artifacts
artifacts copied to clipboard
Releases Please
OCI specs appear to have releases (e.g., image spec and distribution spec). These are great for consumers of the specs because we know where they stand. If something is a feature addition, breaking change, and so forth. And, if someone implements a spec we know by the version when things break.
Can we get released versions of artifacts?
Without releases I can't tell its stability. Is this an experiment that might easily change or something that's stable with staying power?
It's not a spec.. it's going to be a running reference with guidance... The samples will define the dependencies, such as versions of the image and distribution spec.
Can we get released versions of artifacts?
I would expect each artifact type to have versioning, releases, ...
https://github.com/opencontainers/artifacts/blob/master/artifact-authors.md#defining-a-unique-artifact-type
Without releases I can't tell its stability. Is this an experiment that might easily change or something that's stable with staying power?
See here: https://github.com/opencontainers/tob/blob/master/proposals/artifacts.md#versioning--roadmap
We would have to go back to the tob and request a change in scope for the project to add a specification w/versioning.
Hopefully some "optional" tests can be added to distribution for particular artifact samples. Maybe in a subsequent release of images / distribution it (requirements) can be more formal.
Thoughts?
Without releases I can't tell its stability. Is this an experiment that might easily change or something that's stable with staying power?
I would like to better understand what stability here means. We are trying to figure out how these will be consumed by end users so we can figure out what the deliverable is. How do you expect to consume this repository?
I believe the questions are:
Q: if a group creates an artifact, (aka helm), with a specific config.mediaType
, how can they know they won't get broken in the future?
A: This was the delay in identifying the format of artifact mediaTypes. With that complete, and merged, that question should be closed.
Q: how does an artifact author claim ownership of a unique mediaType
?
A: this was the other delay in researching the IANA registration process. The current artifact-authors.md
describes: Registering Unique Types with IANA to resolve this concern. We will have a new updated version of Optional: Publishing the Artifact Type that will define the publishing process. However, this doesn't block an artifact author from defining their type, or owning their unique config.mediaType
. It just means there's no definition of a "clearing house" of types, yet. This is an opportunity to do better, and part of the TOB commitment, but not a blocker.
Q: It's great to have a merged document, but what stops someone from changing it? A: This is where the versioning question likely has the most relevance. There's nothing stopping someone from making a breaking change to a spec either, there's just a process. A specific published/versioned spec shouldn't take a breaking change. There's a process of defining a new version that defines the change in behavior. The artifacts repo does not have this process defined.
So, the question could be: Why not have a versioned artifacts spec, to formalize this commitment?