Hi there,
this draft of linked identities (and the corresponding implementation in
openkeychain) is by me :) cheers Wiktor for bringing it up again!
URIs can simply be represented by plain User IDs and software can
easily detected such URIs if desired.
I'll try to describe some more why I went with a user attribute over
simple user ids.
Most importantly, URI attribute are intended for identities that are
encoded as URIs, not simply user-facing urls that might have some
unspecified association to the user. This means that a typical URI in
there won't be "clickable", which will be confusing at best (and a
security issue at worst) for the user.
For instance, this is one of the uri attributes on my key:
openpgpid+cookie:@https://twitter.com/Valodim/status/592675006880022529
A secondary thought is that, while the content explicitly isn't
restricted, overloading user ids with more stuff adds to the amount of
guesswork implementations have to do to interpret and display them in a
helpful way. For URI Attributes, if an implementation has no handler for
the schema, then not showing them at all is the reasonable thing to do.
On a related note, personally I would be happier if user ids were
name-addr not only by convention, and we made better use of the user
attribute namespace for stuff like telephone numbers or whatever other
types of identities some people want to stuff in their keys.
The need to implement UAT only adds more complexity for a questionable
purpose.
User attributes with unknown attribute id already have to be handled as
opaque packets, so it adds zero complexity to implementations that don't
want to handle them. If an implementation chokes on user attributes in
general, that's a bug in the implementation, not the spec.
Note that these image UAT were introduced due to marketing needs of
PGP or NAT and (iirc) only specified after they had been introduced in
their software.
While this is an interesting historical note, I don't believe it's a
good basis for a decision here. If avoiding use of uats for this reason
is in fact general consensus, they should be deprecated in rfc4880-bis.
If it's not, I don't see anything wrong with assigning new attribute
ids.
- V
_______________________________________________
openpgp mailing list
openpgp(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/openpgp