(j3.2006) interp fodder from comp.lang.fortran

[Cohen Malcolm]

Well let's at least talk about implementations that *TRY* to be 
standard-conforming.

The "intrinsic default" suggested is not conforming to ANY published 
standard ever.  Existing F90 and F95 programs would stop working.

So as far as I am concerned this is simply not plausible at all, nor is it a 
reasonable response of a vendor when faced with "here is something that is 
hard to implement (but is an upwards compatible extension of F95)" to 
implement something completely different that is NOT an upwards compatible 
extension of F95!

Not to mention that a fundamental principle of generic resolution - and 
let's be clear, we are talking about generic resolution of ASSIGNMENT(=) 
here, is that resolution occurs at compile time not execution time. 
Obviously, the words for choosing defined assignment here have violated that 
fundamental principle, so one can reasonably guess that this was 
unintentional.  But equally obviously, one can be very certain indeed that 
we did not intend to remove an existing facility without even mentioning it!

>I would not be surprised if other implementations work the same way.

I am stunned there is even one implementation that took such a contradictory 
interpretation.

Cheers,
-- 
........................Malcolm Cohen, Nihon NAG, Tokyo.