| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2019-05-13 | |||
| 13:22:27 | efried | gtema: Sure thing. How much background do you have on Nova's situation here? | |
| 13:23:07 | gtema | not necessarily much, but very interested in the result and possible impl. options | |
| 13:24:01 | efried | Nova talks to a bunch of other APIs (glance, cinder, neutron, ...) to do its thing. Around queens we did some work so that operators could (mostly) set up those communications using common ksa oslo.config options. But they're still talking through python-*client. | |
| 13:24:37 | efried | which means that, for the most part, we use the ksa config opts to create an adapter to get an endpoint, then throw away the rest of the adapter and pass the endpoint to the python-*client, which uses it to create... an adapter. | |
| 13:24:42 | efried | all of which kind of sucks. | |
| 13:26:22 | gtema | sure, and I am exactly on the front of replacing python-*client to SDK, so should be fitting | |
| 13:27:16 | efried | gtema: so step 1 is getting openstacksdk to understand ksa oslo.config opts, which is wip here: https://review.opendev.org/#/c/643601/ | |
| 13:27:39 | efried | and proven in this series https://review.opendev.org/#/c/643664/ | |
| 13:28:13 | efried | and dustinc has been working in that series to replace python-ironicclient with sdk calls, one by one. | |
| 13:28:54 | efried | That first one just needs tests so we can merge it. Then get it in a release of openstacksdk so nova can require it. | |
| 13:29:21 | mordred | yeah - I've got adding tests to that patch on my todo list for this week | |
| 13:29:21 | efried | so we can continue the work of swapping out client pieces with sdk calls. | |
| 13:29:39 | efried | mordred: maybe gtema is volunteering to take over :P | |
| 13:29:43 | gtema | efried: great. Will be looking at. Need to run away unfortunately. Count me in for SDK parts as told during PTG | |
| 13:29:51 | efried | thanks gtema :) | |
| 15:48:45 | AJaeger | could somebody review https://review.opendev.org/656704 for os-service-types to add testing jobs | |
| 15:49:23 | AJaeger | (update not add) | |
| 15:54:42 | AJaeger | thanks, cmurphy ! | |
| 15:59:21 | cmurphy | yw AJaeger | |
| 16:03:50 | openstackgerrit | Merged openstack/os-service-types master: Update testing jobs https://review.opendev.org/656704 | |
| 16:22:37 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Use cliff formattable columns in volume v1 commands https://review.opendev.org/658856 | |
| 16:53:12 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Use cliff formattable columns in volume v2 commands https://review.opendev.org/658863 | |
| 17:20:12 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Use cliff formattable columns in object storage commands https://review.opendev.org/658871 | |
| 18:15:28 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Add support for vendor hooks https://review.opendev.org/657848 | |
| 21:46:23 | dustinc | QQ fo ryou guys..I see baremetal is implementing patch_node in next release according to docs (https://docs.openstack.org/openstacksdk/latest/user/proxies/baremetal.html#openstack.baremetal.v1._proxy.Proxy.patch_node) but not sure where to find spec for it, is there a spec somewhere? | |
| 22:45:27 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: [Compute]Make column content readable for both human and machine https://review.opendev.org/658933 | |
| #openstack-sdks - 2019-05-14 | |||
| 01:22:27 | openstackgerrit | caoyuan proposed openstack/keystoneauth master: Replace git.openstack.org URLs with opendev.org URLs https://review.opendev.org/655016 | |
| 04:02:48 | openstackgerrit | qian wang proposed openstack/microversion-parse master: Add .idea/* in .gitignore https://review.opendev.org/658956 | |
| 06:49:06 | openstackgerrit | Artem Goncharov proposed openstack/openstacksdk master: Add image.stage methods https://review.opendev.org/652981 | |
| 10:10:55 | openstackgerrit | Merged openstack/openstacksdk master: Extract image download method into a mixin https://review.opendev.org/658346 | |
| 12:52:53 | efried | dtantsur: see question from dustinc about 15h ago | |
| 13:13:28 | mordred | efried, dustinc: what do you mean by "spec" in this context? | |
| 13:14:08 | openstackgerrit | Merged openstack/openstacksdk master: Replace use of log.warn with log.warning https://review.opendev.org/658529 | |
| 13:18:44 | efried | mordred: FYI dustinc is on Pacific time, so it may be a couple hours before he gets back to you. (And I don't know what he meant by 'spec') | |
| 13:20:48 | mordred | efried: neat! I'm also in and out today - so we may or may not manage to connect with each other :) | |
| 13:21:29 | efried | mordred: In general, what's the process when someone wants to enhance one of these proxies? Is a spec (as in rst document describing the change) required? | |
| 13:22:30 | mordred | nope - one just dives in. if it's a big change, checking in with other folks first is a good idea - but we don't have nearly enough active devs to warrant the overhead of specs atm | |
| 13:23:21 | efried | ack | |
| 13:23:22 | gtema | mordred, why do you say so, a 3 of us is not enough? ;-) | |
| 13:23:42 | efried | dustinc: FYI the change appears to have been delivered via https://review.opendev.org/#/c/647730/ | |
| 13:26:43 | efried | dustinc: If you're looking for more information on what the json payload should look like, I would start with the corresponding baremetal API: https://developer.openstack.org/api-ref/baremetal/?expanded=update-node-detail#update-node | |
| 13:27:16 | mordred | gtema: 3 of us is perfect :) | |
| 13:27:28 | gtema | we are nearly almighty | |
| 13:27:34 | mordred | nearly | |
| 13:27:35 | gtema | :D | |
| 13:27:38 | efried | (which, mordred/dtantsur/gtema, would be a good thing to link from any of these proxy methods) | |
| 13:27:49 | efried | (especially the ones that have **kwargs) | |
| 13:27:51 | gtema | for baremetal its dtantsur | |
| 13:27:53 | mordred | efried: I agree - that's not a ba idea at all | |
| 13:29:27 | efried | prolly not | |
| 13:29:38 | efried | since it's in another repo | |
| 13:29:48 | efried | I was thinking of the :oslo.config:`group.option` role | |
| 13:30:02 | efried | but that only works in same repo | |
| 13:31:26 | mordred | yeah - there's ways to do fancy sphinx stuff with multi-repo external links and stuff - and connecting that with zuul for multi-repo source might be neat - but it's also probably easier to just put the links in for now | |
| 13:31:51 | mordred | I could see us easily getting so clever we don't understand how to write docs anymore | |
| 13:32:52 | gtema | ++ | |
| 13:33:10 | efried | I thought we had already passed that point. | |
| 13:33:23 | efried | only stephenfin knows how to write docs | |
| 13:33:26 | efried | job security | |
| 13:33:50 | stephenfin | /o/ | |
| 13:33:53 | stephenfin | \o\ | |
| 13:34:30 | mordred | efried: does that mean stephenfin will write all my docs for me? neat! | |
| 13:34:47 | stephenfin | efried: It could be done by building on top of intersphinx | |
| 13:35:00 | stephenfin | but whether it's important enough is another matter | |
| 13:37:10 | efried | considering we should ideally do it for every proxy method, and probably a bunch of the others, it would be nice if it happened automatically without any input. Like introspected something about the method. Like if we (already) had decorators on the methods indicating which operation they invoke. But we don't. So we'd have to add something to each method anyway. Or something more convoluted like a centralized mapping. | |
| 13:37:23 | efried | so if we're going to do something manual to each method anyway, might as well be adding the actual link. | |
| 13:37:32 | efried | except... | |
| 13:37:38 | dustinc | efried mordred thanks I’m in and out next couple hours but will check that out. | |
| 13:37:40 | efried | stephenfin: how do I noqa a docstring? | |
| 13:39:52 | dustinc | And by spec I meant blueprint so I could use it to find the change so all is good. Thanks! | |
| 13:40:36 | efried | dustinc: You can use some git tricks to find the change, lmk if you want to go over that. | |
| 13:40:56 | stephenfin | """Some string. | |
| 13:40:57 | stephenfin | ||
| 13:41:00 | stephenfin | More stuff. | |
| 13:41:03 | stephenfin | """ # noqa | |
| 13:41:08 | stephenfin | ^ efried (I think) | |
| 13:41:11 | efried | ah, /me tries... | |
| 13:42:06 | gtema | efried, dustinc, do you need placement support in sdk? | |
| 13:42:07 | efried | stephenfin: nope :( | |
| 13:42:20 | efried | gtema: It's already there, proxy-wise. | |
| 13:42:39 | efried | gtema: Other than something broken in the dep chain, this works: https://review.opendev.org/656023 | |
| 13:42:39 | stephenfin | Then I have no idea, I'm afraid. I assume it's the line length thing that's causing issues? | |
| 13:42:55 | efried | stephenfin: Yes, I have a hyperlink in the docstring and the hyperlink is >80c by itself. | |
| 13:43:13 | stephenfin | You could manually wrap inside the docstring. It would make for uglier code but keep flake8 happy | |
| 13:43:19 | stephenfin | Otherwise, use a URL shortner? | |
| 13:43:22 | efried | backslash? | |
| 13:43:23 | gtema | efried: I mean in a normal sdk way (with resources and refular proxy methods) | |
| 13:43:34 | gtema | s/refular/regular/ | |
| 13:43:53 | efried | gtema: I think we've talked about it before and decided it's not worth it, at least for now. | |
| 13:44:04 | gtema | ok, thks | |
| 13:44:17 | efried | gtema: It could be kinda neat, but the straight up APIs are pretty easy to use as is. | |
| 13:44:40 | gtema | that's good | |
| 13:44:59 | efried | Feel like if people have that kind of time, there's more important things to be done :) | |
| 13:45:13 | efried | like figuring out how to put long hyperlinks into the docstrings | |
| 13:45:15 | gtema | :) | |
| 13:46:54 | efried | stephenfin: backslash-wrapping works, and isn't *too* hideous. Thanks. | |
| 13:48:15 | efried | ahcrap. The pep8 error I was getting, it wasn't even in the docstring I was messing with. | |
| 13:48:40 | mordred | efried: BOW DOWN TO THE PEP8 GODS | |
| 13:51:36 | openstackgerrit | Eric Fried proposed openstack/openstacksdk master: Link to baremetal API reference from patch_node https://review.opendev.org/659068 | |
| 13:51:45 | efried | mordred, gtema, dtantsur: Toe in the water ^ | |