Parse port from base url in ModelRegistryClient

[jonburdo]

What you would like to be added?

Support parsing port from base_url in ModelRegistryClient. This should be added to the internal client from the model-registry library as described in this issue: kubeflow/model-registry#2110

Then we will have it for free here for newer versions of model-registry (kubeflow[hub]). We should try to avoid doing the implementation in this repo as that would mean we doing url parsing in ModelRegistryClient and then again in ModelRegistry - this would be complicated and more bug-prone.

The criteria for this issue can be:

• handle model-registry version differences
  • preferably just update the minimum version of model-registry to a version that has this feature one the issue above is done
  • alternatively include warnings in code and docs about older version of model-registry missing this feature
• add examples to README/docs of ways to handle port

  ModelRegistryClient("https://example.org", port=456)  # port kwarg
  ModelRegistryClient("https://example.org:456")        # base_url
  ModelRegistryClient("https://example.org")            # default (`443` for https, `8080` for http)

• document in ModelRegistryClient.__init__ docstring
• add test cases for this

Why is this needed?

If I have model_registry_url = "https://example.org:456" (port in the url), then currently I have to separate this out to do:

ModelRegistryClient(base_url="https://example.org", port=456)

Ideally I would like to be able to simply do:

ModelRegistryClient(base_url=model_registry_url)

Love this feature?

Give it a 👍 We prioritize the features with most 👍

[github-actions]

🎉 Welcome to the Kubeflow SDK! 🎉

Thanks for opening your first issue! We're happy to have you as part of our community 🚀

Here's what happens next:

• Our team will review your issue soon! cc @kubeflow/kubeflow-sdk-team
• If you'd like to contribute to this issue, check out our Contributing Guide for repo-specific guidelines and the Kubeflow Contributor Guide for general community standards

Join the community:

• Slack: Join our #kubeflow-ml-experience and #kubeflow-trainer Slack channels
• Meetings: Attend the Kubeflow SDK and ML Experience bi-weekly meetings

Feel free to ask questions in the comments if you need any help or clarification!
Thanks again for contributing to Kubeflow! 🙏

[jonburdo]

Related discussion:

• feat: add model registry client #186 (comment)
• feat: add model registry client #186 (comment)

[osamaahmed17]

@jonburdo , I can have a look at this issue and would be happy to come with a solution.
/assign

[osamaahmed17]

Hi @jonburdo and @andreyvelich,
Can I start working on this now, or should I wait until the Model Registry PR is merged to main?

[andreyvelich]

@osamaahmed17 We might need to wait until we fix it in the Model Registry client upstream, but feel free to help with this issue first: kubeflow/model-registry#2110

[andreyvelich]

/triage accepted