Deprecation of -fdefault-* options

[Dominique d'Humières]

> Le 25 déc. 2015 à 21:50, Steve Kargl <sgk@troutmask.apl.washington.edu> a écrit :
> 
> On Fri, Dec 25, 2015 at 08:46:40PM +0100, FX wrote:
>> I am strongly against removing user options that are widely used
>> (including in established libraries and software build systems),
>> are well defined and documented (including their downsides), and
>  ^^^^^^^^^^^^^^^^
>> do not pose maintainance issues.
> 
> Well, therein lies the problem.  The -fdefault-* options are
> not well-defined.

For me it is well defined: REAL and DOUBLE PRECISION are promoted, variables with explicit KIND are not.
This allow to set explicit KIND for variables involved in procedures expecting a given KIND, such as ETIME.

Dominique

>> 
>> We can document them better if need be. We can warn about them, maybe.
>> 
> 
> You're assuming users read the documentation.
> 
> -- 
> Steve