The "joy" of #RDAP and especially its "notice" elements is to allow a registry to publish basically that they do NOT guarantee any accuracy and want to not be liable to any problems related to that, yet ICANN contract with registry at https://itp.cdn.icann.org/en/files/registry-agreements/base-registry-agreement-21-01-2024-en.html#specification10.4 clearly defines an #RDAP update time (between #EPP command and RDAP visibility) of "≤ 60 min, for at least 95% of the probes". So eventually consistent…

@pmevzek

when i taught DNS, i always emphasized that the DNS is *loosely* coherent. a few examples of how loose things could be usually both shocked the class and made it more clear. ;)

ICANN clearly wants things to be loosely or performatively coherent more than what most of us would consider *actually* coherent.

@paul_ipv6 My point was more about basically registries saying "no liability, content as is, no guarantees whatsoever", where in the background they have SLAs with ICANN, so there are some guarantees. But other than that, yes, nothing is real time, only illusions. "It is often the case that engineers refer to the DNS as a loosely coherent, distributed database." from https://blog.apnic.net/2021/03/11/some-dns-myths-and-misconceptions/ and mandatory #GeoffHuston at https://www.potaroo.net/presentations/2021-03-04-dns_deep_dive.pdf

@andy Yes :-) I love a love/hate relationship with RDAP. And Whois. And EPP. And HTTPS. And TLS. etc. too many to choose from 🙂

@pmevzek

yup. i alway said "the DNS is a widely distributed, loosely coherent database".

we do much better at widely distributed than we do at acknowledging loosely coherent. :) and sadly, we're just as casual with strong typing as with any other database. TXT. the blob field of the DNS.

@paul_ipv6 Considering databases went from char/varchar/text to xml and now to json datatypes, I expect at least that much from the DNS too, soon. Already planned (JSON type for content) for EDE messages. But I am also deeply enraged by TXT records that at their time were defined as character-strings when they are bytes and now everyone considers that means UTF-8… which was not even invented at time of RFCs 1034/1035.

@pmevzek

pretty sure cricket liu's first nutshell DNS book used that phrasing too.

for other folks reading this thread, beyond cricket, geoff's talks, bert hubert's DNS camel talks are a good cautionary tale also.

these days, i refer to DNS as the DNS SME full employment act.