On Fri, 13 Mar 2015, Pieter Lexis wrote: Thanks for the review Pieter,
I'm very much on the "Yes, this is good"-side of things. 3.1: The MAY in the last sentence is much too weak. We can’t have interoperability without some stronger rules. Suggest moving this whole section into -usage or mentioning that these will be specified in a later document (-usage in this case).
I'm happy to add a much larger recommendation advise in the -usage document, but I did want to mention it in the record format document because it _does_ affect the lookup mechanism that implementors might need or want to support.
5.1: Singling out one RRTYPE to push DNS-COOKIES feels weird. I'm just mentioning it, but as this section is merely a suggestion, it is fine.
I would gladly generalise it, but this document is not allowed to update the core DNS protocol. But I would like implementors to look at this and possibly put that restriction in. I'm actually worried about this, especially since using old software with the generic record syntax would actually not have any such limitation implemented.
Appendix A: Two things: I suggest moving this to -usage and adding pseudocode examples. The latter mostly to encourage more implementations.
I thought about keeping it software agnostic, but in the end figured since gnupg has been around for over a decade, it is kind of similar to using openssl commands as example. But I have no problems making this software agnostic if the WG thinks that is more appropriate. Paul _______________________________________________ dane mailing list [email protected] https://www.ietf.org/mailman/listinfo/dane
