podman-desktop-extension-ai-lab
podman-desktop-extension-ai-lab copied to clipboard
feat: adding container provider connection select
What does this PR do?
Allowing to select the connection the user want to use (if multiple exists) to deploy the Inference Server or Application.
Design is inspired from the Pull Image page of Podman Desktop
Screenshot / video of UI
| Single connection | Multiple started connections |
|---|---|
Demo
https://github.com/user-attachments/assets/916d0533-7f33-4270-97f6-e6461c742006
What issues does this PR fix or reference?
Fixes https://github.com/containers/podman-desktop-extension-ai-lab/issues/1462
[!CAUTION] Requires https://github.com/containers/podman-desktop-extension-ai-lab/pull/1535 for windows
How to test this PR?
- [x] unit tests has been added
How to test
- Start several podman machines
- Go to the
Create Servicepage - assert all started machines are visible
- Select the target machine and create
- assert
pdoman --connection=<machine-name> container lsis listed
I think, Model and Port should be first as they are the primary things user will need to interact with.
I don't understand this this could be hidden to the user
I don't understand this this could be hidden to the user
This is not hidden in the Podman Pulling page, why should it be hidden ? If the user configure a podman remote with more capabilities, it should be able to decide where the resources (recipe / inference server) should be deployed ?
If we only have one, like most cases we do not show this option, it only appear when multiple connections are available.
I don't understand this this could be hidden to the user
This is not hidden in the Podman Pulling page, why should it be hidden ? If the user configure a podman remote with more capabilities, it should be able to decide where the resources (recipe / inference server) should be deployed ?
If we only have one, like most cases we do not show this option, it only appear when multiple connections are available.
Remote would be a use case but I think we should focus on AI app dev features rather that low level technical infra. For example I believe selecting the proper connection based on footprint and load is adding more value
Rebasing after https://github.com/containers/podman-desktop-extension-ai-lab/pull/1535 has been merged