newbie willing to contribute...

[Steve Kargl]

On Sat, Nov 05, 2005 at 04:04:19PM +0100, Pierre-Matthieu anglade wrote:
> 
>  I've run into the problem reported in PR 20244.  I would
>  like to see this bug fixed, and so I've decided to try to
>  contribute to gfortran.  Unfortunately, PR 20244 appears
>  to be beyond my newbie skills.  Is anyone working of this
>  bug, and what bug should a new developer cut his teeth,
>  In the view of being able later to contribute to the implementation
>  of derived type (which result in PR 20244).

Welcome!

PMA check the contributing page about copyright assignment.
http://gcc.gnu.org/contribute.html

I forgot to mention the Fortran standard committee's web site.
You can obtain working drafts of the standards, so that you
can review the documents and provide input to the committee.
To learn more about the committee, google "J3 Fortran".  The
gfortran wiki may be of help here.

Here's a fairly "simple" project.  g77 supports a COMPLEX
intrinsic routine, while gfortran currently does not.  You
use it like

         program a
         double complex z
         z = complex(1.d0, 2.d0)
         print *, z
         end

Note the standard intrinsic routine CMPLX does not preserve the
KINDness on the arguments.  That is, Z = CMPLX(1.d0, 2.d0) takes
double precision, returns a single precision complex value, 
and then the assignment promotes it to double complex.  CMPLX
does take an optional 3rd argument.  Z = CMPLX(1.d0, 2.d0, KND).
Thus, you can map COMPLEX(X,Y) to CMPLX(X,Y,VALUE) where the
value of VALUE depends on the types and kinds of X and Y.  See
the g77.info file for more details.

-- 
Steve