On 23/09/2013 16:50, Peter Saint-Andre wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 9/23/13 9:16 AM, Alexey Melnikov wrote:
On 23/09/2013 15:45, Peter Saint-Andre wrote:
On 9/23/13 7:21 AM, Alexey Melnikov wrote:
On 21/09/2013 03:34, Peter Saint-Andre wrote:
To my chagrin, I just realized that this header field is in
fact registered:
http://www.iana.org/assignments/message-headers/message-headers.xhtml#prov-headers
It's been so long since I looked into this issue that I must have
forgotten.
However, I do plan to get an RFC published so that there is
stable documentation for the Jabber-ID header field.
Hi Peter, Some comments on -10:
Hi Alexey,
Hi Peter,
Thanks for the feedback. When I submitted this I-D to the ISE, I
said that you'd be a potential reviewer. It seems I was right.
:-)
Your ABNF in Section 2:
Jabber-ID = SP *WSP pathxmpp *WSP CRLF
And your example in at the end of 3.2 is:
Following the rules in [RFC5122] and the Jabber-ID header
field syntax, the resulting header field might be as shown
below (note that this representation includes folding white
space, which is allowed in accordance with the ABNF):
Jabber-ID: ji%C5%99i@%C4%8Dechy.example
I think this doesn't match your ABNF I think the ABNF should
be:
Jabber-ID = SP [FWS] pathxmpp [FWS] CRLF
Thanks for the correction. Do most of the header field
definitions allow folding white space?
Yes, but there are some inconsistencies in whether they are allowed
at the end of the value.
FWS at the end doesn't sound like a great idea.
Many header fields include "[CWFS]" (folding whitespace with
optional comments) instead of "[FWS]". For simplicity of parsing I
would avoid using these, unless needed. But if you choose to use
[CFWS], that would still be fine.
Personally I see no need to allow comments. Keep it simple.
So I suggest:
Jabber-ID = "Jabber-ID:" [SP] [FWS] pathxmpp CRLF
I even wonder about the FWS...
Strictly speaking not necessary. But I would actually do
Jabber-ID = "Jabber-ID:" [FWS] pathxmpp CRLF
because SP is allowed by FWS and FWS is more generic anyway.
