zithro wrote: > On 13 Apr 2023 01:15, Dan Ritter wrote: > > zithro wrote: > > > On 12 Apr 2023 22:15, Greg Wooledge wrote: > > > > > > RFCs are there for having a common ground, right ? > > > > Sort of. > > > > At various meetings, a grad student was "volunteered" to take > > notes. Not quite certain of how accurately he had transcribed > > things, he typed up "Request For Comments" at the top and sent > > around copies. > > The student was Jon Postel, right ? (RIP dude, you did us good !)
Steve Crocker, who is still with us. Jon took the role of editor shortly thereafter. The story is told in RFC1000 https://www.rfc-editor.org/rfc/rfc1000.txt > > Since then, the IETF RFC Editor has established that some RFCs > > are for noting what people are doing, some are for making > > proposals, and some are "standards track" which are expected to > > have compatible implementations. > > I guess I wrongly think RFC are standards, whereas it's only a "work in > common" ? "Standards Track" RFCs start as proposals, and if they get consensus and multiple implementations -- especially interworking implementations -- you can treat them as standards. But there are almost always interpretations, options, and experimental extensions. If you want to see a really nailed-down standard, consider WireGuard. > But isn't the IETF "standardizing" protocols ? I mean it in ISO terms. > (playword by chance ^^). Interesting that you should bring up the polar opposite of the IETF in the 'standards' world. The ISO starts by appointing a group of experts, publishes a standard, and charges for access. The IETF brings together interested volunteers, publishes many revisions until a consensus is reached, and gives them away for free. > > > If I'm not wrong, RFC compliance may even be required in some areas (via > > > contracts). > > > > That might happen, but it wouldn't be a great idea from a legal > > standpoint: RFCs are often ambiguous in surprising ways. > > Oh, good to know. Do you have an example at hand ? Nearly everything about IMAP. In the most trivial possible conflict: try to figure out the difference in expected behavior, if any, between folders in the default namespace called Trash, Junk, and Spam? > So I just use the usual software (postfix/dovecot), and should be good to go > ? No fear of loosing mails, either sending or receiving ? Let's say that the postfix/dovecot duo are capable of doing all the standard things that people want a mail system to do, or at worst can be extended in reasonably documented ways. Somebody else on the Internet is always doing something stupid, something malicious, and something arguably correct but different. With postfix and dovecot, you'll have as good a chance as anyone else to get things right. -dsr-
