19:07:06 <dtroyer> #startmeeting openstackclient
19:07:07 <openstack> Meeting started Thu Feb  2 19:07:06 2017 UTC and is due to finish in 60 minutes.  The chair is dtroyer. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:07:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:07:11 <openstack> The meeting name has been set to 'openstackclient'
19:07:16 <stevemar> you should have used "Marketing COmmittee"
19:07:35 <dtroyer> heh
19:07:57 <dtroyer> #topic release status
19:08:13 <dtroyer> So 3.8.1 is out and is the stable/ocata branch
19:08:17 <stevemar> nice
19:08:32 <dtroyer> also osc-lib, os-client-config and cliff are all branched too now
19:08:33 <stevemar> http://docs.openstack.org/releasenotes/python-openstackclient/unreleased.html
19:08:39 <stevemar> #link http://docs.openstack.org/releasenotes/python-openstackclient/unreleased.html
19:08:42 <stevemar> release notes ^
19:09:45 <dtroyer> while on that subject, fixing up release notes in both mitaka and master is in queue, suffering form (I think) the rash of libvirt issues
19:09:58 <stevemar> recheck recheck recheck recheck
19:10:29 <dtroyer> I hate to keep doing that, it encourages others to do that too without understanding the failures… I've seen a bit of that the last couple of days
19:10:56 <dtroyer> #topic reviews
19:11:09 <dtroyer> that said, are there any open reviews we should discuss?
19:11:39 <stevemar> i've been heads down on keystone stuff
19:11:45 <stevemar> haven't looked at open changes for a while
19:11:49 <dtroyer> I've started cleaning up some of the osc-lib/occ workarounds
19:11:51 <dtroyer> kk
19:11:52 <stevemar> i WAS going to look at neutronclient stuff
19:12:18 <stevemar> https://review.openstack.org/#/c/416321/  https://review.openstack.org/#/c/340763/    https://review.openstack.org/#/c/409759/  https://review.openstack.org/#/c/396727/  https://review.openstack.org/#/c/409819/10
19:12:46 <stevemar> i'm worried that neutron commands are drifting a bit from our prescribed command structure
19:13:07 <dtroyer> of course they are… I'll take a minute to go look at them too
19:13:26 <stevemar> dtroyer: "a minute" -- they are huge
19:13:27 <dtroyer> maybe that's my Boston conf talk… how to design a CLI
19:14:23 <dtroyer> there is only so much we can do in plugins it's the resource naming I worry about the most
19:15:25 <dtroyer> also, I added the volume v3 topic to the PTG list, Justin is not going to be there but I think we have a couple of things to decide
19:16:11 <dtroyer> stevemar: what would you think about putting the purge command and other things like that into a plugin repo?
19:16:23 <stevemar> dtroyer: we could
19:16:52 <dtroyer> maybe an admin-y repo that has more sharp edges than the regula user stuff
19:17:14 <stevemar> eh, then half the keystone commands could go there
19:17:33 <stevemar> "admin-y" is not a good logical division
19:17:38 <dtroyer> I don't mean all admin-only, just the not-your-usual-stuff.  purging is a Big Deal
19:18:01 <dtroyer> ok, yes, maybe extended-value-added commands
19:18:13 <dtroyer> purge is the only one I have in mind right now
19:18:17 <stevemar> i'm glad that jordan is hooked in, he maintains ospurge
19:18:26 <dtroyer> ah, cool
19:18:42 <stevemar> dtroyer: "just-get-me-a-vm --image ubuntu"
19:18:51 <dtroyer> he has been pretty helpful lately
19:19:15 <dtroyer> yes, things like that, although I'm not sure I'd want that one in the same plugin as purge :)
19:19:38 <stevemar> true
19:20:27 <dtroyer> the other thing I want to explore is what mordred is doing with "direct REST" in shade, very similar (in concept) to how we did object and some of the image API bits
19:20:44 <dtroyer> it's the low-level API I've wanted since the beginning
19:20:48 <stevemar> dtroyer: alleviate the need on python-*client
19:21:29 <dtroyer> I don't like the idea of yet-another-repo to install, but merging those might be very useful between OSC and shade
19:21:46 <dtroyer> it's on my list of things to talk to mordred in ATL
19:22:06 <stevemar> i'm still confused why shade is in openstack-infra
19:22:25 <dtroyer> initial convenience, now inertia
19:22:37 <dtroyer> it was for nodepool and turned out to be actually useful
19:22:44 <dtroyer> elsewhere
19:23:36 <stevemar> yeah
19:23:46 <stevemar> i would love to de-couple from glanceclient
19:23:57 <stevemar> that things brings in all sorts of junk
19:24:05 <dtroyer> that's the first priority
19:24:21 <dtroyer> and why as mich of it exists in openstackclient.api as does
19:24:38 <stevemar> it does?
19:24:51 * stevemar gives dtroyer a skeptical look
19:24:57 <dtroyer> not a lot, but more than any other except obejct-store
19:25:01 <stevemar> oh look at that
19:25:13 <stevemar> just image_list
19:25:14 <dtroyer> I think we're using that for image list at least
19:25:34 <stevemar> should we use the SDK or bake our own stuff?
19:26:03 <dtroyer> I really really really want a lower-level API than the SDK.  Still.  and what is happening in Shade is confirming that is not a Bad Idea
19:26:33 <stevemar> the sdk doesn't have all the stuff the that glanceclient has :(
19:27:09 <stevemar> https://github.com/openstack/python-glanceclient/tree/master/glanceclient/v2 ugh
19:28:33 <dtroyer> I don't have much else here…
19:28:41 <dtroyer> #topic open discussion
19:29:04 <stevemar> i should make a glanceCLI vs OSC spreadsheet like i did with cinderCLI vs OSC
19:29:15 <stevemar> i think the guys on the team appreciated it?
19:29:42 <dtroyer> yeah, it was a plce to see what is actually done already
19:29:49 <stevemar> if we can get one less CLI in openstack by EOY, i'll be happy
19:30:05 <stevemar> i suspect nova and swift will around for a loooong time
19:30:16 <stevemar> swift will never go away for obvious reasons
19:30:16 <dtroyer> I doubt we'll get anyone of the core projects to remove theirs
19:30:24 <stevemar> keystone left :D
19:30:28 <stevemar> neutron is deprecated
19:30:33 <stevemar> glance should be an easy win
19:30:42 <stevemar> cinder ... *shudders*
19:30:56 <smcginnis> :)
19:31:09 <stevemar> smcginnis: you guys are as stubborn as your project name
19:31:13 <dtroyer> there is still too much "loss of control" felt by the teams to want to give those up
19:31:17 <stevemar> smcginnis: i say that with love
19:31:20 <smcginnis> A bunch of block heads. ;)
19:31:29 <stevemar> ba-dum tsst
19:31:58 <dtroyer> see now I always think f the "smoldering embers" connotation with cinder
19:32:09 <smcginnis> Hah
19:32:10 <dtroyer> but then I was there wneh we pulled it out of nova
19:32:34 <stevemar> dtroyer: pulled it out?
19:32:44 <stevemar> dtroyer: you mean ctrl+c -> ctrl+v ?
19:33:06 <dtroyer> anotherjesse basically did a cp -r nova cinder and started git rm'ing stuff
19:33:13 * mordred haves to dtroyer and stevemar  - reads scrollback real quick
19:33:40 <dtroyer> mordred: you are implicated in yet another plot to overthrow something
19:33:45 <mordred> yay!
19:34:08 <dtroyer> at least I'll have company
19:34:11 <mordred> well, fwiw, we've gotten rid of glanceclient and swiftclient so far and some tiny bits of novaclient. neutronclient is up next
19:34:32 <dtroyer> luckily you don't need every corner case in shade :)
19:35:04 <mordred> indeed.
19:35:22 <stevemar> dtroyer: how are you planning the PTG?
19:35:23 <mordred> in any case, I _definitely want to chat with y'all about this topic at the PTG
19:35:47 <dtroyer> I'm listing stuffs in https://etherpad.openstack.org/p/osc-ptg-pike
19:36:11 <dtroyer> no schedule yet though, but I think other things I need to pay attention to happen on Tuesday so I'm leaning toward doing OSC-y things Monday
19:37:19 <stevemar> sounds good
19:37:35 <stevemar> theres a few usability things we could improve with auth and such
19:38:34 <stevemar> dtroyer: did we ever get any usable metrics on speed?
19:38:48 <dtroyer> I've not returned to that since Austin
19:39:15 <dtroyer> but I'm running dwarf locally on my new NUC and OSc feels _really_ snappy with it :)
19:39:45 <dtroyer> dwarf == very minimal OpenStack APIs wrapped directly around libvirt
19:40:01 <dtroyer> kida what nova was at Bexar release, less multi-node
19:40:25 <stevemar> ah
19:40:29 <dtroyer> lets me use the same tools (OSC!) to manage my dev work
19:40:31 <stevemar> i was just googling that
19:40:46 <dtroyer> https://github.com/juergh/dwarf
19:42:33 <stevemar> dtroyer: he should move to OSC :P
19:42:54 <stevemar> dtroyer: oh oh
19:43:02 <stevemar> dtroyer: did you want to release our meta package?
19:43:16 <stevemar> do a 0.1 ?
19:43:42 <dtroyer> yes, but I was thinking it might wait until after the ff/rc crush
19:43:44 <stevemar> there are no tests for it :)
19:43:46 <stevemar> https://github.com/openstack/openstackclient
19:43:49 <dtroyer> since it isn't goign to be an Ocata deliverable
19:43:56 <stevemar> makes snese
19:44:14 <dtroyer> also, I've wondered about how we should version it
19:44:18 <stevemar> some folks had wanted it
19:44:34 <dtroyer> if that's what we talk about users installing, that's yet one more version to trac
19:45:05 <dtroyer> keep it's versions matched (at leas x.y) between both repos?
19:46:47 <stevemar> hmm
19:46:54 <stevemar> i hadnt thought of that, but that makes sense
19:47:06 <stevemar> maybe do an initial 0.1? then 3.8.0 ?
19:47:16 <stevemar> or whatever py-osc is at
19:47:58 <dtroyer> yeah, 0.y and what would have been 1.0 becomes py-osc
19:48:09 <stevemar> ack
19:48:17 <dtroyer> and that's a good short name for the existing repo, thanks!
19:48:18 <stevemar> we should outline goals that we want to hit in Pike
19:48:26 <stevemar> py-osc vs osc :D
19:48:59 <ankur-gupta-f1> hi sorry I am late
19:49:03 <ankur-gupta-f1> anything to be aware of
19:49:44 <dtroyer> hey ankur-gupta-f1, we're just chatting about random stuff
19:50:06 <ankur-gupta-f1> okay. I did have a question. Any example of how to introduce a new endpoint/service into SDK
19:50:10 <ankur-gupta-f1> thinking Octavia
19:50:16 <dtroyer> mostly what you might need to know is that I'm being slow to merge new things that are not bug fixes or tests for a little while yet.
19:50:34 <dtroyer> it keeps the process a bit cleaner if we need to do a quick bug-fix release
19:50:54 <dtroyer> SDK stuff needs to go to briancurtin
19:51:23 <ankur-gupta-f1> dtroyer: That is what i was thinking. Slow reviews makes sense. Have conveyed it to teamates with remaining neutron commands
19:51:51 <ankur-gupta-f1> will take a look at launchpad later today
19:52:00 <ankur-gupta-f1> how are we in terms of functional/unit test coverage?
19:52:41 <stevemar> ankur-gupta-f1: lots of stuff needs to be converted to the new format
19:52:50 <ankur-gupta-f1> json or regex?
19:52:51 <dtroyer> our test coverage is overall pretty weak still.  Network is in pretty good shape there compared to most of the others
19:52:57 <dtroyer> json
19:53:03 <dtroyer> please
19:53:14 <dtroyer> I think I gave stevemar hives when he saw "import re"
19:53:22 <ankur-gupta-f1> haha
19:53:22 <stevemar> the json format is so nice
19:53:29 <ankur-gupta-f1> improves readability too
19:53:30 <stevemar> dtroyer: a light seizure maybe
19:53:35 <dtroyer> it's turned out to be pretty clean
19:53:53 <stevemar> i'll start hacking at the identity tests
19:53:59 <stevemar> easy enough
19:54:40 <dtroyer> one more tiny-bit-o-news I just remembered:  reno handles editing relnotes from prior releases properly (I didn't know that had been fixed)
19:55:21 <dtroyer> also, when https://review.openstack.org/#/c/427842/1 merges we'll have complete stable series notes
19:55:51 <stevemar> dtroyer: whuaaa
19:56:02 <stevemar> dtroyer: what if you re-name the note?
19:56:50 <dtroyer> I didn't try that, but I think reno matches filenames, so different names at different points in the git history is likely to be bad?
19:57:02 <stevemar> yeah, darn
19:57:02 <dtroyer> within a release series it is still ok
19:57:15 <stevemar> eh
19:57:22 <stevemar> master will still be funky
19:58:03 <stevemar> i'll give it a shot and see what happens
19:58:37 <dtroyer> ankur-gupta-f1: I'm guessing you will not be at the PTG?
19:58:53 <stevemar> dtroyer: reedip will be
19:59:00 <dtroyer> oh, good
19:59:02 <ankur-gupta-f1> I fought very hard. But they won't be. Reedip can rep the neutron team
19:59:12 <ankur-gupta-f1> hes probs a better person to have anyways :p
19:59:19 <stevemar> dtroyer: do you know if huanxuan will be there
19:59:23 <dtroyer> I don't think any of the OSIC guys who have been contributing to OSC will be there :(
19:59:58 <dtroyer> I havent heard about huanxuan recently, I do recall him not thinking he would due to the travle
20:00:14 <stevemar> should have applied for travel support!
20:00:33 <dtroyer> and with that we're out...
20:00:37 <dtroyer> thanks guys!
20:00:39 <dtroyer> #endmeeting