lagache@violet.berkeley.edu.UUCP (05/28/87)
There has been two sorts of replies to my posting about teaching structured FORTRAN on UNIX. The first is to try a structured preprocessor. Unfortunately that idea failed to pass muster with the staff. The main reason being that it is very hard to associate error messages which are referenced to the generated FORTRAN code with the program the student actually wrote. Since this is an introductory course, we felt it was asking too much of the students. The other sort of reply was that courses of FORTRAN should teach the present standard (FORTRAN-77). For historical reasons we have always taught with watfiv dialects (primarily because WATSOFT compilers are very good at catching errors). However, there is a philosophical case to be made as well. With FORTRAN 8.X due in at most 2 1/2 years, there isn't a very strong case to teach a nearly defunct standard. Also providing actual structured flow of control provides an important enforcement to good programming habits, and is much easier to debug. We are not very dogmatic about keeping structured FORTRAN, but there is a general concensus that getting a structured FORTRAN would make everybody's life easier. Edouard Lagache School of Education U.C. Berkeley lagache@violet.berkeley.edu
bzs@bu-cs.BU.EDU (Barry Shein) (06/01/87)
Posting-Front-End: GNU Emacs 18.41.4 of Mon Mar 23 1987 on bu-cs (berkeley-unix) From: lagache@violet.berkeley.edu (Edouard Lagache) > The other sort of reply was that courses of FORTRAN should teach > the present standard (FORTRAN-77). For historical reasons we have > always taught with watfiv dialects (primarily because WATSOFT compilers > are very good at catching errors). However, there is a philosophical > case to be made as well. With FORTRAN 8.X due in at most 2 1/2 years, > there isn't a very strong case to teach a nearly defunct standard. > Also providing actual structured flow of control provides an > important enforcement to good programming habits, and is much easier to > debug. Right, and by the mid 1990's F88 compilers should be readily available... I agree with the opinion to teach whatever dialect of fortran is readily available on the system, such as F77 on Unix. I can only assume you teach Fortran at all because you believe people really use it for solving real problems, well let them learn the versions people are really using. You're flailing against the wind wishing for something that obviously isn't in use yet and then wishing it would also be well supported. They'll pick up the F88 quick enough when it starts to creep into their environment, they'll be close to 30 by then and accustomed to such minor revolutions if they really use it, everyone around them will be in the same boat. Your sentiment is nice, but I think it's not necessary. Just teach them Fortran if you're going to teach them Fortran and stop wishing for something else, teach them what people are using. -Barry Shein, Boston University