| opendevreview | Merged openstack/ansible-collections-openstack master: Fix dns zone update issues in CI https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1011348 | 08:30 |
|---|---|---|
| stephenfin | cardoe: in what context? | 08:37 |
| stephenfin | (the UUID-like type) | 08:37 |
| opendevreview | Jed Preist proposed openstack/ansible-collections-openstack master: Add target_provision_state option to baremetal_node https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1010852 | 09:57 |
| opendevreview | Merged openstack/openstacksdk master: Fix Barbican key-manager microversion negotiation https://review.opendev.org/c/openstack/openstacksdk/+/1011547 | 10:06 |
| opendevreview | Merged openstack/openstacksdk master: Add Barbican secret store migrate helper https://review.opendev.org/c/openstack/openstacksdk/+/1002400 | 10:06 |
| opendevreview | Jed Preist proposed openstack/ansible-collections-openstack master: Add target_provision_state option to baremetal_node https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1010852 | 10:08 |
| opendevreview | Moritz Tanner proposed openstack/ansible-collections-openstack master: image: do a web-download import when uri is given https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1006899 | 10:20 |
| opendevreview | Merged openstack/ansible-collections-openstack master: Handle the cloud parameter as a dict in application_credential module https://review.opendev.org/c/openstack/ansible-collections-openstack/+/955695 | 11:50 |
| opendevreview | Merged openstack/ansible-collections-openstack master: Use **-operator to unpack dict in volume_type_access https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1006027 | 11:59 |
| opendevreview | Merged openstack/ansible-collections-openstack master: Fix: Include target_project_id in RBAC creation https://review.opendev.org/c/openstack/ansible-collections-openstack/+/964684 | 12:00 |
| cardoe | stephenfin: adding typing to functions | 12:13 |
| stephenfin | So I don't believe you can indicate the "shape" of a string with type hints. You'd need a separate validation framework for this like what pydantic has https://pydantic.dev/docs/validation/2.2/usage/types/uuids/ | 12:19 |
| stephenfin | The best we can do is insist on passing around instances of uuid.UUID rather than str | 12:19 |
| stephenfin | also worth noting that a lot of things in OpenStack that look like UUIDs aren't actually enforced as UUIDs. Keystone is the main culprit here but iirc it's not the only one | 12:21 |
| opendevreview | Jed Preist proposed openstack/ansible-collections-openstack master: Add target_provision_state option to baremetal_node https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1010852 | 12:35 |
| opendevreview | Moritz Tanner proposed openstack/ansible-collections-openstack master: image: do a web-download import when uri is given https://review.opendev.org/c/openstack/ansible-collections-openstack/+/1006899 | 13:32 |
| sean-k-mooney | stephenfin: uuid.UUID is not a backward compatible change however | 14:43 |
| sean-k-mooney | but ya i was gong to mention the keyston thing | 14:44 |
| sean-k-mooney | not all identifies are actully uuids event if that is the defualt generated by some backends | 14:44 |
| sean-k-mooney | the same is true of the nova flavor id | 14:44 |
| sean-k-mooney | its a string not a uuid | 14:44 |
| sean-k-mooney | it just defaults to a uuid if you leave the api autogenerate it for you | 14:45 |
| sean-k-mooney | stephenfin: cardoe keystons id cant be a uuid in all case bacue some time it the Acitive directory/kerboeros identifier | 14:45 |
| sean-k-mooney | although if your using federation it alwasy provides a uuid in that one case | 14:46 |
| cardoe | So I was more thinking of places that we take in "instance_uuid" in the SDK... to use a nova example. | 14:46 |
| cardoe | Really I'm asking for Ironic because I'm trying to lift ironicclient to openstacksdk and we've got "node_ident" and "port_ident" | 14:47 |
| cardoe | Which is either a UUID or the name of the node or port | 14:47 |
| sean-k-mooney | so it woudl be possibel to defien a type unition | 14:47 |
| cardoe | But then it's automatically looked up and converted to the UUID | 14:47 |
| sean-k-mooney | that is a string with uuid or name validation via regex i think | 14:47 |
| sean-k-mooney | and use that isntead | 14:48 |
| cardoe | Yeah. | 14:48 |
| sean-k-mooney | the underlying type woudl still be a str | 14:48 |
| sean-k-mooney | but it woudl vaildatiable i think | 14:48 |
| cardoe | I'd like to have something common that can be shared across multiple places. | 14:48 |
| sean-k-mooney | thats kind of what stephenfin was suggesting too i think | 14:48 |
| sean-k-mooney | the other thing you coudl do is you can use the abotiy to have multipel deffintion of the same fuction | 14:51 |
| sean-k-mooney | to have one that accept a sting and a seocnd that asscele a uuid_or_name type | 14:51 |
| sean-k-mooney | that woudl mean that thing that opt into the new type get the validation | 14:51 |
| sean-k-mooney | adn older users can still pass the raw string | 14:51 |
| sean-k-mooney | i.e. sort of liek this | 14:53 |
| sean-k-mooney | https://github.com/openstack/openstacksdk/blob/master/openstack/compute/v2/_proxy.py#L468-L486 | 14:53 |
| cardoe | Yep. I love me some @overload | 14:53 |
| sean-k-mooney | but instead of the empy ... | 14:53 |
| sean-k-mooney | you could provide a body for the stricter one that does the type check and then passse the data through | 14:54 |
| cardoe | Another item I was thinking about... "openstack --version"... there's no consideration of the other clients that might be installed or showing their version | 14:54 |
| sean-k-mooney | well its the version of openstack client | 14:54 |
| sean-k-mooney | not the verion of the plugins | 14:54 |
| sean-k-mooney | so it should not includ the other plugins inits out put but you coudl add a --plugin-versions | 14:55 |
| cardoe | Right. Just saying it makes a difference. | 14:56 |
| sean-k-mooney | i personally just use pip show but i get that you might not have pip installed in all cases | 14:56 |
| cardoe | Just had someone with up to date python-openstackclient but out of date plugins | 14:56 |
| sean-k-mooney | ah | 14:56 |
| sean-k-mooney | having some plugin list or version commands/flags could help with that for sure | 14:57 |
| cardoe | yeah cause I posted on the launchpad asking them for openstack --version and they gave it but of course the lag of async comms | 14:57 |
| cardoe | And then I had to tell them to pip show... which they told me they didn't use pip... so I had to ask how they installed it and then go around and around that. | 14:58 |
| cardoe | If I could have said paste openstack --version and openstack --plugin-versions like you had suggested from the get go it would have been quicker | 14:59 |
| cardoe | Anyway, just something that was top of mind. | 14:59 |
| JayF | consider also adding a --all-versions or similar | 15:00 |
| JayF | just to make that common interaction one command | 15:00 |
| cardoe | Yeah that would be good. | 15:00 |
| cardoe | JayF: the bug was that python-ironicclient doing the lookup of the node that I believe you merged the other day. | 15:00 |
| cardoe | for context | 15:01 |
| JayF | I remember that and am not just nodding as if I remember | 15:01 |
| * JayF nods as if he remembers | 15:01 | |
| cardoe | Ironic API takes in a node UUID only and the CLI doesn't make that clear that it had to be the UUID and we passed along the value. | 15:02 |
| cardoe | I fixed it by making the client lookup the node UUID from the supplied user input. | 15:02 |
| cardoe | But if we had types on those methods, it would have made things more clear. | 15:02 |
| opendevreview | Stephen Finucane proposed openstack/python-openstackclient master: quota: Fix behavior of -c opt for quota show https://review.opendev.org/c/openstack/python-openstackclient/+/1011892 | 15:18 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2026.2: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011900 | 16:44 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2026.1: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011902 | 16:58 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2026.1: Read connect_retry_delay in get_session_client https://review.opendev.org/c/openstack/openstacksdk/+/1011903 | 16:58 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2026.1: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011904 | 16:58 |
| opendevreview | Stephen Finucane proposed openstack/python-openstackclient master: quota: Fix behavior of -c opt for quota show https://review.opendev.org/c/openstack/python-openstackclient/+/1011892 | 17:04 |
| shermanm | hey, I was submitting backports for https://review.opendev.org/c/openstack/openstacksdk/+/1011585, but noticed that for 2026.1 and earlier, | 17:13 |
| shermanm | jgilaber's https://review.opendev.org/c/openstack/openstacksdk/+/986350 wasn't backported yet. I'm happy to submit backports for that as well, but didn't want to do a rudeness :) | 17:13 |
| stephenfin | shermanm: Feel free to backport both as stacked changes :) | 17:39 |
| shermanm | +1, will do | 17:39 |
| stephenfin | cardoe: I missed the rest of that thread but I don't think there is much more we can do currently | 17:42 |
| cardoe | alright. was just curious before I spent time hacking away at it. | 17:42 |
| stephenfin | Long-term, I'd like to copy what pydantic does and make better use of typing.Annotated to express these kinds of things but that's a bit off yet | 17:43 |
| stephenfin | https://docs.python.org/3/library/typing.html#typing.Annotated | 17:43 |
| cardoe | Yeah that'd be great. | 17:43 |
| stephenfin | passing type=uuid.UUID to resource.Body _might_ work, but I've never tested it | 17:43 |
| opendevreview | Stephen Finucane proposed openstack/openstacksdk stable/2026.1: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011904 | 17:47 |
| shermanm | ah, good catch on the stacked cherry-pick by version, instead of each from master backwards | 17:49 |
| stephenfin | yeah, that's just convention to ensure we don't miss any branches | 17:49 |
| shermanm | a follow-up question on these "missing methods on the sdk" fixes. What is the actual authoritative source of what args/methods should be there? | 17:49 |
| stephenfin | err, a little more context? | 17:50 |
| shermanm | e.g. status_code_retry was accepted, but silently ignored | 17:50 |
| shermanm | but how do we know that it's supposed to be there | 17:50 |
| shermanm | same for status_code_retry_delay retriable_status_codes | 17:51 |
| stephenfin | I imagine we should support practically everything that Adapter supports | 17:51 |
| stephenfin | at least where it makes sense to configure said thing globally | 17:51 |
| shermanm | basically I'm imagining an automated test that just enumerates everything on Adapter, and makes sure we can accept it | 17:52 |
| shermanm | to avoid more "this thing is missing" bugs down the road | 17:52 |
| stephenfin | fortunately Adapter is fully typed now, so the list of option is accepts/expects is finite https://github.com/openstack/keystoneauth/blob/master/keystoneauth1/adapter.py#L131-L160 | 17:53 |
| stephenfin | I don't necessarily want a test to validate that, since (a) it's unlikely to change often and (b) when it does, it would break our gate and force a ksa bump, but I imagine a one-off script would be very feasible | 17:54 |
| stephenfin | *I don't think we necessarily | 17:55 |
| stephenfin | actually, we can go one better: openstack.proxy.Proxy (which is what we ultimatly construct an instance of) has a near-identical __init__ signature. We could have test against _that_ since it's in-tree | 17:57 |
| shermanm | is there a clean way to warn if an unsupported option is passed instead of just silently dropping it? | 17:58 |
| shermanm | not sure if we're passing through kwargs in a way that makes it impractical | 17:58 |
| stephenfin | well you can't: kwargs is passed straight through to `constructor` [1], which default to openstack.proxy.Proxy [2], and that accepts a finite set of arguments (no kwargs) [3] | 18:07 |
| stephenfin | [1] https://github.com/openstack/openstacksdk/blob/a0d83104c392c50851c6f6eaddb3cbb44f0d3c03/openstack/config/cloud_region.py#L1311 | 18:08 |
| stephenfin | [2] https://github.com/openstack/openstacksdk/blob/a0d83104c392c50851c6f6eaddb3cbb44f0d3c03/openstack/config/cloud_region.py#L1248 | 18:08 |
| stephenfin | [3] https://github.com/openstack/openstacksdk/blob/a0d83104c392c50851c6f6eaddb3cbb44f0d3c03/openstack/proxy.py#L101-L138 | 18:08 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011917 | 18:09 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: Read connect_retry_delay in get_session_client https://review.opendev.org/c/openstack/openstacksdk/+/1011918 | 18:09 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011919 | 18:09 |
| shermanm | ah, that context helps, thanks! | 18:15 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011927 | 19:04 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: Read connect_retry_delay in get_session_client https://review.opendev.org/c/openstack/openstacksdk/+/1011928 | 19:04 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011929 | 19:04 |
| opendevreview | Merged openstack/openstacksdk stable/2026.2: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011900 | 19:32 |
| opendevreview | Merged openstack/openstacksdk stable/2026.1: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011902 | 20:15 |
| opendevreview | Merged openstack/openstacksdk master: zuul: Re-enable post-review queue https://review.opendev.org/c/openstack/openstacksdk/+/1006281 | 20:24 |
| opendevreview | melanie witt proposed openstack/openstacksdk master: accelerator: Set _max_microversion for Resources https://review.opendev.org/c/openstack/openstacksdk/+/1011952 | 20:51 |
| shermanm | I'm seeing zuul failures for backports to 2025.1 and 2025.2, but they seem to be for pre-existing things? | 21:01 |
| shermanm | I think the cause is this mypy fix not being backported to those, but not 100% sure https://opendev.org/openstack/openstacksdk/commit/bffe676997463b61954c67602e7a3a49cb775900 | 21:01 |
| opendevreview | Boris Bobrov proposed openstack/openstacksdk master: Fix find_domain to resolve from token on Forbidden https://review.opendev.org/c/openstack/openstacksdk/+/981605 | 22:07 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011917 | 22:43 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: Read connect_retry_delay in get_session_client https://review.opendev.org/c/openstack/openstacksdk/+/1011918 | 22:43 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011919 | 22:43 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: Fix mypy type check errors https://review.opendev.org/c/openstack/openstacksdk/+/1011960 | 22:43 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.2: [stable-only] pre-commit: Bump flake8-import-order https://review.opendev.org/c/openstack/openstacksdk/+/1011961 | 22:43 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: Allow configuring connect_retry_delay https://review.opendev.org/c/openstack/openstacksdk/+/1011927 | 22:56 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: Read connect_retry_delay in get_session_client https://review.opendev.org/c/openstack/openstacksdk/+/1011928 | 22:56 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: config: Pass status code retry options to adapter https://review.opendev.org/c/openstack/openstacksdk/+/1011929 | 22:56 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: Fix mypy type check errors https://review.opendev.org/c/openstack/openstacksdk/+/1011962 | 22:56 |
| opendevreview | Michael Sherman proposed openstack/openstacksdk stable/2025.1: [stable-only] pre-commit: Bump flake8-import-order https://review.opendev.org/c/openstack/openstacksdk/+/1011963 | 22:56 |
| opendevreview | Stephen Finucane proposed openstack/keystoneauth master: typing: Accept Adapters in discover https://review.opendev.org/c/openstack/keystoneauth/+/1011642 | 23:08 |
| opendevreview | Stephen Finucane proposed openstack/keystoneauth master: discover: Handle key-manager microversion discovery automatically https://review.opendev.org/c/openstack/keystoneauth/+/1011964 | 23:08 |
| opendevreview | Stephen Finucane proposed openstack/openstacksdk stable/2025.2: [Stable-only] Drop mypy testing https://review.opendev.org/c/openstack/openstacksdk/+/1011967 | 23:31 |
| opendevreview | Stephen Finucane proposed openstack/openstacksdk stable/2025.1: [Stable-only] Drop mypy testing https://review.opendev.org/c/openstack/openstacksdk/+/1011968 | 23:32 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!