We have submitted a new version of the 7228bis proposal. There are two new things in there:
* Section 5.3 is a trial balloon for a nomenclature distinguishing wildly different classes of physical layer bitrates. This probably needs a lot of discussion. One alternative would be to just use a form of log10(bitrate), e.g. B0 for 1..10 bit/s, B1 for 10..100 etc. This is about burst rate (and thus direct serialization latency), but we also could talk about sustained bitrate, message rate etc. * Section 3 extends the now well-known device classes 1 and 2 upwards. This starts by introducing a more large-grained grouping (group "M" for microcontrollers, group "J" for general purpose CPUs) and then goes ahead defining class 10 to 19 for the latter. I'm not very happy with the term "general purpose CPU", but people seem to know what's meant (i.e., "can run Linux"). Also, the actual classes 10 to 19 are completely arbitrary trial balloons. Finally, we probably need a class 3, 4, 5 (and there even has been talk about subbdividing 1, 2). All these could be done in arbitrary ways. But this is not about choosing a color for the bike shed, this is about creating terms (and thus language) for intelligently talking about protocol (and system) designs. Feedback about how this best is accomplished would be appreciated. https://tools.ietf.org/rfcdiff?url2=draft-bormann-lwig-7228bis-01.txt Grüße, Carsten _______________________________________________ Lwip mailing list [email protected] https://www.ietf.org/mailman/listinfo/lwip
