@tchambers For sure. I'm taking notes as I build this for a blog post I'll write detailing all of the gotcha's. That post will be intended for developers that are nervous about the complexity of AP, to show them it's not actually very hard.
#activitypub The Podcast Index ActivityPub bridge now properly handles Mastodon "authorized fetch". If you had trouble following a podcast from an authorized fetch enabled instance, please cancel that follow and try again.
This may also resolve other sporadic follow issues from Sharkey/Misskey instances. Testing from people on those platforms would be welcome.
@tchambers@BeAware Ok, so this looks like enabling that requires request signing for everything instead of just POST requests. That would be where the issue is since I think the bridge doesn't sign requests for things like Actor lookups. I'll start there.
@tchambers I was able to follow from my test account, but when issuing the "latest" command to get the most recent episode I got a signing failure. Let me dig in tonight and see what's going on.
That would allow bringing in the properties needed for both podcast apps and Fediverse apps to treat them the same way.
An example would be the transcript uri. Where does that go? It doesn’t fit into a Document or any of its sub-types cleanly. There will be many such properties.
The existing types are probably fine for most of the subclasses we need. So, the transcript property can just reference an Attachment object or set of Attachments if multiple.
The Podcast object itself can subclass Object of course, and then perhaps things like Podroll can just be a collection of Podcast objects since each will contain a guid and feedUrl.
#activitypub It seems like we will need to define a Podcast ActivityStreams object type. I’ve been trying to see if we can just transmute a podcast into an Audio or Video object type. But the loss of fidelity is so high. I think a dedicated object type is going to be pretty necessary.
"The Times article says that the company exhausted supplies of useful data in 2021, and discussed transcribing YouTube videos, podcasts, and audiobooks after blowing through other resources."