Re: SDP connection attribute optional [was: DTLS-SDP: TLS support added]
Roman Shpount <[email protected]>
| Newsgroups | gmane.ietf.mmusic |
|---|---|
| Message-ID | <CAD5OKxutkKx_JVfs+RMyz-vojL+tHyebCqHVO1k9A6ndnvWjpA@mail.gmail.com> |
On Mon, Apr 10, 2017 at 3:40 PM, Christer Holmberg < [email protected]> wrote: > I suggest to add something like the following, to make sure it’s clear: > > > > *“An offerer and answerer MUST use the SDP 'connection' attribute even if > it* > > * is known that both support the SDP 'tls-id' attribute.* > > * NOTE: As defined in [RFC4145], if the SDP 'connection' attribute is not > explicitly* > > * present, the implicit default value is 'new'.”* > If third party call control is used it is not known that the same end points are communicating. Because of this it is never quite known what is supported by the remote end point when offer is generated. Because of this I would prefer this language to say that "Offerers should not make assumptions about the support of SDP 'tls-id' attribute and MUST always use SDP 'connection' attribute. To avoid ambiguity, answerers MUST always use SDP 'connection' attribute as well. NOTE: As defined in [RFC4145], if the SDP 'connection' attribute is not explicitly present, the implicit default value is 'new'." Regards, _____________ Roman Shpount _______________________________________________ mmusic mailing list [email protected] https://www.ietf.org/mailman/listinfo/mmusic