Skip to content
This repository has been archived by the owner on Jul 18, 2023. It is now read-only.

Yes, we should delete this repository #63

Closed
jdolitsky opened this issue Feb 3, 2023 · 9 comments
Closed

Yes, we should delete this repository #63

jdolitsky opened this issue Feb 3, 2023 · 9 comments

Comments

@jdolitsky
Copy link
Member

jdolitsky commented Feb 3, 2023

Follow up to #46 and #61

This repo causes lots of confusion for people asking "What is an OCI Artifact". If we need better language in image-spec/distribution spec around this definition, then we should do that instead, and redirect this to that page.

The important behavior here has been codified into distribution-spec. Please see https://github.com/opencontainers/distribution-spec/blob/main/spec.md#pushing-manifests-with-subject

Append a descriptor for the pushed image or artifact manifest to the manifests in the referrers list. The value of the artifactType MUST be set in the descriptor to value of the artifactType in the artifact manifest, or the config descriptor mediaType in the image manifest. All annotations from the image or artifact manifest MUST be copied to this descriptor.

@imjasonh
Copy link
Member

imjasonh commented Feb 3, 2023

+1. I volunteered for an AI in October 2022 to identify useful information from this repo, mainly in Artifact Author Guidance, and move it into the image-spec repo, as non-spec guidance. I have totally failed to do that so far.

My findings, from reading through that whole repo:

  • Defining a Unique Artifact Type boils down to guidance about how to construct a unique and valid IANA media type, which I think we can just link to IANA guidance on.
  • Defining layer.mediaTypes boils down to linkable IANA guidance, with the addition of the +gzip convention that should probably go ...somewhere?
  • Optional: Defining Config Schema boils down to "write docs about how folks should consume your config, if you put data there" -- I don't know if anybody is effectively doing this today, so it might be premature to tell folks how to do it.
  • Visualizing Artifacts feels completely out of scope for OCI, IMO
  • There's also some semi-official-seeming list of known media types and links to where they're defined and used. I don't believe this approach scales to the level we want these manifests to be freely used and defined. The list will always be incomplete and folks will just argue over ordering and wording, so I'd rather OCI not try to list them all ourselves and do a bad job.

All in all I think most of that repo can turn into a paragraph or two in either image-spec (as SHOULDs), or in a short doc about best practices for artifact type creators. That guidance should be based on real world usage in the wild.

edit: we should remove whatever content makes it over into image-spec and archive the repo with pointers to image-spec; not delete the repo entirely.

@imjasonh
Copy link
Member

imjasonh commented Feb 4, 2023

  • Defining a Unique Artifact Type boils down to guidance about how to construct a unique and valid IANA media type, which I think we can just link to IANA guidance on.
  • Defining layer.mediaTypes boils down to linkable IANA guidance, with the addition of the +gzip convention that should probably go ...somewhere?

I've proposed these additions to the image manifest spec here: opencontainers/image-spec#1010 -- the +gzip guidance was already present there.

With this specified for image manifests I think we can remove any remaining guidance in the OCI artifacts repo and archive it, pointing to the image-spec repo's image manifest spec.

@imjasonh
Copy link
Member

opencontainers/image-spec#1010 was merged, and I propose we should proceed with archiving this repo, now that most of its useful information is now present in the image-spec repo.

@caniszczyk
Copy link
Collaborator

caniszczyk commented Mar 30, 2023 via email

@sudo-bmitch
Copy link

My own preference is to hold for opencontainers/image-spec#1043 which hopefully won't be long.

@imjasonh
Copy link
Member

I'm fine to wait for #1043, and then after that take it to the TOB to vote.

@jdolitsky
Copy link
Member Author

alright, lets do the vote

@jdolitsky
Copy link
Member Author

opened opencontainers/tob#126

@SteveLasker
Copy link
Contributor

Closing as this is a dupe of #61: The evolution of OCI Artifacts is being covered as a TOB Vote: opencontainers/tob#126

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants