On 2010-08-25 06:26, Alain Durand wrote: > On Aug 24, 2010, at 2:11 PM, Hemant Singh (shemant) wrote: > >> >>> 2) Even if redirect could be useful in such a leaf network, it would >> not change my overall point. This is not an argument to mandate it >>> mandatory to implement for ALL networks. >> So are you asking for the text like this to change from MUST to SHOULD? >> >> >> "A router MUST implement Redirect functionality" >> >> Or are you OK with the MUST in the quoted sentence directly above and >> would like to see a MUST implement and SHOULD NOT enable by default? > > I'd like to see a SHOULD implement. I'd also like to see recommendation to > NOT enabling it for point to point link.
SHOULD, in IETF-speak, means "there may exist valid reasons in particular circumstances to ignore a particular item, but the full implications must be understood and carefully weighed before choosing a different course." It seems to me that in a requirements document, SHOULD needs to be accompanied with a brief description of possible valid reasons to ignore it. Such as Redirect functionality could reasonably be omitted if the implementation in question is limited to support only interfaces for point to point links or non-broadcast multiple access links. If you don't do that, we know that the SHOULD will definitely be misused by sloppy product managers. Brian -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
