| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2017-07-16 | |||
| 18:58:51 | stevemar | amotoki: no need for 484209, i'm pretty sure it's broken :( | |
| #openstack-sdks - 2017-07-17 | |||
| 12:06:50 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Fix masked variable name https://review.openstack.org/484340 | |
| 12:06:51 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Add support for service type aliases https://review.openstack.org/484341 | |
| 12:37:51 | openstackgerrit | M V P Nitesh proposed openstack/python-openstackclient master: now we can add description for role creation in OSC https://review.openstack.org/484355 | |
| 13:43:45 | stevemar | @mordred poke | |
| 13:43:52 | stevemar | eww, my slack is showing | |
| 13:44:45 | stevemar | mordred: IIRC you (or someone from the infra team) had managed to check in docs and create slide decks out of them? this making any sense? | |
| 13:58:56 | mordred | stevemar: yup! we use gertty for the slides - and they mostly translate directly into sphinx docs | |
| 13:58:59 | mordred | stevemar: https://docs.openstack.org/shade/latest/user/multi-cloud-demo.html is an example | |
| 14:06:57 | mordred | efried: I think we should do the max_endpoint_version stuff on top of your current wip patch, right? | |
| 14:07:06 | efried | mordred Yeah. | |
| 14:07:20 | efried | That one ought to be a fairly noncontroversial one. | |
| 14:09:15 | stevemar | mordred: ah, i was thinking about http://inaugust.com/talks/os-client-config.html#/ | |
| 14:09:28 | stevemar | mordred: is the source for that anywhere? | |
| 14:10:45 | mordred | stevemar: ah - yah - so, the ones in my inaugust.com one I do in reveal.js html directly - the source for that is here: http://git.inaugust.com/cgit/inaugust.com/tree/src/talks/os-client-config.hbs | |
| 14:11:17 | mordred | stevemar: it's mildy templated using a build script in that repo that just handles my typical frontmatter | |
| 14:11:20 | efried | mordred Is keystoneauth1.identity.base.BaseIdentityPlugin#get_endpoint a new (as-yet-unreleased) API? | |
| 14:11:58 | efried | i.e. can I remove the `version` kwarg from that one, as part of https://review.openstack.org/#/c/483604/ ? | |
| 14:12:14 | mordred | efried: nope. that's a very used api | |
| 14:12:34 | efried | mordred Okay, but get_endpoint_data in same class? | |
| 14:14:07 | mordred | https://github.com/openstack/keystoneauth/blob/2.21.0/keystoneauth1/identity/base.py#L160-L162 | |
| 14:14:11 | mordred | efried: get_endpoint_data is safe | |
| 14:14:16 | efried | mordred Cool. | |
| 14:14:28 | stevemar | mordred: hmm. i'm wondering if i could make some markdown, translate it to HTML and publish that | |
| 14:14:38 | stevemar | so the source isn't HTML :( | |
| 14:15:18 | mordred | stevemar: yes you can - in fact, you can write slides in markdown for reveal.js directly - also the tool pandoc can convert from rst and markdown formats into html backends | |
| 14:15:42 | mordred | stevemar: although it currently doesn't do as good a job as I want with rst slides written for presentty | |
| 14:16:36 | mordred | so I've got a todo-list item to make improvements to rst->reveal.js conversion so that I can write slides for presentty then occasionally show them as reveal.js instead - and also a sphinx plugin to cleanup the presentation of them when those rst files are included in docs | |
| 14:33:35 | efried | mordred Who's gonna know about ironic microversions? Should I ask in #openstack-ironic? | |
| 14:54:51 | mordred | yah ... | |
| 15:33:05 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: WIP Use CacheControl for caching https://review.openstack.org/484397 | |
| 15:43:24 | openstackgerrit | Monty Taylor proposed openstack/service-types-authority master: Don't dump trailing whitespace https://review.openstack.org/484400 | |
| 15:59:07 | mordred | dhellmann, efried: ^^ both may be interesting. the first one is in response for cdent suggesting we go ahead and do that. it turns out we've got a little bit of a long tail to chase before we can - so I hope we can land the current patch as is and do the change later | |
| 15:59:58 | efried | mordred (min_version='latest', max_version=None) converts to (min_version=None, max_version='latest') -- First, I don't think that's right; I think it should convert to (min_version='latest', max_version='latest'). Second, I thought this behavior was in a doc somewhere, but I can't find it - can you help? | |
| 16:03:24 | mordred | efried: it should be in that spec update, which needs another update now ... | |
| 16:03:59 | efried | mordred Do you agree that (min_version='latest', max_version=None) should convert to (min_version='latest', max_version='latest')? | |
| 16:04:19 | mordred | efried: I think th ekey is that min_version='latest', max_versoin=None or minversion=None, max_version='latest' should both return latest, yeah? | |
| 16:04:24 | mordred | efried: yes | |
| 16:04:43 | mordred | efried: also - I got an answer from ironic on microversions - they only support old form | |
| 16:04:52 | mordred | HOWEVER - I think I've got an answer for the whole thign withouth special casing :) | |
| 16:05:34 | efried | mordred It depends whether 'latest' means 'the latest you have' or 'the actual latest'. | |
| 16:06:34 | efried | mordred In the former case, None/latest could reasonably get you some version that's not the actual latest, whereas latest/latest would fail because we can't satisfy it | |
| 16:06:55 | efried | mordred That may be a nonsensical distinction, I suppose. | |
| 16:07:20 | efried | By definition, the latest we know about is the latest we can act upon; we have no way of knowing if it's really the absolute latest. | |
| 16:07:48 | mordred | efried: there is no concept of "the actual latest" | |
| 16:07:52 | mordred | so yes | |
| 16:08:17 | mordred | "latest" always means "the latest version the cloud you're talking to supports" | |
| 16:08:56 | efried | mordred Was gonna say, if max_version=latest, min_version is irrelevant. But that's not right, cause min_version can still indicate a lower bound, which WOULD cause an error if the latest is lower than that. | |
| 16:09:16 | mordred | yup | |
| 16:11:57 | mordred | cdent: btw - patches up for cachecontrol - and I'm about to have a thing related to the microversion thing | |
| 16:12:22 | mordred | cdent: tl;dr on cachecontrol - we can't do it this week - we have some testing plumbing in our testing libraries we need todo (the code works, we just can't test it) | |
| 16:12:23 | cdent | mordred: i’m basically done for today and out tomorrow doing family stuff, but will look wednesday | |
| 16:12:59 | cdent | yeah, saw your comment, makes sense, and there’s no real rush from my point of view, I just didn’t want to build in some kind of resistance to it over the long term | |
| 16:14:25 | mordred | cdent: ++ - just wanted to follow up with you on the topic | |
| 16:15:00 | mordred | cdent: I'm going to bug efried and dhellmann to get that patch landed for now while you're out -so wanted to make sure you're ok with moving forward with the plan in place to land the patch as soon as we have the testing libraries updated | |
| 16:15:15 | cdent | yes, totally cool with that | |
| 16:15:38 | mordred | jamielennox: speaking of - when you get in, please ping me - I've got a 'fun' challenge - I want to use requests_mock and cachecontrol on the same library, and both are transport adapters which I can't stack | |
| 16:16:55 | mordred | jamielennox: (we want to wind up with requests.session being mocked by requests_mock and cachecontrol then being mounted on that so that cachecontrol is operating the whole time but when cachecontrol decides that it needs to make a remote call it hits requests_mock | |
| 16:18:21 | openstackgerrit | Gage Hugo proposed openstack/keystoneauth master: Enable sphinx todo extension https://review.openstack.org/484417 | |
| 16:36:00 | openstackgerrit | Monty Taylor proposed openstack/service-types-authority master: Remove leading openstack/ from project name https://review.openstack.org/484422 | |
| 17:34:07 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: Add support for getting OpenStack project information https://review.openstack.org/484435 | |
| 17:35:37 | mordred | dhellmann, efried: ^^ that should help both the docs job (pulled in doug's method from that patch) and also the microversion header patch | |
| 17:37:58 | efried | mordred Is there a test for get_service_data? | |
| 17:38:18 | mordred | efried: thanks - I missed that one - one sec | |
| 18:20:33 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: Add support for getting OpenStack project information https://review.openstack.org/484435 | |
| 18:20:34 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: Refactor bulk of data checking into mixin class https://review.openstack.org/484445 | |
| 18:21:01 | mordred | efried: ok. added that test - and another missing test - then got annoyed and refactored a chunk of copy-pasta | |
| 18:23:54 | openstackgerrit | Monty Taylor proposed openstack/service-types-authority master: Remove leading openstack/ from project name https://review.openstack.org/484422 | |
| 18:24:14 | mordred | efried: k - rebased that ^^ it's not actually related to the trailing-whitespace | |
| 19:14:17 | lbragstad | curious if i could get a quick api change sanity check here - https://review.openstack.org/#/c/482190/3 | |
| 19:14:40 | lbragstad | i'm just wondering if ^ is ok to do or not and looking for some advice | |
| 20:02:28 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: Add support for getting a complete list of type names https://review.openstack.org/484473 | |
| 20:03:31 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Add support for service type aliases https://review.openstack.org/484341 | |
| 20:03:32 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Fix masked variable name https://review.openstack.org/484340 | |
| 20:04:16 | mordred | efried, cmurphy: ^^ that should take care of service-type alias support AND supporting old-style microversion headers AND supporting services, like cinder, that have a different microversion name than official service-type | |
| 20:06:31 | cmurphy | mordred: o7 | |
| 21:04:04 | mordred | cmurphy: oh - btw - I made you an initial core on os-service-types since you have context paged in for it | |
| 21:04:41 | cmurphy | mordred: oh neato | |
| 21:51:28 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Fix column names for server list --no-name-lookup https://review.openstack.org/484500 | |
| 22:52:29 | johnsom | Hi folks, FYI, I am going to try to pick up the octavia/load-balancer SDK patches | |
| 22:52:53 | johnsom | Reading the contributor docs now... grin | |
| 22:56:34 | johnsom | Should I use allow_retrieve or allow_get? I see the docs say allow_retrieve but some, neutron ports are using allow_get | |
| 23:05:44 | johnsom | Digging I see a bunch of stuff moving to resource2 so I guess that is the path forward. | |
| 23:24:58 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Remove deprecated_since parameter for interface https://review.openstack.org/484528 | |
| 23:27:43 | efried | So mordred I've been banging my head against these same couple of test cases for the past couple of days. I'm at the point now where I'm actually working to understand all the logic in the new discovery algorithms. | |
| 23:28:27 | efried | mordred I'm pretty desperate not to have to support version+min_version+max_version in all these new APIs. | |
| 23:28:37 | openstackgerrit | Monty Taylor proposed openstack/keystoneauth master: Remove deprecated_since for interface and fix text https://review.openstack.org/484528 | |
| 23:30:13 | mordred | efried: is the code that's up your latest? | |
| 23:30:20 | mordred | efried: if so, I can take a look first thing in the morning | |
| 23:31:09 | efried | mordred Well, nooo, not really. I've got a mess of a sandbox at this point which includes some things that are fixes, some things that are experiments, and a bunch of debug code. | |
| 23:31:16 | mordred | nod | |
| 23:31:32 | efried | mordred But I can try to work something up that represents only the first of those things and post it before morning. | |
| 23:31:57 | mordred | efried: cool - it may be that we're at the point where if you show me what you've got a second set of eyes can help move if along | |
| 23:32:50 | efried | mordred What I'm pretty sure I'm screwing up is the logic around https://review.openstack.org/#/c/483604/6/keystoneauth1/discover.py@905 | |
| 23:33:09 | efried | I *thought* I didn't need all those weird and complex variables anymore. | |
| 23:33:46 | efried | But my _catalog_matches_version is a tad too loose, so I'm not getting latest when I should because a non-latest matches. | |
| 23:34:41 | mordred | efried: yah - I tihnk you still need them - but I think after you normalize version -> min/max you should be able to do update the definition of exact_match | |
| 23:35:31 | efried | mordred Yeah, I think that new def is making sure max_version[0] is equal, but max_version isn't exceeded. | |
| 23:36:28 | efried | So version_match, but with the inequality reversed. | |
| 23:39:19 | mordred | efried: actually - so version=2 -> min_version=2.0 max_version=2.latest right? | |