> > Maybe conditionally mandatory? If the router can be used for routing > > That's what SHOULD is *for* > If you are concerned it won't get implemented, then any weasel room you > leave will be exploited. At that point, the market gets to decide.
<on vacation, so not responding quickly and not in Yokohama> No, SHOULD and conditionally mandatory are 2 very different types of requirements. When designing a test tool to automate testing of requirements, a tool generally issues a warning when a "SHOULD" is not implemented (correctly). But for conditional mandatory, a tool will test for the condition and then issue a failure if the requirement is not implemented (correctly). But I agree that in IETF, I haven't seen use of conditional mandatory. I think that's unfortunate and I think this is the reason many IETF "SHOULD" requirements aren't implemented when they need to be implemented. But I'm not willing to spend more time arguing in this particular case. CE router vendors are a tremendously pragmatic bunch. If it's easy and could cause dissatisfaction with their product if not implemented, they will more than likely do it. This looks pretty easy to me, especially with the availability of a lot of open source code for it. Barbara _______________________________________________ homenet mailing list homenet@ietf.org https://www.ietf.org/mailman/listinfo/homenet
