RE: mailto URIs

On Mon June 20 2005 13:24, Larry Masinter wrote:
> 
> How about the following tack?
> 
> Define the syntax for the address part of as 'mailto-addr',
> and give the syntax as 'uric*', i.e., a sequence of
> (possibly hex-encoded) bytes. 
> 
> Then, in the text, note that
> 
> (URIC-encoded) 'addr-spec' is reliable
> (URIC-encoded) 'mailbox-list' MAY be used, but is often
>    badly implemented

Not only syntax, but also semantics, ought to be specified.  In this
case, the syntax in un-URI-encoded form has to match RFC 2822 syntax
and the semantics ought to be specified as corresponding to an RFC
2822 To field body.  A random sequence of uric octets won't work (not
only because the semantics of a mailto URI are intimately tied to the
message format (RFCs 822, 2822) and mail transport (SMTP, IMAP, etc.)
protocols, but because of URI reserved octets, particularly those that
are separately reserved in separate parts of a URI).

> Then note that if future email standards allow UTF-8
> in some future version of RFC 2822, that the 'mailto-addr'
> could be used with (URIC-encoded) utf-8-addr-spec.

That would be highly speculative on several grounds:
1. backwards compatibility requirements makes it unlikely
2. BCP 18 requirements for language-tagging (satisfied by RFC 2047)
   make it highly improbable (except to the extent that utf-8 is
   a perfectly valid charset to use with RFC 2047 encoding; I assume
   that you meant raw, untagged utf-8)
  
> > Imagine the URL quoting problems to properly do 'mailbox'.
> > Really. I simply can't imagine that many people would do it correctly,
> > even after a few tries.
> 
> 
> URIC encoding is a simple algorithm, really.

Under RFC 2396 rules, where "reserved", "excluded", etc. characters
are explicitly and reasonably clearly defined.  Not so simple for
the RFC 3986 rules, unfortunately.

> It might not 
> be easy to do manually, but it's pretty simple to implement
> taking a full "mailbox-list" and URIC-encoding any reserved
> characters within it. And the reverse -- mapping from a URIC-encoded
> mailto: address to the "to:" line of an email message -- also
> seems pretty straightforward.

Yes, with the caveat above re. 3986 vs. 2396.

Received on Monday, 20 June 2005 18:43:42 UTC