ManMachine
@max@manmachine.me
support for DNS-over-TCP has been explicitly necessary since 2010
it's irritating that we still have to keep explaining this https://lobste.rs/c/hatmxu
@fanf so, RFC 5966 said SHOULD, RFC 7766 turned it into MUST, and they still had to publish an even _more_ emphatic RFC 9210?
"The key words 'SHOULD', 'MUST', and 'GOOD GRIEF WHAT IS WRONG WITH YOU HOW CAN WE MAKE THIS ANY CLEARER' are to be interpreted …"
@simontatham there's a slight subtlety that rfc 7766 is about implementations and rfc 9210 is about deployments, so rfc 9210 is more like HEY OPERATORS, THIS MEANS YOU TOO, FIX YOUR FIREWALLS
it has a very banging-my-head-against-the-wall review of the painful and somewhat erratic history, pointing out that there are lots and lots of other rfcs that depend on dns-over-tcp, and it was in practice required before 2010, just not clearly written down as such
To whom it may concern: I am currently experimenting with Snac, @max is also me, if you get a follow from it, be not afraid.
History