[gfortran, patch] PR29441: Error out for non-standard intrinsics in initialization expression

[FX Coudert]

> We used to only error out for code such as
>  real, parameter :: pi = 4.*atan(1.0)
> if -pedantic was given, even with -std=f95.  This patch fixes this  
> by removing pedantic from the condition leading to the error.

We have nice gfc_notify_std routines, why can't we use them here?  
That will help getting a consistent behaviour between different cases.

> Testcase included in diff, built & tested on i686-darwin.

(FX makes a mental note. Tobias has a MacIntel...)

Would you consider building gfortran packages on a monthly (or less)  
basis? I can send you my completely-automated-build-and-packaging  
script if you want...

FX