Re: [Patch, Fortran, OOP] PR 57305: ICE when calling SIZEOF on an unlimited polymorphic variable

[Mikael Morin <mikael dot morin at sfr dot fr>]

Le 26/08/2013 16:16, Janus Weil a écrit :
>>>> I'm slightly inclined to kindly invite the user to switch to
>>>> STORAGE_SIZE+SIZE instead.  Any other opinion?
>>>
>>> Since the SIZEOF intrinsic has been around for some time in gfortran
>>> (before STORAGE_SIZE was available), I would say we should at least
>>> continue to support it for backward compatibility. And I guess we
>>> should also make it behave reasonably for all inputs. However, it
>>> might be worth to add a note in the documentation that STORAGE_SIZE
>>> and SIZE should be used instead in standard-conforming code.
>>>
>> I thought about it again, and what I'm actually in favor of is
>> diagnosing by default _all_ extensions having a standard replacement.
> 
> By 'diagnosing' you mean to give a warning? This can already be
> accomplished by compiling with "-std=f2008 -Wintrinsics-std", I guess
> (not by default, though).
> 
Yes, it would boil down to using that by default.

> Using only -std=f2008 currently results in errors like:
> 
> undefined reference to `sizeof_'
> 
... which is the best we can do, as sizeof could be a user
externally-defined function.

Mikael