Dear DDJ,
Thank you very much for the writeup about our IBM 1620 restoration work which appeared in "News & Views," DDJ, February 1999. The generous response from DDJ readers has been surprising.
So far, I've received e-mail offering a 1311 disk drive, some spare parts, a few manuals, some software, and a number of personal accounts of the machine. We're still looking for peripherals, especially a 1622 card reader/punch, a 1621/1624 paper tape reader/punch, and a 1443 line printer.
I have one piece of sad news to report. Joe Crespo, the IBM 1620 development engineer heading the restoration effort, died tragically since the article was written. We are continuing with the restoration in Joe's memory but could use some technical advice from any IBM 1620 engineers or field engineers who worked on the machine.
Finally, the restoration team is also developing an extensive "IBM 1620 History Site" that readers might enjoy (http://www .computerhistory.org/IBM1620/).
Dave Babcock
daveb@sgi.com
Dear DDJ,
Thank you for your article "Controlling Motion-Tracking Devices," by Mike Harrington (DDJ, March 1999). I have been in this field for 19 years, and Harrington did a good job of explaining the real-time problems facing VR users of motion trackers.
A motion tracker spewing data continuously at 47 characters every 16 milliseconds, can quickly bring a host to its knees if the serial port driver was written for 10 characters per second from a human typist.
Motion tracking is one area where vastly different technologies compete, as every technology has serious drawbacks: Magnetic trackers are disturbed by nearby metal, ultrasonic and optical trackers need clear lines of sight, inertial trackers drift with time, tilt sensors can't distinguish orientation from position accelerations.
One gotcha in VR trackers is failing to do dynamic testing early. Final integration is a bad time to find out that your tracker can't track both accurately and quickly.
My favorite company killer is the tilt sensor for orientation tracking: It appears to be simple and gives two degrees of freedom of angular measurement, typically elevation and roll. When paired with a magnetic compass, azimuth is also obtained. Thus, you have a simple 3DOF tracker.
However, the tilt sensor does not work well dynamically. The tilt sensor gives the direction of the total acceleration. In the static case, that is the earth's gravity. In a dynamic case, position acceleration adds to the earth's gravity, shifting the direction of total acceleration.
Wear a tilt-sensor VR helmet, and move your head forward and backward, without tilting your head, and watch the virtual world jump up and down in elevation.
Analog Devices points out that accelerometers solve this problem, as the magnitude of the acceleration is then known, permitting corrections to be calculated.
Peter T. Anderson
traneus@emba.uvm.edu
Dear DDJ,
Aside from its obvious intrinsic scientific interest, one can imagine that only a sense of guilt, one engendered by the rarity of DDJ articles that mention Fortran, motivated you to publish "Simulating Severe Weather," by Louis J. Wicker in the March 1999 issue. What else, in heaven's name, could have prompted you to relinquish precious space to an article that doesn't discuss C++ or Java? Perhaps the DDJ computer language police were asleep at their keyboards?
While not quite as egregious as an article discussing Cobol, one that mentions Fortran exclusively merits the condemnation of any worthy programmer, regardless of any other redeeming qualities. One can only hope that the editors of DDJ are not implying that serious programmers actually use Fortran, or even Fortran 95. Certainly you've thought of the repercussions if standardized, simple, elegant, easily understood languages like it or Ada, ones that are easily parallelized and generate compact, fast-running programs were to be adopted on a broad scale. Regular readers of your magazine surely expect you to place the highest priority on the job security of the vast coding armies of C++ and Java programmers and on their continued high renumeration. The same can be said for the continued existence of the mammoth support industry that produces the tools that allow them to ferret out their programming errors. These readers do not expect you to drive to the ranks of the unemployed the legions of authors of thousand plus page tomes on these two languages that weight heavily on leagues of shelf space in your typical bookstore. Furthermore, there must be no small degree of self interest to consider: the continued livelihood of the cadres of writers who wile away their years dissecting these cryptic, endlessly changing languages with the talmudic precision they demand.
I speak only for myself, but I think perhaps I echo the sentiments of many, when I ask for a restoration of sanity to the editorial staff of DDJ.
Norman S. Clerman
opcon@cinti.net
Dear DDJ,
Thanks to Jonathan Erickson in his April 1999 "Editorial" for legitimizing the desire for grown men and women to play with Lego. The Mindstorms kit is the first toy in years that I really wanted for Christmas, and my kids (and understanding spouse) got one for me!
Needless to say, the first thing I asked to myself was, "Can I program this thing in FORTH?" After looking at the Web for resources I came up with legOS and a detailed description of the firmware by Kekoa Proudfoot.
I wrote a FORTH interpreter for the little yellow box in about 40 hours, the results and lots of references to online LEGO Mindstorms resources are at http://www .bmts.com/~rhempel/lego/pbFORTH/default.html. There is a huge community of adults that I believe Lego sorely miscalculated in their production estimates, and we are all happy to share our work.
Ralph Hempel
rhempel@bmts.com
Dear DDJ,
In regard to Jonathan Erickson's April 1999 "Editorial," I first heard about the Lego Mindstorms Robot last summer and have not come down yet! I got mine for Christmas and haven't stopped playing with it. It is the most interesting and educational gizmo I have seen. I am trying to build the cat and mouse robot but like you, I have to wait for school to let out to ask the "experts." I'm not a programmer but the language is easy for anybody. Anybody mention Java for the Mindstorms robot? Thanks for the tip on do it yourself sensors. My browser is open and I'm ready to rock.
Steve Grappone
steve.grappone@insignia.com
Dear DDJ,
Like Al Stevens in his April 1999 "C Programming" column, my in-laws live in a dead spot too. My stair steps hover around 1 or 2 and I have to go out to the driveway as well. I haven't told them yet. I figure living in ignorance is better than living in denial. Anyhow, I liked his comments regarding Java. I'm a full-time Java developer and I really am annoyed with all the hyperbole that hovers around Java. As an object-oriented programming language, Java is really nice. With a bit more evolution, I expect it to become much more popular for the right reasons.
Java right now is too boutique'ish. It's a cute little language that hasn't got valid acceptance.
Other issues that remain undiscussed concern the need for platform independence. What if I just want an alternative from C++/MFC? Symantec's Visual Café (and perhaps others) has a feature to compile Java applications into an executable.
That's really what people want. Use Windows as the operating system but have an alternative to C++/MFC.
Jim Cloughley
James_Cloughley@markham.longview.ca
Dear DDJ,
In his April 1999 column entitled "Java Jive," Al Stevens described his problems with cellular phones. His cell phone problem may be due to the fact that he has a digital phone. This is still relatively new and does not have the infrastructure of the analog phone market. He might ask AT&T for a map of their coverage areas. Chances are that Al is on the edge of one of their areas.
Also, Al made a valid point about the over-hype of Java. I am a Java programmer and while there are some great arguments for the language, I don't think it the answer to all of the world's problems.
However, from his column, it would seem that he wants to say that we will never do without some other base OS (Windows, Linux, and so on). I disagree. There are movements to incorporate a Java-based OS (http://www.jos.org/). It's not hard to imagine a computer of any kind that loads the JVM into memory from BIOS on startup, then executes a startup sequence of something like JOS. He could even have something minimal in small devices like cell phones and large graphical Java-based OS comparable to Windows with Java-based word processors and browsers and what have you. By implementing the VM in the hardware, it might actually be fast enough to run.
Kevin Johnson
kevin@utig.ig.utexas.edu
Dear DDJ,
I read with interest Jon Bentley's "Analysis of Algorithms" (DDJ, April 1999). I heeded Jon's suggestion to study the subtleties of Search1 (algorithm 1) before proceeding. What I noticed immediately (and I'm sure I'm not alone) is that the first iteration of the For loop is wasteful, because at that point i==(m-1), so that the parameters to the swap() function are identical, meaning a vector gets swap()ed with itself. Eliminating all these extraneous iterations would do a lot to improve this algorithm, and subsequent ones.
John Bartley
bz910@freenet.buffalo.edu
Dear DDJ,
Thanks for publishing my article "The CVS Data Format" (DDJ, May 1999). However, I've received dozens of e-mail messages from readers asking about the URL of our web site. Apparently a "-" was dropped. The correct URL is http://www-gtarpa .usc.es.
Cesar A. Gonzalez Perez
phcgon@usc.es
DDJ