Re: What is wrong with SVN?

[Tobias Burnus <burnus at net-b dot de>]

Angelo Graziosi wrote:
> Il 30/09/2013 16.23, FX ha scritto:
> > > What is wrong with this command:
> > > $ svn co svn://gcc.gnu.org/svn/gcc/branches/fortran-dev fortran-dev
> BTW, there is another way to get the fortran-dev branch (for example 
> with rsync)?

You could try the git mirror, http://gcc.gnu.org/wiki/GitMirror – I 
don't think it is otherwise available.

In addition, don't expect too much from the Fortran-dev branch. Quite 
some work went into it to convert gfortran's dimension triplet (lbound, 
ubound, stride [in multiples of element size]) to TS 29113's 
(lower_bound, extent and sm [stride multiplier in bytes]). Additionally, 
the branch comes with the C binding (ISO_Fortran_binding.h). But there 
are still about a dozen files in the test suite which failure, not 
everything has been converted internally to bytes in the stride 
multiplier. Additionally, setting the lower bound internally to 0 
doesn't always work, yet.

Thus, if you want to play with some TS29113 items, Fortran-Dev can be 
useful. But it is far from production ready. I hope that we can polish 
the remaining items such that the branch can be merged early after the 
GCC 4.9.0 release (= beginning of the GCC 4.10 development). But that 
depends how much time the contributors have. (The new descriptor is also 
a prerequisite for deferred-length character array components; but no 
work has yet been made on the branch.)

For TS29113, see ftp://ftp.nag.co.uk/sc22wg5/N1901-N1950/N1942.pdf - 
expect some minor changes in Fortran 2015 related to TYPE(*), 
lower-bounds of DIMENSION(..) and some other additions (e.g. SELECT RANK 
is on my wishlist).

Tobias