CEP request: OCI packaging
- Dominant language
- No language data
- Stars
- 34
- Forks
- 41
- Avg merge
- 1h 46m
- Merged PRs (30d)
- 1
Description
https://github.com/orgs/channel-mirrors/packages has a mostly complete mirror of conda-forge and bioconda in GHCR.io. Each anaconda.org artifact (be it tar.bz2 .conda), has been pushed there under the following conventions:
* Name of the "container": `//`
* Label of the container: `-`
For each name+label combo, we have the following manifest of layers:
```json
{
"digest": "sha256:",
"mediaType": "application/vnd.oci.image.manifest.v1+json",
"size": ,
"config": {
"digest": "sha256:",
"mediaType": "application/vnd.unknown.config.v1+json",
"size":
},
"layers": [
{
"digest": "sha256:",
"mediaType": "application/vnd.oci.image.layer.v1.tar",
"size":
},
{
"digest": "sha256:",
"mediaType": "application/vnd.conda.package.v2",
"size":
},
{
"digest": "sha256:",
"mediaType": "application/vnd.conda.info.v1.tar+gzip",
"size":
},
{
"digest": "sha256:",
"mediaType": "application/vnd.conda.info.index.v1+json",
"size":
}
]
}
```
I *think* the tar and conda layers correspond to tar.bz2 and .conda ones. Always present, but only one is populated. The info layers correspond to the compressed `info/` contents and the `info/index.json` file, respectively.
This should be standardized in a CEP.
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.