ISO/IEC JTC1/SC22/WG5 N1334 To: WG5 From: DIN Working Group for Fortran German National Activity Report 1. Membership and Meetings Current membership of the DIN working group for Fortran is approximately 12, with about two thirds active, but only about half able to attend meetings. Attendance at the last few meetings has typically been between 4 and 6. There have been 2 DIN meetings since the last WG5 meeting in Trollhttan. Due to the small membership and scanty support from employers, the DIN working group has been able to perform a limited review of the latest draft standard document only. However, DIN believes that discussion of the issues and problems identified should not be delayed until the public review period. For "Interoperability with C", we have, for the time being, lost the support of our expert Michael Hennecke (which will not come as a surprise to anyone), although we are still hoping to "reactivate" him some time in the future . . . ;-) 2. Political Issues The DIN working group is worried about the rapid decline in membership within the national Fortran groups (including J3) and has discussed the apparently diminishing role of Fortran repeatedly. In Germany, there is a definite feeling that Fortran 90 introduced many useful new concepts, but too many of them with intolerable inadequacies and restrictions (e.g. modules, operators, pointers, allocatable arrays, derived types), especially after such a long period of development. If these had been remedied in Fortran 95 (even with some delay), Fortran would have stood a chance of regaining its legendary popularity. However, we definitely failed in finishing the task we started with Fortran 90, and a majority of the new features are still incomplete and unsatisfactory today. Moreover, they will still be lacking when Fortran 2000 will be the new standard. In our opinion, a programming language community can hold out for one additional standard if the obvious problems are resolved with the next standard. However, very few programmers will wait for the next 2 or 3 standards to get their problems solved. In our case, Fortran 95 was our LAST chance to solve a large majority of the problems inherent in Fortran 90, and we got nowhere close to that goal (the varying-length character string module irrefutably proves this point). Even Fortran 2000 will not achieve this goal although it will be a big improvement over Fortran 95. In the meantime, it is getting increasingly difficult to teach Fortran and remain credible in the eyes of young students (I try every week), let alone convince people to switch to Fortran once they have learned other languages. Wilhelm Gehrke, by far the most productive author of Fortran books in Germany, has been telling us that sales of his Fortran 95 handbook (published by the computing center in Hannover and sold wholesale only, at a very low price) have dropped sharply compared with the high volume sales of his FORTRAN 77 and Fortran 90 handbooks (which were in the tens of thousands). We are curious whether other countries are observing the same decline in sales of Fortran books (and software?). We do not know whether the situation can be saved. or whether J3 will even be able to finish Fortran 2000 with the remaining work force. We are in the paradoxical situation that our language is both too large and too small at the same time: too large to be kept coherent and to be dealt with by a small committee, and too small because it will still contain, even after the Fortran 2000 standard is published, quite a few restrictions and sometimes spurious limitations. Compared with languages that were developed by one person or a very small team and that were 90 % finished before they got bogged down in a democratic process (e.g. C++ and Java), the development of Fortran has some clear disadvantages. As long as Fortran standards were mainly regulating existing practice, the outcome was somehow more acceptable (even if it was not quite like your favorite solution) because it was at least a viable solution that had been proven by experience. As soon as new features were introduced which had not been time- tested by our community (and we were apparently reluctant to learn from the experience of other languages), things changed, and it is unclear how much experimenting a user community is willing to accept. To conclude, we (being incorrigible optimists) are still hoping for a happy end to the story, but we foresee a fairly harsh judgment in the first public review of the Fortran 2000 draft. If the ballot proves us wrong, it might be because only a minority will have voiced their opinion and the others will have gone into a state of indifference. In the end, the final verdict will be indicated by the size of our community. 3. Technical Issues The DIN working group has started a review of some of the most important new features of Fortran 2000. Some preliminary questions/issues are raised in paper N 1335. 4. Other Activities DIN is in the process of establishing home pages for all the languages in NI22 (SC22) for which there is an active group. Some members of our working group are already providing information about Fortran and links to other relevant material on their home pages. Java has been a point of discussion at many recent meetings. DIN is disappointed that specifications of the Java language, Java virtual machine and other Java technology have generally not been open to public discussion, much less for international standardization. Wolfgang Walter (head of German delegation)