Don’t think there’s many fedi clients that connect over plain http and follow redirects, but it might be worth looking at fixing at the reverse proxy (e.g. nginx) level
Every object has an object identifier and I assume that the majority of #activitypub developers assume that an update to an object returns the same object identifier as the location and that the properties of this object have changed as a result of the update.
However, the update changes an object and it should also receive a new id!
I'm noticing that #ActivityPub Activities contain an id, but I am not certain whether it is a requirement. The AS spec notes that all objects MUST have an id provided, but I'm not 100% sure about activities...
@devnull there can be "transient" objects according to the specification[1], but personally I don't see why short lived objects/activities wouldn't benefit from having IDs.
I think activities could be imagined as transient but they would leave no traces in anyone's outboxes/inboxes. The only sign they were effected would be their side-effects ? But I have no mental model for this at the moment. :D
@evan@context
aaaahr, I let myself get confused.
If an object has properties in an object without an id, then these are transient according to the specification:
...unless they are intentionally transient.
In the case of the actor, this would mean that oauthAuthorizationEndpoint is transient. which is wrong
>Publicly dereferencable URIs, such as HTTPS URIs, with their authority belonging to that of their originating server. (Publicly facing content SHOULD use HTTPS URIs).
-- SHOULD use HTTPS URIs, but not MUST
This means one could design a protocol where IDs are not location-based and it might still be technically called ActivityPub. Some Fediverse servers may even support this protocol with little effort.
I am not sure what is correct to put in kbin-core/issues without cluttering it up with somewhat speculative requests. I have no idea how to implement this or whether it is possible. So I will post here?...
It's true! I actually took a wee peek into the ActivityPub standard earlier and saw that the ID thing is already there like you say. Which I guess is the URLs as posted above?
So that part is done & dusted I guess. :) The part that would remain now would be to make a convention to transform markdown. Markdown is already being transformed.
Does the threadiverse need to maintain perfect compatibility with the rest of the activity pub fediverse? For me, just kbin and lemmy being roughly on the same page would be great. But maybe my viewpoint is too narrow and self centred.
per the activitypub spec on what to do when receiving objects:
> it should dereference the id both to ensure that it exists and is a valid object, and that it is not misrepresenting the object. (In this example, Mallory could be spoofing an object allegedly posted by Alice).
this kinda defeats the purpose of sending the whole object and not just the id, doesn't it?
universal IDs for threadiverse items (re kbin-core #635)
I am not sure what is correct to put in kbin-core/issues without cluttering it up with somewhat speculative requests. I have no idea how to implement this or whether it is possible. So I will post here?...
Best of Wednesday's NFL minicamps: OBJ, Lamar Jackson getting into rhythm (www.espn.com)
We are tracking everything from including rookies, team debuts, notable absences and more at 21 camps.