(j3.2006) F03/0110

[Robert Corbett]

On 08/25/11 20:18, Malcolm Cohen wrote:
>> The example program I gave includes the example function from
>> interpretation F03/0110.  The program, as far as I can see,
>> "uses only those forms and relationships described herein"
>
> Not so, as I pointed out.
You said the program unit is not conforming and therefore the program
is not conforming.  I quoted text in the conformance clause of the
standard that defines what it means for a program unit to be conforming.
The example function satisfies that definition.  The definition might
not be what it should be, but it is what it is.
> I still find it difficult to understand how anyone thinks we can 
> produce quality
> software if obviously non-conforming programs are required to be 
> accepted on the
> grounds that with appropriate input data they might not end up actually
> executing any of the fragments that make them non-conforming.
I agree that compilers should report obvious errors.  Some
programs that are not standard conforming should be rejected,
although the standard leaves considerable leeway in that
regard.  Nonetheless, by the definition of conformance given
in the standard, the example function conforms to the standard.
If a different definition of conformance should be used, the
definition in the standard should be changed.

Robert Corbett