when should it send a feedback in O-mode

Hi qingbaibai,

> In ROHC 3095 5.7.6, "when decompression fails, feedback SHOULD be sent
> only when decompression of several consecutive packets has failed, and
> when this occurs, the feedback rate SHOULD be limited;", "A decompressor
> MAY limit the feedback rate by sending feedback only for one out of every k
> packets provoking the same (kind of) feedback.". Does it mean "k" should be
> consecutive?

I think so. The 2nd sentence uses the MAY verb: it is a proposal of implementation,
not a requirement nor a recommendation. The 1st sentence however uses the
SHOULD verb: it is a recommendation. The proposal shall follow the recommendation.

> In ROHC 3095 5.3.2.2.3, "In the Full Context state: When the CRC check of
> k_1 out of the last n_1 decompressed packets have failed, context damage
> SHOULD be assumed and a NACK SHOULD be sent in O- and R-mode." . Should
> the k_1 be consecutive? If k_1 is not consecutive, should it send a feedback?

I do not think so. If the authors wanted to mean 'consecutive k_1', they could
have said 'When the CRC check of the last k_1 consecutive decompressed packets'.
This is much simplier.

The above answers are my own opinion about the RFC. Not the authors' one. Ask your
questions on the IETF mailing list related to ROHC if you want to be completely sure!

Last but not least, the ROHC library is not RFC compliant yet on feedback rate limits.
See my answer in FAQ #187649 for details.

Regards,
Didier