|
| 1 | + * p 9 (bottom): |
| 2 | + In the unicast case, a participant may |
| 3 | + receive from all other participants in the session using the same |
| 4 | + pair of ports, or may use a distinct pair of ports for each. |
| 5 | + |
| 6 | + COMMENT: |
| 7 | + By default, a RTPSession instance will use only one pair |
| 8 | + of ports. |
| 9 | + |
| 10 | + * p 14 (top): |
| 11 | + A profile MAY define additional |
| 12 | + marker bits or specify that there is no marker bit by changing the |
| 13 | + number of bits in the payload type field (see Section 5.3). |
| 14 | + |
| 15 | + COMMENT: |
| 16 | + At the moment, no support for additional marker bits is present |
| 17 | + |
| 18 | + * About payload data and payload types: |
| 19 | + The library doesn't inspect payload data or payload types. It is |
| 20 | + left to the application using the library to decide if a packet should |
| 21 | + be ignored or accepted |
| 22 | + |
| 23 | + * p 18 (middle): |
| 24 | + o If a particular class of applications needs additional |
| 25 | + functionality independent of payload format, the profile under |
| 26 | + which those applications operate SHOULD define additional fixed |
| 27 | + fields to follow immediately after the SSRC field of the existing |
| 28 | + fixed header. Those applications will be able to quickly and |
| 29 | + directly access the additional fields while profile-independent |
| 30 | + monitors or recorders can still process the RTP packets by |
| 31 | + interpreting only the first twelve octets. |
| 32 | + |
| 33 | + COMMENT: |
| 34 | + No direct support for such extensions are provided |
| 35 | + |
| 36 | + * p 20 (bottom): |
| 37 | + 4. A fourth, OPTIONAL function is to convey minimal session control |
| 38 | + information, for example participant identification to be |
| 39 | + displayed in the user interface. |
| 40 | + |
| 41 | + COMMENT: SDES info is supported by the library |
| 42 | + |
| 43 | + * p 21 (top): |
| 44 | + The |
| 45 | + recommendations here accommodate SSM only through Section 6.2's |
| 46 | + option of turning off receivers' RTCP entirely. |
| 47 | + |
| 48 | + COMMENT: |
| 49 | + For now, the RTCP (sending) will only be disabled entirely |
| 50 | + if no destinations are present. |
| 51 | + |
| 52 | + * p 25 (middle): |
| 53 | + The profile MAY further specify that the control traffic bandwidth |
| 54 | + may be divided into two separate session parameters for those |
| 55 | + participants which are active data senders and those which are not; |
| 56 | + |
| 57 | + COMMENT: this will not be implemented |
| 58 | + |
| 59 | + * p 26 (top): |
| 60 | + This delay MAY be set to half the |
| 61 | + minimum interval to allow quicker notification that the new |
| 62 | + participant is present. |
| 63 | + |
| 64 | + COMMENT: is supplied as an option which is enabled by default |
| 65 | + |
| 66 | + * p 26 (middle): |
| 67 | + An implementation MAY scale the minimum RTCP interval to a smaller |
| 68 | + value inversely proportional to the session bandwidth parameter with |
| 69 | + the following limitations: |
| 70 | + |
| 71 | + COMMENT: not supported |
| 72 | + |
| 73 | + * p 28 (middle): |
| 74 | + SSRC sampling -> not supported |
| 75 | + |
| 76 | + * p 28 (bottom): |
| 77 | + An implementation which is constrained to two-party |
| 78 | + unicast operation SHOULD still use randomization of the RTCP |
| 79 | + transmission interval to avoid unintended synchronization of multiple |
| 80 | + instances operating in the same environment, but MAY omit the "timer |
| 81 | + reconsideration" and "reverse reconsideration" algorithms in Sections |
| 82 | + 6.3.3, 6.3.6 and 6.3.7. |
| 83 | + |
| 84 | + COMMENT: randomization and reconsideration will always be used. omitting |
| 85 | + them is not implemented. |
| 86 | + |
| 87 | + * p 28 (top): |
| 88 | + New entries MAY be considered |
| 89 | + not valid until multiple packets carrying the new SSRC have been |
| 90 | + received (see Appendix A.1), or until an SDES RTCP packet containing |
| 91 | + a CNAME for that SSRC has been received. |
| 92 | + |
| 93 | + COMMENT: Currently, probation is optional. A source is considered |
| 94 | + valid when either a number (depending on probation) of RTP |
| 95 | + packets are received or when a CNAME has been received |
| 96 | + |
| 97 | + * p 42 (bottom): |
| 98 | + 6.4.3 Extending the Sender and Receiver Reports |
| 99 | + |
| 100 | + COMMENT: Extensions of SR and RR are not supported |
| 101 | + |
| 102 | + * p 68 (middle): |
| 103 | + RTP relies on the underlying protocol(s) to provide demultiplexing of |
| 104 | + RTP data and RTCP control streams. For UDP and similar protocols, |
| 105 | + RTP SHOULD use an even destination port number and the corresponding |
| 106 | + RTCP stream SHOULD use the next higher (odd) destination port number. |
| 107 | + For applications that take a single port number as a parameter and |
| 108 | + derive the RTP and RTCP port pair from that number, if an odd number |
| 109 | + is supplied then the application SHOULD replace that number with the |
| 110 | + next lower (even) number to use as the base of the port pair. For |
| 111 | + applications in which the RTP and RTCP destination port numbers are |
| 112 | + specified via explicit, separate parameters (using a signaling |
| 113 | + protocol or other means), the application MAY disregard the |
| 114 | + restrictions that the port numbers be even/odd and consecutive |
| 115 | + although the use of an even/odd port pair is still encouraged. The |
| 116 | + RTP and RTCP port numbers MUST NOT be the same since RTP relies on |
| 117 | + the port numbers to demultiplex the RTP data and RTCP control |
| 118 | + streams. |
| 119 | + |
| 120 | + COMMENT: Currently, the transporter only accepts an even portbase. |
| 121 | + |
| 122 | + * p 68-69: |
| 123 | + It is RECOMMENDED that layered encoding applications (see Section |
| 124 | + 2.4) use a set of contiguous port numbers. |
| 125 | + |
| 126 | + COMMENT: No direct support for this using RTPSession, but it can be |
| 127 | + constructed in principle using several RTPTransmitter |
| 128 | + instances. |
| 129 | + |
| 130 | + * p 69: |
| 131 | + A profile MAY specify a framing method to be used even when RTP is |
| 132 | + carried in protocols that do provide framing in order to allow |
| 133 | + carrying several RTP packets in one lower-layer protocol data unit, |
| 134 | + such as a UDP packet. |
| 135 | + |
| 136 | + COMMENT: For UDP, such a framing mechanism will not be implemented |
| 137 | + |
| 138 | + |
| 139 | + |
| 140 | + |
| 141 | +QUESTIONS: |
| 142 | + * p 28: after BYE packet, ssrc should be removed after an appropriate delay. |
| 143 | + what delay is considered appropriate? |
| 144 | + * p 57: why do CNAMEs have to be forwarded to allow SSRC identifier collisions |
| 145 | + to work? |
| 146 | + -> If the CSRC sources are not senders, their identifiers wont be |
| 147 | + seen in the CSRC list of an RTP packet. By forwarding SDES CNAMEs, |
| 148 | + their identifiers will be present in the SDES chunk and collisions |
| 149 | + can be detected. |
| 150 | + * p 64: Section 8.3 |
| 151 | + * What MTU should be used? Should it be the same for all session members? |
| 152 | + Is it the MTU of the protocol just below RTP or of the link layer? |
| 153 | + * p 33 (BYE scheduling): Should members be increased for each BYE packet or |
| 154 | + for each SSRC in a BYE packet? |
| 155 | + |
| 156 | + * p 35: |
| 157 | + RTP receivers provide reception quality feedback using RTCP report |
| 158 | + packets which may take one of two forms depending upon whether or not |
| 159 | + the receiver is also a sender. The only difference between the |
| 160 | + sender report (SR) and receiver report (RR) forms, besides the packet |
| 161 | + type code, is that the sender report includes a 20-byte sender |
| 162 | + information section for use by active senders. The SR is issued if a |
| 163 | + site has sent any data packets during the interval since issuing the |
| 164 | + last report or the previous one, otherwise the RR is issued. |
| 165 | + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 166 | + -> Is this also true when the sender timeout multiplier differs |
| 167 | + from 2 ?? |
| 168 | + |
| 169 | + * About report blocks: should an entry be made for each SSRC_n in the |
| 170 | + list of report blocks if the RR or SR originates from a valid SSRC? |
| 171 | + -> probably not, since then we don't have the right transport |
| 172 | + address of the SSRC_n |
| 173 | + -> Indeed: see p 61 at bottom |
| 174 | + |
| 175 | + * Is it a valid tactic to use a timestamp unit estimation (from |
| 176 | + two SRs) for jitter calculation if no exact info is available? |
| 177 | + |
0 commit comments