Making the case for smart cards for the average user

[Daniel Kahn Gillmor]

On Tue 2015-04-07 08:39:57 -0400, MFPA wrote:
> I was talking about what happens when the angle brackets are not
> there.
>
> If I generate a key with the UID of:-
>
>    Test20150407 user at example.com
>
> and try to encrypt an email to user at example.com it fails.

The above is neither an RFC 5322 addr-spec nor an RFC 5322 name-addr.
That is, it would not be considered acceptable in the To: line of an
e-mail header:

  https://tools.ietf.org/html/rfc5322#section-3.4

We could invent arbitrary ways to structure a User ID that includes an
e-mail address, but writing code to extract the e-mail address from
these things seems like a lot of heuristics at best, and there are all
kinds of ways that it could fail.

We know how to structure a proper name-addr and an addr-spec, and it's
not difficult.  If you want an e-mail address to be recognizable to
automated tools, you should structure it in a recognizable way.

The above UID is simply a mistake, and i don't think GnuPG should try to
accomodate it.

    --dkg