ISO/IEC JTC1/SC22/WG5/N1249
                                                              1997-02-07

                   GERMAN NATIONAL ACTIVITY REPORT
                  =================================
                   (DIN working group for Fortran)

1. General
----------

Since the July 1996 meeting of WG5 in Dresden, there have been two
meetings of the DIN working group (October 1996 and January 1997). 
Both were held for two days, and were attended by six members of the 
working group. There are about a dozend active members on the DIN 
mailing list, most of which come from universities and research 
centers and represent users of the language. 

The DIN Fortran working group has repeatedly discussed Fortran's 
diminishing role in the world of computer programming.  When top 
managers become convinced (how and by whom?) that Fortran is no longer 
worth being supported or even programmed within their organization, we 
have a political problem and often an informational deficit.  
Participation of vendors in the German group is decreasing, partly 
caused by the fact that none of the vendors do compiler development in 
Germany any more. When large organizations such as CERN in Geneva and 
the GMD near Bonn (and others) cease all official activities in Fortran, 
we should be concerned.  

Regrettably, Karl-Heinz Rotthaeuser had to give up chairing the DIN 
working group for such reasons, and we should all be concerned. At its 
last meeting in January, the DIN working group appointed Wolfgang Walter 
as its new chairman, and would like to take this opportunity to thank 
Karl-Heinz Rotthaeuser for his long-term dedication to the Fortran 
standardization effort.


2. Balloting
------------

Germany has approved both Corrigendum 3 to Fortran 90 and the 
Draft International Standard for Fortran 95 (YES without comments). 

The PDTRs on Enhanced Data Type facilities and Floating Point Exception
Handling have been reviewed and approved (YES with comments). Most of
the comments are editorial. One particular concern is that both PDTRs 
should contain introductory text which conforms to the procedures set 
up by WG5 in document N1152.


3. Work on F2000 requirements
-----------------------------

The German working group has spent most of its time for the work on
F2000 requirements: 

  - the draft PDTR on Interoperability of Fortan and C has been prepared 
    (by the project editor, Michael Hennecke) and reviewed (by the whole
    group), 
  - the WG5 technical subgroups WG5/data and WG5/misc are led by Manuela 
    Zuern and Christian Weber, who also prepared the base documents for 
    the Las Vegas meeting of WG5,
  - the DIN working group members participated in these as well as 
    in the WG5/hpc subgroup.
  - the work for conditional compilation has been informally reviewed.


4. Strategy
-----------

Meeting time has mostly been devoted to deciding on DIN's priorities
for the F2000 requirements, and to discussions on the strategies which
are applied for Fortran standardization. DIN strongly believes that 
the next standard has to be very convincing and hard to ignore, and that
this will definitely be our LAST CHANCE to clean up and complete several
major features which were introduced into Fortran 90 in a way which has 
proved to be too restrictive, incomplete, or otherwise less useful than
anticipated. Among these currently unsatisfactory features are

  - modules and visibility rules, 
  - generic interfaces and overloading rules, 
  - extension operators and operator priorities, 
  - data abstraction facilities in general.

Even if we get these features completed in 2002/3, it will be 12 years 
after they were originally introduced. This stretches the users' 
patience to the breaking point and causes many users to turn to 
alternative languages.

The DIN working group also expresses its concern about an increasing
need for support of low-level language facilities in the area of high
performance computing and interoperability with C. Particularly the
message passing interface de-facto standard MPI is based on many details 
of memory layout of user data, and on asynchronous processing. Fortran 
is currently unable to deal with these issues. For C interoperability, 
the features which are most urgently needed (in addition to the PDTR) 
are the support for procedure variables, and the access of Fortran 
features (esp. data structures, descriptors, ...) from C. DIN strongly 
recommends that these issues be addressed for the F2000 revision.