Date: Tue, 05 Nov 1996 21:15:09 GMTServer: NCSA/1.5Content-type: text/htmlLast-modified: Thu, 12 Sep 1996 23:37:55 GMTContent-length: 8170<title>Andy "Krazy" Glew's Favorite Quotes and Sayings</title><h1>Andy "Krazy" Glew's<br>Favorite Quotes and Sayings</h1>This collection started out as a set of IBM computer cards pastedto the wall of McGill's McConnell Engineering Computer Facility,where I operated the RJE site and provided programming advice,was pasted to the glass wall of my office at Gould, and has followed me around.<p><hr>Summus non Americae, eh?<br>-- The Canadian National Motto<hr>BUGS: it's the PIPELINE, stupid!!!<br>-- Andy Glew<UL><LI>Expressing my observation that many bugs in P6 were caused by incorrectdesign of one dimensional pipelines, not the multi-pipeline decoupled natureof P6.<LI>The title of a talk I gave at the IEEE Asilomar Conference in 1994.<LI>We need more formal methods for the straightforward design of pipelines.</UL><hr>Bypasses - just say no.<br>-- Andy Glew<hr><DL><DT><BIG><STRONG>An Architect</STRONG></BIG><DT><SMALL><SUB>A Micro-Architect</SUB></SMALL></DL>-- Robert Geva<p>(You will not be able to see this without HTML 3, supportingthe "BIG" and "SMALL" HTML commands.)</p><hr>Measure twice, cut once.<br>-- "This Old House"<hr>Virtual - something that isn't there, but appears to be.<p>Transparent - something that is there, but doesn't appear to be.<hr>"In times of profound change, the learners inherit the earth, while thelearned find themselves beautifully equipped to deal with a world that nolonger exists."<br>--- Al Rogers, Global SchoolHouse Network, speaking at CUE<DL><DT>Glew comment:<DD>    I wonder if there is some correlation between "liberal"     and "ability to learn and adapt to new things",    so that liberalism is a trait that has a survival advantage    in times of rapid change,     but which is of less advantage during the long intervals    of less change.    The computer industry may act as an example:    profoundly liberal when it started,    conservative now that the time of great innovations and new ideas has passed.</DL><hr>The 3 most important facts in programming:<ol><li>There are N numbers in the set 0..N - <i>not!</i>.    <br>---paraphrased from Stan Kelly-Bootle<li>Any programming problem can be solved by adding a level of indirection.    <br>---Unknown<li>Any performance problem can be solved by removing a level of indirection.    <br>---Mike Haertel</ol><p>---Compiled by Mike Haertel.<hr>Q: What is the single most important fact in Computer Science?<p>A: The fact that there are N+1 elements in the set {0..N}<p>---Stan Kelly-Bootle<hr>Names are a scarce resource.<br>---Andy Glew<hr>If it ain't broke, find something else to fix<br>-- Tim Olson, AMD<hr>The worst maxim around is "Don't fix it if it ain't broke".You'd better be improving it,because your competitor is.<br>---Wayne Calloway, Pepsico CEO<br>---FORTUNE April 10, 1989<hr>All great ideas eventually degenerate to hardware<br>---Peter Drucker?<p>(The poor ideas remain in software)<hr>Computer memory requirements are growing by 1 address bit / 18 months<br>---IBM, 1987<hr>If your users don't cost more than the appropriate computers,you should get better users(unless they are grad students)<br>---andy@polya.stanford.edu<p>(Justification for running emacs on Cray)<hr><b>ALWAYS code to be reused</b>.Because then your code can be reused.Saying that "this code will never be reused" and coding that way is a self fulfilling prophecy.<br>---Andy Glew<hr>There has never been a good product designed by a committee.<p>Keep a decade behind.<p>I'm all for simplicity.If it's very complicated I can't understandit.<p>Parity is for farmers.<p>---Seymour Cray<hr>Now that I have taught you how to avoid division,the next time I will teach you how to avoid multiplication.<p>---Seymour Cray, talking to numerical analysts.<hr>Let's make smart machines to run slower software faster!<br>---Bob Campbell, uw-beaver!tikal!bobc<hr><dl><dt>Flon's axiom:<dd>There does not now, nor will there ever,exist a programming language in which it is the least bit hardto write bad programs.</dl><hr>Development is maintenance.<br>---Brian E. Marick<hr>"Do you really want to do this?"is a question that should never be askedunless there is a chance that the answer is "No".<p>Otherwise, if you are asked this question 9 times a day,answering "Yes",the 10th time will have you answering out of habit.<hr><dl><dt>The 10th axiom of Systemantics<dd>When a fail-safe system fails,it fails by failing to fail safe.</dl><p>---John Gall, Systemantics<hr>Scientific applications grow a factor of 100 every 10 years<br>---Giles<hr>Perfection is the enemy of the dissertation<br>---Ivan Sutherland<hr>Perfection is the enemy of completion.<br>---Andy Glew<hr>If all else fails, follow the instructions.<p><dl><dt><dd>No! No! No! <b>Read</b> the instructions, bearing in mind they're probably fucked up anyway!<br>---Andy Glew</dl><hr>Indirection is the right direction.<br>---Andy Glew<hr>Never put off until run time what can be done at compile time.<dl><dt><dd>You wouldn't believe the shit I got from late binding advocatesthe first time I posted this to a Gould newsgroup!<br>---Andy Glew.</dl><hr>Never put off until execution time what can be done at decode time.<br>---Andy Glew<hr>Recurses! Called again...<p>Recurse you, Mad Baron!<hr>It's easier to apologize than to ask permission.<br>---Grace Hopper<hr>We need a coalition to stamp out smart I/O devices.<br>---Chris Torek<hr>Most "smart" I/O controllers are, in fact, "smart-ass" I/O controllers.<br>---Rob Pike (paraphrased)<hr>An educated man knows of a synonym for every concept.<hr><dl><dt>The Titanic Effect<dd>The severity with which a system failsis proportional to the intensity of the designers'belief that it cannot.</dl><hr>In the early 1970s 8K of ROM was equal to 8 bits of register space.What a reason for microcode!<br>---Info from Patterson, 1985<hr>Instructions should not bind together operationswhich an optimizing compiler might otherwise choose to separatein order to produce a more efficient program.<br>---Brian Case<hr>It's just a shift to the left<br>And then a shift to the ri-i-i-i-ght<br>Get a load of those MIPS<br>And the code's real ti-i-ight<br>This lockstep pipeline<br>Really drives me insa-a-a-a-a-ane<br>Let's do the pre-charge again!<br>---Brian Case<hr>Measurement of repetition is not equivalent to repetition of measurement.<br>---Eugene Miya<hr>The best is the enemy of the good.<hr><dl><dt>RISC<dd>Relegates Important Stuff to Compilers</dl><hr>The maxim<cite>Nothing avails but perfection</cite>may be spelled<cite>P A R A L Y S I S</cite><p>---Winston Churchill<hr>Turning off the type and range checking after the programis debugged is like throwing away the lifeboats after the maidenvoyage.<p>(Or like travelling without a convoy through a submarine war zone.)<hr>The biggest (and most common) mistake in computer design is not providingenough address bits.<br>---Bell and Mudge,  "Evolution of PDP-11"<hr>Rules of Thumb<ul><li>Every 3 years transistor account on a chip increases 2X.<li>Every 3 years chip  speed increases by 2X.<li>Every 3 years DRAMs increase in size by 4X.<li>Every 10 years DRAMs get 30% faster.<li>Every 3 years Disks increase in size by 2X.<li>Every 10 years Disks get 30% faster.<li>1MB of main memory is required for every MIP.    <dl>    <dd><i>No way! Disproved on microcomputers!</i>    </dl><li>1Mb/s of bandwidth is required to feed every MIP.<li>Programmers increase the size of programs 2X every year.</ul><p>--- UNIX International, May 7, 1990<hr><dl><dt>Rule for debugging:<dd>When in doubt, print it out!</dl><hr>Buying clothes more than once a decade is a waste of time<br>--- Andy Glew<hr><hr>$Header: /u/g/l/glew/public/html/RCS/sayings.html,v 1.8 1996/09/12 23:34:44 glew Exp $