commented:
Heck yeah! I'll take a look at this as soon as possible and port it to
Go.
On a very quick glance I had, the identities were not really based on
ActivityPub actor URLs but on their webfinger double @ resource
identifiers. I'm not sure if it makes any difference to the underlying
data, but I think having the URLs as identifiers would be a better
decision, because it would drop the need for implementors to have yet
another thing to do.  (I'm saying that because WebFinger is not
actually linked to ActivityPub in any way, but it's indeed widely used
as a user discovery mechanism in the Fediverse)

  commented:

On a very quick glance I had, the identities were not really based on
ActivityPub actor URLs but on their webfinger double @ resource
identifiers.

I thought so too, but it looks like it auto-canonicalizes for you. See
this part of pkd-crypto.

  commented:
OK, cool. I still think that handling strictly URLs would be better,
people can transform the webfinger resource to a URL outside of the
library, but it's just a 0.X version, I'm sure these things come up as
the development advances and the library increases robustness.

  commented:
Comment removed by author
  commented:
I’m not sure if this was added after your comment but this is in the
post:

Why don’t you force the use of Actor URLs instead of double @
identifiers?
The data that is stored in the transparency log always contains the
canonical URL for an Actor.
The user-facing APIs (HTTP API, client software, etc.) permits strings
like @soatok@furry.engineer, but these strings will be canonicalized
for you (i.e., to https://furry.engineer/users/soatok for my handle).
This is a convenience feature and is only available so long as
WebFinger for the host instance is online. I wanted to make this
software as easy-to-use as possible while still being strict about
Actor identification under-the-hood.


  commented:
Yeah, I deleted my comment that said as much because Soatok told me on
Signal that he was adding it to the blog post, and it would have been
redundant.
.