Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2026-08-26
14:56:28 opendevreview Stephen Finucane proposed openstack/openstacksdk master: Temporarily restore doc/requirements.txt https://review.opendev.org/c/openstack/openstacksdk/+/1002517
15:08:16 TheJulia stephenfin: so are you thinking that we should have a choice of None just to allow invocation of "no minimum", is that tht takeaway impression your having?
15:08:17 TheJulia https://review.opendev.org/c/openstack/keystoneauth/+/988544/3..4
15:10:20 stephenfin I was suggesting we should have choices, but I wasn't sure what the value of the unset choice would have to be
15:10:43 stephenfin either `choices=['1.2', '1.3', '']` or `choices=['1.2', '1.3', None]`
15:11:58 stephenfin I could test it myself but I was hoping you might have deployed the change somewhere and could easily test it
15:31:56 TheJulia unfortunately I'm in no place close to really be able to do much to test without a quick smoke test, but I'll see what I can hammer out in the next hour or so
15:54:37 opendevreview Oria Weng proposed openstack/openstacksdk master: Fix EndpointGroupProject resource base path https://review.opendev.org/c/openstack/openstacksdk/+/1001902
17:37:31 cardoe stephenfin: https://review.opendev.org/c/openstack/openstacksdk/+/1001423 is the change that needed the bump
17:39:16 stephenfin could you bump ksa in requirements.txt real quick?
17:39:41 stephenfin ignore me
17:40:19 stephenfin we do deps in pyproject.toml now (literally the thing we're just talking about in #openstack-release. it's getting late here) 🤦
17:43:57 cardoe Thanks
17:46:07 opendevreview Doug Goldstein proposed openstack/python-openstackclient master: Add a "cloud cache delete" command https://review.opendev.org/c/openstack/python-openstackclient/+/1001433
17:55:10 cardoe stephenfin: so I've got a handful more "cloud" commands
17:55:25 cardoe So using my Kubernetes example for how we work with things.
17:55:54 cardoe We've got domains and then projects in the domain. But a lot of our federated users will have an inherited role for a domain. So they'll have that role on all projects within the domain
17:56:04 cardoe So we use projects like kubernetes namespaces
17:56:34 cardoe So I've got to not break clouds.yaml, I've got another file context.yaml but if I upstream it I would fold it into clouds.yaml potentially.
17:57:31 cardoe But there's a context object
17:58:10 cardoe It's got a "current" and "previous" key
17:58:41 cardoe so I can run "openstack cloud use <blah>" instead of having OS_CLOUD exported into my environment
17:58:58 cardoe I prefer OS_CLOUD over what's in that context for proper behavior
17:59:24 cardoe You can also do "openstack cloud use -" and it'll flip to your previous one
18:00:33 cardoe I've also got "openstack cloud list" which shows you all your entries in clouds.yaml
18:00:41 cardoe just by name and shows you what's your current one
18:01:17 cardoe and since I was saying we use projects like namespaces... we've got an override which I think would need some re-thinking to be upstream worthy
18:01:32 cardoe But most folks have their clouds.yaml pointing at a domain they're working with
18:02:03 cardoe and then they can do "openstack project use <project-name>" and it'll override with OS_PROJECT_NAME taking preference
18:02:20 cardoe and "openstack domain use <domain>" and it'll override with OS_PROJECT_DOMAIN_NAME taking preference
18:02:51 cardoe This has a bit of an intersection with Dan's work.
18:06:35 cardoe Another one I use a bunch is "openstack whoami" which gives me information about who I am and what auth I've got.
18:19:10 TheJulia Just chiming in, but that sounds awesome
18:19:48 TheJulia since I've often felt "going and looking at env-vars, or translationg that cross to OS_CLOUD config feeling like an anti-pattern, espescailly if I'm doing anything advanced authentication wise
18:21:40 cardoe TheJulia: do I get bonus points for even patching starship to make my shell prompt show the current state?
18:21:55 cardoe it already supported reading OS_CLOUD env var
18:23:21 opendevreview Julia Kreger proposed openstack/keystoneauth master: Add minimum tls version and caller cipher controls https://review.opendev.org/c/openstack/keystoneauth/+/988544
18:23:32 TheJulia stephenfin: ^
18:23:55 TheJulia None is the default value when not set which gets interpretted, fwiw.
18:41:37 opendevreview Merged openstack/openstacksdk master: Temporarily restore doc/requirements.txt https://review.opendev.org/c/openstack/openstacksdk/+/1002517
18:52:06 opendevreview Merged openstack/openstacksdk master: Fix EndpointGroupProject resource base path https://review.opendev.org/c/openstack/openstacksdk/+/1001902
19:53:16 cardoe stephenfin: I fail at pep8. :(
20:18:38 opendevreview Merged openstack/python-openstackclient master: Bump openstacksdk minimum version https://review.opendev.org/c/openstack/python-openstackclient/+/1002467
20:20:56 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Keep the unscoped credential of a federated plugin https://review.opendev.org/c/openstack/openstacksdk/+/1001423
20:20:57 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Add a way to discard the authentication cached for a cloud https://review.opendev.org/c/openstack/openstacksdk/+/1001431
22:01:36 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Keep the unscoped credential of a federated plugin https://review.opendev.org/c/openstack/openstacksdk/+/1001423
22:01:36 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Add a way to discard the authentication cached for a cloud https://review.opendev.org/c/openstack/openstacksdk/+/1001431
22:07:37 opendevreview Oria Weng proposed openstack/python-openstackclient master: Migrate 'endpoint group' commands to SDK https://review.opendev.org/c/openstack/python-openstackclient/+/1001447
22:10:57 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Add a way to discard the authentication cached for a cloud https://review.opendev.org/c/openstack/openstacksdk/+/1001431
22:13:43 opendevreview Oria Weng proposed openstack/python-openstackclient master: Migrate 'endpoint group' commands to SDK https://review.opendev.org/c/openstack/python-openstackclient/+/1001447
22:21:27 opendevreview Doug Goldstein proposed openstack/openstacksdk master: Add a way to discard the authentication cached for a cloud https://review.opendev.org/c/openstack/openstacksdk/+/1001431
#openstack-sdks - 2026-08-27
09:48:11 stephenfin cardoe: just to confirm: are you planning to file a FFE for the websso changes?
09:49:55 stephenfin Also, I like the idea of the `openstack cloud *` commands you raised just so long as we can't get in a situation when OSC sees a different "state" for a cloud entry from what's in clouds.yaml. That would be very confusing
10:07:56 opendevreview Stephen Finucane proposed openstack/python-openstackclient master: volume: Replace server arg for volume attachment create https://review.opendev.org/c/openstack/python-openstackclient/+/1002622
10:09:38 opendevreview Merged openstack/openstacksdk master: Keep the unscoped credential of a federated plugin https://review.opendev.org/c/openstack/openstacksdk/+/1001423
12:11:17 cardoe stephenfin: I posted the FFE to the ML
13:21:50 stephenfin ty
14:36:03 stephenfin cardoe: err, which ML did you use?
14:36:14 cardoe openstack-discuss
14:36:47 cardoe https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/R6JZUE7AU77JHBRNILDRJ7MJXBHGEEBG/
14:37:28 stephenfin okay, corporate gmail routed it to spam. Thanks Google
14:43:08 cardoe Is that why none of my emails show up on the ML?
14:43:23 cardoe or I should say why nobody responds to me
14:43:47 cardoe I tried to write up something about the scopes but I missed the mark... I'm going to reply again.. https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/W2IE6PN5BU4G4VNUE6KYGAH2JE2LU7GS/
15:16:35 stephenfin (Un)fortunately (?) I see the TC proposals 😅 It was just the SDK one that went to spam
16:21:50 cardoe So I'd need to bump ksa to 5.17.0 no?
16:23:15 cardoe I guess https://review.opendev.org/c/openstack/keystoneauth/+/1000759 would need to be workflowed first.
#openstack-sdks - 2026-08-28
00:18:31 opendevreview Kyuyeong Lee proposed openstack/python-openstackclient master: Network: Simplify ip_availability_details access https://review.opendev.org/c/openstack/python-openstackclient/+/1002713
12:04:19 cardoe stephenfin: can you workflow the patch?
13:42:49 stephenfin cardoe: done. I'm off today but if you're able to propose a openstack/releases patch once your patch and TheJulia's patch have landed, I'll check in later today and +1 that also
13:43:03 cardoe I will do so.
13:43:13 cardoe I can review TheJulia's patch as well.
13:43:28 cardoe Maybe in a cycle or two I'll propose myself as a reviewer. :-D
13:44:03 cardoe stephenfin: enjoy the weekend
14:42:35 opendevreview Merged openstack/keystoneauth master: Add a v3websso authentication plugin https://review.opendev.org/c/openstack/keystoneauth/+/1000759
15:13:02 opendevreview Merged openstack/keystoneauth master: Add minimum tls version and caller cipher controls https://review.opendev.org/c/openstack/keystoneauth/+/988544
15:18:25 opendevreview Merged openstack/python-openstackclient master: Network: Simplify ip_availability_details access https://review.opendev.org/c/openstack/python-openstackclient/+/1002713
15:22:44 opendevreview Merged openstack/openstacksdk master: Add a way to discard the authentication cached for a cloud https://review.opendev.org/c/openstack/openstacksdk/+/1001431
15:22:47 opendevreview Merged openstack/python-openstackclient master: Add a "cloud cache delete" command https://review.opendev.org/c/openstack/python-openstackclient/+/1001433
18:40:15 opendevreview Silvia Wachira proposed openstack/python-openstackclient master: share: Add functional tests https://review.opendev.org/c/openstack/python-openstackclient/+/994970
21:01:51 opendevreview Oria Weng proposed openstack/python-openstackclient master: Migrate 'endpoint group' commands to SDK https://review.opendev.org/c/openstack/python-openstackclient/+/1001447
23:00:07 rm_work opendevreview: just noticed this -- doesn't this do the opposite of what we'd want with the specific example of the websso plugin cardoe ?
23:02:09 rm_work stephenfin: with the openstack in-tree stuff merging, are we closer to getting gophercloud moving? and do I need to make any adjustments on my PRs? I've done the maximum due diligence I can on those I think, TBH I do not really do Golang so they are 99% AI code really, all I could do is look at the patterns and guide it a bit <_<
23:08:33 cardoe rm_work: what do you mean?
23:11:49 rm_work so the issue I was raising with the re-use of the entrypoint was that if you had the old out-of-tree websso plugin installed, AND the one you are merging into keystoneauth directly, it was unclear which would load
23:12:10 cardoe It prefers out of tree plugins.
23:12:14 rm_work with that patch, it will DEFINITELY load the old out-of-tree plugin, which is going to be deprecated and almost certainly not the one we'd want loaded
23:12:17 rm_work right
23:12:18 rm_work that's bad
23:12:27 rm_work in our case at least for websso
23:12:43 cardoe The point was that vexxhost wants to continue to use theirs.
23:13:17 rm_work ok, so then ... if the out-of-tree plugin is installed in an old env, it'll (quietly?) take over
23:13:25 rm_work or will it be noisy, that would at least be good
23:13:54 cardoe And in the other patch it was decided that people should be able to overload in-tree versions with out of tree.
23:14:10 rm_work I guess in my case I can try to remember to make sure I remove it from my own envs, but i forsee getting support requests if it ever breaks and being very confused until I realize people still have the old one
23:14:25 cardoe It’s not making any noise but we could probably do that.
23:14:57 rm_work yeah the entire basis of my comments about that since the beginning has been "some people might load up the old out-of-tree plugin and that would be bad"
23:15:13 rm_work so this just seems like it was "solved" by doing the opposite of what I was hoping, lol
23:15:24 rm_work but I do get the point about overriding
23:15:36 rm_work it just almost makes me think we SHOULDN'T have re-used the entrypoint

Earlier   Later