Sure we could just link to the bencoded torrent file, but the point of this is to federate the contents of a torrent file, not just the infohash and magnet link. To make a direct Torrent object that can be extended with other terms and be made a general way of federating data.
The infohash can be a tag in a Link, when the torrent itself is just a Link. However this defines a Torrent object which has an infohash as one of its properties, as any other object might.
The magnet link can be made into a Link object, thats no prob, although I dont know what benefit that would have, semantic or functional.
Yes, I understand that Torrent object is supposed to be extendable, but why not add url property to it like PeerTube does? The benefit is that you can use standard properties instead of introducing new ones.
It's always better to use standard properties because that enables interoperability. For example adding content and url would make your Torrent compatible with my software without any changes on my side.
ok, if that is a standard way to represent urls i could do it that way. any prior art i can cite?
edit: if we are making it an object, is there a standard way to do mime/etc. type? having multiple types of related entities like CIDs and etc. would be nice to have too
@silverpill i guess one question is what the semantics of url would be compared to e.g. alsoKnownAs - like it looks like they are using url to mean “different versions of the same thing” with additional qualifications like mime type and extra metadata for different encodings, which traditional skos-like terms don’t really capture well. like technically all things in linked data are either a literal or a URI, so a url field is a bit ambiguous of a name, but i like the thing that peertube is doing with it and think it’s better than having a term for every possible kind of other representation like the current text of the FEP does with bencoded