> -----Original Message----- > From: regext [mailto:regext-boun...@ietf.org] On Behalf Of Patrick Mevzek > Sent: Monday, December 11, 2017 11:34 PM > To: regext@ietf.org > Subject: [EXTERNAL] Re: [regext] FW: I-D Action: draft-hollenbeck-regext- > rdap-object-tag-05.txt > > Hi Scott, > > On Tue, Oct 24, 2017, at 14:04, Hollenbeck, Scott wrote: > > > A New Internet-Draft is available from the on-line Internet-Drafts > > > directories. > > > > > > > > > Title : Registration Data Access Protocol (RDAP) > Object > > > Tagging > > One comment on the draft itself. > While I think I can understand the need, I still feel very uneasy by the > solution of tackling two values together by a given separator. > In fact it shows even in the history of the document, where you had to > change the separator multiple times. > > Even beside the fact that tilde looks like a dash inferior lookalike, I > think that you would get problems whatever separator value is used. This > shows in many sentences of the text. > > Were other solutions already explored?
Thanks for the comments, Patrick. I changed the character more than once only because we were trying to find one that was guaranteed (more or less) to not be part of the values currently in use. At least one of my earlier proposals was, in fact, a collision. As noted in the text, the concept described in the document is already implemented by multiple registries and is based on current RIR practice. It's not new. No, I didn't consider other solutions because I thought the existing RIR practice would work just fine because it's working just fine right now. > Like one the following two: > - instead of adding the service provider to the current handle, why not > having a new RDAP attribute, like handle_provider to store only this > value? > - or, even more radical, having the current handle element not a string > anymore but a dictionary/map with one or two keys, like value (mandatory, > would be the current text in the element) and provider (optional). > > Obviously these 2 solutions involve schema changes so are more difficult > to put in place, but I see them are more future-proof. As you said, both are more radical and thus probably more than we need. I'm just trying to find a simple way to add structure to an identifier so that queries can be bootstrapped in the same way other RDAP queries can be bootstrapped. Scott _______________________________________________ regext mailing list regext@ietf.org https://www.ietf.org/mailman/listinfo/regext
