APL and the Object-Oriented Paradigm: Another Cold War Ahead o! Us? Martin Gfeller Reuters Risk M a n a g e m e n t Sot~ware Kieinstrasse 6 CH-8008 Zurich Switzerland E-mail: mgf@ipsaint, ipsa. Reuter.COM Fax: +41 1262 17 92 At APL92 it became apparent t h a t some APL implementers are t h i n k i n g about adding Object-Oriented (0-0) extensions to their implementation of the language. I t h i n k this move is good and necessary for APL, because: ¢ I also like to encourage interested users to contribute to the design; especially users who have exposure to o t h e r 0 - 0 s e ~ w a r e or to the c u r r e n t m a i n s t r e a m of computing could bring in fresh ideas. There are several suitable forums for design contributions; for example, the comp. l a n g . a p l newsgroup on Usenet. I also have assembled e-mail addresses of major vendors, and can forward contributions to them. I don't t h i n k we need to strive for complete consensus at this point, but we must agree on the major framework. ¢ O-O adds encapsulation and information abstraction to APL, simplifying large system construction. An APL-to-C Compiler for the IBM RISC System/6000: Compilation, Performance and Limitations ¢ Encapsulation eases interfaces to external systems and languages. ¢ New operating systems support 0 - 0 applications and m o d u l a r application components (applets), and are thernse]ves implemented using O-O techniques. 0 - 0 is required to t a k e advantage of such systems, which is m a n d a t o r y for mains t r e a m computing in the future. ¢ You c a n ' t sell a n y t h i n g t h a t isn't O-O these days .... Wai.Mee Ching Computer Science Department IBM Thomas J. Watson Research Center Y o r k t o w n Heights, New York 10598 Dz-Ching Ju IBM Santa Teresa Laboratory ,5,55 Bailey Avenue San Jose, CA 95141 Problem: "Cold War" Abstract We have developed an APL-to-C compiler for the IBM RISC Systera/6000 T M workstations by porting our Bysterrt/370 APL-to-C compiler to the RISC System/6000 series. We describe the compilation process (i.e. t r a n s l a t i n g from APL to C) and our experience in using the APL/C compiler. We report the performance of the compiler in terms of compilation time and the execution time of the compiled code. We compare the speedup of the compiled code over t h a t of the interpreter w i t h the speedup of the APL370 compiler on the mainframe on several examples. We also discuss some subtle restrictions the compiler imposes on the application code and the reasons w h y t h e y e x i s t . However, i f each implementor designs an O-O extension t~o APL on t h e i r own, we're bound for incompatible language dialects in a m u c h more severe way t h a n w i t h nested arrays. APL simply cannot afford a n o t h e r "cold war" as it had with the i n t r o d u c t i o n of nested arrays. Communication The APL community has to communicate, and share ideas and designs---~efore a n n o u n c i n g an implementation! Going-ahead may give a vendor a short-term m a r k e t share, but the long t e r m outlook on APL could be bleak. C o m m u n i c a t i 6 n is the responsibility of the vendors, but also of interested users, who can contribute to the design. AlGer APL92, I encouraged major APL vendors to share t h e i r design ideas. Most vendors consider the Minnowbrook conference as the best venue for this, but it is perhaps too singular in time and attendance for a subject t h a t is so important for the future of APL. I ntroduction In F e b r u a r y 1990, IBM Corporation introduced the IBM RISC System/6000 series of workstations which implements the superscalar RISC (Reduced I n s t r u c t i o n Set Computer) a r c h i t e c t u r e [1] and brought powerful machines to peoples' deskside [7]. The RISC Systend6000 series w o r k s t a t i o n has APL Quote Quad

/lp/association-for-computing-machinery/an-apl-to-c-compiler-for-the-ibm-risc-system-6000-compilation-GYaHGYdbvb