After taking another look at how the feature is written, I agree it needs a rework. Not exactly sure where it needs to end up. The current format is also lacking the ability to define something like MultiKey did:
This would lead to some level of type safety being enforced through json-ld. This is certainly not useful for something that varies a lot, e.g. a Note, but I think is quite appropriate for cryptographic material.
I think these are both really good points. I think the current version of FEP-2e40 is too over engineered. I completely ignored that having other processes to create vocabularies might be a possibility when writing it.
Iāve reworked the file format a bit, the current result can be seen at
In order to test it, I added the terms from the security vocabulary commonly used to it, see here and can say that the tests work to find mistakes. At least the mistakes, I tend to make.
Iāve now updated the index page. Most things about references now work as intended:
From a technical standpoint: The examples as shown are incompatible with Mastodon. One would need to use:
{
"@context": [
"https://purl.archive.org/fedi-2023/context.json",
"https://www.w3.org/ns/activitystreams",
],
"id": "https://example.com/note/1",
"type": "Note",
"content": "This is a note containing a #tag",
"tag": {
"type": "Hashtag",
"name": "#tag",
"href": "https://example.com/hashtags/tag"
}
}
if one wants Mastodon to understand the objects.
From a more philosophical stand point: Iām no longer sure if a centralized context is a good idea. It might be a better idea to use āFediverse Vocabularyā to just collect definitions of URIs used.
The entire thing needs thinking. Fortunately, the linked Fediverse Vocabulary provides an excellent base on how stuff might be build.