Dr. Dobb's Journal January 1998
Michael is editor-at-large for DDJ. He can be contacted at mswaine@cruzio.com.A lexically wounded reader, call him Kilroy, hobbled over my electronic doorstep after stepping on an unexploded acronym that I had carelessly left lying around in one of my recent columns, and inquired politely,
"Mike, what does FUD signify?"
"Oops," I replied, lowercasing it to avoid adding injury to injury. "I'm really sorry about that, Kilroy. I know it's hard enough to keep up with all the official jargon without having to know all the slang jargon as well." I briefed him, briefly, on the meaning and derivation of the slang acronym FUD. "It stands for 'fear, uncertainty, and doubt.' I believe it was originally applied to certain of IBM's practices," I said, "like heading off an upstart in a new market by hinting that IBM might be planning to enter the market. Now it seems to be applied more often to Microsoft's policies."
Potentially confusing acronyms should be exploded (expanded, glossed) harmlessly at first sight, before unsuspecting readers can trip them. Electronic spellcheckers and grammarcheckers don't flag unexploded acronyms; a spellchecker might spit FUD back as unrecognized, but after it did that a few times you would tell it that FUD was a perfectly acceptable word and that it could keep its opinions to the contrary to itself. It's not impossible that a grammarchecker could detect an obvious acronym and, like one of those dope dogs, sniff around in its immediate vicinity for something that looked like an expansion of it, say "PG&E (Pacific Gas and Electric)" or "FORTRAN (formula translation language);" and bark if it didn't find what it was sniffing for. But the fact that no grammarchecker does this is probably evidence of what I suspect -- that it wouldn't work well.
That's why we have people. One of the reasons, anyway.
I'm not a big fan of spellcheckers or grammarcheckers (preferring people), but I would pay actual money for a good FUDchecker. Here's how a FUDchecker might work, demonstrated on some randomly selected text, say some remarks to the press by, oh, let's call him Mr. Ionaire. Bill Ionaire. The FUDchecker's output, and attempt to extract the actual assertion behind a statement, is in brackets.
"The notion that Sun says that we are not being compatible, that is just not factual." [We passed the test suites.]
"We do a better job passing those Java tests than Sun or anybody else that is in the implementations of those things do." [Okay, maybe we didn't pass the test suites, but neither did anybody else.]
"We certainly are doing the best job we can in compatibility." [Nobody could pass those test suites.]
"Partly it gets to the notion of: Is this an open standard or not?" [You lose your right to enforce license agreements when you submit a product to ISO for review.]
"They have always hated PCs. They think PCs give people too much freedom, too much power, and they campaign against the PC." [They're evil.]
"The PC industry...is the most competitive business that there is." [Microsoft is the most competitive company that there is.]
No, see, even this imaginary FUDchecker doesn't remove the FUD. It just emphasizes it. Darn stuff's hard to get rid of. Obviously Mr. Ionaire is Mr. Gates, speaking to the press following the Sun lawsuit. After I FUDchecked Bill's remarks, I read the license agreements between Sun and Microsoft and compared them to the claims in Sun's suit. There's plenty of FUD on both sides in this dispute.
Why is Bill so defensive? Well, his company is definitely under attack. Microsoft wants to view Java as merely a language. Sun sees Java as a language, a platform, an operating system, and a lot of other things. But Sun is using Java as a weapon. "Write Once, Run Anywhere" is a battle cry.
I've tried off and on to banish sports and military metaphors from my writing, but it never works. The reason is simple: I keep writing about the actions of people who think, and who live their lives, in terms of these metaphors. It's hard to use gardening metaphors when you're talking about people who dress up as Samurai warriors to appear on the cover of Red Herring.
Some day, these poor souls will be able to get the treatment they so badly need at the Larry Ellison Clinic or the Scott McNealy Development Center or the Bill Gates Hospital for the testosterone imbalanced. Until then, the sports and war metaphors seem like just accurate reporting. Java is a weapon.
Even Microsoft agrees that Java is a language, and while some DDJ readers have let me know, in no uncertain terms, that they consider it a less than perfect language, it's certainly improving rapidly. Witness: Symantec is bringing out a fast native x86 compiler that should bring Java performance within hailing distance of C++.
Sun will fold its "Hot Spot" optimizing compiler into the next version of the JVM.
Beyond that, things are moving rapidly on the Java-as-a-dessert-topping front.
For the embedded systems market, Rockwell Collins Inc. has unveiled a microprocessor that directly executes the Java instruction set. And according to Alexander Wolfe at EETimes, "Sun has tapped out the picoJava core that will form the heart of its first Java-specific chip. The processor incorporates a hardware-based Java Virtual Machine (JVM), which reportedly delivers a ten-fold performance increase over a conventional equivalent software JVM."
But interesting as these developments are to us mere mortals, to the generals at Sun the real developments are the movements of the pieces on the Big Board in the War Room. (Okay, I admit that Sun doesn't really have a Strangelovian War Room with a Big Board showing the missiles converging on Microsoft. Maybe they'll put one in when they move into the booby hatch.)
(One more parenthetical aside, for Kilroy: Sun is purchasing the Agnews Development Center in San Jose, a former booby hatch, to use for its new high-tech campus. Read into that whatever significance you will. And pardon the politically incorrect term.)
Java is intended to take marketshare, even markets, from Microsoft. This is even true where it looks like Sun is trying to create a new market that Microsoft claims not to be interested in. I'm thinking of Network Computers. Sun thinks that the future for thin clients is Network Computers running HotJava Views. Microsoft's position is that these guys hate PCs because they give people too much freedom, and that the future for thin clients is NetPCs running Microsoft software. Microsoft = freedom. QED.
Well, that's clearly the design initiative (see http://java.sun.com/marketing/collateral/design_hjv.html). HotJava Views is "JavaSoft's first user environment that runs on top of JavaOS," according to Sun, but not by any means the last. It implements what Sun calls the WebTop UI.
Don Gentner of JavaSoft has been talking about HJV and its WebTop UI, suggesting that the plan may be a bit more inclusive. In a talk at Stanford entitled "The WebTop: HotJava Views and Beyond," he said:
The WebTop is a new model for accessing and distributing both personal and shared information. The implementation of this model is just beginning, with the emergence of Network Computers and Java-enabled web browsers and operating systems.
He described HotJava Views as a "WebTop environment and suite of applications for a Network Computer": described how WebTop abandons some of the characteristic features of current GUI interfaces, like the desktop metaphor, file systems, and the distinction between applications and documents; and hinted at how "this approach could be expanded to provide the full WebTop vision of access from any device, anywhere."
Any device, anywhere?
Hey, that's Microsoft's turf.
The month's mail included a resurrected Lindley Link, the alumni newsletter from Indiana University's computer-science department. My alma mater has just purchased a Cray Origin2000 supercomputer from the Cray Research subsidiary of Silicon Graphics. It's nice to see that SGI is selling some significant hardware; they've had some difficulties lately. I remember walking past their HQ one night recently, returning from a concert at Shoreline Amphitheater, and seeing all the lights on. Bravura, I remember thinking. Trying to convince the world that they're not in trouble. (But everything in perspective: I'll talk about Apple later in the column.)
The Cray Origin2000 is a 64-processor beast (upgradable to 128), and signals a change in the supercomputer field. It's designed to be compatible with desktop workstations and to run workstation software with minimal modification. The Lindley Link description made it sound like a really fast PC with 15 gigs of RAM and a terabyte disk, so I went over to SGI's web site.
The desktop software that this SMP (symmetric multiprocessing) supercomputer runs isn't DOS stuff, of course; it's SGI software. The same OS (Cellular IRIX) that SGI uses on its workstations runs on the Cray Origin2000. Sort of an SGI-specific Write Once, Run Anywhere. Gee, I wish I were doing illumination modeling of a suspension bridge or studying the fundamental processes of chemistry, so I could justify one of these. Maybe I can justify a visit to my alma mater.
The phrase "alma mater" comes from Alma Mater Studiorum, the proper name of the Bologna University, in Bologna, Italy. This is, my source tells me, the oldest university in the world at 908 years and counting. I learned this from Alex Martelli, a DDJ reader at Cadlab, a California corporation with roots and other protrusions in Bologna.
DDJ reader David Kuller also works in Italy, in a stylish studio in Milan, where he does web work, among other things. He wrote recently to, among other things, remind me about Magic/l.
Why don't you mention the Magic/l software development system from Loki Engineering (Cambridge, MA)? I think it was probably the best development environment by far, even to this day. I believe that Loki got absorbed by Lotus in 1987. The main man was Arni Epstein. Their manual (by now a collectors item) was like a bible with tons of hidden anecdotes on how to write software.
Magic/l was a threaded interpreter/compiler sort of like Forth but with a syntax that read more like Pascal. It had practically all of the UNIX command-line functions. It also had a built-in assembler, as well as compile time arguments (to build classes). I developed loads of software with it from 1983 to 1989. They had versions that ran on practically all hardware platforms. It had a full 32-bit addressing scheme as well as inline assembler, not to mention a memory management system identical to that of the Mac.
Did you know that the original Mac hardware engineers used Magic/l to get the machine debugged and running? Unfortunately, they never showed it to the software team, so it never got considered as the development system of choice.
I built a fully user extensible video graphics titling and animation production machine with it in 1986 marketed by 3M as the Panther. Anyway, the entire astronomy field used it in the early '60s -- if I remember correctly, Arni developed it on a PDP-1 at the Harvard planetarium as an interactive assembler for peeking and poking.
I didn't know that.
I also get mail from outside Italy, like this from Neville Black:
In your November 1997 "Programming Paradigms" column, you quote Michael Bedward saying that a New Scientist article asserted that a "Bombe" emulator on a Pentium took 18 hours to do what the original hardware did in 15 minutes.
I thought that was wonderful. [But] the New Scientist piece is not online.
Neville did, however, find an article by N. Shaylor at http://www.geocities.com /CapeCanaveral/Hangar/4040/bombe.html, which said:
In order to clarify how the Bombe worked I wrote a small simulator in C...In [tests] it takes about 8 minutes on a 100-MHz Pentium when the program is correctly optimized. In contrast the real Bombe would have taken something like 15 hours.
"Sadly," Neville concludes, "the Pentium wins."
Then again...Murray Lesser wrote to say:
In your piece in the November 1997 issue of DDJ, you said (reemulation): "Whatever the motives and purported benefits, is there any reason to believe that end users will trade performance for other benefits, like ease of use?"
There appears to be evidence for that belief. Perhaps it is because the majority of computer "power users" are just ignorant, but why do so many of them insist on taking the performance hit that comes from running the "rewritten for Windows" version of a perfectly good DOS program? With the proper hype, they can probably even be sold emulation!
Evidence: I would remind you that WordPerfect has long been available in two flavors, one for DOS and one for Windows. For a long time, the DOS version was advertised as the "high performance" version and the Windows version as the "ease of use" one. However, most people bought the Windows version. At the moment, while Corel still sells WP 6.1 for DOS (even giving "competitive upgrade" pricing), they no longer advertise it. There now seems to be so little interest in the DOS version that I couldn't even find any third-party book teaching WP 6.1 for DOS when my wife installed it in her DOS 5.0 machine; the Dummies people told me that while they have a new edition of their book for WP 6.1 for Windows, they haven't put out one for WP for DOS since version 6.0.
I didn't think I would publish any more responses to the simple little puzzle about the connection between Steve Jobs and the phrase "anywhere but here," but David Rowland gets a belated no-prize for this:
You offered a "simple little puzzle" at the end of your column. It was too easy, at least for me. "Anywhere but here" is the title of the well-regarded first novel by Mona Simpson, who is Steve Jobs's biological sister. Mona is an old friend and coworker of my wife and gave us our first baby shower. The baby is now 16.
Steve's kids have been busy. I went over to Apple for the press briefing on the release of Rhapsody Developer Release, which went out to developers on October 13. There were fewer dogs and ponies than in past Apple events, mostly just demos that didn't crash.
Apple is right to brag about its progress with this release of its next-generation operating system. The Apple and NeXT software teams didn't get integrated into a Rhapsody team until March, and in seven months they managed to get a product out the door. That's impressive enough. That they also succeeded in riding over the NIH GUI hurdle to meld NeXT and Mac features is all the more impressive.
Two acronyms and a sports metaphor, sigh. All right, NIH means not invented here (an acronym of scorn) and GUI I know I don't have to explain. Back to the Rhap.
Developers seeded with an early version of Rhapsody claim that Apple has actually delivered its own version of write once, run more or less anywhere. AAA+ Software raved about being able to "develop apps on Rhapsody and deploy them on Windows NT without any compiling or linking under Windows." And one hears raves about the experience of developing on Rhapsody, talk of developing robust, commercial-grade apps in a timeframe that just isn't possible on any other platform. We'll see if that's so when these developers deliver their products next year. Meanwhile, Apple delivered its fourth quarter results -- worse even than expected. The troops are mired in the mud. There will be casualties. Meanwhile, back home at Stately Swaine Manor, a large truck has just delivered a PowerPC with Rhapsody Developer Release installed.
I'll keep you posted.
DDJ