Replies: 2 comments
-
Such a "config-specific metadata" could all be specified in a {
...
"@type": "sc:Dataset",
"name": "dataset_name",
"croissant": {
...
}
} What I think could be useful in this
At some point I think it also makes sense to keep a version of the config itself and a changelog. However maybe such information would be better fitted outside of the config itself (CHANGELOG.md + version file in github?). Please let me know what you think. |
Beta Was this translation helpful? Give feedback.
-
+1 separately storing croissant-related info in a metadata dictionary would be much cleaner, and keeping it in the config itself is the solution that at this stage makes more sense to me! Following #91 (comment), the croissant source could also be a field in the croissant metadata dictionary. |
Beta Was this translation helpful? Give feedback.
-
A Croissant config describes a dataset, so fields like
description
are about the dataset.However there are no fields to describe the config itself.
Topic came up in #45 (comment): the PR introduced a config and a README.md file next to the config to precise that the config was not describing the dataset fully. A comment was rightfully made that the warning could go in the description field of the config, as not to scatter the information.
The problem with that approach is that it mixes data about the dataset and the config itself, and so I thought it might be handy to have metadata fields describing the config itself, such as for example the status (draft, ready for consumption, config version?), or just general meta descriptions, todos, warnings, etc.
What is the sentiment around such meta metadata?
Beta Was this translation helpful? Give feedback.
All reactions