Sean Donelan wrote:
How much notice is adequate when changing widely used behavior on the Internet?
When the ARPANET changed from NCP to TCP/IP how many days advance notice was needed?
Umm, hard to remember that long ago, and haven't looked at IENs in ages, but should memory serve me correctly, the first TCP/IP documents I read were circa 1978, and the cutover was circa 1983. However, I wasn't directly involved -- my email eventually went through a gateway at UMich to ARPAnet -- and the network I was cobbling together on a NOAA+EPA grant used bisync, X.25 (for the satellite links), and an assortment of serial protocols (such as 5-bit baudo coded weather data) that we mixed together on Alpha somethingorothers and Interdata 7/32s, first with the proprietary OS, and then with a new-fangled thingy called "Unix". (Now, we'll drag out the old stories for the kiddies.) The folks at Merit fed me the TCP/IP documents, and were a lot more closely involved in the cutover.
When the Internet changed from IPv4 to IPv6 how many days advance notice was needed?
Well, we finished the initial design in 1993 (I was part of the original design team), and the cutover will be (heavy sigh) probably never.... After all, it's time to design the next generation!
When the COM and NET zone behvior was changed to eliminate NXDOMAIN how many days advance notice was needed?
<sarcasm> Oh, there was advance notice? </sarcasm> I wrote my first DNS implementation in 1987. I know it's still in use on a number of old routers and dialup access boxen. My guess would be another 16 years, or so, to clean up the entire mess. Easier to eliminate the problem at the source! -- William Allen Simpson Key fingerprint = 17 40 5E 67 15 6F 31 26 DD 0D B9 9B 6A 15 2C 32