Adding quite a few intrinsics
Janne Blomqvist
jblomqvi@cc.hut.fi
Wed Nov 9 16:50:00 GMT 2005
On Wed, Nov 09, 2005 at 04:59:59PM +0100, François-Xavier Coudert wrote:
>
> > In fact, I think BM:s definition is so good that we should adopt it,
> > and spell it out in the documentation, so that our users will
> > understand the distinction too.
>
> Well, I don't agree with this definition. That would mean that code
> that compiled out of the box with g77 will not compile with gfortran
> unless you give it the -std=legacy flag, which:
>
> 1. sounds like saying "please dump that big concrete block on me",
> and most probably people will be reluctant to do it
>
> 2. will break configury scripts, like the GNU auto* tools
>
> 3. will get an endless crown of people asking on our list why their
> code doesn't run :)
You're right. As a compromise, perhaps the default (i.e. -std=gnu or
no -std= flag at all) should be "accept everything, but warn about
legacy/obsolete features"?
--
Janne Blomqvist
More information about the Fortran
mailing list