Adam Ruppe wrote:
Lerdorf's quote strikes me as actually being somewhat close to what
Walter is talking about. Web applications don't focus on making a
thing that never fails, but instead achieve reliability by having an
external watchdog switch to backups - that is, a fresh copy of the
program - when something goes wrong with the first one.
Doing this gives them adequately good results at low cost.
That part is true, but the part about catching exceptions thrown by programming
bugs and then continuing the same process runs completely counter to any
sensible notion of reliability.