dokc / operator-feature-matrix

Opearator Feature Matrix - Features specification for Kubernetes data operators
https://dok.community
Apache License 2.0
25 stars 4 forks source link

The base name of the image is given as its short name instead of a complete name #11

Open edithturn opened 1 year ago

edithturn commented 1 year ago

The base name of the image is given as its short name in the spec https://github.com/dokc/operator-feature-matrix/blob/main/postgres/spec/feature_matrix.md.

Someone could set the container registry to something other than the docker hub, so if they do an unqualified search, they could wind up with the wrong image. Could it be better to write the full name? e.g., use “us-docker.pkg.dev/my-project/my-repo” instead “debian” or “docker.io/library/fedora” instead of "fedora"

ahachete commented 1 year ago

Are you referring to the feature genc/coios (Container images OSes)?

If so, take into account that this is the feature description and it's intended for end-users to understand what principles is the operator based on, in this particular case which is the base image used for constructing the container images.

This value is never intended to be something to use for deployment scripts or anything like that, it's just meant to be informative for the user.

jzb commented 1 year ago

@ahachete Thanks for the clarification. Does that assume that there's a canonical source of all base name images that users can refer to for the basename? e.g., that "fedora" is a universal base image name that end users can assume is the same image?

(I think that's probably true in practice today but I'm not sure if it will remain true.)

ahachete commented 1 year ago

I think at this point it may be hard for us to determine all the possible scenarios, and while it may be a nice effort to try to normalize names and provide a limited set of values (an enum) as of today I believe it's more practical to leave it more open as a string.

Once we have enough submissions we can re-evaluate and see if we "know" all the possible values and switch to an enum at a later revision of the OFM spec.

Let me know if this is good for you and we can proceed to close the issue or there's something else you'd like to address now for this issue. Thanks!

jzb commented 1 year ago

Sorry I don't think I ever got a notification of reply. We can close this. Thanks!