[SystemSafety] C++ and Pointers

[Derek M Jones]

Peter,

> They do, you know. But I grant you that the phenomenon is often obscured.

So, yes, that 90% is a made-up number.

I don't have any data for problems that strong typing catches, or
could catch.  Based on rough percentages of fault classifications,
e.g., errors of omission, commission, logic, arithmetic, wrong
variable, etc; I would say it is less than 10%.

The only work I know of that tries to answer the question of
what faults might be detectable, using the concept of type, if the
code was riddled with the appropriate annotations, is this:
earlbarr.com/publications/typestudy.pdf

-- 
Derek M. Jones           Software analysis
tel: +44 (0)1252 520667  blog:shape-of-code.coding-guidelines.com