Posted	Nick	Remark
#openstack-sdks - 2017-11-29
14:42:20	Shrews	mordred: oh right. is that process described anywhere?
14:46:54	mordred	Shrews: it's pretty much making a patch to openstack/releases ... you'll find a file, deliverables/queens/shade.yaml - that has entries you can copy-pasta
14:46:54	mordred	Shrews: you need a git sha - so doing a pull on master is usuallya good first step
14:52:07	celebdor1	Does anybody know if shade support heat tags? I couldn't find it with simple grepping
14:52:24	Shrews	mordred: ah, the README there has the process... but the PTL or release liaison has to submit the request
14:53:15	mordred	celebdor1: I don't think it does, no - but adding it would be a great idea
14:54:31	celebdor1	mordred: It is indeed
14:54:42	celebdor1	I need it for queens though :P
14:55:16	mordred	celebdor1: :) it's pretty quick to add things to shade and get them released - I don't think getting it in for queens should be a problem
14:57:28	mordred	Shrews: remote: https://review.openstack.org/523890 Release shade 1.25.0
14:58:55	celebdor1	mordred: thanks for the info
14:58:59	celebdor1	:-)
14:59:30	mordred	Shrews: also, I have learned this morning that there are a few openstack projects, octavia-dashboard, senlin-dashboard, tricircle, bilean, senlin, heat and a couple others - that an sdk release is going to break
15:00:22	mordred	Shrews: so I think I need to put 'get patches landed to each of those projects' to the list of things we do before we cut an sdk release with the new changes in it
15:00:23	Shrews	mordred: that's fun
15:02:17	mordred	Shrews: oh - actually, no - we kept compat for the thing that would break them
15:02:30	mordred	so we can release, then change the interface they all use
15:11:49	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Sort image update results before comparing https://review.openstack.org/523893
15:12:02	mordred	Shrews: that ^^ should fix the unittest failures that are happening
15:13:37	mordred	Shrews: I'm guessing the jsonpatch release on november 25 is to blame, fwiw
15:48:20	openstackgerrit	Monty Taylor proposed openstack-infra/shade master: Avoid tox_install.sh for constraints support https://review.openstack.org/508061
15:48:22	openstackgerrit	Monty Taylor proposed openstack-infra/shade master: Fix basepython setting in tox.ini https://review.openstack.org/523915
16:16:24	mordred	incoming patch bomb ... sorry for the rebase ... it was needed to get in the jsonpatch sorting fix
16:16:33	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Add ability to work in other auth contexts https://review.openstack.org/522970
16:16:33	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Add helper property to get the current user id https://review.openstack.org/522971
16:16:34	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Remove reference to context-managers from release note https://review.openstack.org/522972
16:16:34	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Document current_user_id in a release note https://review.openstack.org/522973
16:16:35	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Fix py35 and pypy tox env https://review.openstack.org/523383
16:16:35	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Cleanup objects that we create on behalf of images https://review.openstack.org/522974
16:16:36	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Add method to cleanup autocreated image objects https://review.openstack.org/522975
16:16:36	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Complete move of baremetal machine tests https://review.openstack.org/522976
16:16:37	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Treat clouds.yaml with one cloud like envvars https://review.openstack.org/522977
16:16:37	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Protect against p_opt not having prompt attribute https://review.openstack.org/522978
16:16:38	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Update make_rest_client to work with version discovery https://review.openstack.org/522979
16:16:38	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Added nat_source flag for networks. https://review.openstack.org/522980
16:16:39	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Shift image tests from test_operator https://review.openstack.org/523258
16:16:39	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Switch baremetal nics/ports tests over https://review.openstack.org/523355
16:16:40	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Start using Connection in openstack.cloud https://review.openstack.org/523259
16:16:40	openstackgerrit	Monty Taylor proposed openstack/python-openstacksdk master: Remove get_legacy_client https://review.openstack.org/523483
16:16:55	mordred	Shrews: those should, you know, pass this time
16:17:41	sdague	cdent: I don't know that there were any real epiphanies from folks. I think a new interesting spin on it was with graphql, someone asked if that solved the problem all by itself
16:18:53	sdague	which, it solves a related problem in theory, though in practice, if your applications don't consume the API directly, but through a 3rd party SDK, as we've seen, graphql probably just always asks for everything, then you are back to REST resources.
16:19:08	cdent	I’ve read elsewhere of graphql being a way out of a versioning problem, if you start there
16:19:25	cdent	but if you’re not already there…
16:20:54	sdague	yeh
16:23:07	sdague	also, I think graphql seems to work best when the front end has intimate knowledge of the costs of asking for things on the backend. So as a layer between the github ui and services, it's awesome
16:27:39	mordred	sdague: yah - or, as soon as they finish adding all of the things to the graphql layer, it'll help zuul's github driver make fewer API calls and thus avoid rate-limit issues
16:28:15	sdague	mordred: sure
16:28:46	mordred	sdague: but I agree, in general, with the assessment about intimate knowledge - in this example, the zuul thing is a cost optimization - it's certainly not an easier/better way to write the first version of the thing
16:29:07	sdague	mordred: and, you are hitting the API raw, because you need that level of granularity
16:29:17	mordred	yup
16:29:26	sdague	but, if you went through someone's random SDK
16:29:35	sdague	you are unlikely to get those optimizations
16:29:40	mordred	yup. exactly
16:30:11	mordred	that said - I'd kind of LOVE a graphql api option for openstack service ... for similar reasons
16:30:49	mordred	finding the port that's ona subnet on a network is basically doing three full list calls and doing a client-side in-memory join
16:31:02	sdague	mordred: honestly, building a graphql porcelin for all of openstack would be a totally legit thing to do
16:31:38	sdague	it would be different enough that I don't think it would be confusing between the rest resources at the point sevices and what it returned
16:32:56	mordred	sdague: yah. although having a per-service graphql as well that the porcelin could use for push-downs for efficiency would also be nice ... but yes, a graphql openstack porcelain that could do "I want to create a floating IP on a port from the server that is on this subnet of this network"...
16:33:29	mordred	would be awesome to do as one call
16:41:58	openstackgerrit	Monty Taylor proposed openstack-infra/shade master: Fix basepython setting in tox.ini https://review.openstack.org/523915
16:56:42	openstackgerrit	Merged openstack/python-openstackclient master: Set correct designate endpoint in docs https://review.openstack.org/523567
17:17:33	openstackgerrit	Dean Troyer proposed openstack/python-openstackclient master: DNM: Test -tips variations https://review.openstack.org/523938
18:07:32	openstackgerrit	Monty Taylor proposed openstack-infra/shade master: Fix basepython setting in tox.ini https://review.openstack.org/523915
18:07:33	openstackgerrit	Monty Taylor proposed openstack-infra/shade master: Avoid tox_install.sh for constraints support https://review.openstack.org/508061
18:08:26	Kvisle	Hi! I can see that https://review.openstack.org/#/c/522004/ looks good with review, but it doesn't seem to be merged yet. Do I need to do anything?
18:10:13	Shrews	Kvisle: No. Just waiting for mordred to have a chance to look it over first.
18:10:22	Kvisle	Ah =)
18:14:18	mordred	Kvisle: +A - thanks!
18:15:31	Kvisle	\o/
18:20:29	Shrews	mordred: that's an interesting error you got there on that image sorting fix. rsync seems to think there's a MITM attack
18:21:30	mordred	Shrews: yay!
18:21:34	Shrews	http://logs.openstack.org/93/523893/1/gate/openstack-tox-py27/8ddfac1/ara/result/38e984ce-db99-4dac-8399-5e4002a0025b/
18:25:03	mordred	Shrews: that sounds like a thing we should bring up over in infra land
18:25:12	Shrews	k
18:35:09	thingee	mordred: I'm still interested on working on oaktree. I read through the brain dump email and appreciate you writing that up, finally ;)
18:37:28	thingee	mordred: I would like to keep my core seat please.
18:40:07	cdent	I’d like to work on oaktree too, but I’m struggling to think I’ll have the time :(
18:51:47	openstackgerrit	Merged openstack-infra/shade master: Implement availability_zone_hints for networks and routers. https://review.openstack.org/522004
19:06:37	Shrews	just for the record, if oaktree attempts to handle more than just openstack clouds (which is already insanely complex), it will be doomed to failure. <EOM>
19:35:30	cdent	Shrews++
19:43:18	SamYaple	Shrews: all we gotta do is abstract a little bit more. easy peasy
19:44:52	openstackgerrit	Harald Jensås proposed openstack/python-openstackclient master: Allow setting network-segment on subnet update https://review.openstack.org/523974
19:48:02	thingee	Shrews: +1
20:18:38	openstackgerrit	Dean Troyer proposed openstack/python-openstackclient master: DNM: Test -tips variations https://review.openstack.org/523938
20:32:38	openstackgerrit	Dean Troyer proposed openstack/python-openstackclient master: DNM: Test -tips variations https://review.openstack.org/523938
20:55:44	openstackgerrit	Harald Jensås proposed openstack/python-openstackclient master: Allow setting network-segment on subnet update https://review.openstack.org/523974
21:04:56	mordred	thingee: woot. sorry it took me so long to get that down on paper in any useful way
21:05:06	mordred	SamYaple: agree
21:05:39	mordred	cdent: I'm going to give you core for now, just because - if it doesn't work out, no biggie
21:06:20	cdent	mordred: I think I can handle that responsibly. Thanks.
21:09:35	openstackgerrit	Merged openstack/python-openstacksdk master: Sort image update results before comparing https://review.openstack.org/523893
21:09:35	openstackgerrit	Merged openstack/python-openstacksdk master: Add ability to work in other auth contexts https://review.openstack.org/522970
21:09:37	openstackgerrit	Merged openstack/python-openstacksdk master: Add helper property to get the current user id https://review.openstack.org/522971
21:09:38	openstackgerrit	Merged openstack/python-openstacksdk master: Remove reference to context-managers from release note https://review.openstack.org/522972
21:09:38	openstackgerrit	Merged openstack/python-openstacksdk master: Document current_user_id in a release note https://review.openstack.org/522973
21:10:53	SamYaple	mordred: ill try to get on the rest of the reviews after i land in denver
21:11:07	SamYaple	keep them small and ill keep reviewing :P
21:18:18	mordred	SamYaple, Shrews: I think we need a tag or a topic or something to mark "this already landed in shade or os-client-config and is just a cherry-pick to get them up to date" vs. "this is actual code that needs actual review other than 'do the tests pass'"
21:19:35	openstackgerrit	Dean Troyer proposed openstack/python-openstackclient master: DNM: Test -tips variations https://review.openstack.org/523938

Earlier   Later