On Thu, 2006-03-02 at 13:14 -0500, Glenn Gobeli wrote: > I realize, and have seen mailings to same, that SIP (RFC 3261) is quite > free in allowing white spaces to appear in headers on either side of a > colon (e.g., "Subject : lunch"). I have also seen mailings > related to LWS.
As far as I know, this freedom is always explicitly described in the BNF of the standard in question. See, for example, the use of HCOLON in section 25.1 of RFC 3261. > My specific concern is with SDP (RFC 2327) coding of the fields > containing a colon and white spaces around same. Section 6, paragraph 2, > makes a point of clearly stating that "Whitespace is not permitted > either side of the `=' sign." No such statement is made concerning a > colon appearing later in the field. For example, > b=<modifier>:<bandwidth-value> > k=<method>:<encryption key> > a=<attribute>:<value> The BNF in RFC 3227 Appendix A seems to be very specific where whitespace may appear. The only place where there seems to be some freedom is in the value component of an "a=" line. > In addition, the RFC states that any > attribute not understood should be ignored, so I believe that clearing > the session is inappropriate either way; the observed behavior of a > terminal adaptor. If by "atribute" they mean "a=" line (as is suggested by the BNF productions attribute = (att-field ":" att-value) | att-field att-field = 1*(alpha-numeric) att-value = byte-string then the statement "any attribute not understood should be ignored" suggests that extraneous whitespace in other lines could be reason to declare SDP to be unparsable. OTOH, common sense indicates that terminating an existing session is inappropriate even if the SDP is garbage (assuming a session is already in progress). Instead, the re-INVITE should be rejected with (say) 488 (Unacceptable session description). Similarly, an initial INVITE should be rejected with a suitable error code. Dale --- interop.pingtel.com -- the public SIP phone interoperability test server _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
