When adding a User ID to your OpenPGP key, you are typically asked (for example by GnuPG) to provide a name, a comment, and an email address (where each part is optional).
This format (Name (Comment) <email address>
) is described by RFC 4880, section 5.11 (User ID Packet):
By convention, it includes an RFC 2822 [RFC2822] mail name-addr, but there are no restrictions on its content.
While it says it’s only a "convention" and there are "no restrictions on its content", the first sentence of this sections reads:
A User ID packet consists of UTF-8 text that is intended to represent the name and email address of the key holder.
I assume (please correct me if I’m wrong) that "intended to represent the name and email address" is not normative. So that would mean:
- I may specify any kind of content (it does not have to represent a name and/or an email address).
- If I specify a name and an email address, I don’t have to follow the address format described in RFC 2822.
Is this correct?
If so, I could specify any kind of URI (using angle brackets to delimit it). As the URI scheme would be included (contrary to the "convention" format), it could, for example, prevent people mistaking an XMPP address for an email address.
Examples of possible User IDs:
<mailto:[email protected]>
(my email address)<xmpp:[email protected]>
(my JID)<tel:+1-201-555-0123>
(my telephone number)<acct:[email protected]>
(my Security SE account)<https://example.org/>
(my website)
Is this sensible, maybe even recommended? Will it break something?
I especially wonder if tools (like keyservers, mail clients or key chains) will still find/associate a key with an email address if the full URI is specified (<mailto:[email protected]>
), and if they won’t interpret an XMPP address (<xmpp:[email protected]>
) as email address.