Greg Shepherd wrote:
This, surprisingly, can't be said enough. I've repeatedly heard the
argument that someone's running code is verification that the idea
is sound,
If "sound" means no ambiguity or contradiction is found
in the specification under the environment assumed by the
code, you are right.
But, it may merely be that ambiguities or contradictions
are overlooked by implementers.
Moreover, there may be ambiguities or contradictions under
other environments.
Worse, lack of ambiguity or contradiction in the specification
only means it "sound" but not good, useful nor worth adopting.
So, existence of running code does not mean so much.
But, as no code is required to make something PS,
specifications with ambiguity/contradiction often
become PSes.
and therefore the solution should be adopted.
Totally wrong.
Masataka Ohta
