[TUHS] Unix clones

Doug McIlroy doug at cs.dartmouth.edu
Sun Apr 9 01:22:23 AEST 2017


> That's interesting that that sort of thing dates back (at least) to the Labs.

Research couldn't hold a candle to Development on making smooth transitions.
You don't take a telephone switch offline to change a file format or the like.
The development cycle used to be about three years: one year for design,
one for implementation, and one to build a hybrid to bridge the transition.
At 2AM on Sunday, you'd install the hybrid on one of the dual cross-checked
 processors at a time, so the switch was never interrupted. Later you'd
dispense with the hybrid the same way.

Doug


More information about the TUHS mailing list