[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[AVT] misleading RFC 3611 statement on RTCP packet stacking



Some people are reading RFC 3611 as overriding RFC 3550 on RTCP packet stacking.

This is what Sec. 1.1 of RFC 3611 says:

"XR packets supplement the existing RTCP packets, and may be stacked with other RTCP packets to form compound RTCP packets [9, Section 6]."

Reference 9 is RFC 3550, which says:

"all RTCP packets MUST be sent in a compound packet of at least two individual packets"

That is, RFC 3550 says that RTCP packets MUST be stacked into a compound packet, and some people are reading RFC 3611 as weakening that mandate to say only that RTCP packets MAY be stacked into a compound packet (but also MAY NOT, at least in the case of the XR packet).

As one of the authors of RFC 3611, I can say that it was not my intention to override RFC 3550. Nor was it the intention of the two of my eight co-authors who I have discussed this with. Rather, the purpose of this sentence was to remind the reader that XR packets, like any RTCP packets, are designed to be stacked, and to helpfully point the reader to the RFC 3550 section that deals with RTCP packet stacking for further details.

To the best of my recollection, no discussion in AVT leading up to approval of RFC 3611 indicates that any group participants at the time believed that RFC 3611 would override RFC 3550 on RTCP packet stacking.

Still, I understand how people are reading RFC 3611 in this way. It does say "may be stacked", and though "may" is not capitalized, it is nonetheless an RFC 2119 keyword, meaning that the item is truly optional.

Co-authors Nick Duffield, Ramon Caceres, and myself would like to open discussion on how to address this situation. We start by suggesting that we file an RFC Errata report, indicating that the sentence should read instead:

“XR packets supplement the existing RTCP packets, and stack with other RTCP packets to form compound RTCP packets, as described in Sec. 6 of RFC 3550 [9].”

By dropping the "may", we drop any hint that this RFC overrides the RFC 3550 stacking mandate. And by adding "as described in Sec. 6 of RFC 3550" we make clear the informational nature of this sentence.

Timur Friedman