Flying With Python (Strong versus Weak Typing)

[Tim Peters]

[Mike Silva]
> Is a factor of 100 difference in error rates of certified, fielded
> aviation software relevant?

If it concludes Python is the winner, yes; else it's riddled with
methodological flaws, if not outright deception <wink>.

Provable correctness doesn't have much of a following in Dynamic Language
Land, perhaps because such languages are so effective at tackling problems
where "the solution" isn't known-- perhaps not even recognizable --in
advance.  Dijkstra was famous for saying that testing can't demonstrate the
absence of bugs, only their presence, and he was right.  That's what XP
*does* in this part of the world, but it's not ashamed of it.  Provable
correctness is far too expensive for most of the world's software to even
contemplate as a design goal.

That doesn't mean a Python program can't be proved correct, although
exploiting the full power of the language would make that very difficult for
a program of any appreciable size.

then-again-i-started-my-career-flying-on-planes-designed-by-
    legacy-fortran-programs-and-lived-to-type-about-it-ly y'rs  - tim