Intrinsics arguments evaluated multiple times

[Brooks Moses]

FX Coudert wrote:
>> I've got some other ideas for things that need to be tested with  
>> the intrinsics as well (for instance, we don't have much of a  
>> library testsuite; most all of the tests get constant folded, and I  
>> suspect there are some cases where we only test the library and not  
>> the constant folding), which this could be added in with.
> 
> There already is a PR about that in the libfortran component. (Why  
> can't we have a libGfortran bugzilla component?) The one we're  
> talking in this thread and the newly filed PR is purely front-end.

Yeah -- I'm think more about "added in with" in the sense that someday 
when I have a colocation of sufficient time and sufficient peevedness, 
that I'll sit down with the standard and go through and write systematic 
tests for all the intrinsics, both in library and front-end form.

I hope that this will happen sometime before "the standard" in that 
sentence would be the Fortran 2025 version.  :)

>> So, file an enhancement PR, and then add me (brooks@gcc....) to the  
>> CC list so it won't get lost if/when I have time to work on  
>> this.  :)  (The "if" part of that if/when is why I'm not saying to  
>> go ahead and assign it to me; I don't know if I'll have time for  
>> this or not.)
> 
> Done.

Thanks!
- Brooks