirements? 07:30:34 tonyb: even if it's a sha? 07:30:39 tonyb: ceilosca is weird 07:30:41 It can't be, deterministicly, coinstallable if it's from git 07:31:00 tonyb: it's basically a patch on top of ceilometer 07:31:01 prometheanfire: but that isn't representable in u-c 07:31:19 It's not using it as a python2 module 07:31:30 hmm 07:31:34 dirk: but how is in fundamentally different from neurton or horizon? 07:31:49 tonyb: in that case ya... 07:32:13 tonyb: they don't hardcode the Sha maybe? 07:32:42 I need to look at it in more detail. That patch was only the first shot after a few min 07:32:54 pip install git+git://github.com/aladagemre/django-notification.git@2927346f4c513a217ac8ad076e494dd1adbf70e1 07:32:57 I was hoping someone else would test it 07:33:19 not sure about reqs.txt 07:33:27 prometheanfire: nope, https://github.com/openstack/monasca-ceilometer/blob/master/test-requirements.txt#L2 07:34:01 well, they are not pointing to a sha 07:34:06 that should be rejected 07:34:39 dirk: my issue is that with your chnage we then will allow nova to add git+hhtps://some.random.server/aweome/lib 07:34:47 and we'll never see it as a problem 07:35:15 What they're doign is fine for them but they're not managed so they shoud just switch the job to non-voting 07:36:30 ya, without enforcement of at least a sha (tags could be redefined) I don't like it 07:36:34 *if* ceilometer were on pypi we could just list it which is a thing we're trying to get to but there are challenges 07:36:47 tonyb: good point 07:39:24 so I guess pause to reflect and discuss again next week? 07:39:40 +1 07:39:50 ya, I changed my vote for now, but sounds good 07:39:54 I think won't fix might be the right answer 07:40:33 # other items (closed discussion)? 07:40:36 #topic other items (closed discussion)? 07:41:48 if nothing else I think it's time for open discussion 07:42:22 +1 07:42:43 I have no other times 07:43:04 #topic open discussion 07:43:09 for a min then close 07:43:20 other than I might be adding a tw py36 job (non-voting) at some point in the future 07:43:27 when infra merges the depending patchsets 07:43:32 dirk: cool 07:43:38 this way we can run a bit against python 3.6 and find new problems 07:43:51 dirk: If you do that you'll need to massgae u-c to include py3.6 07:44:08 tonyb: yeah, was looking at that 07:44:12 that sounds fine, my main question is what OS? 07:44:18 otherwise u-c will ignore you and you'll get $latest versions for stuff 07:44:19 Opensuse 07:44:23 is 18.04 defaulting to it? 07:44:28 tw == tumbleweed 07:44:29 that works too :D 07:44:37 prometheanfire: Yeah we'll get it then alos 07:44:49 ah, thought they were sticking with 3.5 07:44:51 I am not sure, but it seems likely that Ubuntu updates to that version as well 07:44:54 nice to hear, 3.6 is awesome 07:45:31 Yep, first one that is actually faster than py27 ;-) 07:45:42 lol 07:45:49 just use pypy 07:46:03 #endmeeting