Okay so this is definitely an initial emotional reaction more than a strong Considered one,
but what's going on with this URL format?
<guid isPermalink="false">
at://did:plc:x4qyokjtdzgl7gmqhsw4ajqj/app.bsky.feed.post/3kh6noefhpq24
</guid>
@gaditb did: is an IRI format which specifies a resolver (plc?) and a resource key; at:// is a protocol which takes a DID and indicates that it should be accessed using the AT protocol and then that the path should be followed (side-note: IETF does not like the name at: but has no mechanism by which to disallow it)
so this is saying
Access [PLC DID] with AT and then access the [app.bsky.feed.post] collection and then access item [ID] within it
@Lady Yeah no that's the bizarre thing I mean.
The "app.bsky.feed.post" bit, based on its contents, sure looks like IT is the "where to go for the data you don't presently have".
Whereas a did:... is an opaque identifier. It davka DOESN'T point to any further info besides itself! What does "access [PLC DID]" even mean? Access it through WHAT? Where would there be anything interpreting to be an authority defining a "app.bsky.feed.post" collection?
@Lady (Oh right, IRI.)
Like how they are USING it is obvious. "contact the owners of the bsky.app URL and ask them for a post in a feed, specifically the feed of account did:plc:x4qyokjtdzgl7gmqhsw4ajqj , the item 3kh6noefhpq24 .
But that's... not the semantic meaning of what they put together into that IRI. What the heck is going on. What did they THINK they were doing.
@gaditb what they're semantically saying is "ask the account plc:… for this post they made on bsky.app"; it's to support account portability (if they ever get a PLC resolver which allows that??)
@Lady But what is the listener you are you asking. What would the PLC be resolving to.
@gaditb see for yourself: https://plc.directory/did:plc:x4qyokjtdzgl7gmqhsw4ajqj
it's a centralized resolver service (plc.directory) that tracks a bunch of metadata about the thing in question
@gaditb the at protocol then specifies how to turn that response plus the path into something meaningful
@gaditb (DID registry is here if you are curious <https://www.w3.org/TR/did-spec-registries/#did-methods>. i think this whole thing is very silly and reinventing problems we all agreed were a bad idea like ten years ago)
@Lady I don't know what you're talking about, "silly". This is great and good: https://xn--3n8h.amy.gy/
@gaditb DIDs have a separate registry for all of the DID types that they maintain themselves, so you have to check the registry and look up wtf PLC is and how to resolve it
literally they just reinvented URNs but with a registry they own themselves. you can't resolve them without knowing what they mean
@gaditb this is very fucked up, just imo, but that's what it is