| *** __ministry is now known as Guest8014 | 01:35 | |
| opendevreview | Artem Goncharov proposed openstack/codegenerator master: Fix identity resource link code https://review.opendev.org/c/openstack/codegenerator/+/933765 | 09:31 |
|---|---|---|
| frickler | stephenfin: gtema: there have been multiple sdk job failures now on the reqs bump to latest sdk, so I'm worried that there is an actual regression there, maybe one of you can take a closer look? https://review.opendev.org/c/openstack/requirements/+/932687 | 11:23 |
| frickler | hmm, interestingly no failure in the combined u-c bump, so maybe only bad luck after all https://review.opendev.org/c/openstack/requirements/+/924547 | 11:32 |
| gtema | frickler: ack, looking | 11:55 |
| stephenfin | adamcarthur5: If you could get 928919 passing, that would be super. Alternatively, if you can think of a better approach then please feel free to explore that | 13:38 |
| stephenfin | Virtually all Python web framework appear to provide a routing mechanism and the concept of controller for API endpoints/resources | 13:40 |
| stephenfin | gtema can obviously explain codegenerator better than I can, but both that and the schema validation tests I'm adding to various projects rely on iterating through the routes on a router to find the controllers and ultimately the methods the handle a given HTTP verb + path | 13:41 |
| gtema | correct. For Ironic I can easily access all the routes (http verb + path) and only need to get supported request body, query parameters, information about microversion (either one function that supports different bodies per microversion or one function per microversion) and response description | 13:43 |
| gtema | things like https://opendev.org/openstack/ironic/src/branch/master/ironic/api/controllers/v1/chassis.py#L46 or https://opendev.org/openstack/ironic/src/branch/master/ironic/api/controllers/v1/runbook.py#L72 make it incredibly hard to access the schema and params individually | 13:47 |
| opendevreview | Merged openstack/ansible-collections-openstack master: Enable glance-direct interop image import https://review.opendev.org/c/openstack/ansible-collections-openstack/+/932372 | 13:51 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-collections-openstack master: Remove default value for interface https://review.opendev.org/c/openstack/ansible-collections-openstack/+/889729 | 15:04 |
| opendevreview | Merged openstack/ansible-collections-openstack master: Allow wait: false when auto_ip is false https://review.opendev.org/c/openstack/ansible-collections-openstack/+/905308 | 15:53 |
| adamcarthur5 | Yeah stephenfin I can get those tests passing. And I'm with gtema, the way Ironic does micro versions makes it tricky to do anything "smarter" than what the current plan is | 16:07 |
| adamcarthur5 | It's also a plan that the Ironic folks were happy with at the PTG, so let's just stay the course. | 16:08 |
| gtema | adamcarthur5 - what about adding decorators that do not influence ironic but expose the data for the openapi generator? | 16:08 |
| adamcarthur5 | Hmm.. decorators that are there purely for the generator? Or that actually do something in the code? | 16:21 |
| adamcarthur5 | If I am not mistaken, stephenfins commit is using the ironic validation decorator already, and my plan was to follow suit? | 16:23 |
| gtema | yes sure, I mean if we can't enforce changing way of MV handling in ironic it is still more useful to have decorators only for openapi rather then producing wrong specs (or not producing at all) | 16:24 |
| gtema | very initially we were talking also about validation decorators with possibility to switch off validation. This is surely not the most ideal solution, but if there is no other way - why not | 16:25 |
| JayF | I thought this trick might be useful to sdk cores; I looked into it after pushing a oslo.utils change that failed lint: https://jay.jvf.cc/posts/openstack-pre-commit/ tl;dr you can setup the pre-commit hooks to automatically go into new repos | 16:39 |
| gtema | JayF - we already use pre-commit in all repos (maybe missed some of the rarely used) | 16:50 |
| JayF | Yeah; this is about not having to configure the local hook everywhere | 16:50 |
| gtema | ah ok | 16:50 |
| JayF | so people like me who work across lots of repos won't forget to run `pre-commit install` | 16:51 |
| gtema | indeed | 16:51 |
| gtema | thks | 16:51 |
| adamcarthur5 | Hmm. I feel like having decorators that are not enforced just adds even more change for error? Unless I am misunderstanding something | 17:16 |
| adamcarthur5 | I think for us in Ironic, that is the biggest thing we want to fix. We want as much validation as we can get. If we find a particularly hard API to model through validation, I would consider having a validation-off decorator | 17:18 |
| JayF | Eh, some of the value is forcing us to identify the changes between microversions more easily. I'd be sad if we had to punt any cases in Ironic. | 17:19 |
| adamcarthur5 | +1 | 17:23 |
| gtema | sounds great for me. If you want to have validation in the form that can be introspected - I am only supporting that | 17:57 |
| JayF | Yeah I told adam I basically had two goals for this: 1) the actual schema/docs generation but also 2) having ironic microversions look more like other projects, and no longer be a mess of `if x:` | 17:59 |
| tom | hello sirs | 17:59 |
| tom | can you help me understand whi can't i open oftc.net from my ipv6 | 18:00 |
| tom | checked my ip on https://mioip.info and it's ok | 18:00 |
| tom | also on ifconfig.io | 18:00 |
| adamcarthur5 | stephenfin looks like a recheck is all that was needed: https://review.opendev.org/c/openstack/ironic/+/928919 | 21:01 |
| adamcarthur5 | JayF and me suspect you got caught in the CI wave for latest openstack release. | 21:01 |
| adamcarthur5 | I am going to quickly test locally and then we can get this one merged, and I will move on to the next one :)) | 21:01 |
| JayF | adamcarthur5: I think you meant to say you tirelessly worked to resolve CI issues so his job would pass /s ;) | 21:01 |
| adamcarthur5 | Don't worry JayF, I am spending all that extra time I saved re-writing Ironic into Rust :) | 21:11 |
| JayF | nice, we'll finally be Secure(tm) | 21:13 |
| gtema | Welcome to the Rust gang adamcarthur5 ;-) Hopefully we can finally convince TC majority to approve Rust | 21:25 |
| JayF | Honestly don't hate it as an additional idea, just nobody get crazy about rewriting things :D | 21:27 |
| JayF | I only volunteered to help write Ironic once ;) | 21:27 |
| gtema | I doubt people are crazy to simplify rewrite openstack at once. But we shouldn't block introduction where it makes sense | 21:28 |
| JayF | fwiw, I believe you can use rust in any openstack project now at your option; we made the point that the platform we publish is not meant to exclude other stuff, just cover things that have to be handled | 21:29 |
| JayF | In any event, do good things and let the policy follow. Most fear of approval of /anything/ in openstack is that we don't have enough people to do the work. | 21:30 |
| gtema | Well, unless rust is accepted as language I don't see how we can achieve that | 21:30 |
| adamcarthur5 | > Well, unless rust is accepted as language I don't see how we can achieve that | 21:53 |
| adamcarthur5 | What do you mean by accepted here (sorry, still adjusting to IRC, ignore my quote attempt) | 21:53 |
| JayF | I assume the fact we don't have a rust version specified here: https://governance.openstack.org/tc/reference/runtimes/2025.1.html -- but I don't agree that means it can't be used. | 21:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!