On Mon October 25 2004 00:42, Martin Duerst wrote:
I tried to adapt the syntax below to the changes I made.
The problem with URIs/URI references is solved by RFC2396bis.
I don't think [FWS] and *WSP are needed, the syntax given for
similar headers in RFC 2822 doesn't explicitly mention these,
either.
Regards, Martin.
Actually, RFC 2822 does include [FWS] etc., but they are part of
the lower-level construct specifications, so aren't readily apparent
when looking only at the field definitions; they do become visible
if one expands "domain", "dot-atom", "date-time", etc. That
does not apply to the constructs used in the proposed field, since
the ABNF for those constructs does not explicitly include [FWS]
etc. Incidentally, the successor to RFC 2822 may change in that
respect; a proposed revised grammar has been discussed on the
ietf-822 list.
Given the draft ABNF
archived-at = "Archived-At:" '<' URI '>' CRLF ; URI not empty
and the example implicit in the text
As an example, the URI
"http://www.w3.org/mid/0I5U00G08DFGCR(_at_)mailsj-v1(_dot_)corp(_dot_)adobe(_dot_)com",
containing the message id
"0I5U00G08DFGCR(_at_)mailsj-v1(_dot_)corp(_dot_)adobe(_dot_)com",
is redirected to the URI of this message in the W3C mailing-list
archive at http://lists.w3.org/Archives/Public/uri/2004Oct/0017.html.
implies that
Archived-At:<http://lists.w3.org/Archives/Public/uri/2004Oct/0017.html>
is legal but
Archived-At: <http://lists.w3.org/Archives/Public/uri/2004Oct/0017.html>
^ N.B. SP here
Archived-At: (w3 list archive)
<http://lists.w3.org/Archives/Public/uri/2004Oct/0017.html>
etc. are not legal.
Concerning that example, please note that a message identifier (RFC 2822)
includes angle brackets; i.e. the message id is presumably
"<0I5U00G08DFGCR(_at_)mailsj-v1(_dot_)corp(_dot_)adobe(_dot_)com>". When using
mid
schemes or conversions which implicitly involve msg-id to mid or
mid-like constructs, it might be advisable to note the difference in
angle bracket inclusion and/or to refer to the mid RFC (2387) as a
normative or informative reference. The angle bracket issue has also
been recently discussed on the ietf-822 list.
Note also that RFC 2822 has provisions in the parse (obs-) syntax
for accommodation of whitespace between the field name and the colon
delimiter (which was allowed in earlier versions of the text message format).
You might wish to consider whether to provide for that for parsing,
particularly in the obs- rule for X-Archived-At.