SULLIVAN@suny-bing.CSNET (05/21/87)
I don't quite understand the acrimony in the criticism of DEC's modula compiler. Granted, I haven't used it very much, yet. I would like to make the following points: 1. The business about invisibility in the implementation module of identifiers imported in the definition module would be serious, if it were the case. It certainly is not in my copy of the compiler! 2. I can't really comment about failure of timestamping. I use make and don't worry about such things. This is, after all, a Unix system. 2. Input/Output. There is room for much debate here. Certainly the facilities provided are very different than those one usually finds with a modula system. They are very similar to what one expects in other Unix compilers. On the other hand, i/o as suggested by Wirth (and provided with many systems) is as unnatural as anything I can imagine. I detest writing "WHILE Done DO ..."! I am running the DEC compiler on a Unix system for research, and the Hamburg compiler on a VMS system for my classes. I have written a modules for both which are identical in function, and mimic Pascal i/o as closely as it's possible to do without overloaded parameters. One of the nice features of modula is that the i/o system can be replaced if it is found to be distasteful. 4. The compiler at least seems to be more robust than the Hamburg compiler (which crashes on input records longer than about 120 characters, and has other similar problems). I haven't tried the Cambridge compiler, so I can't compare. (Where can I get it? Does it cost money?) Fred Sullivan Dept. Math. Sci. SUNY @ Binghamton sullivan @ suny-bing.csnet