On Thursday, 7 August 2014 at 03:54:12 UTC, Ola Fosheim Grøstad
wrote:
«The __assume(0) statement is a special case.»
So, it does not make perfect sense. If it did, it would not be
a special case?
It doesn't have to be a special case if you define it in the
right way - in terms of control flow. Then the interpretation of
assume(false) as unreachable follows quite naturally:
instead of defining assume(x) to mean that x is an axiom, define
assume(x) to mean that P=>x is an axiom, where P is the
proposition that control flow reaches the assume statement.
so assume(false) actually means P=>false, or simply !P
and !P means !(control flow reaches the assume), as desired.