Some comments: - 4 page 5: "It (TCP) SHOULD NOT be used only for zone transfers and as a fallback."
this SHOULD NOT is very hard to implement without dubious interpretations (i.e., the idea is right but the current wording could lead to unexpected/unwanted results). - 5 page 4: "both clients and servers SHOULD support connection reuse" this should not be applied to servers as they MUST use the same connection than queries are received (cf last statement of 4 page 5). IMHO this is mainly a wording issue. - 5 page 6: the last statement of RFC 5966 was removed, perhaps because it could be applied to the TCP fast open? - 5 page 5-6: a problem is not covered: what a server should do when a client closes the TCP connection before pending responses are sent. I.e., what to specify for the server side of lack of connection signaling? Note the new 6 page 5 makes this more likely. - 6 page 5: either the should for parallel processing has to be upper case or another word has to be used. Note the new 7 uses a RECOMMENDED key word for a statement including this. - 8 page 7: IMHO 8 (TCP Fast Open) should be dropped or moved. - 9 page 7: please move this section (summary of pros and cons) in an appendix as it is clearly not normative. - 11 page 9: I am repeating what I said at the mic: if we don't allow servers to close TCP connections with any timeout, including 0 seconds, we'll open servers supporting TCP (so all servers :-) to easy DoS attacks. Regards [email protected] _______________________________________________ DNSOP mailing list [email protected] https://www.ietf.org/mailman/listinfo/dnsop
