aye- in my “implementation branch” for the ZYRE pr- i made sure to implement that way (eg: if you set the key, it starts ignoring v2 beacons).
for the purposes of “the protocol” i kinda defaulted towards SHOULD because i don’t pretend to know how it might break other stuff. your initial reaction was similar to mine- it should be at-least SHALL if not MUST [heh, say that 3x fast..]. very open to MUST- just tend to leave such strict language for things that are important and maybe should be considered at the last mile (eg: if someone was to implement it differently outside of the C lib..). i’m almost indifferent since i implemented it as MUST- if only cause it also makes configuration much simpler.. but defaulted to SHOULD/SHALL to solicit advice from others that may have done this already and found ways MUST may break things. > On Aug 17, 2017, at 9:32 AM, Luca Boccassi <[email protected]> wrote: > > The only comment I have, as I've written on the PR, is whether the > requirement to ignore v2 packets when CURVE is enabled should be > upgraded from SHOULD to MUST, given the security implications. -- wes wesyoung.me
signature.asc
Description: Message signed with OpenPGP
_______________________________________________ zeromq-dev mailing list [email protected] https://lists.zeromq.org/mailman/listinfo/zeromq-dev
