[cap-talk] definition of the term "safe language"

[Sandro Magi]

On 09/04/2010 3:04 AM, Mike Samuel wrote:
>> Not how you're using "unexpected results". If you take it to mean
>> "results inconsistent with the safety properties guaranteed by the
>> spec", then I agree, and in fact Java implements exactly these safety
>> properties and thus is safe.
> 
> Nonsense.  As I pointed out earlier, if any implementation that agreed
> with its spec were safe then C would be safe.

It's not merely a case of agreeing with its own spec, the spec has to
provide abstractions and safety properties. In a very real sense, C
provides no such abstractions, no safety properties that cannot be
circumvented, and that's why it's unsafe. This is nothing like the
alleged floating point abstraction violation. There is no safety
property associated with floating point strictness or portability.

Just because Java does not provide a floating point abstraction that you
expect does not make it unsafe.

Sandro