On 20/09/2018 13:25, Stewart Bryant wrote: > The problem with the many proposed alternative versions of Master/Slave that I have seen over the years, is that they fail to express the technical importance of the absolute relationship between the two entities. > > The term master/slave is used when it is technically required that the instruction is executed without equivocation. Indeed in hardware-land, dithering over what to do (metastability) is so catastrophic that many technical measures need to be taken to avoid it. > > If all the master-slave flip-flops in the Internet were replaced with do-it-if-I-feel-like-it flip-flops, we would not have an Internet. > > In RFC-land we are mirroring the long-standing language of the hardware designers, and having a common terminology that transcends all aspects of logic design seems to me to be a net benefit to the internet as a whole. Yes, we always need to take context into account! I fully agree with with Stewart and Riccardo (previous reply) on this. -- Petr Špaček @ CZ.NIC
