Non-conforming code in the testsuite (do_1.f90)

[Brooks Moses]

FX Coudert wrote:
>>- Document, in a comment, that this is an extension?
> 
> After reading the clf thread, I'd go for this. The comment should 
> mention that "the Standard says that such code is processor-dependent, 
> and gfortran implements such that the loop iterates 11 times", or 
> something like that. If I understood correctly, this is not an 
> *extension* ;-)

Agreed.  Generally, people on clf refer to these things -- that is, 
matters of how processor-dependent requirements are handled by the 
compiler -- as "quality of implementation" issues.

(For example, the standard allows a compiler which prints out "This 
program exceeds the capacity of this compiler" for everything it's 
given.  Actually compiling some programs is certainly not an 
"extension", but it isn't strictly a requirement of the standard either.)

It's perfectly reasonable for the gfortran test suite to test that it is 
providing the expected quality of implementation, as well as merely 
giving strict conformance to the standard.

- Brooks