Xref: utzoo comp.arch:10544 comp.lang.misc:3051 Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!agate!eris.berkeley.edu!mwm From: mwm@eris.berkeley.edu (Mike (I'll think of something yet) Meyer) Newsgroups: comp.arch,comp.lang.misc Subject: Re: Double Width Integer Multiplication and Division Message-ID: <26116@agate.BERKELEY.EDU> Date: 8 Jul 89 02:25:02 GMT References: <57125@linus.UUCP> <1989Jun24.230056.27774@utzoo.uucp> <13961@haddock.ima.isc.com> <1395@l.cc.purdue.edu> Sender: usenet@agate.BERKELEY.EDU Reply-To: mwm@eris.berkeley.edu (Mike (I'll think of something yet) Meyer) Organization: Missionaria Phonibalonica Lines: 43 In article <1395@l.cc.purdue.edu> cik@l.cc.purdue.edu (Herman Rubin) writes his usual continuing diatribe against modern (or, in the case of C, not-so-modern) programming language designers. Herman wants a language that is both notationally flexible enough to allow him to use whatever notation is comfortable for him, and sufficiently smart that he can write code that obviously maps 1-1 onto machine hardware for the machine he's writing it for. Since modern programming language designers are more concerned with features like portability, maintainability and reusability (and the ability to say *ability :-), they tend to ignore the things he wants. Notational flexibility leads to nearly unreadable code, and incredible support libraries, the antithesis of readability and maintainability. The ability to write code specific to a machine architechture is obviously antagonistic to the goal of portability. Herman isn't the only person running around claiming some major area of CS is going in the wrong direction. He isn't even the only one to argue about it a lot, and do nothing to prove his point. He may even be right. However, flaming about it on USENet isn't going to do anything other than make AT&T a lot of money. Herman, you need to do three things. 1) Write or find a language that is what you want. 2) Arrange to give it, along with enough documentation for it to be useable, to anyone who wants it. 3) Write a paper acceptable to a refereed journal (or SNot) describing the language, and why it's superior to everything else available. I predict that you'll be roundly ignored, as what you want is really little more than a flexible template macro processor on top of an assembler. Spiffy assemblers have been proposed before (Johnson at Bell Labs wrote one; Whitesmith sold one with their C compilers), and were roundly ignored.