klaiber@DEWEY.UDEL.EDU (12/12/86)
I just read Bob Campbell's message from Dec 11 and I really feel I should give some comments on that: He seems to be VERY fond of compiler options and in particular suggests several compiler options to beef up the language a bit - his main point being that a compiler not supporting these features still can compile these programs, as compiler options are nothing but comments to it. This is one of the WORST proposals I have seen on this bboard so far!! The idea is to standardize the language, not to add tons of different extensions that can somehow be activated through the use of some obscure compiler-options. Now if one really wants to add features to the language, then I feel it shoud be done in an ORDERLY and STANDARDIZED manner - and so far I have yet to see any language where ALL compiler options are nicely standardized! Some more comments: - INLINE code : I don't think it is a good idea to add a complete assembler to existing modula-compilers; I'd rather prefer a clean interface to modules/procedures compiled (assembled) in other languages. - EXCEPTIONS: It is well-known that the exception mechanism in PL/I (dynamically associated exception-handlers) had some serious drawbacks which made it practically useless. Thus, one tends to favour statically associated exception-handlers (as in CLU or ADA). I don't really think one could include an exception-mechanism like the one found in these two languages by just providing some library module. The setjmp()/longjmp() functions in C are in any case HIGHLY UNSTRUCTURED and violate the philosophy of modern structured languages. So the bottom line is: If you want extensions, then you better had do it RIGHT in the first place, else leave them out! Alexander Klaiber (klaiber@dewey.udel.edu) <usual disclaimers etc...>