[SystemSafety] OpenSSL Bug

[David MENTRE]

Hello,

Le 14/04/2014 23:59, Derek M Jones a écrit :
>> Suppose technical methods T are known to avoid, definitively, mistakes
>> of type X,
>
> more empirical evidence,

No. As Peter said, the method T *definitely* avoids the mistake, by 
construction.

To take just one example, consider the "=" vs. "==" issue in C's if 
construct (e.g. "if ((a = check()) == SUCCESS) { ... }").

In C++, Ada and other languages, "if" construct only takes a Boolean as 
first test parameter (contrary to C that takes arbitrary expression), so 
an error like "if ((a = check()) = SUCCESS) { ... }" is systematically 
caught by the compiler with a type error if type of "a" is not boolean. 
In C, you could use a home defined bool type and have a static checker 
checks that the use of this bool type is properly made according to some 
typing rules.


>> and T are practical.
>
> and yet more empirical evidence,

No. In my previous example, the proposed approach is demonstratively 
practical because you can write *exactly* the same expressions as before 
(e.g. "if ((a = check()) == SUCCESS) { ... }"). The additional typing 
adds no additional burden on the programmer.

Sincerely yours,
david