]> Oracle

rjsparks@nostrum.com

RFC3261 constrained several SIP header fields whose grammar contains the “name-addr / addr-spec” alternative to use name-addr when certain characters appear. Unfortunately it expressed the constraints with prose copied into each header field definition, and at least one header field was missed. Further, the constraint has not been copied into documents defining extension headers whose grammar contains the alternative. This document updates RFC3261 to state the constraint generically, and clarifies that the constraint applies to all SIP header fields where there is a choice between using name-addr or addr-spec. It also updates those extension SIP header fields that use the alternative to clarify that the constraint applies (RFCs 3261, 3515, 5002, 5318, 5360, and 5502).

defines several header fields that contain URIs to allow both a form that contains the bare URI (addr-spec) and one that provides a name and the URI (name-addr). This subset, taken from the ABNF specified in shows the relevant part of the definition of the syntax of the “From” header field:

The prose in section 20.20 of , which discusses the “From” header field, constrains how the production may be used by saying:

Section 20.39, which discusses the “To” header field contains no such constraining text. This constraint is specified slightly differently, but with the same intent, in the introduction to section 20:

). ]]>

Unfortunately, this can be read to only apply to the Contact, From, and To header fields, making it necessary to provide the constraint explicitly in the prose discussing any other header field using the name-addr or addr-spec alternative. As extension header fields were standardized, the specifications sometimes failed to include the constraint. Many errata have been entered to correct this omission. When the constraint was called out, the form has not been consistent. This memo updates the specifications of SIP and its extensions to clarify that the constraint to use the name-addr form applies anywhere there is a choice between the name-addr and addr-spec production rules in the grammar for SIP header fields. It is important to note that a message formed without honoring the constraint will still be syntactically valid, but would be interpreted differently. The characters after the comma, question mark, or semicolon would be interpreted as header field parameters or additional header field values as discussed in section 7.3.1 of .

This text from the introduction to section 20 of :

). ]]>

is replaced with:

, , , , and are updated to include:

This memo has no considerations for IANA.

The updates specified in this memo clarify a constraint on the grammar for producing SIP messages. It introduces no new security considerations. One pre-existing consideration is worth reiterating: messages produced without honoring the constraint will be mis-interpreted by the receiving element.

Brett Tate identified this issue in several extension documents, submitted several corresponding errata, and drove the discussion that led to this memo. Substantive comments leading to this text were provided by Paul Kyzivat, Gonzalo Camarillo, and Dale Worley.

Please remove this section in its entirety before publication as an RFC. Please replace any instances of RFCxxxx with the RFC number assigned to this memo. This memo, if it is approved, obviates Errata 3744, 3894, and 4648-4652 inclusive.

&RFC3261; &RFC5234; &RFC3515; &RFC5002; &RFC5318; &RFC5360; &RFC5502;