Maintenance item 0260 has 4 sub-issues described within. Discussions on 0260 concluded that ONLY the first two issues were addressed, and the remaining two issues should be re-submitted as a new maintenance request. This is the re-submission of the remaining issues. See https://www.802-1.org/items/377
The MatchRecoveryAlgorithm can be used by both the Base recovery function (7.4.3) and the Individual recovery function (7.5). The frerSeqRcvyIndividualRecovery defines which one applies. Behavior of MatchRecoveryAlgorithm differs and current text does not reflect those differences. The issues described here are related to the text describing an example scenario "If a Talker or a relay system fails in such a way as to repeatedly transmit packets with the same sequence_number subparameter".
Issue1: "If a Talker or a relay system fails in such a way as to repeatedly transmit packets with the same sequence_number subparameter (perhaps repeating exactly the same packet), those packets will continue to be discarded, at least until the sequence_number wraps around." This sentence is not correct. Wraps around of sequence_number does not really matter. The result depends on whether or not individual recovery is used. If Talker fails as described, then in case of individual recovery, all repeated packets are discarded. In case of NOT individual recovery, all except one repeated packets are discarded per RemainingTicks interval. If a relay system fails, then in case of individual recovery, all packets are discarded. In case of NOT individual recovery, if there is a correct member stream, then the failed member stream pollutes the correct member stream. Therefore, I would say that in case of MatchRecoveryAlgorithm is used as base recovery, it is highly recommended to use it together with individual recovery.
Issue2: Text refers to C.10 also containing incorrect statements. "The Sequence recovery functions (7.4.2) in the two systems receiving the Member Streams 1 and 2 will discard the repeated packets until the sequence_number subparameter on the good Member Stream 2 wraps around after 65 536 packets. Then, whichever packet 5 is received first will be relayed to the next stage. It could be the new, good, Member Stream 2's packet 5 or the old, bad, Member Stream 1's packet 5." Statement on packet 5 is wrong. What happens with packet 5 depends on which Sequence recovery function (7.4.2) is used on the member streams. In case of VectorRecoveryAlgorithm (7.4.3.4), the bad packet 5 is always forwarded (as it arrives before and accepted by the history window). In case of MatchRecoveryAlgorithm (7.4.3.5), the failed stream pollutes the correct stream (see issue1).