by Darius Kazemi, September 4 2019
RFC-247 is titled “Proferred Set of Standard Host Names”. It's authored by Peggy Karp of MITRE and dated October 12, 1971.
The technical content
This RFC is the first one to make me laugh out loud since the early days of RFCs. The document opens:
In RFC #226, BBN's TENEX list of Host names was set up as a strawman set of standard Host names. Comments received since then (an RFC actually generated comments!!!) have influenced me to propose the following general rules for forming Host names.
You'll recall that her RFC-226 garnered at least five responses in RFC form as well as one or two others that were written at least partially in response.
She takes feedback into account and proposes a different host site naming scheme. She gives an algorithm for choosing the name! She proposes a name be of the form
If a site name is shorter than 5 characters, use the full name. If not, use the standard acronym, if that is shorter than 5 characters. If not, use the standard abbreviation, if shorter than 5 characters. Failing that, use the first four letters (for example, MITRE is always MITRE, so she would suggest “MITR” for the site name). Failing that, the site itself should pick something appropriate shorter than 5 characters.
The machine is a designation for the computer being connected to and uses similar rules for picking the name.
The RFC ends with a potential 2-byte implementation where the first byte maps to a site from some kind of known table, and the same goes for the machines.
I'd argue that the word “proferred” in the title was very intentionally chosen to invoke Steve Crocker's seminal RFCs, RFC-54 and RFC-123. These RFCs were the foundations of protocols that went on to define the early ARPANET and I think this was a smart rhetorical move by Karp in response to the rather intense response she got to her RFC-226. I believe her making light with her “
an RFC actually generated comments!!!” is another classic move used to defuse a tense situation.
I think the other authors in conversation with Karp had a good point that machines will change often so it might not make sense to name a host after its machine. Also maintaining two tables, one for site-to-integer mapping and one for machine-to-integer mapping, seems like a lot of work and highly error-prone! So I'm not a fan of what is being proposed here.
How to follow this blog
I'm Darius Kazemi. I'm an independent technologist and artist. I do a lot of work on the decentralized web with ActivityPub, including a Node.js reference implementation, an RSS-to-ActivityPub converter, and a fork of Mastodon, called Hometown. You can support my work via my Patreon.