Modeling spec: attributedTo type under 'Repository'

Specify the mailto alternative re-using the wording from the Commit model.

Also add an attributedTo field in the associated example.


(08:25:32) dachary: zPlus: would you be so kind as to review this (very) minor change to the spec ?

It turns out the review for this proposed change is not considered trivial. Since fr33domlover has been unresponsive this year it may take a while.

(08:36:28) zPlus: dachary: “if their actor URI is unknown, it MAY be their email address as a mailto URI” this has significant implications and is not a minor change. In theory, an URI can be anything so there’s nothing special about a mailto URI. An Actor is well defined in ActivityPub, it’s basically a resource where you can send messages to and get messages from. Actors’ URIs can in theory be a HTTP link, mail, P2P link,
(08:36:30) zPlus: whatever. The problem is a practical one, in the sense that an implementation needs to know how to message that specific URI. The ActivityPub spec is only focused on HTTP (GET, POST) and I think it’s wise to stick to it (for now at least). You should always have an actor before sending an Activity. You can also already mailto URIs if you want, but pretty much every other ActivityPub implementation won’t
(08:36:32) zPlus: understand them
(09:02:29) dachary: zPlus: understood, thanks for that background information.
(09:04:59) dachary: maybe I could borrow the rationale that led to this exact wording for the attributeTo property of the Commit Model? peers/forgefed: Main repository containing the forgefed spec and website. - Free code hosting . Or would it be enough to say that since the two properties are identical in nature they should be discribed in an identical way?
(09:19:30) zPlus: dachary: I think fr33domlover has added that bit on the Commit object… fr33domlover are you around to comment on the “MAY be their email address as a mailto URI” in the attributedTo property of Commits? dachary I think the rationale was that whe you look at a git repo history, Commits object can contain all sort of values for “committer”, for example an empty email string, or a bogus one. Repositories on the
(09:19:32) zPlus: other hand always have a user that has created it.
(09:22:17) dachary: zPlus: thanks for bringing fr33domlover into the conversation!
(09:25:05) zPlus: dachary: in general there is no reason to restrict URIs by their schema, so for example “http/mailto ok, everything else not ok”. I think the idea was that when you create a Commit object, there might not be an Actor for it because you don’t know who committed it, so you could use its email address instead. In the latter case however you cannot dereference the ActivityPub actor
(09:25:41) dachary: it makes sense

OT: I think you may need to reconsider the approach to ForgeFed. Are original creators still interested in maintenance? Shouldn’t you consider their work as a versioned Draft, and continue elsewhere with follow-up versions? And ensure along the way that that process is less likely to get stuck by inactivity of those involved? Just saying :slight_smile: (you might move to separate topic to discuss further)

1 Like

We are on the same train of thought it seems :slight_smile: The NLnet grant applicaition for the 1st October deadline was along these lines. I felt it would be a good match to work with zPlus on that. But it turned out the timing was a little too short and it did not happen.

1 Like